Sie sind auf Seite 1von 255

The International Electromagnetic Field (EMF) Dosimetry Project was initiated at a NATO

Advanced Research Workshop on Radio Frequency Radiation Dosimetry in Slovenia in 1998.


The mission of the project is to promote and develop high quality EMF dosimetry for the
assessment of human exposure and for in vitro and in vivo experimental systems. The intention
is to create an internationally accepted Dosimetry Handbook which will be a living and
substantially on-line document with integrated software tools and guides for dosimetry
measurements and calculations. The primary benefactor of this project would be the public
health, via assurance of the quality and transportability of human and experimental data; the
ability to acquire robust scientific information is dependent upon accurate and precise dosimetry.

The previous versions of the handbook, dealing mainly with CW signals, have provided the
dosimetric bases of human and experimental RF studies as well as current exposure guidelines.
The use of telecommunications technologies, particularly those incorporating pulse-modulated
(and in the future ultrawideband) signals, has become more common since the last version of the
handbook was published (in 1986). Dosimetry specific to such systems will be a major topic for
the International EMF Dosimetry Handbook. The major improvements in computational
dosimetry in the last 15 years also indicate that a new Handbook is due.

MCL hosts the Handbook's website and co-ordinates the work of the leading international
EMF/RF dosimetry experts who will contribute to the writing of the Handbook. Many of these
experts attended the Slovenia meeting and have already agreed to participate in the Project if
support can be found for its management and administration. Such support has been provided by
the UK DTI.
Project aims and objectives

1. Develop a plan to update and expand 4th edition of the Radiofrequency Radiation
Dosimetry Handbook: The last edition of the Radiofrequency Radiation Handbook was
published in 1986. Advances in technologies have enhanced the capability to measure
and predict energy absorption during exposures to RF electromagnetic fields. Also, the
global reach of the Internet make it possible for the next version of the document to be
widely available in an enhanced electronic form in addition to a traditional paper version.
2. Recruit international scientists and engineers to serve as project leaders and contributors
to the new handbook: Understanding energy absorption during exposure to
electromagnetic fields requires a team approach. Physicists, engineers, biologists,
physiologists, and health physicists will be recruited to write the chapters essential for an
integrated Dosimetry Handbook. As an open document, contributions from many sources
are welcomed, but its structure and validity will be ensured by the project leaders.
3. Establish an Internet presence for International EMF Dosimetry project: By employing an
open international forum, the Internet, the project should proceed more rapidly, at less
cost, with the results accepted internationally.
4. Develop an open copyright statement that will allow authors of text and software to retain
the copyright to their contributions: this approach will encourage the development of a
new handbook that could continually evolve with developing technologies. The ability to
retain an open source copyright would encourage software contributions.
5. Develop working relationships with potential dosimetry user societies, including: Health
Physics Society, International Labour Organisation, WHO and ICNIRP as well as UK
national bodies: The new dosimetry handbook will address the professional needs of
individuals responsible for protecting public and occupational health. The new handbook
will be user-friendly from an introductory level up to an advanced research level.

Keywords: EMF, dosimetry handbook, online, international, software


Background

During the past 20 to 30 years, society has become more dependent upon the use of technologies
incorporating electromagnetic fields, and particularly telecommunications. These new and useful
technologies have been embraced by the public but it has become increasingly important to be
able to measure the emitted fields accurately and predict the amount of energy absorbed by
biological tissue and cell cultures. Such issues were the impetus behind the development of the
original Radiofrequency Radiation Dosimetry Handbooks (1976, 1978, 1980, 1986).

The original handbooks were written at a level that was easy to comprehend by the novice
researcher and their contents have withstood the test-of-time. Specific absorption rate (SAR)
values predicted by more recent computational approaches are in many instances quite
comparable to those presented in the most recent dosimetry handbook. More than 2000
hardcopies of the Radiofrequency Radiation Dosimetry Handbook have been distributed and it is
one of the cornerstone publications in bioelectromagnetics research.

Over the past few years, the explosive increased use in wireless communication devices has
resulted in a flurry of meetings to develop or revise exposure standards. The ability to measure or
predict energy absorption during exposure to radiofrequency EMFs is the basis of these
standards. Recent advances in technologies utilised in dosimetry measurements permit better
mapping of EMF absorption, and assessment of the biological responses to EMF exposure.
Increased computer power permits the prediction of localised SAR values at higher resolutions.
Reports produced by respected committees, such as the Independent Expert Group on Mobile
Phones, highlight the importance of scientific replication of experimental results. Such
replications are possible only when accurate dosimetry has been conducted in the original report.
One of the primary goals of the Dosimetry Handbook has been to provide instruction in such
methodology. Accurate experimental dosimetry in conjunction with well-replicated experiments
will provide the best possible scientific evidence on which to base public health policy.

The new applications of EMF technologies and the marked improvements in computational
dosimetry since 1986 will require substantial additions to the Dosimetry Handbook. To be useful
from an introductory level up to the advanced research level, the new dosimetry handbook must
be user friendly and maintain the high quality of scientific data reported in the four previous
versions. The content will be expanded to the entire frequency range used for communications
and elated applications and will reflect the tremendous advances in computational and analytical
dosimetry since 1986. The new handbook will include data for researchers conducting in vivo or
in vitro experiments, epidemiologists, electrical engineering professionals, medical professionals,
and those involved in compliance testing and establishing exposure standards.

Development of the handbook will employ the Internet as an open international forum.
Technologies such as hypertext markup language (HTML), World Wide Web (WWW), virtual
reality markup language (VRML), and JAVA TM will be incorporated to permit sharing of ideas
and results on an international basis. This aspect of the new Handbook would allow it to be used
as an on-line resource by researchers: standard, verified dosimetric models of animals or
computational model of exposure facilities could be accessed remotely, removing the need for
dedicated dosimetric effort whilst ensuring good quality dosimetry. As an international group,
contributors to International EMF dosimetry project could also facilitate efforts to harmonise the
EMF exposure standards that are currently unique to each country .

The need for a new version of the dosimetry handbook is evident. Questions will continue to be
raised concerning the way in which the emitted energy from new and emerging devices interacts
with biological systems. Furthermore, the evaluation of a system's effectiveness and operational
safety will continue to be a challenge to those developing and operating this new technology The
new version of the Handbook must not only be up to date, it must be able to evolve at a rate that
matches the development of EMF technology and devices.
References

Durney, C.H., Iskander, M.F., Massoudi, H., Allen, S.J., and Mitchell, J.C. Radiofrequency
Radiation Dosimetry Handbook, 3rd Edition, USAFSAM-TR-80-32, 1980.

Durney, C.H., Johnson, C.C., Barber, P.W., Massoudi, H., Iskander, M.F., Lords, J.L., Ryser,
D.K., Allen, S.J., and Mitchell, J.C. Radiofrequency Radiation Dosimetry Handbook, 2nd
Edition, USAFSAM-TR-78-22, 1978.

Durney, C.H., Massoudi, H., and Iskander, M.F. Radiofrequency Radiation Dosimetry
Handbook, 4th Edition, USAFSAM-TR-85-73, 1986.

Johnson, C.C., Durney, C.H., Barber, P.W., Massoudi, H., Allen, S.J., and Mitchell, J.C.
Radiofrequency Radiation Dosimetry Handbook, 1st Edition, USAFSAM-TR-76-35, 1976.

Klauenberg, B.J and Miklavcic D. (eds). Radio Frequency Radiation Dosimetry and its
Relationship to the Biological Effects of Electromagnetic Fields. Proceedings of the NATO
Advanced Research Workshop , Gozd Martuljek, Slovenia, October 1998.
NATO Science Series 3. High Technology - Vol. 82. Kluwer Academic, Dordrecht, 2000.

Mason, P.A., Klauenberg, B.J., Chadwick, P., Gajsek, P., Walters, T.J., Hurt, W.D. and Ziriax
J.M. International EMF Dosimetry Project. In Biological Effects, Health Consequences and
Standards for Pulsed Radiofrequency Fields: Proceedings of ICNIRP/WHO International
Seminar, Erice. Sicily, November 1999.
EMF Dosimetry Handbook Copyright

The Handbook itself will not "own" any of the contributions in a formal sense.

Contributors must agree to their work being published by the Handbook. They will retain the
rights to use all or portions of their contributions in future publications.

Reproduction or quotation of the published materials by others will be encouraged, with the
proviso that the quotation or copy be accompanied by a complete citation, including the author of
the material being copied or quoted.

Any material that contributors supply must be either non-copyright or copy-righted to the
contributor with permission for free dissemination.

Where a chapter contains material originating with a third party, permission to publish under
these terms must be attained by the Chapter contributor prior to the chapter being delivered to the
Handbook project.

We wish to encourage the use of on-line computational/analytical models. As far as possible we


would like these to be published under the same copyright agreement as the rest of the chapter.
However, it is recognised that for some organisations and individuals such models represent
intellectual property and a commercial resource. Where models cannot be published without
constraint, two possibilities exist: models could be run "on-line" without the ability of the user to
download the model, or links could be provided to third party sites. We would not allow such
links to be used as advertisements or solely to generate business for the model-holder.
Mechanisms of electrostimulation: Application to electromagnetic field
exposure standards at frequencies below 100 kHz.

J. Patrick Reilly, Metatec Associates, 12516 Davan Drive, Silver Spring, MD,
20904, USA and The John Hopkins University Applied Physics Laboratory Laurel,
MD, USA
1. Introduction
The term “dose” as applied to electromagnetic safety parallels its use in pharmacology; it refers
to the quantity of an agent (in this case electrical energy) that can potentially result in a
biological effect. Past editions of this Dosimetry Handbook have focused on methods for
measuring or calculating the absorption of electromagnetic energy by the human body, that is,
the administered “dose.” In this work, I intend to go beyond an exposition of the absorption and
distribution of electrical energy in the body.

It would be a laudable achievement if I could specify a safe “dose” of electrical energy and
explain how to measure it. But the subject is far too complex to be reduced to a number, or even
a single table of numbers. Instead, we are faced with a myriad of physical and biological
relationships that account for human reactions to electrical energy – relationships that involve
spatial and temporal characteristics of the electrical forces within the body, factors related to the
human subject, the method of application of the energy, and the environment. To further
complicate the problem, we are confronted with variations in the outcomes of biological
experiments that defy causal explanations, and for which we are consequently reduced to
formulations based on probability to describe adverse biological reaction thresholds. Even the
definition of “adverse” is not obvious and requires careful delineation.

Considering these complications, I believe the subject of electrical dosimetry is best approached
through an exposition of the biophysical forces and mechanisms that account for human
reactions, whether adverse, beneficial, or benign. I have attempted to take this approach in this
chapter for the portion of the electromagnetic spectrum below 100 kHz. The investigator who
develops standards for electrical exposure needs an understanding of underlying biophysical
mechanisms to help interpret experimental data, extrapolate from particular experimental
conditions to more general conditions that may require regulation, and devise methods whereby
the important quantities can be measured or calculated.

Numerous mechanisms have been advanced to account for human reactions to electrical energy.
Among these, it is important to distinguish between established and proposed mechanisms. An
established mechanism is defined here as one having the following properties: (a) it can be used
to predict biological effects in humans; (b) it can be explicitly modelled using equations or
parametric relationships; (c) it has been verified in the intact human; (d) it is supported by strong
evidence; (e) it is widely accepted among experts in the scientific community. Mechanisms not
having these characteristics are classified as proposed. I have identified established and
proposed mechanisms based on these criteria in my recent book [Reilly, 1998a], and in other
publications; [Reilly, 2000a, 2000b, 2002]. I will draw from these and other publications in this
chapter.

Of the established mechanisms, the one that has the most impact on standards at the relatively
low frequencies treated in this chapter is an excitable tissue effect, referred to here as
electrostimulation; another established electrical mechanism for biological reactions is a thermal
one. At frequencies below 100 kHz, electrostimulation reaction thresholds will typically be
lower than thermal reaction thresholds. Above 100 kHz, thermal effects typically exhibit lower
thresholds of reaction than do electrostimulation effects. However, with pulsed waveforms of
low duty factor, the frequencies at which electrostimulation thresholds are lower than thermal
thresholds can extend into the megahertz region.

Electrostimulation produces short-term effects, that is, it results in acute reactions that are
manifested within seconds, (usually a fraction of a second) after the exposure begins. It
dominates over thermal thresholds at frequencies below 100 kHz and as low as 1 Hz. Below
that, magnetohydrodynamic mechanisms can be the most sensitive ones responsible for the
human reactions.

Although there are many questions remaining about electrostimulation effects, we largely
understand the underlying mechanisms, we can verify theoretical mechanisms in humans and
animals, the experimental results are robust, and we can define biological end points in the intact
human. It is therefore valuable to define limits to human exposure based on our understanding
of acute excitable tissue effects.

Other mechanisms of interaction that fit into the proposed category relate to long-term or chronic
exposure effects [Olden, 1999; Reilly, 1998a]. These mechanisms are typically mentioned in
connection with hypotheses concerning adverse health effects, including cancer, reproductive
effects, and nervous system disorders from chronic exposure to low-level electric and magnetic
fields. Standard-setting and advisory groups, while not dismissing long-term exposure
mechanisms as irrelevant, have concluded that the evidence and body of knowledge concerning
them is presently insufficient to derive a human exposure limit [ICNIRP, 1998; IEEE, 2002].
Progress in research on proposed mechanisms should nevertheless be monitored and evaluated as
to whether any one can be included in the list of established mechanisms.
1. Principles of nerve and muscle excitation
Every biological cell maintains a potential difference between its interior and exterior; usually
the interior is negative with respect to the exterior. Nerve and muscle cells respond to electrical
stimuli by becoming “excited,” a state in which the neural membrane undergoes a marked
change of conductivity that leads to a large change in the cellular potential. The excited state is
triggered when the potential difference across the cellular membrane is sufficiently reduced from
its normal resting state. This potential change, called an action potential, propagates along the
nerve’s axon. In an afferent neuron (e.g., a nerve conveying information from a sensory
receptor to the brain) the action potential normally travels to the spinal cord and thence to the
central nervous system (CNS). In an efferent nerve cell (e.g., one conveying information from
the brain to muscle cells), the action potential is initiated in the CNS, from whence it propagates
to muscle connections called motor end plates. Communication from one nerve cell to another
or at the motor end plate takes place across junctions called synapses, most typically by means of
chemical agents called neurotransmitters.

These normal processes can be activated or modified by electrical forces introduced into the
body through applied current or electromagnetic induction for medical diagnosis or therapy. If
uncontrolled, the same forces can be detrimental.

Excitable tissue effects are typically observed shortly after the application of the stimulus, often
within milliseconds to seconds. These "acute" effects stand in contrast to responses to chronic
electromagnetic exposure effects that many investigators have studied at much lower exposure
levels for possible implications on human health.

1.1. Cellular polarisation

Biological cells normally maintain a potential, Vr, in which the interior of the cell is negative
with respect to its exterior. Typical values of Vr for nerve and muscle cells are -65 and -90 mV
respectively. Considering the membrane potential (≈ 0.1 V) and thickness (≈ 10-8 m), the
electric field developed across the resting membrane is around 107 V/m. The conductivity of the
excitable membrane is controlled by the this enormous electric field. Disturbances from the
resting condition can lead to profound changes in the membrane's electrical properties, and
ultimately initiate the functional responses of nerve and muscle.
Figure 1. Representation of current flow around elongated cell placed in medium having a
uniform electric field (uniform current density). The membrane is assumed to be semi-
permeable to current flow.

Figure 1 illustrates the distribution of current flow around an elongated cell within a medium
having a uniform electric field (i.e., uniform current density). The cell is presumed to be
oriented with its long axis parallel to the undisturbed field. The flux lines suggest that the
current through the membrane, and hence the disturbance of membrane polarisation, is greatest
at the ends of the fibre. The anode-facing end of the cell will be hyperpolarised, and the cathode-
facing end will be depolarised. In an alternating field, the sites of hyperpolarisation and
depolarisation alternate every one-half cycle of the field oscillation.

We can analyse the potential disturbances of the elongated cell using the theory of electrical
coaxial cables. Consider a cable of length 2L in a longitudinal static field of strength E. The
steady-state solution for membrane voltage is given by [Sten-Knudsen, 1960]:

(1)

where X = x/λ, x is the longitudinal distance from the centre of the cell, λ is the space constant,
and 2L is the length of the cell. X = 0 is taken as the centre of the cell, and the ends are at ± L.
The space constant λ, also known as the electrotonic distance of the membrane, defines the
distance along the membrane that a steady-state voltage disturbance due to point current
injection will decay to e-1 of the value at the disturbed location. Space constants for invertebrate
nerve are in the range 0.23 - 0.65 cm [Rall, 1977].

Figure 2. Normalised membrane voltage of a finite cable immersed in a static field of


strength E. Cable length = 2L. Voltage has odd-valued symmetry about X = 0.

Figure 2 illustrates Eq. (1) for several cable lengths. Since Vm has odd-valued symmetry about X
= 0, only one quadrant of the function needs to be illustrated. The maximum membrane voltage
occurs at the ends of the fibre, and has the value

(2)
For very long cells L → ∞, and the voltage at the cell's terminus is Eλ, a value which is closely
approached even for fibres of modest length. For instance, with L/λ = 2 (total length = 4λ), the
membrane voltage at the ends is ±0.964Eλ.

As evident in Eqs. (1) and (2), the fundamental force for membrane polarisation is the in-situ
electric field, E, rather than current density, J. Although it is also possible to describe
electrostimulation effects in terms of current density, as has been a common practice in the past
[Bernhardt, 1988; ICNIRP, 1998; IEEE, 1999], the in situ electric field is a more fundamental
descriptor. Of course, we can relate the two by J = Eσ, where σ is the conductivity of the
medium. However, the conversion introduces an additional parameter (σ) about which there may
be some additional uncertainty in an applied situation. The calculation of the in situ electric field
is less sensitive to assumptions of tissue conductivities compared to internal current density.
Consequently, it is preferable to express membrane polarisation effects, including nerve and
muscle excitation, in terms of the in-situ E-field rather than current density. To my knowledge,
the IEEE low-frequency standard [IEEE, 2002] is the first to specify basic restrictions for the
general public in terms of the in-situ electric field.

1.2. Polarisation of nerve cells within an electric field

A nerve cell is an extremely elongated cell: the length of a sensory nerve innervating the
fingertip or toe has a length of about one metre. Figure 3 illustrates modes of stimulation of a
nerve cell, designated as end, bend, and spatial gradient modes [Reilly, 1998a; Reilly and
Diamant, 2003]. The illustration shows a myelinated nerve, which, due to its significantly lower
threshold as compared with an unmyelinated nerve, is a good choice for electrical stimulation
models.

An action potential is initiated by depolarisation of the cellular membrane from its resting
potential. Depolarisation occurs at points along the membrane experiencing current efflux. As
illustrated in the figure, current efflux could occur at a site where the nerve is terminated, such as
with a sensory receptor or motor end plate, where the nerve undergoes a sharp bend, or where a
spatial gradient of the electric field exists. In practice, all three of these modes can be take place
at one time. The site where excitation first occurs will be the one in which the depolarisation is
maximal, and this site determines the threshold of excitation.
Figure 1 Modes of neural stimulation. Excitation is initiated at points of maximal current efflux
across neural membrane. Potential excitation sites consist of fibre terminals, sharp bends,
and maximal gradient of E-field. Arrows indicate current flow. The wrappings along the
long process (the nerve axon) consist of insulating myelin; the uninsulated portions are
called “nodes of Ranvier.”

We model stimulation of myelinated nerve using an equivalent circuit model (Figure 4) which
contains circuit elements for the electrical conductivity at each node of Ranvier [McNeal, 1976].
Figure 2 Equivalent circuit models for excitable membranes. The response near the
excitation threshold requires that the membrane conductance be described by a set
of nonlinear differential equations. [After McNeal, 1976].

The terms Ve,n are potentials at the exterior of each node of the myelinated axon with respect to a
distant electrode in the medium, Vi,n are the interior potentials, Cm is membrane capacitance at
the node, and Rm is its resistance. The membrane potential of this circuit model can be
expressed as

(3)

which is a discrete form of the cable equation, where Vn is the membrane potential at node n.
The term , which is the driving function for membrane polarization
change, is a second difference (approximately the second derivative) of the spatial potential
measured along the long axis of the nerve fibre, or, equivalently, the first derivative of the
longitudinal electric field. The term Ii,n representing the ionic current flowing across the
membrane (i.e., through the element Rm in Fig. 4) is governed by a set of nonlinear differential
equations applying to the myelinated nerve membrane [Frankenhaeuser and Huxley, 1964].

Equation (3) has been developed as a computer model consisting of an arbitrary number nodes,
and with a threshold criterion based on propagation of an action potential; it is described as the
Spatially Extended Nonlinear Node (SENN) model [Reilly et al., 1985, 1998a], which is an
extension of a previously developed model [McNeal, 1976]. The SENN model alternates the
magnitude of extra-nodal potentials (Ve,n) between threshold and no-threshold conditions to
determine the threshold of excitation within 1%. It accommodates arbitrary spatial and temporal
distributions of the extra-nodal potentials, Ve,n.

Because equation (3) requires a finite gradient of the longitudinal E-field to cause a change in Vn,
one might suppose that excitation of a nerve is impossible in an E-field which lacks a spatial
gradient. If we were to restrict our attention to a mathematically ideal nerve fibre of infinite
length, such a conclusion would be correct. However, a nerve that terminates or bends (Figure
3) will experience a second derivative of the external potential function, even if the field itself
lacks a significant spatial gradient. The end and bend sites are typically where excitation is
initiated when current is introduced into the biological medium through cutaneous electrodes or
through magnetic induction.

Within a uniform field, thresholds of excitation are inversely proportional to fibre diameter. This
occurs because the nodal separation is proportional to fibre diameter such that d = 100 D, where
d is node separation, and D is fibre diameter [McNeal, 1976]. Consequently, the voltage
difference from node to node is directly proportional to fibre diameter. For example, a 10-µm
fibre would have an internode spacing of 1 mm. The membrane potential change at the terminal
node is approximately equal to Vm = Ed. Compare this result with the cable relationship
mentioned above, for which Vm = Eλ.

The distribution of myelinated fibre diameters in peripheral nerve effectively covers the range 2 -
20 µm. Since the lowest thresholds correspond to the largest fibres, we use a 20 µm fibre to
model minimum response thresholds for peripheral nerve stimulation.

1.3. Strength-duration law of excitation


Figure 3 Strength-duration relationships derived from the myelinated nerve model: current
thresholds and charge thresholds for single-pulse monophasic and for single-cycle
biphasic stimuli with initial cathodic phase, point electrode 2 mm radially distant
from the centre of a 20-µm fibre. Threshold current refers to the peak of the
stimulus waveform. Charge refers to a single phase for biphasic stimuli. [From
Reilly et al., 1985].

Figure 5 shows thresholds of excitation based on the SENN model of a point electrode 2 mm
radially distant from a 20-µm nerve fibre and with waveforms consisting of a monophasic square
wave, a biphasic square wave, and a single cycle of a sine wave. Thresholds are shown in terms
of charge units (right vertical axis), and peak current (left vertical axis). Threshold charge is
determined by QT = IT tp, where IT and QT are peak current and charge thresholds, respectively,
and tp is the phase duration (the duration between zero crossings of the stimulus waveform).
For the monophasic stimulus, peak current thresholds fall to a minimum plateau called rheobase
as the pulse width is increased above 1 ms. At short durations (<10 µs), thresholds converge to a
minimum charge value. Stimulus energy is not a pertinent descriptor of excitation thresholds, as
some have erroneously supposed.

Strength duration functions

When thresholds are plotted against phase duration, the result is termed a strength-duration (S-D)
function. For a monophasic stimulus, the S-D function conforms to the empirical relationships

(4)

(5)

where Io and Qo are minimum (rheobase) thresholds of current and charge, and τe is a parameter
known as a strength-duration time constant. Equations (4) and (5) are redundant in that one can
be derived from the other. Note that two parameters completely describe the S-D law of the
excitable tissue: the rheobase and the strength-duration time constant. The relationships of Eqs.
(4) and (5) can also be derived from theoretical models involving polarisation of a segment of an
idealised linear membrane [Reilly, 1998a].

Strength-duration parameters

As noted above, an S-D curve is defined by two parameters: rheobase, and an S-D time constant,
τe. Rheobase can be defined in terms current applied to an electrode contacting the biological
medium (I), the current density within the medium (J), or the electric field within the medium
(E). No matter which of these metrics we choose to use, the form of the S-D curve (Eq. 4) is the
same when expressing the ratio of the threshold quantity to the rheobase (the left-hand term in
Eq. 4). In cases of direct electrode contact, we may know the conducted current, but not its
density and distribution within the body. In such cases, it is preferable to express the threshold
and rheobase metrics as current values, IT and Io. With magnetic field induction, the induced
electric field is the metric of choice, in which case the threshold and rheobase quantities are
designated ET and Eo.

Although the S-D relationships discussed here are universal properties of excitable tissue, both
τe and Eo vary with the type of tissue being stimulated and the method of stimulation. The
smallest values of τe apply to nerve excitation; values for direct muscle excitation are
approximately 10 times greater, and those for synaptic processes are about 100 greater. The
parameter τe is not just a property of the tissue being stimulated, but also of the locality of the
stimulation. The smallest values apply to the most focal application of current, such as with a
point electrode adjacent to the nerve fibre. As the stimulation current is applied more gradually
across the spatial dimensions of the neuron, the time constant increases.

Terminated axon models

When a small active electrode is placed near a nerve fibre, the spatial derivative of the electric
field along the fibre appears as a driving force in Eq. (3). However, in chance electrical
exposure, including exposure to environmental electric and magnetic fields, nerve excitation
typically occurs where the derivative of the in situ electric field is insufficient to account for the
reaction. Such cases can be modelled as excitation of a terminated or bent nerve fibre within a
locally constant field, as illustrated in Fig. 3.

Table 1. Excitation requirements for end and bend modes of stimulation.

Bend Bend
Angle Node Eo (V/m) τe (µs)
(deg) (#)
0 1 6.15 120.7
90 2 8.55 115.7
90 4 9.84 105.1
90 6 9.96 103.9
90 8 9.96 103.9
180 2 6.50 93.4
180 4 5.45 96.3
180 6 5.10 101.5
180 8 4.98 103.9
Thresholds apply to a 20-µm nerve fibre
within a constant E-field that is oriented
parallel to the nerve beyond the bend point
(21-node fibre modelled). Source: Reilly
and Diamant, 2003.

Table 1 [Reilly and Diamant, 2003] lists rheobase thresholds and S-D time constants from the
SENN model of a 20-µm fibre that is either sharply bent, or terminated at a node; thresholds are
given in terms of the electric field external to the affected fibre. Although the lowest tabulated
threshold applies to a fibre with a sharp 180° bend at a location distant from the terminus, such a
condition would not be realistic for practical fibre trajectories. For the remaining cases, the
straight, terminated fibre provides the lowest practical threshold. Note that gradual bends would
necessitate higher excitation thresholds, since the second derivative of voltage would necessarily
be lower as compared with a sharp bend. In light of these observations, the terminated fibre is a
good model for determining the minimum thresholds of nerve excitation in chance electrical
exposures. Accordingly, the SENN model suggests Eo = 6.15 V/m, and τe = 121 µs for a 20 µm
myelinated nerve.

For the straight, terminated fibre in a constant electric field, rheobase thresholds are inversely
proportional to fibre diameter, and the strength-duration time constant is independent of fibre
diameter. Since a 20 µm fibre diameter is at the outer limits of fibre diameters found within the
human peripheral nervous system, the thresholds for the fibre modelled in Table 1 correspond to
minimum thresholds of excitation within the peripheral nervous system.

The SENN model values correspond well to experimental data. Median experimental values of
τe with magnetic stimulation are reported in the range 146 - 152 µs [Barker et al., 1991;
Mansfield and Harvey, 1993; Bourland et al., 1991a], although larger values have also been
reported [Havel, 1997; Nyenhuis et al., 1990; Bourland et al., 1999]. Values of τe with contact
current stimulation encompass a fairly wide range that includes the values observed with
magnetic stimulation. The threshold electric field with a coil encircling the forearm was found to
be 5.9 V/m (Havel et al., 1997), which is quite close to the SENN model value of 6.15 V/m. In
addition, an underlying neural excitation assumption of 6.15 V/m correctly reproduces the
distribution of let-go current thresholds in adults [Sweeney, 1993]. Furthermore, thresholds of
excitation with pulsed magnetic stimulation calculated with Eo = 6.15 V/m are reasonably
consistent with experimentally determined thresholds, as explained in Sect. 7.1.
It is often convenient to express S-D functions as asymptotic boundaries rather than a complex
function, as in Eq. (4). For instance, the asymptotic approximation to excitation thresholds can
be expressed as

ET = Eo for tp ≥ τe (6a)

ET = Eo(τe/tp) for tp ≤ τe (6b)

where ET is the threshold of the in situ electric field, Eo is the rheobase, tp is the duration of the
excitation pulse (also called phase duration).

Table 2 lists various parameters that will be used in this chapter’s development of electrical
dosimetry. For the present, I will comment only on the S-D parameters that have been discussed
up to now. The other parameters listed in the table will be discussed presently.

Table 2. Models for established thresholds of reaction: median in situ E-field thresholds.

Reaction Eo peak τe fe

(V/m) (ms) (Hz)


Synapse activity alteration, brain 0.075 25.0 20

10-µm nerve excitation, brain 12.30 0.149 3350

20-µm nerve excitation, body 6.15 0.149 3350

Cardiac excitation 12.0 3.0 167

(a) Interpretation of Table as follows:. Ei = E0 for ; for

.
Also, for ; for .

(b) (V/m-pk) refers to the temporal peak of the electric field.

(c) Adapted from Reilly [1998a]

The 3rd row of Table 2 lists S-D parameters for peripheral nerve excitation. To determine basic
restrictions in electromagnetic field standards, it is conservative to assume a small value of τe,
rather than a large one. Consequently, Table 2 adopts a value of τe = 149 µs as suggested by an
average of the lower experimental values mentioned above. The theoretical rheobase of
Eo = 6.15 V/m for a 20-µm fibre is considered a median within a distribution of minimum
thresholds in healthy adults.

1.4. Sinusoidal and other biphasic stimuli

A biphasic wave is generally less effective for excitation than a monophasic wave of the same
magnitude and duration. This occurs because the biphasic phase reversal tends to reverse a
developing action potential initiated on the initial phase. The result is higher thresholds of
excitation for biphasic stimuli, as is evident by comparing the monophasic and biphasic square
wave thresholds in Fig. 5.
Figure 4 Thresholds of excitation vs. phase duration for three waveforms: monophasic
square; continuous sinewave; single cycle sinewave. SENN model results for
terminated 20-µm diameter fibre in constant electric field. [From Reilly and
Diamant, 2002].

To further illustrate biphasic effects, Fig. 6 also shows SENN model thresholds for three
functions. The horizontal axis indicates the phase duration, tp, defined as the duration of the
monophasic wave, or the half-cycle time of an oscillating square wave or a sine wave function.
The mode of excitation modelled is a 20 µm diameter myelinated axon terminated within a
locally constant electric field oriented parallel to the fibre – conditions that are associated with
minimum excitation thresholds.

Strength-frequency functions

Sinusoidal thresholds can be depicted versus frequency through the relationship f = 1/(2tp). The
resulting curve appears as a U-shaped strength-frequency (S-F) function with a minimum plateau
that is approximately equal to the rheobase of a monophasic squarewave stimulus, and with
rising thresholds at both low and high frequencies. The low-frequency upturn typically occurs at
about 10 Hz for a variety of excitable tissue types; it is a consequence of the slow rate of rise of a
low-frequency sinusoidal wave, and is known as accommodation. The high-frequency upturn is
designated fe in Table 2. A U-shaped threshold curve can be derived from the SENN model
when it is exercised with sinusoidal waveforms; it is also evident in data from a variety of
laboratory experiments [Reilly, 1998a].

In Fig. 6, as tp decreases (increasing frequency), thresholds eventually converge to a function


proportional to tp-1 (i.e., are proportional to frequency).

If the stimulus waveform consisted of an oscillating square wave rather than a sinusoid,
thresholds would not have the low-frequency upturn, although high frequency thresholds would
still rise in proportion to frequency as with sinusoidal waveforms (Fig. 5). The low frequency
upturn would also not occur if the sinewave stimulus were initiated at a peak rather than a zero
crossing because the waveform would appear similar to the initial phase of a square wave.
Figure 7 Excitation thresholds as a function of number of cycles of sinusoidal stimulation.
Stimulus duration stepped in half-cycle increments out to four cycles, and full
cycle increments beyond that. Dashed lines indicate duration of stimulus. Point
electrode, 2 mm distant from the centre of a 20-µm fibre. [From Reilly, 1998a].

Figure 7 illustrates SENN model results that show the variation of thresholds with number of
cycles of sinusoidal stimulation, where the waveforms starts at a zero crossing [Reilly, 1998a].

The threshold at one-half cycle of stimulation is relatively low because the waveshape is
monophasic. At one cycle of stimulation, however, the waveshape is a charge-balanced biphasic
wave, which results in a relatively high threshold. Thresholds alternate with half-cycle
increments in a sawtooth pattern, gradually falling until reaching a minimum plateau when the
total duration of stimulation is about one millisecond, which is effectively equivalent to
continuous stimulation. Similar threshold relationships with the number of cycles of sinusoidal
stimulation have been demonstrated experimentally in both human [Budinger et al., 1991] and
bovine [Reinemann et al., 1999] subjects.

The minimum threshold for continuous sinusoidal stimuli converges to a value approximately
equal to the threshold of a monophasic square wave having a duration equal to the phase duration
(half-cycle time) of the sinusoidal wave (Fig 6). This fact allows us to approximately designate
the same peak threshold for both a monophasic square wave and a continuous sine wave of the
same phase duration, as in Table 2.

Asymptotic expressions for the S-F function can be expressed as

ET = Eo for f ≤ fe (7a)

ET = Eo(f/fe) for f ≥ fe (7b)

fe = 1/(2τe) (7c)

The phase duration and frequency relationship expressed in Eq. (7c) has been determined using a
theoretical model of myelinated nerve [Reilly and Diamant, 2002]. Because of the nonlinear
electrodynamics of excitable tissue, Eq. (7c) differs from linear systems for which a relationship
τ = 1/(2πf) would be anticipated.

Table 2 lists S-F parameters in the 4th column. The rheobase values for the

S-F function correspond to those for the S-D function. The parameter fe in the S-F function is
derived from the S-D time constant as in Eq. (7c).

Quasi sinusoidal stimulation


Standards for electrical exposure often refer to sinusoidal frequency as a parameter. Strictly
speaking, one can never test a physical system with a pure sine wave, which is a mathematical
construst that exists in time from minus to plus infinity. In practice, any stimulus must have a
beginning and an end, whether it is from a computer model or a physical test. The fact that the
applied waveform is actually gated on and off may be important. The mathematical abstraction
of a "sine-wave" can only be approximated.

In the SENN model results of Fig. 6, sinusoidal stimuli start with zero phase at t = 0, i.e., at a
zero crossing. Were the sine wave to start at a different phase, the threshold results might be
quite different, particularly at frequencies below 10 Hz. For example, the low frequency
threshold upturn would not exist if the sine wave were initiated at 90°. In that case the stimulus
would appear as a suddenly switched pulse of long duration, and would lead to a rheobase
threshold, rather than an elevated one. Consequently, one can not validly test a so-called "dc"
response by switching on a constant current.

Thermal versus electrical thresholds with sinusoidal stimulation

As noted above, thresholds for continuous sinusoids beyond a critical frequency increase in
proportion to frequency. An example of this is seen in studies of human electrocutaneous
perception [Dalziel and Mansfield, 1950; Chatterjee et al., 1986]. These studies show that above
100 kHz, thresholds achieve a maximum plateau as a result of thermal perception due to tissue
heating. One should not assume, however, that 100 kHz always marks the divide between
mechanisms dominated by thermal and electrostimulation effects. For pulsed stimuli, the
frequency above which thermal perception dominates electrical thresholds depends greatly on
the fraction of on-time (duty factor). This occurs because the heating capacity of electrical
current (i.e., its rms value) is proportional to the square root of the duty factor. Consequently,
one can extend the frequency where thermal perception dominates electrical perception by using
pulsed sinusoidal stimuli of low duty factor [Reilly, 1998b].

This point is illustrated in Fig. 8, which is an adaptation of the experimental data of Chatterjee
and colleagues [1986].
Figure 8 Median finger touch perception threshold of sinusoidal current with duty factor,
df, as a parameter. Perception curve for 100% duty factor (df = 1) determined
from experimental data on adult males. [Data for df = 1 after Chatterjee et al.,
1986].

The curve labelled df = 1 represents the average experimental threshold of perception for finger
contact; the threshold is given in Fig. 8 as peak current rather than rms current, as originally
presented in the cited paper; the horizontal axis represents the frequency. Since the thermal
threshold depends on the rms value of the applied current, it follows that the peak thermal
current is related to duty factor as
( 8)

where IT,d is the thermal threshold with an arbitrary value of df, and IT,1 is the threshold with df =
1. Equation (8) ensures that IT,d has a constant rms value, regardless of the duty factor (with
minor perturbations if the pulse does not contain an integral number of half cycles). The
remaining curves in Fig. (8) apply to values of df ranging from 0.1 to 0.001. Note that the
frequency separating the thermal and electrostimulation perception is advanced in inverse
proportion to df1/2.

How high in frequency do frequency-proportional thresholds apply? Experiments with


sinusoidal stimulation of rats show reasonable correspondence of frequency-proportional
thresholds up to 1 MHz, the highest frequency tested [LaCourse et al., 1985]. The strength-
duration law has also been theoretically demonstrated down to 2 ns using the SENN model
[Reilly and Diamant, 2002a], thereby suggesting that the strength-frequency curve may be
extended to several hundred megahertz. Other experimental data verify the strength-duration
law for human sensory thresholds with monophasic pulsed stimuli as short as 0.1 µs, the shortest
duration tested in humans [Reilly, 1998a], and to 1 ns in frog neuromuscular stimulation
experiments [Rogers et al., 2002].

1.5. Electrical stimulation of muscle and cardiac tissue

The most sensitive means of exciting skeletal muscle is via electrostimulation of the motor
neurons that innervate it. Consequently, thresholds for muscle stimulation normally follow those
for nerve excitation – with transcutaneous stimulation of normal muscle tissue, S-D time
constants of approximately 0.1 - 0.2 ms are observed. However, muscle tissue can be directly
stimulated if the innervating nerves are desensitised, such as with the application of curare
[Mortimer, 1981], or other means of denervation [Sunderland, 1978]. In such cases, time
constants for motor reaction of several milliseconds are observed, although rheobase remains
approximately unaffected.

The heart is a special type of muscle tissue. The heart is most sensitive to electrical excitation (a
premature beat) during the diastolic (relaxed) period. It is most sensitive to fibrillation during
the partial recovery period after the systolic (contractile) phase. Since excitation thresholds are
much lower than fibrillation thresholds, the former is appropriate for electrical safety
considerations.
Experimental data from several sources suggest that rheobase for the heart is about 12.0 V/m at
the median of a statistical distribution among healthy canines [Reilly, 1989a]. At the one-
percentile rank, the rheobase is approximately 6.0 V/m – a factor of two smaller.

Excitation thresholds of the heart exhibit an S-D law, much like that for nerve excitation, except
that the time constants are shifted to much larger values, similar to those observed with skeletal
muscle stimulation. Like nerve tissue, time constants for cardiac excitation increase as the
stimulus current is applied less focally to the stimulated tissue [Reilly, 1989a, Fig. 6.2]. Typical
values of τe for cardiac tissue with large area stimulation is about 3 ms, which is the value listed
in Table 2 – a factor of ten or more greater than τe for nerve stimulation.

1.6. Nonsinusoidal waveforms∗

Customarily, electromagnetic field standards specify maximum permissible exposure (MPE)


limits to sinusoidal fields, and these limits are specified as a function of frequency. In applied
situations, however, the exposure waveform of interest may not be strictly sinusoidal, such as
with pulsed sinusoidal waveforms, mixed frequency waveforms, those having harmonic
distortion, and a great variety of transient functions that may or may not be repeated. Examples
of devices producing nonsinusoidal electromagnetic waveforms to which workers or the general
public may be exposed include metal detectors, pulse modulators of radars, magnetic resonance
imaging devices, video displays, anti-theft devices, and the charging circuits of cellular
telephones [Jokela, 2000].

To develop electrostimulation standards, the analyst needs to identify the threshold of excitation
as a first step. How is one to proceed with such a task when confronted with the myriad of
possible exposure waveforms? One approach is to apply the temporal variation as a driving
function to a nerve model that includes the nonlinear electrodynamics of the neural membrane,
such as embodied in the SENN model. Unfortunately, this may beyond the resources of most
individuals who are interested in the application of standards. It would be desirable to define a
relatively simple method that could be used for evaluation or that could be incorporated into an
electronic sensor. This problem is not as intractable as it may seem, as noted below.

In some cases, the tests described here may be overly conservative. Such cases may occur when
the waveform appears as a low frequency wave on which is superimposed a short duration
impulse. The degree of conservatism would increase as the impulse becomes shorter in duration,
and greater in amplitude. A more precise test would require evaluation of the threshold of a
specific waveform with a neural excitation model, such as the SENN model [Reilly and Diamant,
2002b].
Evaluation method using waveform peak and phase duration

The principles discussed in Sect. 2.4 suggest one means for estimating the excitation potential of
a variety of pulsed waveforms. Recall that peak thresholds for a monophasic squarewave and a
continued sinewave are approximately the same (Fig. 6). This observation leads to a simple,
conservative algorithm for evaluating many other complex waveforms, particularly those
displaying occasional prominent peaks: (1) Determine the peak value of the test wave; (2)
determine the phase duration, tp, associated with the peak (i.e., the duration between zero
crossings on either side of the peak); (3) determine the peak value of the threshold for a
continuous sine wave with equivalent tp; (4) equate that value to the peak threshold of the test
waveform. While this method is not exact, it provides a relatively simple, and usually
conservative test of whether a test waveform is below the excitation threshold [Reilly and
Diamant, 2002].

Figure 5 Synthesised threshold waveform using Fourier components with relative


amplitudes of the threshold values of the individual sinewaves and phased for
maximum peak (odd and even harmonics, i = 1 – 20), 60 Hz fundamental

Figure 9 can be used to illustrate this method. The illustrated waveform was synthesised by
summing 20 individual sinewaves (Fourier Series components); the relative magnitude of each
component was proportional to the sinusoidal threshold determined from the SENN model (Fig.
6), beginning with a fundamental frequency of 60 Hz, and continuing to the 20th harmonic (1200
Hz). The resulting waveform was then used as a time series to exercise the SENN model. The
magnitude of this waveform, which is plotted in Fig. 9, was adjusted so that it was just at the
SENN model threshold. In this case, the peak of the threshold waveform is 7.85 V/m.
The phase duration of the waveform peak in Fig. 9 equals 0.72 ms. Using the relationship f ≈
1/(2tp), one would ascribe an equivalent frequency of 1389 Hz. The SENN model threshold at
that frequency is 7.80 V/m, which is within 2% of the threshold determined by exercising the
SENN model with the waveform in Fig. 9. This procedure proves to be a conservative test for a
wide variety of waveforms [Reilly and Diamant, 2002]. It is most conservative when the in situ
waveform is biphasic and of short duration.

Evaluation based on frequency components

An alternative method for evaluating nonsinusoidal waveforms requires knowledge of the


frequency components of the test waveform, either through Fourier decomposition, or perhaps
through other a priori means. Formulae for applying single frequency exposure limits to
multiple frequency waveforms have been expressed in standards in terms of the following
inequality [ICNIRP, 1998; IEEE, 2002]:

(9)

where S1 is an evaluation metric, Ai is the amplitude of the ith frequency component of the
exposure waveform, and MPEi is the single frequency maximum exposure limit, which is
usually set at some fraction of the excitation threshold. According to the cited standards, if the
inequality is satisfied, then the tested waveform is acceptable. The NRPB applies Eq. (9) to
frequencies from 0 to 12 MHz; ICNIRP applies it from 1 Hz to 10 MHz, and the IEEE applies it
from 0 to 5 MHz.

To appreciate the significance of Eq. (9), consider a case in which the Ai values are scaled such
that the waveform from which they were synthesised is exactly at the threshold of excitation. In
that case, S1 = 1 would indicate that the test is neither conservative nor permissive, S1 < 1 would
indicate a permissive test (i.e., the result would inform the analyst that the test waveform is
subthreshold, even though it is at threshold), and S1 > 1 would indicate a conservative test
(would inform the analyst that the waveform is suprathreshold).
In a recent study [Reilly and Diamant, 2002] we tested Eq. (9) with waveforms having vastly
different characteristics. One such example is shown in Fig. 9. We treated the MPEi terms as
the SENN model threshold for each frequency component present in the test waveform, and the
Ai terms as the magnitude of each Fourier component. In this example, Eq. (9) yielded a value of
1.03, from which the analyst would conclude that the test waveform would have to be reduced by
3% to be below the excitation threshold. Since the test waveform was at the SENN model
threshold, the test criterion would be conservative by 3% in this instance. In other waveforms
examined in the cited reference, the procedure yields typically conservative results, although in
some cases more so than in the above example.

Evaluation with linear signal processing methods

An investigation was made into linear signal processing algorithms for assessing compliance of
test magnetic field waveforms with the ICNIRP standards [Jokela, 2000]. The proposed method
could be implemented with a sensing coil that responded either to flux density, B, or to the time
derivative of flux density, dB/dt, followed by a first-order RC filter. The filter could be either a
high-pass or low-pass type, depending on whether the device was processing the B or dB/dt
waveform. The peak value of the device’s output waveform is to be compared with an ICNIRP
amplitude value for sinusoidal magnetic fields. Considering that it uses a single RC stage, and
that the ICNIRP standard has more than one frequency corner below 100 kHz, a safety factor
implicit in the method varies with the frequency or time scale of the test waveform.

The author concluded:

“On the basis of an electrophysiological model [Reilly, 1989; 1998a] ... a first-order RC-
type filter response provides a reasonable worst-case model for the response of a
myelinated nerve fibre to the magnetic field stimulation, even though the linear RC
model is only a very crude substitute for the complex non-linear stimulation model.
Comparisons of the calculated stimulation thresholds with the proposed peak limits
suggest that the weighted exposure restriction approach is in good agreement with the
basic exposure criteria of the ICNIRP. At the same time, the excessively strict
restrictions arising from the application of the multiple-frequency rule (Eq. 9) are
avoided.”
The author illustrated the proposed peak restriction procedure with data measured from various
sources of non-sinusoidal magnetic fields.

It should be noted that Jokela’s method is an ad hoc one, designed specifically for the ICNIRP
magnetic field standards. Jokela’s proposed method has recently been endorsed by ICNIRP
[2003]. It would be desirable to investigate the application of linear processing methods, such as
the one proposed by Jokela, for more general application, and with multiple frequency corners in
the linear filter.

2. Electrostimulation of the central nervous


system
2.1. Excitation of CNS neurons

Electrostimulation of the central nervous system (CNS) has been long studied for medical
purposes, including, pain control [Hosobuchi, 1985], electroconvulsive therapy [Malitz and
Sackeim, 1986], functional control and prostheses [Agnew and McCreery, 1990; Ueno et al.,
1990], auditory and visual prostheses [Brindley and Lewin, 1968; Umematsu et al., 1974] , and
various forms of functional diagnosis [Amassian et al., 1987; Mykelbust et al., 1985; Yeomans,
1990]. Such applications include stimulation by direct electrode contact of the brain and spinal
cord [Agnew and McCreery, 1990], transcutaneous methods [Geddes, 1987; Grandori and
Rossini, 1988; Levy et al., 1984; Marsden et al, 1983], and focal magnetic induction via pulsed
current into small coils [Amassian et al., 1989].

In most studies involving electrostimulation of the CNS, the mechanism of neural excitation
occurs through stimulation of myelinated axons, as discussed in Sect. 2. The diameter
distribution of neurons within the brain, however, is shifted to smaller values as compared with
peripheral nerve. Consequently, a fibre diameter of 10 µm is more appropriate for analysis of
sensitive neurons responding to brain stimulation, and the rheobase for excitation of a sensitive
brain neuron is greater than that for a peripheral neuron. The SENN model predicts rheobase for
a 10-µm myelinated nerve fibre within a uniform electric field to be 12.3 V/m – twice that of a
20-µm peripheral nerve fibre, as noted in Table 2.

Whereas fibre diameter significantly affects rheobase, the SENN model predicts that the S-D
time constant (also noted in Table 2) is unrelated to the diameter of a myelinated nerve. To
illustrate this point, the S-D curve developed with stimulation of the visual cortex in the cat
[Ronner, 1990] has a time constant of about 140 µs, as determined by the ratio of the rheobase
charge to rheobase current [Reilly, 1998a] – a value quite close to that determined with the
SENN model, and one within the range of values observed in many studies of peripheral nerve
excitation.

Stimulation of the motor cortex.

Since the brain contains a wide distribution of neuron diameters at various orientations relative to
an induced electric field, it may require a sophisticated experiment to detect reactions at the
threshold where some brain neurons begin to be excited. However, at sufficient suprathreshold
levels, reactions may become obvious and can be clearly adverse.

The experiments of Hayes [1950] are enlightening in this regard. He applied 60 Hz electric
current just above and ahead of the ears of anesthetised spider monkeys, while measuring the
electric field within the brain via electrodes inserted through holes drilled in the skull. With
current at 58 mA, Hayes reported strong jerking motions of the monkey with each application of
the current. The attendant electric field measured within the monkey's brain ranged from
17 - 23 V/m, depending on the location of the measurement probe. Assuming that the reported
currents are rms values, the peak fields would be in the range 24 - 32.5 V/m. These values are
about 2 to 3 times the assumed rheobase values noted for excitation of a 10-µm nerve fibre
(Table 2).

From the experiments of Hayes, we can derive scaling factors to determine the relationship
between the electric field in the brain and the applied current for electrodes placed on the surface
of the skin at the temples as: E/I = 293 V/m/A at the cortex, and 400 V/m/A at the midbrain.
These conversion factors are in excellent agreement with those of Rush and Driscoll [1968;
1969] using a physical model of a human skull and brain in a saline bath, and an analytic model
consisting of three concentric spheres having dimensions and conductivities specific to human
extracranial tissue, cranium, and brain tissue.

Rossini and colleagues [1985] applied pulsed stimuli to various areas of the head and scalp of 23
human subjects in areas above the motor cortex. He reported current levels needed to cause
motor stimulation equivalent to moderate voluntary contractions of the hand or leg. To properly
interpret his results, one must take into account the different waveshapes and electrode
configurations he used. One phase of his experiments used "unifocal" electrodes (an "active"
electrode at the apex of the head, and a ring of "return" electrodes around the forehead) . The
electric field in the brain under his active electrode is thought to mimic what one would expect
with large electrode separations on the head, such as with temple-to-temple electrodes. For this
case, Rossini used 100 - 150 µs pulses, and corresponding thresholds were in the range 60 - 106
mA.

The pulse widths used by Rossini are sufficiently short that his reported current thresholds would
be above rheobase. If we assume for the S-D time constant τe = 128 µs, then from Eq. (1) we
conclude that 100 and 150 µs pulses would be above rheobase by factors of 1.84 and 1.45,
respectively. If we apply the larger factor to Rossini's reported thresholds, we calculate rheobase
current in the range 32.6 - 57.6 mA. Using a scaling factor of 293 V/m/A (noted above from
the experiments of Hayes), the corresponding field induced in the cortex of the brain would be
in the range 9.55 - 16.9 V/m. If instead we had calculated rheobase by assuming that Rossini's
pulse width was 150 µs, rheobase at the lower end of his threshold range would be calculated as
12.1 V/m. We thus conclude that his most sensitive subject exhibited motor responses that
correspond to a rheobase field between 9.55 and 12.1 V/m within the motor cortex. Considering
the number of subjects tested (23), the lowest measured threshold can be attributed to about a
4.3 percentile rank within a distribution of subject thresholds.

Other experiments reported by Rossini with much shorter pulse widths and closely spaced
electrodes yielded much greater threshold values. For instance, with pulse widths in the range 27
- 48 µs and with electrode spacing of a few centimetres, excitation thresholds were in the range
440 - 940 mA.

Electrical induction of brain seizures.

Grand mal seizures can be induced with relatively high stimulus levels. Seizures induced during
electroconvulsive therapy (ECT) are sufficiently intense that patients are routinely anesthetised
and given muscle relaxants to avoid bone fractures from intense muscle contractions that might
otherwise occur [Glenn and Weiner, 1985].

The waveforms used in ECT vary widely from one instrument to another, and single instruments
sometimes provide the practitioner with a selection of waveshapes, including pulsed and
sinusoidal functions [ECRI, 1990]. The frequency and pulse repetition rate also varies among
devices. The amplitude and duration of the stimulus for both sinusoidal and pulsed waveforms
are usually selectable. Although the ECT practitioner is provided with this vast array of choices,
the dosimetry and optimisation of effective ECT parameters has been little studied, although a
few systematic studies do exist.
The distribution of seizure thresholds was determined by Weaver and Williams [1986] using
repeated trains of pulses having widths of 1 ms and a train duration less than a few seconds. In
2003 treatments, the mean (and median) thresholds were 587 mA. The most sensitive 1% of
patients exhibited thresholds of approximately 300 mA. Using the scaling factors noted above,
the cited current values would be interpreted as an electric field in the brain between 172 and 235
V/m at the 50 percentile rank, and between 88 and 120 V/m at the one percentile rank. Through
Eq. (4), we apply these rheobase values to a 300 µs pulse by multiplying the cited values by 1.11
(assuming τe = 128 µs).

The duration of the stimulus affects the seizure threshold. Stimulus duration, which is defined
here as the total time during which the stimulus is applied, should not be confused with pulse or
"phase" duration appearing in Eq. (4). The later refers to the duration of an individual pulse, or
to the half cycle time of a sinusoidal stimulus. Practitioners of ECT typically use stimulus
durations from a fraction of a second to a few seconds (usually less than 4). Weaver and
Williams found that with trains of 1 ms pulses, a stimulus duration of 1.0 s resulted in many
more seizures than durations in the range 0.5 - 0.75 s. It is opined by some that the relevant
stimulus parameter for ECT is the product of the current amplitude and the stimulus duration
[Sackheim, 1991]. It is not clear whether one can apply this rule beyond the few seconds that
comprises the upper limit of stimulus duration available in most ECT instruments.

Factors that lower seizure thresholds are: young age; hydration; previous seizures within the last
few minutes; sedative hypnotic withdrawal; irritative brain diseases; amphetamines and other
stimulants; antidepressant drugs; phenothiazines and lithium; unilateral electrodes; the drugs
pentylenetetrazol, pitressin, reserpine; sensory stimulation; and good stimulus electrode
application. Factors that raise seizure thresholds are: old age; dehydration; previous seizures
within last few days; sedative hynoptics and most anesthetics; poor stimulus electrode
application; decreasing electrode distance; frontal electrode(s); and diffuse nonirritative brain
disease in some patients [Glen and Weiner, 1985].

2.2. Synaptic mechanisms

Excitable cells communicate with one another across junctions called synapses. An action
potential that has travelled along the presynaptic cell affects the post synaptic cell through the
release of specialised chemicals called neuro transmitters in the case of chemical synapses, or
electrically (ephaptic transmission) in the case of electrical synapses across a gap between the
pre- and post-synaptic cell. Chemical synapses are the more sensitive and common of these two
modes of intracellular communication in the CNS. The neurotransmitter that is released from the
chemical synapse flows across the gap, where it binds to receptor sites in the postsynaptic cell.
If the postsynaptic potential is sufficiently depolarised by the neurotransmitters and if other
conditions are met, an action potential will be launched in the post synaptic cell.
Both temporal and spatial integration contribute to the postsynaptic potential of neurons within
the CNS [Dudel, 1989; Shepherd, 1998], and this integration takes place in an nonlinear fashion.
With temporal integration, sequences of action potentials contribute to the post synaptic
potential. With spatial integration, a particular neuron may combine the inputs from thousands
of other neurons that form synapses on that one neuron. Typically, single neurons form
branching structures, sometimes consisting of thousands of individual branches (called
dendrites), each one of which communicates with other neurons across synaptic junctions. In
the cortex of the human brain, the density of synapses can be as large as one billion per cubic
millimetre [Shepherd and Koch, 1998].

Because of the complexity of synaptic circuits, it is not possible to devise an electrostimulation


model based on a single neuron, such as we did in applications to peripheral nerves. Although
the literature on synaptic mechanisms is enormous, we lack a model that can be used to
determine the CNS effects of electrostimulation by externally applied electric fields.
Consequently, we rely on experimental data, albeit sparse, on electrostimulation of synaptic
processes.

Experimental data show that an important property of the synapse is that a relatively small
change in presynaptic potential can have a much larger percentage change in postsynaptic
potentials [Katz and Miledi, 1967]. If an electric field exists around a neuron, its cell body and
dendrites will become polarised as explained by cable theory (Sect. 2.0), and these potentials
may be amplified in the post synaptic cell through spatial and temporal integration. Thus, we
might expect much lower thresholds of synaptic effects than observed with nerve excitation.
These effects could be either excitatory or inhibitory, that is, could result in the excitation of a
neuron that would otherwise not have been excited, or could inhibit excitation of a neuron that
would otherwise have been excited.

Saunders and Jefferys [2002] found that published data from in vitro studies support the view
that CNS effects resulting most probably from synaptic interactions occur at induced electric
fields as low as 1 V/m – a factor of about ten below the postulated median threshold of excitation
of sensitive brain neurons. They further argued on theoretical grounds that the CNS in vivo may
be more sensitive to extremely low frequency electric fields and currents than in vitro
preparations. Spatial and temporal integration of interacting groups of neurons were thought to
be responsible for the greater sensitivity of in vivo exposures.
Experimental in vivo data on electrostimulation of synapses from in situ electric fields come
primarily from studies of electrostimulation of the retina, discussed below. These data indicate a
much lower rheobase than that observed for nerve excitation. With respect to the S-D time
constant of synapse electrostimulation, time constants of synaptic mechanisms are demonstrated
from a small fraction of a millisecond in synaptic “microcircuits”, to hundreds of seconds in
larger hierarchal structures [Shepherd and Koch, 1998]. Despite this enormous range, observed
S-D time constants applicable to electrostimulation of synapses within the retina fall within a
relatively narrow range, as discussed below.

2.3. Synapse electrostimulation within the retina

Whereas the nerve cell requires membrane depolarisation of approximately 15 to 20 mV to


initiate an action potential, synaptic processes can be affected by altering the presynaptic
membrane potential by less than one mV, and possibly as little as 60 µV, as with electrical
stimulation of synapses in the retina [Knighton, 1975a,b] – a factor 250 times lower than
minimum neural excitation thresholds. Consequently, the synapse is a potentially sensitive site
for neural interaction with applied electrical stimuli.

An example of a synaptic polarisation effect is attributed to the phenomenon of electro- and


magnetophosphenes, which are visual effects resulting from electric currents or magnetic fields
applied to the head [Adrian, 1977, Barlow, 1947a, 1947b; Baumgart, 1951; Bergeron et al.,
1995; Budinger et al., 1984; Carstensen, 1985; Clausen, 1955; Lövsund et al., 1980a,b; Silny,
1986]. Experimental evidence suggests that phosphenes result from modification of synaptic
potentials in the receptors and neurons of the retina [Lövsund et al., 1980a, 1980b; Knighton,
1975a; 1975b], rather than excitation of the optic nerve or the visual cortex, although visual
sensations with stimulation of the visual cortex have been demonstrated with much stronger
stimuli [Brindley and Lewin, 1968; Brindley and Rushton, 1977; Ronner, 1990]. Phosphenes are
most sensitive to current or an in situ electric field that is oriented in a radial direction with
respect to the retina. This finding suggests that the electric field interacts with radially oriented
neurons in accord with cable theory (see Sect. 2.2).

Using data from magnetophosphenes [Lövsund et al., 1980a,b] the corresponding induced E-
field in the head at the most sensitive frequency tested (20 Hz) is 0.079 V/m-rms as calculated
with an ellipsoidal model of the head (see Appendix A). At the retina, where the electrical
interaction is thought to take place, the calculated field is 0.053 V/m-rms, which is consistent
with the current density threshold of 0.008 A/m2 at the retina determined for electro-phosphenes
[Lövsund et al., 1980b] assuming the conductivity of the brain is 0.15 S/m. The peak of 0.075
V/m is associated with an rms value of 0.053 V/m, and that peak value is noted as rheobase in
Table 2. The internal E-field corresponding to phosphene perception at the optimum frequency
is a factor of 100 or so below rheobase thresholds for neural stimulation.
Experimental strength-duration data show that τe for phosphenes using electrodes on the temples
is approximately 14 ms [Baumgart, 1951; Bergeron et al., 1995] and for electrically evoked
potentials in the frog's eye, τe is in the range 14 - 36 ms [Knighton, 1975 a, 1975b]. These
values are consistent with the phosphene data described above, but are about 100 times greater
than corresponding values for peripheral nerves.

Relatively few data exist on synaptic polarisation effects by applied electric fields. Considering
this dearth of data, I have made use of experimental data on synaptic effects from applied electric
fields and I have assumed parallels with nerve excitation properties. One class of these
properties concerns strength-duration and strength-frequency characteristics. For example,
experimental data show a S-D time constant (τe) for phosphenes using electrodes on the temples
is about 14 ms [Baumgart, 1951; Bergeron et al., 1995]. Knighton [1975a, 1975b] developed S-
D curves for electrically evoked responses, and found τe in the range 14 to 36 ms (Fig. 10).
Figure 10 Strength-duration curves for electrically evoked potentials in the retina of the
frog’s eye. Curves represent various experimental procedures. [Adapted from
Knighton, 1975].

An average strength-duration time constant for synapse effects is τe = 25 ms, which is listed in
Table 2. Using the relationships noted for nerve excitation, a strength-frequency constant of fe =
20 Hz is expected above which in situ electric field thresholds should rise. This rise is indeed
observed in the case of electrophosphene thresholds, although the rate of rise is greater than that
observed with nerve excitation (Clausen, 1955; Adrian, 1977). Magnetophosphene strength-
frequency curves reported by Lövsund and colleagues (1980a; 1980b) show a minimum at 20
Hz, and rising thresholds at lower frequencies, in accord with electrophosphene data. Thresholds
above 20 Hz vary somewhat with the experimental parameters (background illumination and
wavelength, subject visual acuity). Considering electro- and magneto-phosphene strength-
frequency and strength-duration curves in total, it is reasonable to adopt a threshold curve similar
to that found in electrostimulation of nerve and muscle, but with a much lower strength-
frequency constant (or, equivalently, a larger strength-duration time constant), and with lower
rheobase. Additional study of CNS synaptic interaction effects is needed to clarify these
assumptions.
Frequency sensitive thresholds for phosphenes have been experimentally tested only to a
maximum frequency of about 75 Hz. I have made the conservative assumption that synaptic
polarisation thresholds follow a frequency-proportional law above 20 Hz to a frequency of at
least 760 Hz (above which peripheral nerve excitation limits dominate the magnetic field MPEs,
as will be explained subsequently).

A contrary view was expressed in a recent standard published by the International Commission
on Non-Ionizing Radiation Protection [ICNIRP, 1998]. The ICNIRP standard assumed that the
critical frequency for frequency-proportional thresholds of electrophosphene-related effects is
1000 Hz, rather than 20 Hz as assumed in a recent IEEE standard [IEEE, 2002]. This
assumption, together with the ICNIRP assumption that such reactions include the spinal cord
(see 3.4 below), account for lower basic restrictions in the ICNIRP standards at frequencies
above about 35 Hz as compared with the IEEE standard. Note that no information exists on
frequency constants for synaptic effects, other than the retina studies noted above.

2.4. The spinal cord versus brain as the site of synapse electrostimulation

The spinal cord, which also contains synapses, performs many functions, such as control of
posture and reflex activity. It is logical to inquire whether low electrical thresholds associated
with synapse stimulation might be present in the spinal cord.

In pursuit of this question, note that tests have been conducted with human subjects whose torsos
were subjected to the strong switched gradient fields of experimental MRI systems (See Sect.
7.1). Perception was sometimes preferentially reported in the small of the back at stimulus levels
corresponding to nerve stimulation thresholds in accord with expectations from an elliptical
induction model. These tests showed no observable effects below the neural threshold for
perception. The lack of an observable effect below electrical perception thresholds suggests one
of three possible explanations. One is that spinal synapse interactions did occur, but they were
imperceptible to the subject. Another is that the induced field in the spinal column was below
synapse interaction thresholds, even though the levels just outside of the spinal column were
roughly two orders of magnitude above synapse thresholds observed in the retina. A third is that
stimulation thresholds in the spinal cord were significantly greater than what has been assumed
for synaptic effects in brain neurons (Table 2).

Considering the lack of data to suggest observable effects from stimulation of the spinal cord at
the levels attributed to synapse thresholds, I have concluded that protection standards should
focus on synaptic effects in the brain, rather than the spinal cord [Reilly, 2002]; this view has
been adopted in the recent low-frequency exposure standards of the IEEE [IEEE, 2002].

Not all scientists agree with this view. Indeed, standards published by the International
Commission on Non-Ionizing Radiation Protection [ICNIRP, 1998a], include the spinal cord as
well as the rest of the torso in basic restriction protection levels which were derived from
electrophosphene reactions. One consequence of this decision is that the ICNIRP basic
restrictions for exposure of the torso are lower than those of the IEEE in some portions of the
frequency spectrum.

Subsequent to the publication of their standard, ICNIRP published responses to questions


[ICNIRP, 1998b]. One question and answer were as follows:

Q. Is the basic restriction of 10 mA/m2 based only on the threshold for acute effects in
the central nervous system, or does it apply to other tissues in the trunk of the body?

A. The basic restriction of 10 mA/m2 is intended to protect against acute exposure


effects on central nervous system tissues in the head and trunk of the body, with a safety
factor of 10. ICNIRP recognises that this basic restriction may permit higher current
densities in body tissues other than the central nervous system under the same exposure
conditions.

It would appear from this reply that 10 mA/m2 applies only to the brain and spinal cord, and that
higher basic restrictions may be tolerated elsewhere in the trunk. However, the standard does not
specify what these higher values may be.

3. Temporal and spatial relationships in


electrostimulation-based standards
Exposure standards should be written with a view to tests for compliance. Such tests might
include calculation methods, measurement procedures, or devices that could, at least in principle,
demonstrate whether exposure limits of a standard are met. Two practical issues in such a
demonstration would include questions of temporal and spatial averaging – what are the extents
of time and space over which measurements may be taken?

3.1. Temporal relationships in compliance tests


Electromagnetic exposure standards typically include limits in terms of rms metrics. An rms
measure is a mathematical operation on a series of measurements (or a temporal sequence of
data) in which the square root of the arithmetic mean of the squares of the measurements or data
is taken. The mean referred to in this definition is associated with a time period over which the
measurements are to be taken. For repetitive waveforms, this is simply the repetition period.
For a repetitive waveform of mixed frequencies having integer relationships, the averaging
period would be the repetition period of the lowest frequency component.

Another aspect of temporal averaging treats the question: is there a cumulative effect such a
single instantaneous measure of exposure does not sufficiently represent the hazard? With
ionising radiation, for instance, it is widely accepted that the health hazard is related to
cumulative, rather than instantaneous dose. Is there an integration effect with
electrostimulation?

To explore this question, I investigated the threshold effect of a train of pulses using the SENN
model [Reilly, 1998a, Chapt. 4]. The excitation threshold of repeated pulses of the same
polarity decreases with the number of pulses, but only if the multiple pulses occur over a time
period less than a few tenths of a millisecond. The rheobase of a pulse train having a sufficiently
rapid repetition rate, however, converges to that for a single long pulse. For repetition periods
beyond about 0.5 ms, the individual pulses of a train appear as independent excitatory events.

A nonlinear type of integration effect is seen with sinusoidal waveforms. For sinusoidal stimulus
waveforms, thresholds of nerve excitation evaluated at half-cycle increments oscillate between
gradually falling maxima at odd numbers of half cycles, and minima at even number of half
cycles, and converge to a single minimum threshold at about 1.3 ms of stimulus duration (Fig.
7). This suggests a maximum integration time of 1.3 ms for electrostimulation of nerve.

We lack data on integration effects with electrostimulation of muscle and nerve synapse similar
to that offered by the SENN model, although single cell models have been developed for the
electrodynamics of various tissues in the heart [Reilly, 2002, Chapt. 3]. Lacking appropriate data
on such integration effects, I infer these properties from a comparison of the S-D time constants
of nerve, muscle, and synapse. The S-D time constants for muscle and nerve synapse exceed
those for nerve stimulation by factors of 20 and 168, respectively (Table 2). Consequently, a
measurement averaging duration of 200 ms (≈ 168 x 1.3) would encompass the maximum
integration duration needed to characterise minimum nerve, muscle, and synapse excitation
thresholds.
3.2. Spatial relationships in compliance tests

Recent exposure standards [IEEE, 2002] specify basic restrictions in terms of the in situ electric
field. When determining compliance with the basic restrictions, the analyst must consider an
averaging distance, da, over which the in situ electric field should be measured. A related
parameter is the required distance over which the electric field must exist for efficient
electrostimulation.

The relationship between the threshold of excitation and the distance over which the field exists
(de) has been determined from the SENN model [Reilly and Diamant, 2002]. A minimum
threshold was obtained with de of seven or more internode spaces at the terminus of the axon.
With de of one internode space, the rheobase threshold was twice the minimum value of 6.15
V/m for a field existing over the affected axon (Table 1). I shall refer to this minimum threshold
as a reference field, Er. With de = 2, 3, 4, and 5 internode spaces, the threshold exceeded Er by
34, 14, 7, and 3%, respectively. In the same study, we showed that when a spatially-limited field
existed at a central portion of the nerve axon, the excitation threshold was greater than when it
existed at the terminus.
Figure 11 Errors in threshold measurements resulting from assumed averaging distance, da;
20 µm nerve fibre; stimulation at fibre terminus; de = distance over which the in
situ field exists. [From Reilly and Diamant, 2003].

To illustrate a rationale for choosing da, consider an electric field of extent de at the terminus of a
20-µm axon and just at the threshold of excitation. Assume the field is measured with various
values of da, shown on the horizontal axis of Fig. 11; the vertical axis gives the percentage, p, by
which Ea deviates from Er, where p = 100[(Ea/Er) – 1]. With an averaging distance of 5 mm, p is
at most 24% (with de ≈ 5.5 mm) and at least –17% (with de = 2 mm). Recall that positive p
values are conservative, i.e., result in an overestimate of the threshold field, and negative values
are permissive. By increasing da above 5 mm, positive p values can be reduced, but at the
expense of much more deviant negative values. An averaging distance da = 5 mm is neither
unduly conservative nor permissive.

Note that an overestimate of the measurement is conservative, i.e., would signal the analyst that
the induced field is suprathreshold, even though it was at the threshold of excitation; an
underestimate of the measurement would be permissive, i.e., would signal that a larger field
could be tolerated.
That an averaging distance of 5 mm is a reasonable choice suggests a simple metric for
determining compliance with basic restrictions – the potential difference at a spacing of 5 mm
within the biological medium. Such a potential difference could, in principle, be determined
with computational models, phantom models, or in vivo measurements in animals exposed to
electromagnetic fields.

4. Contact current and spark discharges


Electric and magnetic fields induce electrical forces in conducting objects, as well as in the
human body. A person who contacts such an object may conduct current while contact is made
with the object, and possibly may receive spark discharges just before and after physical contact
with the energised object. We call these indirect effects, to distinguish them from the fields that
are induced in the body by the direct action of the environmental field. The severity of such an
encounter will depend on the size of the induction object, as well as the degree of insulation from
ground of the person and conducting object.

4.1. Contact current perception

Human sensitivity to electric current applied to the skin (electrocutaneous stimulation) depends
on a host of factors, including: the stimulus waveform, the size of the electrode, the hydration of
skin beneath the electrode, the location of contact on the body, the skin temperature, tactile force
of contact, and the size of the subject. I have discussed the effects of these variables previously
[Reilly, 1998a]. In addition, we observe variations in sensitivity among subjects that defy causal
explanations, and for which we must resort to statistical representations.

One of the variables mentioned above is the location on the body at which the stimulus is
applied. In our own research, my colleagues and I have observed that there is approximately a
four-to-one variation in electrical thresholds across the points we tested on the body, and they are
generally ranked with respect to the distance from the stimulus point to the brain. The threshold
at the finger tip is approximately in the middle of this distribution, and this location is a natural
contact point in many applied situations.

Both sensory and motor reactions with electrocutaneous stimulation are due to neuroelectric
excitation. Consequently, mechanisms and waveform relationships discussed in Sect. 2,
including S-D and S-F relationships, apply to electrocutaneous stimulation as well.
The threshold current into a contact electrode varies inversely with the contact area. A touch
contact area of 1 cm2 is a reasonable assumption for a light fingertip contact, whereas a much
larger contact area (≈ 15 cm2) is more representative of a grasped contact.

Numerous experiments with perception of sinusoidal current reveal a strength-frequency law


with a minimum plateau below a critical frequency, fe, above which thresholds converge to a
frequency-proportional law when the current is of a continuous nature [Reilly, 1998a]. Based on
nerve excitation models, S-D and S-F constants are connected by fe = 1/(2τe). Consequently,
factors leading to small values of τe would increase fe. Experimental values of fe vary
significantly, although the factors accounting for this variation are not well understood. The
constant fe = 3 kHz is a reasonable estimate of the upper S-D frequency parameter within
experimental and theoretical values, and that value has been used in a recent IEEE standard
[IEEE 2002].

Table 3. Example rheobase contact current perception thresholds for adult men and women.
Median RMS values calculated from the data of Chatterjee et al. [1986].

Contact method Males Females

(mA) (mA)

Touch 1.03 0.83

Grasp 2.77 2.67

Table 3 lists rheobase thresholds of perception for touch and grasp contacts as extrapolated from
experimental data [Chatterjee et al., 1986]; the data apply to the median among a population of
healthy men and women, and are consistent with other experimental data [Dalziel and Mansfield,
1950]. Chatterjee and colleagues tested thresholds only to a minimum frequency of 10 kHz; an
adaptation of their threshold curve is labelled in Fig. 8 as df = 1. As expected from neuroelectric
properties (Sect. 2.0), measured perception thresholds below 100 kHz and above fe are directly
proportional to frequency. This fact allows us to extrapolate Chatterjee’s curve below 10 kHz to
arrive at an estimate of rheobase, provided we know the S-F constant, fe. The rheobase values in
Table 3 are based on fe = 3 kHz.
Table 3 should be interpreted according to the following equations

IT = Io for f ≤ fe (10a)

IT = Io(f/fe) for f ≥ fe (10b)

where IT is the median threshold current at frequency f.

The perception thresholds listed in Table 3 for women are lower than those for men. This
observation is consistent with experiments conducted in my own laboratory [Larkin et al, 1986].
However, we determined that what seemed to be gender related was actually related to body size,
not to gender, per se – small individuals tend to have lower thresholds than large ones, and
women tend to be smaller than men. If we compare men and women of similar body stature,
their electrical sensitivity is statistically indistinguishable.

4.2. Electric field induction

Since environmental electric fields induce in situ electric fields and body currents, it might seem
logical to conclude that the induced field should be limited so as to preclude direct
electrostimulation effects. In practice, however, contact current and spark discharges (indirect
electrostimulation) limit environmental electric fields to values significantly lower than what is
required to directly induce in situ electric fields at the levels in Tables 1 and 2.

Indirect stimulation effects occur through charge transfer between a person and a conducting
object within the field. With sufficiently strong fields, an individual can perceive spark
discharges just prior to the moment of direct contact and just after breaking contact with
conducting objects that are well insulated from ground. It is also possible to perceive current
through direct contact with such objects.

The contact current component, Ic, for an erect person touching a grounded conductor in a
vertically polarised electric field is [Reilly, 1998a]
Ic = 9.0 x 10-11 h2 f E (11)

where h is the height of the person, f is the frequency of the field, and E is the environmental
field strength. For fields at the relatively low frequencies treated in this chapter, and in which
the environmental field magnitude varies over the area that would be occupied by the body, the
field strength in Eq. (11) may be replaced with the average environmental field over the area in
which the body is placed [Kaune, 1981; Deno and Zaffanella, 1982].

4.3. Spark discharges

A person contacting a conducting object within a strong electric field may experience spark
discharges just before and after the moment of physical contact; during physical contact, current
will be conducted by the subject at the point of contact. The intensity of spark discharges will
depend on the voltage developed on the induction object and its capacitance to ground; the
magnitude of the contact current will depend on the effective area of the induction object, and
the frequency of the field. These effects are described by electrical induction equations [Reilly,
1998a, Chapt. 9].
Figure 12 Perception threshold for stimuli induced within a 60 Hz electric field; electrode
tapped with the fingertip. Curves for k = 1/2 and 1/4 indicate effect of leakage
resistance. The curve for k = 1 applies to an induction object that is perfectly
insulated from ground. [From Reilly and Larkin, 1987].

My colleagues and I studied human reactions to such stimuli at 60 Hz in relation to the electric
fields produced by high voltage power transmission lines [Reilly and Larkin, 1987]. Figure 12,
taken from that study, illustrates median perception thresholds among healthy adults. The
stimulus in every case mimicked the combination of spark discharges and contact current that
can be experienced within a 60 Hz electric field. The horizontal axis of the figure is the
capacitance of the discharging object. The induction object can be any conducting object that is
adequately insulated from ground, or it can be the human body itself. In the latter case, the
capacitance of the human body standing on, but insulated from ground, it is typically in the range
100 – 150 pF. Spark discharges and contact current can occur when the insulated person touches
a grounded conductor.

The parameter k in Fig. 12 takes into account leakage between the induction object and ground,
where Vo is the open-circuit voltage measured on the object, Is is the short-circuit current, ω =
2πf, and Co is the capacitance of the induction object (i.e., the horizontal axis of the figure). For
a well-insulated object (large leakage resistance), k = 1, which applies to the worst-case
situation. The fact that thresholds are lower with reduced leakage resistance (k < 1) does not
mean that reduced resistance makes the shock exposure worse. If the induction electric field is
held constant, decreased leakage resistance reduces the induced voltage even more than it does
the threshold. For example, comparing k = 1/4 with k = 1, the induced voltage is reduced by the
multiple 0.25, but the perception threshold at Co = 3,200 pF is reduced by the multiple 0.52 (as
seen in Fig. 12).

5. Statistical relationships in
electrostimulation
Electrical thresholds can vary significantly from one person to another. Although we can
identify factors accounting for some of this variation, much of it is not well understood, and we
must resort statistical formulations. The statistical distribution of electrical reaction thresholds is
typically represented by a lognormal distribution, i.e., one in which the logarithm of a statistical
variate has a normal distribution.

The mean of a lognormal distribution always exceeds the median. The mean-to-median ratio of
a lognormal distribution, ρ, is expressed by [Hastings and Peacock, 1975]

(12)

where σ is the variance of the natural logarithm of the statistical variate. For a distribution in
which the ratio of 50% to 1% values equals three, the mean-to-median ratio is 1.12, i.e., the
mean exceeds the median by 12%. This relationship is useful in cases where an experimental
mean is given, rather than a median.
Experimental thresholds correspond well to the lognormal distribution in many instances of
electrostimulation, although it is often necessary to replot published data on lognormal
coordinates to demonstrate this. The lognormal distribution requires two parameters to describe
it. I have found it convenient to describe lognormal distributions in terms of the median, and a
slope parameter defined as the ratio of the median to the one-percentile rank, x50/x1.

Table 4. Example lognormal parameters for electrostimulation

Electrical Reaction Species x50/x1 citation

Forearm perception, contact current human 3.0 a

Fingertip perception, contact current human 2.0 a

Perception & pain time-varying mag. fields human 1.9 b


Electroconvulsive therapy seizure human 2.0 c

Contact current perception bovine 2.3 d

Ventricular fibrillation canine 2.0 e

a = Larkin et al. [1986]; b = Nyenhuis et al., [2001]; c = Weaver & Williams, [1986]; d = Reinemann et al., [1998];
e = Dalziel [1968];

Table 4 summarises lognormal slope parameters derived from published data from a number of
sources [Dalziel, 1968; Larkin et al., 1986; Nyenhuis et al., 2001; Reinemann et al., 1998;
Weaver and Williams, 1986]; the slope parameter has been determined from a lognormal fit to
the published data between the median and one-percentile ranks. Note that a slope parameter of
3 represents an observed maximum slope, although a more typical condition would have a slope
parameter of about 2.
Table 5. Normalised distribution of electrical reaction thresholds using log-normal
model for healthy adult population (male & female).

Percentile Rank (%) Threshold multiplier Threshold multiplier

perception & pain ventricular fibrillation


99.5 3.45 2.33
99 3.11 2.14
95 2.24 1.67
90 1.85 1.51
75 1.40 1.24
50 1.00 1.00
25 0.72 0.80
10 0.54 0.66
5 0.45 0.60
1 0.32 0.47
0.5 0.29 0.43

Perception distribution based on human experimental data for arm contact. Ventricular fibrillation
distributions from healthy dog hearts.

Source: Reilly [1998a]

Table 5 provides examples of log normal models (medians normalised to 1.0) applicable to
sensory stimulation of the forearm of healthy adult humans, and to ventricular fibrillation (VF) in
healthy dogs [Reilly, 1998a]. Experimental data for fingertip perception more closely follow the
VF values. Compared with data from healthy animals, a much broader distribution of VF
thresholds has been reported for direct electrode contact to the hearts of human patients
undergoing open-heart surgery for valve replacement [Watson et al., 1973]. Thresholds for
persons in a pathological state or under drug treatment have not been otherwise tested.
Figure 13 Lognormal distributions. The vertical axis follows a standard normal model; the
horizontal axis applies to the statistical variate normalised by its median.
Lognormal distributions are represented as straight lines on this plotting format.

Figure 13 shows the form of lognormal distributions, which are represented by straight lines on
the plotting format. The horizontal axis applies to the statistical variate normalised by its median
value (x/x50) It is tempting to extrapolate the distribution model of Fig. 13 to arbitrarily small
percentile ranks. However, experimental evidence is insufficient to support extrapolation much
below the rank of about 1% due to limitations in the numbers of subjects represented in available
experimental data.
Variations in thresholds from one individual to another are not well understood. The only
significant physiological parameter that has been correlated with electrical thresholds is body
size and related parameters, such as gender, and age [Larkin et al., 1986; Reilly and Larkin,
1987; Reilly, 1998b]. The correlation is such that small individuals tend to have lower
thresholds. A body size relationship is found in sensory reactions, let-go thresholds, and
ventricular fibrillation. Experimental evidence indicates that thresholds of pain in humans, and
VF thresholds in animals vary approximately with the square-root of body weight, although other
relationships have been proposed [Reilly, 1998a]. Let-go thresholds in humans vary
approximately in proportion to body weight [Dalziel, 1972]. Consequently, small individuals,
especially children, would be most susceptible to electrical stimulation effects. On the other
hand, the magnitude of current induced by electric or magnetic fields diminishes with decreasing
subject size. And with contact current, the small individual typically has a greater inter-limb
resistance than a larger person. Because of these compensating factors, the effect of body size is
not expected to be great. Indeed, a study of the relationship between magnetic field perception
thresholds and morphological factors (subject gender, girth, weight, and age) demonstrated a
lack of significant correlation with any of these factors [Nyenhuis, 2001].

6. Exposure to environmental fields


The fundamental dosimetric metric for electrostimulation effects is the in situ electric field.
Restrictions so stated are known as basic restrictions. However, it is often simpler for an the
individual who must enforce a standard to have dosimetry requirements stated in terms of the
environmental field. When deriving environmental limits from basic restrictions, it is often
expedient to make simplifying assumptions that are relatively stressful, but not unrealistic.
Nevertheless, environmental limits may sometimes be unnecessarily restrictive. To guard
against excessively restrictive environmental limits in dosimetry standards, it is desirable to
provide in a dosimetry standard both basic restrictions and environmental limits and to include a
statement along the following lines [IEEE, 2002]:

Lack of compliance with environmental limits does not necessarily imply lack of
compliance with basic restrictions, but rather that it may be necessary to evaluate
whether the basic restrictions have been met. If the basic restrictions are not exceeded,
then the environmental limits may be exceeded.

To derive an environmental magnetic field from in situ E-field magnitudes, it is necessary to


apply an induction model. Traditional methods used to predict whole body energy absorption
during magnetic field exposure include the use of ellipsoid shapes arranged to mimic an animal
or man [Reilly, 1991]. During the past several years, high-resolution anatomical models have
been developed to enhance the capability to predict localised energy absorption, such as within a
single organ or part of an organ [Dawson and Stuchly, 1998; Dimbylow, 1998; Gandhi and
Kang, 2001; Gandhi et al., 2001].
6.1. Magnetic field exposure

Detailed anatomical induction models

The development of high-resolution anatomical induction models has tremendously enhanced


our understanding of energy absorption during electromagnetic field exposure. These models
make use of high-resolution three-dimensional specifications of tissue conductivities, along with
advanced computational methods. With these models it has been possible to compute the
electric field and current density induced within all organs of the body and with high resolution
[Dawson and Stuchley, 1998; Dimbylow, 1998; Gandhi and Kang, 2001; Gandhi et al., 2001].

Although the high-resolution model might seem the perfect tool for dosimetry studies, a number
of questions about these models remain. One question concerns the issue of variability in model
calculations among different analysts. A comparison of induced electric field calculations
obtained by several investigators using a similarly detailed anatomical model and similar
numerical techniques [Dawson and Stuchly, 1998; Dimblylow, 1998; Gandhi, 2000] showed
differences of over 5:1 in the maximum field in critical organs; organ averages were usually
reasonably consistent, although differences as great as 2:1 were noted. Since the basic
restrictions pertaining to electrostimulation effects depend on the maximum field in particular
organs, large variations in reported maximum values make it difficult to apply presently
available detailed models to dosimetry standards. Although variations in the calculated induced
electric field can be attributed to differences in voxel size [Mason et al., 2000], as well as to
differences in assumed permittivity values of the tissues [Hurt et al., 2000], the source of the
variation of results among investigators has not presently been identified.

Considering these variations, it is necessary to validate results from high-resolution


computational methods with measurements in living animal models. For example,
measurements of the electric field induced in the spinal cord of an intact animal might be
compared with results obtained from a high-resolution computational model.

Despite these considerations, high-resolution anatomical models are extremely useful in


assessing the distribution of induced fields within the body under a great variety of conditions
related to the environmental field, and the position and attitude of the human body within the
field. Undoubtedly, standard computational approaches will soon be defined for these models,
and they will be validated with experimental results.

Before detailed anatomical models can be relied on to determine environmental limits from basic
restrictions, it may be necessary to reinterpret experimental data that were originally used to
determine in situ thresholds on the basis of simpler homogeneous conductivity models. Such
reinterpretation would ensure the use of the same induction model for the calculation of in situ
thresholds from exposure data pertaining to the environmental field, and environmental
limitations based on basic restrictions. Some of the experimental data that may be subject to
reinterpretation are discussed in the following section, as well as in Sect. 3.

Ellipsoidal induction model

Lacking results from a validated high-resolution model, a recent standard [IEEE, 2002] used an
ellipsoidal induction model to associate in situ electric field specifications with an environmental
field. This involves an ellipsoidal model of the head and torso of a large individual, with
uniform conductivity, and a constant magnitude and relative phase of the field over the body
dimensions as described in Appendix A. In all calculations, a worst-case assumption has been
made for the direction of the field relative to the body.

Figure 14 Example of position of the heart within three cross-sections of the body; a and b
are semi-major and –minor axes of equivalent cross-section ellipsoids.
Numbered locations are identified with calculated electric field intensity (V/m)
for magnetic exposure within the ellipse of dB/dt = 100 T/s. Direction of field is
perpendicular to cross section. View of internal organs shows conductive paths
through the heart. [From Reilly, 1998a].
Figure 14 illustrates the ellipsoidal model applied to a large man [Reilly, 1998a, Chapt. 9]. Body
dimensions in applied situations may differ from this example, and calculations would vary
accordingly. Furthermore, the location of the heart within the torso will vary as the body
position is changed from prone to erect, and will also move during the cardiac pumping cycle.
Calculated electric fields in Fig. 14 have been carried out by representing body cross sections as
ellipses. Numbered points indicate positions where the E-field (indicated in V/m) has been
calculated according to Eq. (A.1) with dB/dt = 100 T/s.

Calculated E-fields generally increase as the measurement location is moved toward the
periphery of the body, and are especially great along the minor axis of the ellipse. Points of
maximum E-field on the perimeter of the body are points 14, 9, and 3 in the three cross sections;
the maximum E-field at the heart occurs at points 10, 6, and 1.

An underlying assumption in the data in Fig. 14 is that the incident magnetic field is constant in
both magnitude and phase over the torso cross-section. A somewhat more conservative
calculation can be made by assuming the entire body is uniformly exposed to the magnetic field
by using an equivalent body ellipse equal to the height of the person. As noted in Appendix A,
such an assumption somewhat increases the calculated E-field. For instance, the calculated
increase in the E-field was 11% at point (10), and 14% at point (14).

Another implicit assumption is that the internal conductivity of the body is homogeneous. In
reality, the body is composed of various organs having diverse conductivities, as indicated in
Fig. 14, and these variations will distort the internal E-field with respect to a homogeneous
structure. For instance, in the sagittal cross section, the E-field will be enhanced between the
skin and the spinal column as induced current is crowded around the low-conductance vertebrae
and the curvature of the torso in the lower lumbar region. Indeed, Schaefer and colleagues[1994]
report that sensory perception from magnetic stimulation is enhanced in regions where bony
projections are just below the surface of the body.

It should be clear from Fig. 14 that the heart is part of a large electrically conductive circuit that
includes the blood vessels, diaphragm, liver, and intestines, and it should not be treated as an
isolated conducting organ suspended in a non-conductive medium (the lungs) as some have
opined. With longitudinal magnetic exposure (Fig. 14c), the heart is part of a circumferential
electrical circuit that includes the pericardium and muscle lining the thorax.
Using the ellipsoidal model, an in situ field of 6.15 V/m (the assumed median threshold for nerve
excitation) has been calculated to be induced in the periphery of the torso with whole-body
exposure to dB/dt = 37.5T/s (see Table A.1, item 3). That theoretical value applies to conditions
of exposure that minimise the excitation threshold, namely: a very large adult; constant
magnitude, direction, and relative phase of the incident field over the dimensions of the body; a
monophasic square-wave shape of the in situ electric field. In most cases, experimental
conditions deviate from the optimal parameters resulting in greater thresholds than the minimum
ones.

One of the cited optimal conditions was a monophasic square wave shape for the induced electric
field. Note that the induced field follows the waveform of the time derivative of flux density,
dB/dt, which is necessarily biphasic for a magnetic pulse; the mean of dB/dt is zero if the rise and
fall magnitudes of flux density are equal, although the rise and fall times need not be equal. If
the induced waveform is such that the phase reversal is either delayed or is gradual, then the
threshold can be effectively the same as would apply to a monophasic waveform. Consequently,
it is generally conservative, but not unreasonable, to assume a monophasic waveshape for dB/dt.

The conservatively derived theoretical value of 37.5 T/s may be compared with experimental
thresholds conducted with pulsed magnetic field exposure of the human torso in MRI studies
[Bourland et al., 1990, 1991a, 1991b, 1997; Budinger et al., 1991; Cohen et al., 1990;
Mouchawar et al., 1991; Nyenhuis et al., 1990; Schaefer et al., 1994, 1995; Yamagata et al.,
1991], as previously reviewed [Reilly, 1998a, Sect. 9.7]. Mean perception thresholds of 60 T/s
were reported by two investigators [Budinger et al., 1991; Cohen et al., 1990], and a minimum
threshold of 45 T/s was reported by another [Bourland et al., 1990]. Higher thresholds were
reported by others, but, like the above cited studies, these involved sub-optimum waveforms or
conditions not conducive to minimum rheobase values.

Simulated MRI fields used in experiments discussed above varied considerably in amplitude and
relative phase over the dimensions of the human torso. The optimum field metric for
electrostimulation is not clear when such nonuniformity exists. Recent studies report perception
thresholds in terms of the spatially averaged exposure, rather than the spatial peak as in most of
the studies mentioned above. Using a spatial average metric, an average rheobase value of the
perception threshold was reported at 25 T/s in one study involving 65 subjects [Hebrank, 2000],
and 28.8 T/s in another study involving 84 subjects [Nyenhuis et al., 2001].

Cardiac excitation thresholds using magnetic stimulation have been determined in dogs. Early
results [Mouchawar et al., 1992; Yamaguchi et al., 1992] indicated dB/dt thresholds in excess of
what would be predicted from the models used here (Tables 6 and A.1), although this excess
could be explained by the use of sub-optimum exposure conditions in the cited studies [Reilly,
1993]. More recent test results with dogs [Schaefer et al., 2000] conformed well with the models
used in this chapter when scaled from animal to human dimensions. It was also established that
the addition of a 1.5 T static field to the time-varying excitatory field does not alter cardiac
excitation thresholds [Bourland et al., 1999].

Table 6. Models for established magnetic dB/dt thresholds of reaction:

whole body exposure; median thresholds(a).

Reaction τe fe

(T/s-pk)(b) (ms) (Hz)


Synapse activity alteration, brain 1.45 25.0 20

10-µm nerve excitation, brain 237 0.149 3350

20-µm nerve excitation, body 37.5 0.149 3350

Cardiac excitation 88.7 3.0 167

(a) Interpretation of Table as follows: for ; for . Also for


; for .

(b) (T/s-pk) refers to the temporal peak of the magnetic flux density.

Table 6 presents dB/dt ( ) thresholds based on the theoretical and experimental data described
above. Appendix A describes the methods whereby the external field thresholds of Table 6 are
derived from the in situ parameters of Table 2. Thresholds are computed from the parameters of
Table 6 as

for (13a)

for (13b)

where tp is the phase duration of the waveform. Alternatively, the limits can be determined
as

for (14a)

for (147b)

Flux density, B, listed in Table 7 can be computed from the Table 6 criteria using the
relationships for sinusoidal fields

(15)

(16)

where is the minimum (rheobase) threshold value of dB/dt, and Bo is the minimum threshold
value of B. Median flux density thresholds are computed from Table 7 as

for f ≥ fe (17a)

for f ≤ fe (17b)
Table 7 Median magnetic flux density thresholds; whole body exposure.

Reaction Bo Ho fe

(mT-rms) (A/m-rms) (Hz)


Synapse activity alteration, brain 8.14 6.48x103 20

10-µm nerve excitation, brain 7.97 6.34x103 3350

20-µm nerve excitation, body 1.27 1.00x103 3350

Cardiac excitation 59.8 4.76x104 167

(a) Interpretation of Table as follows: for ; for .

The flux density limits in Table 7 are based on the assumed in situ limits of Table 2 evaluated at
the site of interaction. For instance, the brain exposure limits are based on the estimated field
induced in the outer perimeter of the cerebral cortex; cardiac excitation applies to the field
induced in the apex of the heart; and peripheral nerve limits are based on the maximum induced
field in the periphery of the torso.

6.2. Static or quasi-static magnetic field exposure

Whereas Equation (17b) indicates that flux density thresholds would increase to infinity as the
frequency approaches zero, an upper limit on flux density is required in dosimetry standards to
avoid adverse effects from magnetohydrodynamic forces on moving charges within a magnetic
field. Such movement is typically associated with the vascular system, although observable
effects can also result from the rapid movement of the body or eyes within a strong static field.
The physical effects are Hall voltages or Lorentz forces.

With static magnetic fields, reactions under laboratory conditions include a 17% increase in
human cardiac cycle length at 2 T [Jehesen et al., 1988]. The authors opined that the observed
effect is probably harmless in healthy subjects, but that its safety in dysrhythmic patients was not
certain. Other observations included a 0.2 - 3% change in blood velocity between 1-10 T
[Dorfman, 1971; Keltner, 1990]. A host of adverse effects were noted at 1.5 T, including:
vertigo, difficulty with balance, nausea, headaches, numbness and tingling, phosphenes, and
unusual taste sensations; much more marked reactions were noted at 4 T [Schenck et al., 1992].
Other effects include benign enhancement of the cardiac T-wave in rats at 4 T [Gaffey and
Tenforde, 1981, Tenforde et al., 1983].

The studies of Schenck and colleagues report adverse effects in a significant number of subjects
at 1.5 T, which is a reasonable choice as a median threshold for adverse effects. A peak value of
1.5 T is associated with a slowly varying sinusoidal field of 1.06 T-rms.

7. Adverse reactions to electrostimulation


The purpose of electromagnetic field standards is to avoid adverse reactions, not just perceptible
ones. A recent standard has defined an adverse effect as one that is detrimental to the health of
an individual due to exposure to an electric or magnetic field, or a contact current [IEEE, 2002].
This same standard takes the position that aversive or painful electrostimulation is considered an
adverse effect [IEEE, 2002].

7.1. Adverse nerve reactions

Painful sensations from magnetic stimulation of peripheral nerve are reported at multiples above
perception thresholds of 1.3 [Budinger et al., 1991], 1.6 [Bourland et al., 1997], and 1.48
[Nyenhuis et al., 2001; Schaefer et al., 2000] – an average multiple of 1.45. The mean threshold
for intolerable pain was observed at a perception multiple of 2.05 [Schaefer et al., 2000]. I
calculate the median rheobase threshold for painful sensations as Eo = 6.15 x 1.45 = 8.92 V/m
(peak). Based on a log-normal probability model of human perception thresholds of electrical
stimuli (see Sect. 6), a conservative estimate of a one-percentile pain reaction threshold for
healthy adults would be a factor of 3 below the median, resulting in a rheobase of 2.97 V/m.

In the case of contact current stimulation, unpleasant and painful sensations are elicited at greater
multiples above perception than with magnetic stimulation. Based on experimental data from
several sources [Reilly, 1998a, Table 7.3], painful cutaneous stimulation is estimated to occur at
a multiple of 2.4 above the perception threshold; unpleasant sensations are estimated to occur at
a multiple of 1.7; the ratio of pain to unpleasantness thresholds is about 1.4.

That smaller pain-to-perception ratios are found with magnetic stimulation than with contact
current stimulation may be explained by the fact that in magnetic stimulation, the distribution of
induced current varies only gradually with respect to body dimensions. Consequently, at a field
level where some neurons first begin to be excited, a small increase in the field may excite
neurons over a large area. If pain is magnetically induced in some area of the body, it is likely to
be in an extended area. In contrast, cutaneous stimulation is more focal. Suprathreshold
stimulation in a large area may be more painful than in a small area, and that might account for
the differences in pain-to-perception ratios between magnetic induction and small-area contact
current.

7.2. Adverse cardiac reactions

Cardiac excitation is not necessarily hazardous, although ventricular fibrillation (VF) is a life-
threatening condition that is usually fatal unless it is quickly reversed with specialised
defibrillation equipment. Minimum thresholds for VF typically exceed those for excitation by a
factor of 50 or more if the excitation is a single event. However, if the heart is repeatedly
excited, the VF threshold drops such that the margin between VF and excitation thresholds may
be reduced to a factor as little as two when the stimulus is applied during the vulnerable period
within the cardiac cycle [Reilly, 1998a, Chapt. 6].

7.3. Adverse synapse reactions

In connection with phosphene threshold experiments, Lövsund and colleagues [1980a, p. 330]
state:

Virtually all the volunteers noted tiredness and some reported headaches after the
experiment. Some experienced after-images which were generally of only short duration
following exposure to the magnetic field. In one case, however, they persisted up to ten
minutes after the experiment. Individual volunteers reported spasms of the eye muscles,
probably arising from stimulation by the field.

These findings were similar to those of Silny [1986], who reported headaches, indisposition, and
persistent visual evoked potential (VEP) alterations at flux density levels above phosphene
thresholds, but still well below nerve excitation thresholds (by a factor of 23).

Clearly adverse reactions that may be attributable to CNS reactions (tiredness, headaches, muscle
spasms, persistent after images) are reported in connection with phosphene threshold
experiments. It is unlikely that the phosphenes themselves were causing the reported adverse
reactions. A plausible explanation is that the adverse effects were due to electrostimulation of
brain neurons in accord with the synapse mechanism discussed previously.
The ability of sub-excitation fields to alter neuronal response has also been reported after
exposure of hippocampus slices from the rat brain to magnetic fields [Bawin et al., 1984; 1986]
in which induced E-field intensities were as low as 0.75 V/m peak – a factor of 16 below the
threshold of 12.3 V/m for excitation of a 10-µm neuron. The rate of maze learning in living
mice was significantly reduced by exposure to flux densities at and below 0.75 mT at 50 Hz
[Sienkiewicz et al., 1998a,b]. Although the cited studies did not establish a synaptic mechanism,
they do support the view that CNS effects, including adverse ones, are possible well below
thresholds of excitation of brain neurons.

7.4. Adverse magnetohydrodynamic effects

Magnetohydrodynamic effects and forces on charges due to rapid body motion in strong static
and quasi-static fields produce a variety of biological effects. As reviewed in Sect. 7.2, adverse
reactions, including nausea, vertigo, and taste sensations associated with head movements, occur
at 1.06 T-rms (1.5 T-peak) in approximately 50% of human subjects at frequencies below one Hz
[Schenck, 1992].

8. Development of exposure standards


The preceding paragraphs of this Chapter treat thresholds of human reactions to
electrostimulation. This information alone is not sufficient to derive dosimetry standards;
indeed, the analyst must consider additional factors, as discussed below. The discussion that
follows adopts the rationale of a recent standard [IEEE, 2002], the development of which in
recent years has occupied a part of my life far beyond any expectations I had when I undertook
this responsibility. I must admit my bias in this approach, and restate that some of the
conclusions, although they have satisfied the vast majority of IEEE committee members (now
known as ICES – the International Committee on Electromagnetic Safety), are not universally
accepted. I also recognise that some of the ideas discussed below cannot be demonstrated by
syllogistic logic, but rather require scientific judgment.

One of the issues addressed by ICES that required judgment concerned the question of whether
single or multiple tiers should be used in the standards. The committee at an early stage decided
to afford protection to individuals in the general population and to groups in “controlled
environments,” which were defined as:

areas accessible to those who are aware of the potential for exposure as a concomitant of
employment, to individuals cognizant of exposure and potential adverse effects, or where
exposure is the incidental result of passage through areas posted with warnings, or
where the environment is not accessible to the general public and those individuals
having access are aware of the potential for adverse effects.
It was assumed that for the controlled environment, education and various mitigating measures
could be taken to reduce the probability of adverse reactions of exposed individuals, although the
exposure limits were intended to protect against adverse effects for almost all people.

The two-tier approach adopted by ICES for its low-frequency standard differed significantly,
however, from the approach used by other standards [e.g., ICNIRP, 1998; IEEE, 2002]. Other
standards typically define adverse reaction thresholds, and then apply a safety factor for
application to exposed individuals in work environments, or in controlled environments. They
then apply an additional safety factor for application to the general public.

The approach taken in the ICES low-frequency standard instead defined an acceptable level
applicable to all people, in which a probability factor and safety factor had been applied to
median adverse reaction levels. Under some conditions in controlled environments, higher limits
were allowed. ICES considered this approach acceptable because its standard was based on
avoidance of short-term reactions that would be immediately apparent to the exposed individual,
rather than chronic exposure health effects as sub-perception levels, and where cumulative
exposure might be significant. It was assumed that, because the short-term reactions would be
apparent to exposed individuals, they could remove themselves from the environment, modify
their activities, or could take other action to avoid the exposure entirely.

8.1. Basic restrictions

Table 8 Factors for converting median thresholds to maximum permissible exposure levels
(MPEs) as adopted by ICES [IEEE, 2002]

A B C D E F G

Safety factor Rheobase basic restr.

(RMS values)
Reaction Locus med. rms Adverse Prob. Gen. Contr. Gen. Contr.
rheobase
mult. Mult. public Environ Public Environ.
Eot
Fa Fp Fs Fs Eob Eob
(V/m)
(V/m) (V/m)
Synapse Brain 0.053 1.0 0.333 0.333 1.00 5.89x10-3 1.77x10-3

Alter.
10-µm Brain 8.70 1.0 0.333 0.333 1.00 0.970 2.90
neuron

excite.
20-µm Body 4.35 1.45 0.333 0.333 1.00 0.700 2.10

neuron (percept.) (pain)

pain
20-µm Hands, 4.35 1.45 0.333 1.00 1.00 2.10 2.10

neuron feet, (percept.) (pain)

pain wrists,

ankles
Cardiac Heart 8.49 1.00 0.333 0.333 0.333 0.943 0.943

excite. apex

Table 8 demonstrates the approach that ICES used to derive basic restrictions [IEEE 2002]:
column A lists the reaction under consideration; column B lists the locus of stimulation; column
C lists median rheobase excitation thresholds, Eot, from Table 2, but converted from peak to rms
values using the conversion E(rms) = E(peak)/ ; column D lists multipliers, Fa, applied to
column C that convert from a median excitation threshold to a median adverse reaction threshold
(Sect. 8); column E lists multipliers, Fp, that convert from a median threshold to a low-
probability one (Sect. 6.0); columns F list safety factors, Fs, applied to the general public and in
the controlled environment; columns G list rheobase in situ fields, Eob = EotFaFpFs, which are
the rheobase basic restrictions.

Table 9 Basic restrictions applying to various regions of the body(a,b)

General public Controlled

environment
Exposed tissue fe Eo Eo

(Hz) (V/m-rms) (V/m-rms)


Brain 20 5.89x10-3 1.77x10-2

Heart 167 0.943 0.943

Hands, wrists, feet & ankles 3350 2.10 2.10

Other tissue 3350 0.701 2.10

(a) Interpretation of Table is as follows: for ; for


.

(b) In addition to the listed restrictions, exposure of the head and torso to magnetic fields
below 10 Hz shall be restricted to a peak value of 167 mT for the general public, and 500 mT
in the controlled environment.

Source: [IEEE, 2002]

Table 9 lists basic restrictions appearing in the ICES standard. Interpretation of the table is as
follows:

Ei = Eob for f ≤ fe (18a)

Ei = Eob(f/fe) for f ≥ fe (18b)

where Ei is the allowable in situ electric field at frequency f, and Eob is the rheobase basic
restriction listed in columns G of Table 8.

Basic restrictions listed in Table 9 are in terms of in situ induced electric fields; the mode of
induction, however, can be through the action of the environmental magnetic or electric field, or
contact current. In addition to induced electric field specifications, it is also necessary to restrict
the in situ magnetic field to avoid adverse reactions due to magnetohydrodynamic effects from
very low frequency magnetic fields (see 7.2), as noted in the notes below Table 9. It is not
necessary to specify magnetic field basic restrictions at greater frequencies, because potential
adverse effects would be related to the induced electric field, rather than the in situ magnetic
field itself.

The following paragraphs summarise the rationale for the multipliers in Table 8.

Adverse reaction factor

Pain is considered an adverse response with peripheral nerve excitation. An adverse reaction
multiplier of Fa = 1.45 is applied to the nerve excitation threshold to derive a pain threshold (see
8.1). With synaptic effects, brain stimulation, and cardiac excitation, excitation itself is
considered adverse as noted in 8.3; hence the adverse reaction multiplier of Fa = 1.0 is applied to
the excitation threshold for these reactions.

Probability factor

A probability factor, Fp, is applied to convert from a median threshold to a low-probability one.
For a lognormal distribution in which the slope parameter (median-to-one-percentile ratio) is 3,
the multiplier of 0.333 applied to the median threshold corresponds to a one-percentile most
sensitive subject. Whereas a slope parameter of 3 is observed in some cases (e.g., contact current
perception on the forearm), with other reactions (magnetic field perception, cardiac VF, brain
ECT thresholds), the slope parameter is very close to 2.0 (see Sect. 6). With a slope parameter of
2, a multiplier of 0.333 applied to the median threshold results in a 0.01% probability rank.

Safety factor

A safety factor multiplier of Fs = 0.333 provided a margin of protection for exceptionally


sensitive individuals, uncertainties concerning threshold effects due to pathological conditions or
drug treatment, uncertainties in the reaction thresholds, and uncertainties in the induction
models. In the case of the hands, wrists, feet, and ankles, ICES chose Fs = 1 for the general
public in recognition of the narrow cross sections and preponderance of low conductivity tissue
that tend to enhance the in situ E-field in these areas in comparison with other areas of the body.
The committee reasoned that because these regions lack critical function when compared with
the vital organs, a greater localised electric field could be permitted. In the case of the controlled
environment, Fs = 1 for all of the reaction types except for cardiac excitation under the
assumption that a small probability of discomfort would be acceptable in the controlled
environment for some mechanisms, but that cardiac excitation would be unacceptable for all
individuals. The safety factor Fs = 1 was justified for the indicated exposures as explained
above.

If the safety factor Fs = 0.333 is to be compared with that applied in standards at higher
frequencies [e.g., ICNIRP, 1998; IEEE 2002], note that a divisor of 3 applied to the magnitude of
the induced field is equivalent to a divisor of 9 in the specific absorption rate (SAR) because
SAR is proportional to the square of the induced field.

8.2. Maximum permissible exposure levels

Sophisticated test equipment or computational capabilities may sometimes be required to assess


whether basic restrictions are met. So that standards can be more easily interpreted, it is
desirable to define maximum permissible exposure (MPE) levels in terms of the environmental
field, rather than the induced in situ field. Most users can more easily demonstrate compliance
with a standard through environmental field measurements.

Environmental magnetic field MPE limits


Figure 15 Median thresholds for adverse stimulation from magnetic field exposure (broken
lines) and recommended maximum permissible exposure limits (solid lines);
whole-body exposure to spatially constant field.

Figure 15 illustrates the derivation of MPE levels for magnetic fields. The figure shows median
thresholds of adverse reaction (broken lines), and MPE levels (solid lines) with whole body
exposure. The MPEs have been derived from the minimum adverse thresholds at each
frequency, decremented by the appropriate probability and safety factors in Table 8. The curve
for synapse alteration has been extended to 1000 Hz. The MPE curves have been derived from
the lowest adverse reaction threshold across the frequency spectrum as follows: 0 - 0.153 Hz,
magnetohydrodynamic effects; 0.153 - 759 Hz, synapse alteration; above 759 Hz, peripheral
nerve pain. Note that the MPEs in the controlled environment correspond to low probability
reaction thresholds (≤ 1%). The limits applicable to the general public are lower by a factor of
three.

Table 10 Magnetic maximum permissible exposure (MPE) levels: exposure of head and
torso.

General public Controlled environment


Frequency range B(rms) H B (rms) H
(rms) (rms)
(Hz) (mT) (mT)
(A/m) (A/m)
<0.153 118 9.39x104 353 2.81x105

0.153 - 20 18.1/f 1.44x104/f 54.3/f 4.32x104/f

20 - 759 0.904 719 2.71 2.16x103

759 - 3350 687/f 5.47x105/f 2060/f 1.64x106/f

3350 – 105 0.205 163 0.615 490

(a) Limits from 0 – 3000 Hz correspond to ICES standard, [IEEE, 2002].

(b) MPEs refer to spatial maximum.

(c) f is frequency in Hz.

Table 10 lists MPE values for whole-body exposure to sinusoidal magnetic fields. Recall that
the listed values incorporate conservative assumptions such that adherence to them insures that
the basic restrictions are not exceeded. Since the MPEs have been conservatively derived, it is
possible that one may exceed them, and still not exceed the basic restrictions.

The upper frequency of the ICES low-frequency standard was 3 kHz. However, the frequency
limits in Table 10 have been extended to 100 kHz for purposes of this chapter. The extension is
consistent with the principles expressed previously in this chapter.
Above 100 kHz, thermal thresholds are lower than those for electrostimulation when the
exposure waveform is a continuous sinusoid. At these higher frequencies, published standards
define MPE limits on the environmental rms field that are inversely proportional to frequency
[e.g., ICNIRP, 1998; IEEE 2002]. The reason for this decrement is to ensure that the rms value
of the induced field (i.e., its heat-producing capacity) remains constant. However, if the
exposure waveform is a pulsed sine wave, the crossover frequency between thermal stimulation
and electrostimulation increases, and it may be necessary to specify additional compliance
specifications, as noted in Sect. 2.6.

Environmental electric field MPE limits

In the absence of contact current and spark discharges (Sect. 5), environmental E-fields can
sometimes be perceived through vibration of body hair caused by the action of the field on
charged hair follicles. With a sufficiently strong field the sensation can be annoying to some
people. For instance, at 20 kV/m in an outdoor environment, 50% of standing adults can
perceive a 60 Hz field, and about 5% will consider the sensation annoying [Reilly, 1978; Deno
and Zaffanella, 1982]. Although 20% of subjects perceived a 60-Hz electric field at 9 kV/m, less
than 5% could detect electric fields of 2 or 3 kV/m [Reilly, 1978]. With hands raised above the
body, the median perception threshold is 7 kV/m.

The most significant electrostimulation effects from environmental electric fields result from
contact currents and spark discharges. These effects can be most severe when a grounded person
touches a large conductive object that is well-insulated from ground and is within a strong
electric field. The magnitude of the effect depends on the effective area of the induction object,
and its capacitance to ground in accord with well-known physical principles [Reilly, 1998a,
Chapt. 9].

It is not possible to absolutely protect against all possibility of adverse stimulation without
mitigating the induced charge on the object when very large (or long) objects are situated near
sources that produce strong electric fields that are very extended spatially, such as is the case
with high-voltage power transmission lines. For instance, one might postulate a long fence wire
on insulated posts running parallel to a high-voltage transmission line. In such cases, it is
preferable to avoid electrostimulation by properly grounding the conducting object, (as stated in
other safety codes) rather than by limiting the electric field to an impractically small level.
A situation that is difficult to mitigate against is where a person who is insulated from ground
touches a grounded object within a strong electric field. The contact current in such a case
would follow Eq. (11). The ICES standard specified its electric field MPE level for the general
public so as to limit the contact current to 0.5 mA when a person of height 1.75 m touches a
grounded conductor within a vertically polarised field. By solving Eq. (11) for these conditions,
ICES determined an electric field limit given by

E = 1.84x106/f (19)

If this law were to extended to zero frequency, the electric field limit would approach infinity.
An upper limit must be placed on the maximum permissible E-field to limit the probability of an
adverse reaction to a spark discharge.

Table 11 Environmental electric field MPEs, whole body exposure as defined in the ICES
standard [IEEE, 2002].

general public controlled environment

Frequency range E Frequency range E

(Hz) (V/m-rms) (Hz) (V/m-rms)


1 - 368(c) 5,000(a,d) 1 - 272(c) 20,000(b,e)

368 - 3000 1.84x106/f 272 - 3000 5.44x106/f

3000 - 105 614 3000 - 105 1813

(a) Within power line rights-of-way, the MPE for the general public is 10 kV/m under normal
load conditions.

(b) Painful discharges are readily encountered at 20 kV/m, and are possible at 5 -10 kV/m without
protective measures.

(c) Limits below 1 Hz are not less than those specified at 1 Hz.

(d) At 5 kV/m induced spark discharges will be painful to approximately 7% of adults (well-
insulated individual touching ground).
(e) The limit of 20 kV/m may be exceeded in the controlled environment when a worker is not
within reach of a grounded conducting object. A specific limit is not provided in this standard.

(f) MPEs above 3 kHz were not specified in the standard [IEEE, 2002].

The field limitations in Table 11, which were provided by ICES for protection against adverse
contact current, vary in inverse proportion to frequency as in Eq. (19). A maximum permissible
field of 5 kV/m is specified in Table 11 for the general public to limit the probability of painful
spark discharges. ICES estimated that spark discharges would be painful to approximately 7%
of individuals who are 1.8 m tall, are well insulated from ground, and who touch a grounded
object within a 5 kV/m field. Unpleasant spark discharges can also occur when a grounded
person touches a large conductive object that is well-insulated from ground and is situated
within a strong field.

In the controlled environment where the MPE is limited to 20 kV/m, painful spark discharges,
but not contact currents, can be readily encountered at the stated limit for an insulated person at
ground level touching a grounded conductive object In such strong fields, ICES recommended
that workers should limit the probability of painful spark discharges by appropriate use of
protective clothing, grounding measures, contacting techniques, or other work practices that
consider these environmental electric field effects. In the controlled environment, conductive
suits could be worn that shield the body from high environmental electric fields, thereby greatly
reducing indirect electrostimulation.

Power line rights-of-way fall somewhere between the definitions of "controlled" and
"uncontrolled" environments for the general public in that public activity can be circumscribed
by the utility, but that public access is often allowed for the public benefit. Consequently, ICES
specified a limit of 5 kV/m for the general public in regions off the right-of-way, but allowed an
intermediate field of 10 kV/m within the right-of-way under normal load conditions.
Experimental data using spark discharge stimuli on human subjects [Reilly and Larkin, 1987;
Reilly, 1998b] can be applied to this exposure. In a field of 10 kV/m, about 50% of adult
subjects (1.8 m tall) who are well insulated from ground would experience painful discharges
when contacting a grounded conductor. The stated probability would increase with taller
subjects, and decrease with shorter ones. It is also decreased by imperfect insulation of the
person with respect to ground.

Static or quasi-static electric fields

The maximum permissible environmental electric field in Table 11 is capped to limit the
probability of painful spark discharges. This limit could, in principle, be extended to arbitrarily
low frequencies since even a single discharge can be painful. However, at a sufficiently low
frequency, the time constant, τh, at which a human can maintain a charge will begin to limit the
magnitude of the induced charge. The time constant is given by the product of the capacitance
and resistance to ground of the person. For example, consider a resistance of 1000 MΩ, which is
applicable to 10% of people with normal footwear on dry ground [Reilly, 1979, 1998b, p52], and
a capacitance of 150 pF. These assumptions result in a time constant of 150 ms, which is
equivalent to a frequency of 1 Hz below which the induced voltage in a given field would fall,
and the permissible exposure could rise. However, for people on well insulated surfaces, longer
time constants would be possible. The validity of this observation is apparent considering that
one may experience an unpleasant carpet spark a second or more after the charge has been
acquired.

These observations may be applied to the standards of Table 11 as follows. For leakage
resistance of 1000 MΩ, the allowable maximum limits below 1 Hz could be increased
approximately in inverse proportion to frequency; for greater resistances, the applicable
frequency would become lower.
Appendix A: Magnetic field induction
model
The magnetic induction model used in developing this chapter treats an exposed cross section of
the body as an elliptical shape, with homogeneous conductivity. A solution for this model,
applicable to situations where the wavelength of the field is much greater than body dimensions
was published by Durney et al., [1975], and expressed in applied form by Spiegel [1976]. The
present form used here is the one expressed by Reilly [1991]. A general expression for the
induced E-field due to an incident B-field that is constant in magnitude and relative phase over
the ellipse is

(A1)

where au and av are unit vectors along the minor and major axes, respectively, (a,b) are the semi-
major and -minor axes, respectively (u,v) is the location within the exposed area, and is the
time rate of change of the magnetic flux density in a direction perpendicular to the cross section.
In the calculations that follow, the magnitude of the induced field, E, is expressed, rather than its
vector components. The coordinate system is such that the minor axis of the ellipse is along the
u-direction, and the major axis is along the v-direction.

Table A1 summarises the exposure conditions used to determine data expressed in Table 6.
The entries of Table A1 are interpreted as follows. The second column expresses the exposure
condition. For instance, the entry in the first row is interpreted as excitation of a 10 µm neuron
located in the brain, with a magnetic field perpendicular to the sagittal cross section. The third
column gives the semi-minor and semi-major axes of the ellipse. The fourth column gives the
location within the cross section where the E-field is evaluated. The fifth column is the assumed
rheobase value of Eo (from Table 6). The last column gives the values of determined from
Eq. (12). In this formulation, it is assumed that an ellipse is fitted to the torso, body, or head in
one of three orientations. Consequently, the reference system (u,v) is tied to the fitted ellipse,
and not to one specific reference system with respect to the body.

In items (1) and (2), the assumed ellipse is not supposed to represent the actual size of the brain,
but rather the size of an ellipse that encloses its outer perimeter (the cerebral cortex) where the
magnitude of the induced E-field is greatest. The ellipse enclosing the brain has semi-major and
-minor axes that are 1.5 cm smaller than the assumed head size to account for the distance of 1.5
cm between the cortex and the scalp. Items (3) and (5) treat the exposure as uniformly covering
the entire body; items (4) and (6) assume only the torso is exposed. The latter points are
included to demonstrate that there is but a modest difference (about 10%) between worst-case
exposure of the entire body versus exposure of only the torso with respect to peripheral nerve
and cardiac stimulation.

The points u,v are selected to correspond to the worst-case exposure point for each of the
assumed scenarios. In the case of the brain (Items 1 & 2), the cortex is where the induced E-field
is greatest, and sagittal exposure provides the greatest magnetic induction loop. For items (3)
and (5), an ellipse is fitted to the entire body viewed in the sagittal cross section. In the case of
the heart, the location of greatest sensitivity to electrical stimulation is its apex [Roy et. al, 1987],
and the greatest induced field at that location is found with sagittal exposure [Reilly, 1991]. The
points (u,v) in items (5) and (6) correspond to the apex of the heart.

The exposure ellipses in Table A1 correspond to a large (but not extreme) body size for adults
based on anthropomorphic data [SAE, 1979]. It is conservative to assume large body
dimensions.

Table A1 Elliptical exposure model used to compute magnetic induction.

b,a u,v peak Eo

Item Exposure (cm, cm) (cm, cm) (V/m) (T/s-pk)


1 10-µm nerve, brain, sagittal 9, 10.5 9, 0 12.3 237.1

2 synapse, brain, sagittal 9, 10.5 9, 0 0.075 1.45

3 20-µm nerve, body, sagittal 17, 90 17, 0 6.15 37.5

4 20-µm nerve, torso, coronal 20, 40 20, 0 6.15 38.4

5 heart, body, sagittal 17, 90 14, 18 12.0 88.7


6 heart, torso, sagittal 17, 40 14, 18 12.0 98.6

7 leg 9, 42 9, 0 6.15 71.5

(a) b,a represent semi-minor and semi-major axes, respectively, of ellipse fitted to particular body part, viz: the
brain in items 1 and 2, the torso in item 4, and the whole body in items 3 and 5. (u,v) represents the location
within the ellipse where the induced field was evaluated, where u and v are measured along the minor and major
axes, respectively.

∗ Unless otherwise noted, the term “waveform” as used in this chapter refers to the temporal variation of the in
situ electric field, the in situ current density, or the conducted current – quantities that are proportional to one
another. For the relatively low frequencies treated in this chapter, the induced electric field is proportional to
the temporal derivative of the environmental electric or magnetic field. Consequently, the term “waveform”
also applies to the time derivative of the environmental field.
EMF DOSIMETRY HANDBOOK

INTERNATIONAL GUIDELINES FOR QUALITY EMF (RF or


ELF) RESEARCH

Sheila Johnston PhD

Neuroscience Consultant

10 Queen’s Mews,

London, UK,

W2 4BZ
Please note this is a summary document drawing on the expertise of the many international expert
authors quoted directly and indirectly. I am only the gatherer of their statements. I acknowledge them
fully.

Please note that each Specific Research Study Design (Epidemiological, Human Volunteer, In Vivo, In
Vitro) can be read independently, but I would advise for full understanding that you read the complete
document.

EMF DOSIMETRY HANDBOOK

INTERNATIONAL GUIDELINES FOR QUALITY EMF (RF or ELF)


RESEARCH
Introduction

IEEE & ICNIRP

IEEE ICES TC-95, SubCommittees 3 and 4

ICES EMF Literature Surveillance

IEEE ICES EMF Standard Setting

ICES SC-3

ICES SC-4

ICNIRP, IARC and WHO EMF Project Collaboration

ELF Reviews: IARC-2002, ICNIRP-2003, WHO EMF Project-2005

RF Reviews: IARC-2006, ICNIRP-2007, WHO EMF Project-2008

Revision of the ICNIRP Guidelines 0-300 GHz -1998

COST 281

GUIDELINES FOR QUALITY EMF (RF / ELF) RESEARCH

Publication

Replication and EMF Standard Setting Criteria

General Experimental Design Criteria


Hypothesis

Good Laboratory Practice

Methodology

EMF Exposure Systems

RF Exposure Metrics

ELF Exposure Metric

Results Analysis

Conclusion

Publication

Scientific Responsibilities

Replication

Meta Studies

General Research Priorities

Dosimetric Units

RF

ELF

Threshold Studies

SAR as a Measure of Temperature Increase

Modulation

Human RF Dosimetry

Meta Studies

Specific Research Designs

Introduction
Traditional Evaluation of Research

New Evaluation of Research: Molecular Epidemiology

The Epidemiological Study Design

Introduction

Case Control and Cohort Studies

Case Control Studies

Cohort Studies

Spatial Epidemiology (Quoted from Elliott and Wartenberg, 2004)

Bias

Countering Bias

Dosimetry

Data Analysis

Human Volunteer Study Design

Introduction

Hypothesis

GLP

EMF Exposure Systems

RF Exposure Metrics

Human RF Exposure in the Near Field

Human RF Exposure in the Far Field

ELF Exposure Metrics

Results Analysis

Conclusion
Publication

Scientific Responsibilities

Replication

Meta Studies

In Vivo Study Design

Introduction

Hypothesis

Good Laboratory Practice

Methodology

Traceable Dosimetry

Data collection and quality assurance

Results Analysis

Conclusions

Publication

Scientific Responsibilities

Replication

Meta Studies

In Vitro Study Design

Introduction

Hypothesis

Good Laboratory Practice

Methodology

Exposure Conditions
Results Analysis

Conclusion

Publication

Scientific Responsibilities

Replication

Meta Studies

REFERENCES

Appendix A

IEEE ICES SC-4 TRIAGES

For Example IEEE SC-4: In Vivo Triage: 16 Screens Are Summarized Below:
Introduction

International expert evaluations of the published literature are carried out according to guidelines
for quality EMF (0-300 GHz) research. As technology progresses, it is our duty as scientists and
engineers to continually review and monitor new findings and to update and revise our safety
standards and guidelines accordingly (Chou & D’Andrea, 2003; ICNIRP, 1998; Gajšek et al., 2002;
ICNIRP, 1998).
A comprehensive and critical review of the extant scientific database of electromagnetic field (EMF)
published literature is updated periodically by panels of current qualified experts, as recognized by
the international scientific community.
An important task of the international expert panels is to assess the relevant extremely low

frequency (ELF) and radiofrequency (RF) accumulating papers (2200+ papers) for standard setting

for human exposure limits to protect the population against any adverse health effects.

While exposure to EMF may cause biological effects, without any known adverse consequences, the
standards are based on established threshold levels for adverse health effects, i.e., levels of
exposure above which adverse health effects have been established and below which adverse
health effects have not been established.

IEEE & ICNIRP

There are currently 2 international expert panels that evaluate the ever-increasing EMF literature according to criteria for quality EMF (0-
300 GHz) research to set the standards and guidelines.

1. The Institute of Electrical and Electronics Engineers Incorporated (IEEE)[1], International


Committee on Electromagnetic Safety (ICES), Piscataway, New Jersey. ICES membership is open
to all interested persons; membership of the central governing and the technical committees
(TC-95 and TC-34) stands at more than 150 professionals representing 24 countries.
[http://grouper.ieee.org/groups/scc28/]. ICES develops standards through an open consensus
process that is transparent at every level. In addition to standards prescribing safety levels, ICES
also develops standards and recommended practices for implementing these standards, e.g.,
safety programs, methods for exposure assessment.
2. The International Commission on Non-Ionizing Radiation Protection (ICNIRP) Oberschleissheim,
Germany (14 members from 13 countries, see board 2004-8). [http://www.icnirp.de/]. New
members are appointed by the existing members.
Although IEEE ICES standards and ICNIRP guidelines are advisory (voluntary), and carry neither a
mandate for compliance nor a mechanism for enforcement, these documents frequently serve as
the bases for guidelines and regulations set forth by regulatory and other agencies in diverse
countries that have issued their own standards, which, in some cases, may differ substantially from
those of ICES and ICNIRP.
Because of the international expertise in EMF research of IEEE ICES and ICNIRP members, it is
valuable to focus attention on their processes of evaluating the relevant literature for the purpose
of establishing safety criteria.
These expert panels are multidisciplinary and include, for example, epidemiologists, neurologists,
biologists, toxicologists, oncologists, and psychologists who are appropriately specialized medical
scientists, and physicists, engineers, and statisticians. Only peer reviewed scientific studies are
included in the review. While anonymous peer review for publication adds confidence in the study
results, additional review is necessary to evaluate the study design, conduct of the experiment and
the statistical analyses, and to compare various aspects of the results with the results of other
studies to reveal the consistency of the results. National expert reports that for the most part are
not published in scientific journals may also be considered for review. But conference abstracts are
of little value as they generally receive no peer review, contain sparse information, and cannot be
considered as the final outcome of an experiment (Repacholi, 1998). The task of the expert panels is
to assess the entire ELF and RF scientific database (2200+ papers) for standards setting for human
exposure to avoid any established adverse health effects.

IEEE ICES TC-95, SubCommittees 3 and 4

ICES EMF Literature Surveillance


Subcommittee 3 and Subcommittee 4 (SC-3 and SC-4) of Technical Committee 95 (TC-95) develop
standards for safety levels with respect to human exposure to electromagnetic fields. SC-3 covers
the frequency range of 0 to 3 kHz; SC-4 covers the range of 3 kHz to 300 GHz. These subcommittees
review the EMF literature continuously and periodically they may publish review papers in the
scientific peer reviewed literature for the purpose of updating the standards (cf. BEMS Supplement).
The revision process established by the ICES is a continuing rigorous and open scientific process that
is transparent at all levels and includes the opportunity for input from all stakeholders.
The EMF literature surveillance covers health effects in two separate exposure frequency ranges, 0-3
kHz and 3 kHz - 300 GHz, because adverse biological effects (associated with exposures above the
exposure limits) are manifested primarily as induced in situ electric field stimulation in the lower
frequency range and as tissue heating in the upper range, respectively. There is some overlap of the
effects associated with induced in situ electric field stimulation and with heating in the intermediate
range 3 kHz to 100 kHz (or higher for pulsed fields).
IEEE ICES EMF Standard Setting
The IEEE standards are living documents and in accordance with IEEE rules must be reaffirmed or
revised every 5 years. The IEEE C95 standards are based on IEEE ICES members’ anonymous peer
reviews of all the published scientific studies of health effects. They provide a wide margin of safety
to workers and the public from adverse health effects from exposure to EMFs. The IEEE EMF safety
limits are issued as two separate standards.[2] As indicated above, SC-3 and SC-4 have independent
literature review panels who evaluate the papers in a parallel process for the ranges, 0-3 kHz and 3
kHz -300 GHz, respectively. Each subcommittee independently drafts and approves the standard
that applies to its frequency range.
Each standard is first balloted by the subcommittee, which requires at least 75% of the ballots
returned and at least 75% affirmative ballots following ballot resolution (during which time all
negative ballots, comments and changes to the draft resulting from ballot resolution are circulated
to the balloting group to allow members to comment, change or reaffirm their vote). This process is
repeated at the ICES Committee level after which the draft standard is submitted to the IEEE
Standards Association Standards Board (SASB) Review Committee (RevCom). The SASB RevCom has
oversight to ensure that due process has been afforded to all, and that the rigid IEEE SASB Rules for
standards development and balloting have been followed. When satisfied, the RevCom recommends
to the SASB approval of the standard. The SASB procedures provide a mechanism for appeals based
on violations of the IEEE balloting process; technical appeals are referred to the Committee and
Subcommittees. Once approved by the SASB, the standards are published, usually within 2-3
months.
The ELF standard (C95.6) was approved in September 2002 and published in October 2002. When
C95.6 was balloted by ICES, 93% of the ballots were returned with a 90% approval.
The current RF standard (C95.1) was approved in 1991, reaffirmed in 1997, an amendment was
published in 1999 and a second amendment was published in 2004.
ICES SC-3
TC-95 SC-3 was responsible for the development of IEEE C95.6-2002. During development of this
standard, SC-3 was chaired by Kent Jaffa (USA) and had about 75 voluntary ELF research members
from 11 countries. The results of the SC-3 review are incorporated within IEEE Std C95.6-2002.
The SC-3 review of the literature continues with Thanh Dovan (AU) and Philip Chadwick (UK) as SC-3
co-chairs. In addition, SC-3 members have published review papers in Health Physics, 83:3, 2002,
from an international conference in Brussels, 2000, entitled “The EMF exposure guidelines science
workshop”.
The IEEE ICES sponsored a one-day ‘Short Tutorial Course on IEEE Standard C95.6’ in June and
December 2004 in conjunction with their annual and semi-annual meetings. The same short-course
was presented to the Canadian Electrical Association in March 2004 and in Dublin in June 2005. J.
Patrick Reilly and Kent Jaffa, both of the USA, present this short course.
TC-95 is working toward harmonisation of the C95.6 standard with ICNIRP and to that purpose,
within the ELF range, J.P. Reilly has published a paper in Health Physics, “An analysis of differences in
the low-frequency electric and magnetic field exposure standards of ICES and ICNIRP,” (Reilly, 2005).
ICES SC-4
A complete revision of the RF standard (C95.1-1991) by SC-4 is now in progress. The revision is
based on the peer reviewed literature identified by the Literature Surveillance Working Group
chaired by Lou Heynick (USA, 1978-2005) who continuously identified the literature for review. This
process has been taken over by Joseph Morrissey and new papers are now entered on the WHO
EMF Project website
The membership of SC-4 stands at 122 members representing 24 countries and is co-chaired by C-K.
Chou and J.A. D’Andrea (both USA). Chou and D’Andrea coordinated the literature evaluation and,
along with the literature evaluation working group chairs and other designated experts published 12
RF and health, review papers in, the Bioelectromagnetics journal, Supplement 6, 2003 (Chou and
D’Andrea, 2003).
They have also incorporated a thorough literature review within the draft revision of IEEE Std C95.1-
1991/1999. This standard has undergone a complete revision, which was approved by SC-4 in
March and is now (August 2005) undergoing Sponsor (ICES) ballot. The SC4 literature evaluation
process is explained in detail in Appendix A 1.6 of the IEEE Std PC95.1-2005 Briefly adapted from
IEEE Std PC95.1-2005:
Working groups (WGs) evaluate engineering, epidemiology, in vivo, and in vitro aspects of the
research. Additionally, a WG on mechanisms assesses the role of mechanisms of interaction in
standard setting. The Engineering WG is tasked with assessment of the exposure systems, field
characteristics and measurements, dosimetry, specific absorption rates, induced currents and fields,
and temperature/humidity measurements. The sufficiency of the information provided in each
publication, to allow a full understanding of how the experiment was performed, is paramount. The
chair of each WG is responsible for providing copies of each paper to two independent reviewers,
together with specially designed and approved review forms called Triages. These ‘Triage’ forms
(available in Appendix A, below) are in a computer format that requires numerical scoring by
individual reviewers for entry into a computerized database. When a review is completed, the
reviewer gives the paper an overall technical merit rating on a 5-point scale. The rating scale is: Very
High = 5; Moderately High = 4; Acceptable = 3; Low = 2; and Very Low = 1. For ratings of 1 or 2, a
request is made for justification in writing by the reviewer. This is not requested for ratings of 3 and
above, which are considered acceptable. Strong discordance between the two reviews of a given
paper requires a third independent review. Periodically, the chair of each WG submits a summary of
the reviews completed to the Chair of the Risk Assessment WG (RAWG). All of the reviews are
performed by volunteers who are randomly selected from within each working group. The
identification of each reviewer will remain confidential.
ICNIRP, IARC and WHO EMF Project Collaboration

The ICNIRP, IARC, and WHO EMF Project expert members work in phased collaboration with each
other to evaluate the EMF literature.
1. The International Agency for Research on Cancer (IARC), with their secretariat in Lyon, France, is
part of the WHO. (Classifications of carcinogens: ELF, 2001). [http://www.iarc.fr/]
2. The World Health Organisation (WHO) EMF Project have their secretariat in Geneva
with members from 40+ countries. [http://www.who.int/peh-emf/en/]
The ICNIRP / IARC/ WHO EMF Project, EMF literature reviews cover health effects in the two
separate exposure frequency ranges of 0-100 kHz and 100 kHz - 300 GHz.
The IEEE and WHO/ IARC/ICNIRP have not yet harmonized on the intermediate division between the
two ranges but they may in the near future (Reilly, 2005).
On each of the two frequency ranges IARC evaluates the cancer literature, makes a classification and
writes a Cancer Monograph.
On each of the two frequency ranges the ICNIRP panel reviews the biological literature and
publishes its review as a Blue Book.
Consequently the WHO EMF Project panel prepares a review on each of the two frequency ranges
and writes an Environmental Health Criteria (EHC) Monograph.
Finally once they have fully reviewed and evaluated the literature on both the 0-100 kHz and the
100 kHz-300 GHz ranges, the ICNIRP panel will revise the EMF guidelines for limiting human
exposure over the entire EMF range, 0-300 GHz, as a paper published in Health Physics (1998;
2008?).

ELF Reviews: IARC-2002, ICNIRP-2003, WHO EMF Project-2005


Presently IARC has evaluated the cancer literature on ELFs and published an IARC Monograph ‘Non-
ionizing Radiation, Part 1: Static and Extremely Low Frequency Electric and Magnetic Fields’ (Vol. 80,
2002) and has made a cancer classification 2B for ELFs (19-26 June 2001). The IARC Panel (19-26
June 2001) consisted of 21 members from 11 countries and 4 observers (chosen by the secretariat)
and the IARC secretariat.
Following this, the ICNIRP panel 2000-2004 (12 members, the Chair A. McKinlay, vice chair and
chairman emeritus), along with invited consultants have written a review of the biological scientific
literature concerning exposure to static and low frequency EMFs 0-100 kHz (2003) including
dosimetry (2 members, 5 consultants), experimental investigations of EMF biological effects (of
cellular, animal and human experiments)(7 members, 7 consultants), and epidemiology (2 members,
4 consultants). Their report is published as the ICNIRP Blue Book entitled Exposure to Static and Low
Frequency Electromagnetic Fields, Biological Effects and Health Consequences (0 - 100kHz), (2003).
The WHO EMF Project panel are completing their update of the previous Environmental Health
Criteria (EHC) Monograph 35 on ELF Fields, (1992) including comprehensive risk assessment and
policy recommendations, expected in late 2005.

RF Reviews: IARC-2006, ICNIRP-2007, WHO EMF Project-2008


After IARC completes the meta analyses of results of the 13 country Interphone Study (2000-2005) a
similar progression of literature evaluations by the panels of IARC, ICNIRP and the WHO EMF Project
will be initiated for RF exposures in the range 100 kHz to 300 GHz.
IARC will evaluate the cancer-related literature of the RF bands and make a classification of RF
possibly in 2006 and publish an RF IARC Monograph.
Then ICNIRP will evaluate the biological science and publish an RF Blue book.
And WHO EMF Project will write an EHC Monograph on RF to update the previous EHC monograph
137.

Revision of the ICNIRP Guidelines 0-300 GHz -1998


Following the completed cycles of IARC/ ICNIRP/ WHOEMF Project evaluations of both the ELF and
RF published literature, ICNIRP will update the ‘ICNIRP guidelines for limiting exposure to time-
varying electric, magnetic, and electromagnetic fields up to 300 GHz’ (Health Physics 1998), possibly
in 2008. Thus we may have about a 10-year cycle to update the ICNIRP EMF guidelines for limiting
human exposure according to the current literature evaluation process.

COST 281

The fifth international RF literature review group is the European Cooperation in the Field of
Scientific and Technical Research- Telecommunication Information Science and Technology, (COST
281), with their secretariat in Bonn, Germany. The COST 281 science members are drawn from the
25 signatory European countries. http://www.cost281.org.
The COST 244 (1996-2000) continued on as COST 281 from 2001. The COST (281) members
continuously review the scientific literature on radio telecommunications and health and report to
the European Commission.
Previously, COST 244 members wrote two comprehensive reviews of the RF literature [Possible
Health Effects Related to the Use of Radiotelephones: Proposals for a research programme by a
European Commission Expert Group, 1996 (McKinlay); McKinlay, 1997; 1999 (Veyret)].
COST 281 members are not directly involved in setting standards; however they do advise the EC on
RF health effects literature for the purpose of policy and evaluating proposals for further research
funding.
GUIDELINES FOR QUALITY EMF (RF / ELF) RESEARCH

Publication

The first hurdle to demonstrate quality in research is publication; papers considered for health risk
assessment must be published in a peer reviewed journal.
When evaluating submitted papers for journal publication anonymous reviewers and editors may
use the following criteria to comment on the technical content of the papers: Does the TITLE
accurately and briefly portray the content? Does the ABSTRACT clearly convey the problem(s),
findings, and conclusion(s)? Does the INTRODUCTION clearly and concisely define the problem or
issue? Is the METHODOLOGY clear, complete, and adequate for the biology and dosimetry? Are the
RESULTS clearly and concisely presented, making proper use of tables? Is the DATA ANALYSIS
complete and valid, using appropriate statistical methods and tests, and error analysis? Is the
DISCUSSION valid, relevant, and concise? Are the CONCLUSIONS drawn fully, valid and stated
concisely? Are any key citations missing in the REFERENCES? Does the list include all the references
used in text? (Chou, 2003)

Replication and EMF Standard Setting Criteria

Since the paper will have already met the publication criteria, the panel members review the paper
to see if it can meet replication criteria and, if replicated, can meet the EMF standard setting
criteria. Review panels approach the literature in a structured way. Panels have a chairman and
section heads with their committee members representing expertise in epidemiology, in vivo
(animal and human laboratory studies) and in vitro research (tissue cultures) as well as
engineering/dosimetry, statistics and biophysical mechanisms of interaction. The section heads
summarise their areas of review for the chairman. We base guidelines on scientific data related to
adverse health effects meeting the literature evaluation criteria and we require information to be
consistent from multiple studies and disciplines.
Overall criteria for general experimental design, general research priorities as well as criteria for the
specific areas of research (epidemiology, human acute, in vivo and in vitro) are listed below. The
review criteria are derived from numerous published papers, books and standards and guidelines
and the five ICES SC-4 Triages. Some key references are listed directly below:
Papers: Hill, 1965; Blundell, 1996; Rothman et al., 1996; Cardis and Rice, 1997; Repacholi and Cardis,
1997; Repacholi, 1998; Rothman and Greenland, 1998; Repacholi and Greenebaum, 1999; Lin, 2002;
Adair, 2003; Brodsky et al., 2003; Chou, 2003; Gajsek et al., 2003; Habash, 2003; Habash et al., 2003;
Rushton and Elliott, 2003; Kheifets et al., 2003; Ahlbom et al., 2004; Elliott and Wartenberg, 2004;
Foster and Repacholi, 2004; Feychting et al., 2005; Greenland, 2005; Johnston and Scherb, 2005.
Books: EPRI, 1994; NIEHS, 1998; IARC Vol 80, 2002; ICNIRP, 2003; Stravoulakis (Ed), 2003 (Johnston
chapter, 2003); Ahrens and Pigeot, (Eds) 2005.
Standards and Guidelines: ICNIRP, 1998; IEEE Std C95.6-2002; IEEE Std C95.1-1991/1999; IEEE Std
1528-2003; Draft IEEE Std PC95.1- 2005.
Five Triages: Developed by ICES members for consistent literature review in SC-4 [listed in full in Appendix A].

General Experimental Design Criteria

Hypothesis
• The project should test a clearly defined hypothesis, using a detailed protocol that would lead to
quantitative information directly or indirectly relevant to assessment of health risk from ELF and
RF exposure and allow any other independent laboratory(s) to reconstruct the study and
replicate the findings for validation.

Good Laboratory Practice


Good Laboratory Practice (GLP) should be used throughout the design and conduct of any study
where possible and practical, but especially with large and long-term studies (see, e.g., FDA, 1993,
NTP 1992[3]).
• A specific protocol, consistent with the GLP guidelines, should be established and documented.
Any changes instituted during the course of the study should also be documented.
• The protocol should include randomized, double blind, symmetric handling of specimens and
their sources, except when precluded by the nature of the experiment or biological system.
• The protocol should include all appropriate controls (positive, negative, cage controls, sham-
exposed etc.).
• Investigators should be blind to whether they are working with exposed or control materials;
human subjects in laboratory experiments should be similarly unaware of their exposure status.
Quality assurance (QA) procedures should be included in the protocol, including traceable dosimetry
and monitoring of the programme by both a team from within the experimental staff and an
independent group, as required by GLP [Repacholi et al., 1998; Repacholi and Greenebaum, 1999;
Schönborn et al., 2000, 2004; ICNIRP, 2003; IEEE Std C95.6-2002; IEEE Std C95.1, 1991/1999; IEEE
Std 1528-2003; Draft IEEE Std PC95.1-2005].
Methodology
• Well-characterized biological systems or assays should be used, preferably ones that are well
established (validated) from the scientific literature that can be used reliably by other
laboratories (Vijayalaxmi and Obe, 2004; Obe and Vijayalaxmi, 2005).
• The biological system used should be appropriate to the end-point(s) studied.
• In cognitive behavioural research the assessment of the metrological quality (validity and
reliability and variability) of all the psychological tests (HCN, 2004); and measurements
[Electroencephalograms (EEGs)(Angelone et al., 2004), Positron Emission Tomography (PETs)
(Blodgett et al., 2005) Functional Magnetic Resonance Imaging (fMRIs) (Cabeza and Nyberg,
2000) etc] employed must be stated.
• Threshold responses derived from using at least 3 levels of dose and duration of exposure data
where possible are sought in addition to sham-exposed controls (‘Sham/Sham’). Appropriate
positive controls as well as non-RF heating controls can be very important to help assess the
metric of the response of the biological system and the potential effects of RF heating should be
included.
• The a priori estimated statistical power of the experiment, based on prior knowledge and the
number of tests planned, should be sufficient to reliably detect the expected size of the effect
(often as small as 10-20%) (Triages).

EMF Exposure Systems


• It is essential for high quality research that accurate assessment of RF and ELF exposure is an integral part of all studies and that each
research team include scientists and engineers skilled in traceable EMF dosimetry, in sufficient detail for replication (Triages).

• Computational dosimetry provides the quantitative link between internal dose quantities for direct effects and external fields that can
be measured. (Quoted from NRPB, 2004).

• A comprehensive uncertainty[4], variability and artifact dosimetric analysis is required to achieve consistent interpretation of the
results (quoted from Kuster et al., 2004; IEEE Std 1528-2003).

RF Exposure Metrics
• Current and future research should be applicable to mobile telephone systems in use. For RF
research the focus should be on signal protocols in use, e.g. second, third and fourth generation
signals (2G, 3G, 4G) including their modulation patterns. For radars, the frequency and pulsing
regimes should be applicable to current and emerging systems. [derived from Kuster and
Balzano, 1996; Schönborn et al., 2000; IEEE Std C95.6-2002; ICNIRP, 2003; Foster and Repacholi,
2004; Kuster et al., 2004; Nikoloski et al., 2005; Andersen, and Foster Presentations, COST 281
Zurich Feb 17-18, 2005; Reilly, 2005]
• Since adverse effects are typically set on the basis of a 1°C temperature rise, it is essential in the
design of local RF exposures (in vitro, in vivo and human acute experiments) that temperature
rise as well as SAR is measured and verified using numerical modelling of the exposures. There
are some recognized uncertainties indicated by the range of the modelling data relating
temperature rise with localized SAR (Quoted from Draft IEEE Std PC95.1- 2005, Section
C.2.2.2.1.2).
ELF Exposure Metric
• In relation to dosimetry of ELF local exposures, most internal cellular effects associated with low-
level ELF exposure are linked with the induced electric field stimulation and therefore the internal
induced electric field should be determined in ELF research where possible [IEEE Std C95.6-
2002; ICNIRP, 2003; Reilly, 2005].
• Human exposure to (external ELF) magnetic fields is measured in flux density (B) in milliTesla
(mT) and magnetic field strength (H) in amps per meter (A/m) respectively (IEEE Std C95.6-
2002).

Results Analysis
• Results should be analysed and interpreted on the basis of the prior stated hypothesis (Triages).
• Other tests of significance post hoc should be stated as such and an appropriate ‘multiple
comparisons’ correction (I.E. Bonferroni, Tukey, or Empirical Bayes Adjustments) be applied.
Post hoc results can be considered hypothesis generating for new research but cannot be
considered as the direct or main result of the presented research. (Keppel, 1982; Steenland et
al., 2000).
• The known reliability and validity of the tests employed and the normal variance of these tests
stated in the methods should be referred to in the analysis of the results. The variance in the
scores due to experimental exposure needs to be stated in the interpretation of the results
taking into account the normal variance of the test scores without treatment, and variance of
the test scores under sham control, positive control, and negative control treatment conditions.
[IEEE Std C95.6-2002; Draft IEEE Std PC95.1-2005; HCN, 2004; Foster and Repacholi, 2004;
Triages].

Conclusion
• The conclusion should be drawn on the basis of the prior stated hypothesis of the study
(Triages).
• Knowledgeable interpretation of the results must consider the established science (outside of
the EMF area) in the field of investigation such as biophysical mechanisms of interaction,
genotoxicity, cancer, cognition, sleep, etc. [Repacholi, 1998; ICNIRP, 1998; IEEE Std C95.6-2002;
ICNIRP, 2003; Foster, 2003; D’Andrea et al., 2003; Vijayalaxmi and Obe, 2004; Obe and
Vijalalaxmi, 2005; Draft IEEE Std PC95.1-2005]
• Results should be viewed with respect to previously accepted scientific principles before
ascribing the results to new theories. Research findings pointing to previously unidentified
relationships should be carefully evaluated and appropriate additional independent studies
should be conducted before the findings are accepted (Repacholi, 1998; Adair, 2003; Foster,
2003).
• Theories (e.g., for mechanisms of interaction) should make sufficiently concrete predictions that
they can be tested experimentally and be capable of being verified, if correct (Schwan, 1988;
Repacholi, 1998; Adair, 2003; Foster, 2000; Foster, 2003; Foster and Repacholi, 2004; Draft IEEE
Std PC95.1-2005).
Publication
• Well-designed and -conducted studies should be published regardless of the outcome, since
negative results are as useful as positive studies in the context of the EMF database for health
risk evaluation (Kheifets et al., 2003).

Scientific Responsibilities
Scientists are required to take responsibility for:

• the sensitive issue of data protection and confidentiality,


• the new opportunities, such as the increasing availability of computerized data,
• the incorporation of molecular epidemiological methods to aid the investigation of mechanistic
pathways and gene-environment interactions,
• and the development and utilization of sophisticated statistical approaches.
[Quoted from Rushton and Elliott, 2003; see also Conrads et al., 2004; Dorman, 2005; Espina et al.,
2005; Gilham et al., 2005; Greenland, 2005; LaPorte, 2005]

Replication
• Preferably, the experiments should be repeated and the data confirmed independently, or the
claimed effects should be consistent with results of similar experiments, for which the biological
systems involved are comparable (Repacholi, 1998).

Meta Studies
• Since the technologies of electricity and radio telecommunications serve billions of people it is
appropriate to address the remaining health concerns where possible in large multi-centred
international studies [Cardis and Kilkenny, 1999; Haarala et al., 2004; COST 281, Directive,
March 11, 2004].
• Where possible, feasibility studies (pilot studies) for multi-centred studies should be undertaken
with input from international experts to verify and validate every aspect of the methodology of
the EMF field study under question. The feasibility studies should be open for further scientific
comment and published in peer reviewed journals before these large studies are funded and
carried out (Cardis and Kilkenny, 1999; COST 281, Directive, March 11, 2004).
• Simultaneous epidemiological and experimental replications should use the same or similar
methodologies including standardized questionnaires and should report data to provide
quantitative results that can be compared and or combined for meta analysis of all the
independently collected data.
• In meta studies the data collection and analyses should be done according to the hypothesis and
methodology of the published protocol.
• Meta analyses should take into consideration prior distributions for the unidentified bias
parameters used in the original sensitivity-analysis model [Quoted from Greenland, 2003; 2005;
see further details in ‘The Epidemiological Study Design’ below].
General Research Priorities
Review of the new research proposals should give a high priority to studies with ELF and RF levels, frequencies, modulation and pulse
characteristics relevant to human exposures from new technologies with endpoints relevant to human health. [Repacholi, 1998; Foster
and Repacholi, 2004; Andersen, Foster, presentations, COST 281, Zurich, February 2005]

Dosimetric Units
RF
• SAR remains the major RF dosimetric quantity; modulation should not be added to a study
unless adequate statistical power can be maintained [ICNIRP, 1998; ICNIRP, 2003; Foster and
Repacholi, 2004; COST 281, Zurich, February 2005; DRAFT IEEE Std PC95.1-2005].

• Some research, not necessarily a full set of studies, would be warranted for new RF technologies
that employ new modulation schemes (changes in peak relative to average signal level and
changes in frequency content of a signal) if the potential for public exposure is high. This
includes UMTS now being rolled out [Foster and Repacholi, 2004; Andersen, Foster,
presentations, COST 281, Zurich, February 2005].

ELF
• In the low frequency range the standard is based on electrostimulation which is defined as
induction of a propagating action potential in excitable tissue by an applied electrical stimulus;
electrical polarization of presynaptic processes leading to a change in post synaptic cell activity
[Quoted from IEEE Std C95.6-2002].

• In most biological experiments of low-frequency field (ELF) effects the induced electric fields are
poorly known. It is necessary to improve macroscopic dosimetry and, particularly in the case of
in vitro studies, also to examine the microscopic distribution of the induced electric field [Quoted
from ICNIRP, 2003; see also IEEE Std C95.6-2002; Reilly, 2005].
Threshold Studies

• There is a lack of well-replicated studies that reveal the existence of biological effects or adverse
health effects from low-level RF or ELF exposure below, at, and above public and occupational
guideline limits [ICNIRP, 2003; COST 281, Zurich, February 2005].

• Similarly there is a lack of well-replicated studies that reveal the existence of biological effects or
adverse health effects from low-level RF or ELF exposure at various durations of exposures
similar to real life durations of exposure conditions including intermittent exposures [Ivancsits et
al., 2005; Diem et al., 2005].

• We need to establish the human threshold SAR values where RF exposure has a biological and
an adverse effect [COST 281, Zurich, February 2005; IEEE, ICES, COST 281: 2004/09 Thermal
Physiology Workshop, Paris].

• We need to establish more precisely the ELF thresholds for human electrostimulation values
where ELF exposure has a biological and an adverse effect in various tissues [IEEE Std, C95.6-
2002; ICNIRP, 2003; Reilly, 2005].
SAR as a Measure of Temperature Increase
• There is general agreement that more human RF dosimetry research is needed to determine the
validity of the SAR modeling data as a measure of temperature rise (Quoted from Draft IEEE Std
PC95.1-2005 Section C.2.2.2.1.2). [See also: ICNIRP, 1998; Van Leeuwen et al., 1999; Wang and
Fujiwara, 1999; Bernardi et al., 2000; Schönborn et al., 2000; Wainwright, 2000; Bernardi et al.,
2001; Gandhi et al., 2001; Hirata et al., 2002; Yioultsis et al., 2002; Hirata and Shiozawa, 2003;
NRPB, 1997, 2004; Kuster et al., 2004; Nikoloski et al., 2005].
• A future goal is to develop appropriate techniques and thermal models for accurately predicting
the thermo physiological responses of human beings who are exposed to RF fields at specific
frequencies, field strengths, and field characteristics and to validate some predictions with
existing human exposure data [Stolwijk and Hardy, 1977; Adair and Berglund, 1986, 1992; Adair
et al., 1998, 1999, 2001, 2003, 2005; Kheifets et al., 2003; Foster and Adair, 2004; IEEE, ICES,
COST 281: 2004/09 Thermal Physiology Workshop, Paris]. Much data exists that describes the
regulatory response changes in the human body as a function of environmental variables, work,
exercise, age, fitness, clothing insulation, and other characteristics of each individual. [WHO
Geneva Workshop, 2002: Adverse Temperature Levels in the Human, International Journal of
Hyperthermia, Vol 19(3), 2003]. Much of this material is amenable to comparison with data
derived from RF-exposed humans and animals. [IEEE, ICES, COST 281: 2004/09 Thermal
Physiology Workshop, Paris].

Modulation
• We need to note the modulation of RF and ELF signals exactly. Different modulations could
have the same average value. Presently there is no convincing evidence that there is a
difference between continuous and modulated RF signals in their effects. But we need to get
more clear evidence of the different modulations. Are there separate classifications according to
biological effects? Is lower or higher SAR more significant at producing effects? [COST 281,
Zurich, February 2005]

• Experimental EMFs should be fully characterized and re-measured periodically.


Waveform, pulse shape and timing, frequency spectrum, harmonics and transients
from both continuous sources and from switching exposure systems on and off
should all be measured where appropriate (Kuster, 1996; Schönborn et al., 2000;
Nikoloski et al., 2005).
• Background fields, such as ambient, equipment-derived, and crossover fields from other
exposure systems, are also important and need to be characterized. Time-varying and static
components should be measured, as well as the polarization and directions of the fields.
• Field modulation introduced by experimental factors such as motion of sample shakers should
be noted and measured whenever possible. Positioning of cultures or animals within exposure
systems should be noted and randomized where appropriate (Repacholi, 1998).
Human RF Dosimetry
• The position of humans in exposure systems should be noted and stereotaxically (3-D) defined
for replication. (Excell et al., 1996, 1998; Vaul & Excell, 1999).

• Reports of SAR increases due to the presence of nonmagnetic electrodes during RF exposure of
up to a factor of 16 have been reported (Angelone et al., 2004). During RF exposure while
recording EEG, better modelling of the human head SAR is required (Huber et al., 2003). For
instance you cannot use a plastic shell head phantom the same as for mobile phone compliance
when measuring the effect of metal leads (for EEG recording) on SAR because of the insulating
plastic phantom shell since there would be no electrical connection between the leads and the
fluid inside the phantom. (CK Chou personal communication; for SAR modelling methodology
see Angelone et al., 2004).
• Further modelling of SAR of children is required using realistic head and body phantoms for both
near and far field exposures [ICNIRP, 2003; NRPB, 2004; Draft IEEE Std PC95.1-2005].

Meta Studies
• Meta studies are required on RF effects of UMTS and 4G signals on direct and established
measures of human brain function and the possible mechanisms involved, using well validated
measurements [Haarala et al., 2004; HCN, 2004; Angelone et al., 2004; Kuster et al., 2004].
• In vitro meta studies investigating DNA breaks, genomics, proteomics and molecular signalling
pathways during ELF and RF exposures are a priority because of remaining positive studies in the
cytogenetic literature (Vijayalaxmi and Obe, 2004; Moulder et al., 2005; Obe and Vijayalaxmi,
2005. reviews).

• Possible long-term biological effects (up to 30 years; IARC, 2002; ICNIRP, 2003; Ahlbom et al.,
2004; Feychting et al., 2005) from mobile phone exposure of the public require investigation and
should be given priority. Cohort studies including children are recommended. (Stewart et al.,
2000; ICNIRP, 2003; Ahlbom et al., 2004; Feychting et al., 2005; Moulder et al., 2005).

• The incorporation of molecular epidemiological methods to aid the investigation of mechanistic


pathways and gene-environment interactions should be included. [Rushton and Elliott, 2003;
see also Conrads et al., 2004; Dorman, 2005; Espina et al., 2005].

Specific Research Designs

Introduction
Traditional Evaluation of Research
Evaluation of research literature on the effects of ELF or RF exposure is reached in a complex
confirmatory interplay of human epidemiological studies, human acute studies, in vivo bioassays
(animal lifetime studies), animal acute studies and in vitro tissue culture studies. Traditionally,
investigations in human beings of associations between exposure levels and adverse health effects
can utilize either human acute or epidemiological studies. And it was biologically plausible and
prudent to regard EMF exposure studies in animals, as evidence of lack of risk or risk in humans.
This was and is dependent on whether there is evidence for extrapolation of the science from
animals to man, based on known functional and structural homologies [i.e. Spatial memory: Kandel
et al., 2000; Kandel, 2001; Cell division: Lee and Nurse, 1987]. Traditionally, cellular research was
considered as the possible source of a plausible mechanism for any biological effect due to exposure
to RF or ELF signals [Repacholi and Cardis, 1997; Repacholi, 1998].

New Evaluation of Research: Molecular Epidemiology


Great advances in cellular research have given us a new understanding of the interplay of cellular
research of genes, proteins and cell signalling in single living cells, and the functioning of the whole
human body.
For instance molecular profiling for the treatment of individual patient's tumours is currently being evaluated in clinical trials at the
National Institutes of Health, National Cancer Institute (Espina et al., 2005).

And high-resolution serum proteomic features are being implemented for various kinds of cancer
detection (Conrads et al., 2004).
These molecular advances have an impact on EMF research priorities. Molecular profiling may
become essential at all levels of research namely in epidemiology, human acute studies, animals
studies and tissue cultures.
Molecular epidemiology is based on general epidemiology and utilizes the same designs (i.e., case
control and cohort studies) as those employed by general epidemiology. However, molecular
epidemiology utilizes molecular biology to define the distribution of disease in a population (i.e.,
descriptive epidemiology) and identify its potential etiologic determinants (i.e., analytical
epidemiology) [Quoted from Dorman JS, Director, Collaborating Center for WHO Multinational
Project for Childhood Diabetes (DiaMond), U of Pittsburgh; see also LaPorte RE, Director, Molecular
Epidemiology and DNA Technology Transfer http://www.pitt.edu/~rlaporte/who.html]
The following sections are on specific research designs in epidemiology, human acute studies, in vivo
and in vitro research designs respectively.

The Epidemiological Study Design

Introduction
Epidemiologic investigations of possible associations of EMF exposure with risk of chronic disease
pose unique and substantial difficulties. Among them are difficulties specific to an outcome studied,
assessment of exposure, and interpretation of findings and long latency periods required to detect
any cancer risk (up to 30 years), (IARC, Vol 80, 2002).
• In light of the complexity of the topic (covering medicine, biology, epidemiology,
neuroscience, psychology, physics, engineering, statistics) studies should be designed
and implemented using expertise from all the relevant disciplines.
• Information necessary to evaluate the feasibility of a given study should be collected
and analysed by experts in all the related fields and published in a peer reviewed journal
(Cardis and Kilkenny, 1999).
• The design must be maximally efficient in reaching the study objective and utilizing
resources.
• The hypothesis must be explicitly and clearly stated before the onset of the research.
• The study protocol should meet the letter and spirit of all relevant regulations and have
prior approval of all relevant review boards.
• Subjects and controls must have given informed consent.
• The method of ascertaining cases of adverse health must be stated.
• Case identification must be independent of exposure. Definition of cases should be objectively,
and histologically confirmed.

• Controls should be appropriate to the specific aim and design.


• Controls must be matched (to cases) individually, on age and sex, within study region, and on
the basis of frequency, and be population based –i.e. ‘representative’
• Minimization of non-response or non-participation is required to achieve the necessary sample
size and minimize bias through selective non-response.
• Total number of original subjects and controls must be stated as well as those removed from the
study due to non-response or death.

• An adequate population sample size must be used based on previous calculation of the
statistical power. Expected number of cases must be adequate in the study populations to show
a relatively small effect of exposure to EMF emissions, if there is one, for instance, from
electrical power devices, or mobile phones.
• Study populations must be well defined before the onset of the research.
• Study designs should recognize that the exposure metric for possible effects of weak ELF and
weak RF fields is uncertain and usually proxy. Subjects' exposures, particularly historical
exposures that are often determined via surrogates, should be validated from specific
measurements where possible. Data should include as much information relevant to alternate
metrics as possible to aid future research. [Beaglehole et al., 1993; Bracken et al., 1993; Ahlbom,
(ICNIRP), 1996; ICNIRP, 1998; IEEE Std C95.6-2002; ICNIRP, 2003; Ahlbom et al., (ICNIRP) 2004;
Feychting et al., 2005; IARC, Vol 80, 2002; Draft IEEE Std PC95.1-2005; Neubauer et al., 2005;
Reilly, 2005].
• The results must be calculated to evaluate the original hypothesis. It is on the basis of the stated
original hypothesis that the power of the study is calculated and the size of the sample is set.
• The authors should report the basic data on which the conclusions are drawn (IARC, Vol 80,
2002).
• Post hoc comparisons on subsets of the data may be hypothesis generating only. Appropriate
corrections for multiple comparisons must be applied (Keppel, 1982; Steenland et al., 2000;
Johnston and Scherb, 2005).
• Meta analyses should take into consideration prior distributions for the unidentified bias
parameters used in the original sensitivity-analysis model. Accounting for uncontrolled
confounding and response bias under a reasonable prior (distribution) can substantially alter
inferences about the existence of an electromagnetic field effect. Analyses with informative
priors (distributions) for unidentified bias parameters can help avoid misinterpretation of
conventional results and ordinary sensitivity analyses [Quoted from Greenland, 2003; 2005; see
also Blettner & Schlattmann, 2005].

• Studies should include collection of blood samples to create a bio-bank for molecular biological
studies on brain tumours and other diseases.
• In new studies, improvement of the exposure assessment is crucial. (Feychting et al., 2005).
Personal monitoring systems are being developed for base site exposures.
Case Control and Cohort Studies
• Case control and cohort studies can give a relatively close approximation to the biologic model
in investigating environmental health issues because both individual person characteristics and
exposures are studied in the individual environment. They can investigate for instance whether
risk of cancer is raised in human populations according to their EMF exposure. In other words
they investigate whether or not the possibility of an association between EMF exposure and
cancer exists in human populations (IARC Vol 80, 2002; Elliott and Wartenberg, 2004).
Case Control Studies
• Case control studies provide point data for cases and a set of controls. The population is
undefined in general, and cases and controls should be comparable. Cases and controls are
selected according to certain criteria by the investigator. The risk measure is odds ratio (OR)
(Ahrens and Pigeot, 2005).
• Case control studies are prone to selection and other biases, are moderately expensive and
time-consuming to carry out, and are not feasible in all situations (Elliott and Wartenberg,
2004). [For example see Cardis and Kilkenny, 1999, The Interphone Study; and MTHR Swerdlow
et al., 2002-2008, a case control study of risk of leukaemia in relation to use of mobile phones.
http://www.mthr.org.uk/research_projects/funded_projects.htm]
• Epidemiological case control studies should gather toxicological information on other factors as
well as EMF in cases of diseases of unknown causes to look for feasible associations that may be
strongly associated with the disease (UKCCS Investigators, 2000; IARC, Vol 80, 2002). [I.E. In the
UK childhood cancer study (UKCCS) childhood leukaemia has been shown to be associated with
reduced exposure to infection in the first few months of life, (Gilham et al., 2005)].
Cohort Studies
• A cohort approach would allow studies with different types of outcomes.
• Cohort studies define an exposed and a non-exposed population. Incident diseases or causes of
death in the exposed and non-exposed populations are taken from a national register. The risk
measure is Relative Risk (RR) (Ahrens and Pigeot, 2005):

• Cohort studies, although not subject to selection bias, are prone to other biases, including losses
to follow-up, and generally more expensive and time consuming to carry out than case control
studies (Elliott and Wartenberg, 2004). [For examples see Johansen et al., (2001, 2002) and Elliot
et al., Cohort study of mobile phone users (pilot study), 2002-2004, in press
http://www.mthr.org.uk/research_projects/funded_projects.htm]

Spatial Epidemiology (Quoted from Elliott and Wartenberg, 2004)


• Exploratory studies such as spatial epidemiology use aggregate data, such as geographic
correlation studies and offer an alternative approach for generating, prioritizing, and analyzing
data to address specific hypotheses of disease etiology and causation.

• Spatial epidemiology is the description and analysis of geographic variations in disease with
respect to demographic, environmental, behavioural, socioeconomic, genetic, and infectious
risk factors.
• Although they too are prone to biases and misclassification (Elliott and Wakefield, 2000), they
are generally easier, quicker, and less expensive to conduct than case control or cohort studies.
• Sensitivity to detect areas at high risk is limited when expected numbers of cases are small.
• One example of this approach is with use of a dedicated system such as that developed by the
Small Area Health Statistics Unit (SAHSU) in the United Kingdom (Elliott et al., 1992); this has
recently been adopted in other European countries as part of the European Health and
Environment Information System (EUROHEIS).
• One ready means of investigating the relation of RF exposure to disease is the replication of
analyses in different areas based on routine data, as is done in the United Kingdom through
SAHSU and in Europe through EUROHEIS [For an example see Elliott et al., 2003-2005 Case
control study of cancer incidence in early childhood and proximity to mobile phone base
stations, MTHR: http://www.mthr.org.uk/research_projects/funded_projects.htm]
• Problems include the large random component that may dominate disease rates across small
areas. This can be dealt with by using Bayesian statistics to provide smooth estimates of the
disease risks (Elliott and Wartenberg, 2004).

Bias
• The healthy group (worker) effect can result in comparison bias.
• Recall bias occurs when cases and controls may recall exposure differently.
• Interviewer bias can arise when the interviewers may ask questions differentially.
• Selection bias could arise when sampling probability varies, and when there is loss to follow up,
and there are non-responses.
• Information bias can arise from non-differential misclassification and differential
misclassification.
• Confounding could be due to mixing of effects such as age, sex, and competing exposures.

• Ecological fallacy could result from uncontrolled factors related to disease and exposure [Ahrens
and Pigeot, 2005; H. Scherb presentation Feb 2005 COST281, Schriesheim].

Countering Bias
• To control for biases there should be double-blinding (subjects and investigators) where
possible.
• The study should represent the population or use total ascertainment (i.e. using national
registers).

• The study should use stratification of participants (for instance by age, sex, education and social
economic status).

• One can counter bias by use of standardization (direct or indirect) (Ahrens and Pigeot, 2005;
Scherb, 2004).

Dosimetry
• Data on different levels of exposure, its duration and temporal location should be identified
(Balzano, 1999).
• The exposure must be recorded in traceable detail for replication in multiple centres of
simultaneous study. Quantifiable exposure measurement is preferred over qualitative exposure
data.
• Exposure gradients should be developed.
• This dosimetry should be taken into account both at the design of the study and during analysis.
Dose must be independent of control and experimental subjects, and should be measured on an
individual basis (IARC, Vol 80, 2002).

• Adequate and reliable measures of exposure for each study subject are needed.
• Categorising exposure into groups can lead to misclassification, and produces a bias towards the
null hypothesis -underestimating the real effects.

• More research on validation of body-worn human RF dosimeters needs to be done before


studies around mobile phone base stations are feasible (Neubauer et al., 2005).

Data Analysis
• The authors should report the number of exposed and unexposed cases and controls in a case
control study and the numbers of cases observed and expected in a cohort study and from this
group the number used in the statistical evaluation of the primary hypothesis.
• Both in study design and analysis, control for confounding variables is required. Data on
potential confounders should be collected for statistical analysis to minimize or subtract out the
confounding factors where possible. Identification of confounding factors is recognized as
difficult; there is often limited knowledge about causal factors of adverse health endpoints.
• Lack of appropriate action to reduce the impact of these sources of error can decrease the
credibility and the final weight given to the results of the study. (Repacholi and Cardis, 1997).

• The methods of statistical analysis should be appropriate for the evaluation of the hypothesis of
the study and clearly described.
• When sophisticated or non-standard analytical procedures are used, researchers should provide
a simple descriptive analysis of the data. The number of subjects and controls, and the effects
of potential confounding factors that were part of the investigation should all be reported.
• Meta analyses should take into consideration prior distributions for the unidentified bias
parameters used in the original model [Greenland, 2003., 2005].

• To correct for false positives, empirical or semi-Bayes methods of adjustments for multiple
comparisons (post hoc) are recommended when a large number of comparisons have been
made (Steenland et al., 2000).
• In summary, we look for 9 factors or viewpoints in epidemiology, as summed up by Sir Bradford
Hill in 1965, namely: Strength, Consistency, Specificity, Temporality, Biological gradient,
Plausibility, Coherence, Experimental, evidence, and Analogy. He stressed that from these nine
different viewpoints we should study association before we try causation.
• He states that no tests of significance can answer those (9) questions. Such tests can and should
remind us of the effects that play of chance can create, and they will instruct us in the likely
magnitude of these effects. Beyond that they contribute nothing to the 'proof' of our hypothesis
[Quoted from Bradford Hill, 1965].

Human Volunteer Study Design

Introduction
Please note there are several references to details in the ‘General Experimental Design Criteria’ and
the ‘General Research Priorities’ above, in the details of the ‘Human Volunteer Study Design’ below.
The advantage of human volunteer experiments is that they indicate the likely response of other people
exposed under similar conditions.
Disadvantages of volunteer studies include:

• the innocuous nature of the effects that can be investigated,


• the short duration of investigation,

• the small number of subjects usually examined


• the ethical constraints,

• and the subjects are usually healthy adults who may not reflect the responses of potentially
more susceptible members of society (Quoted from NRPB, 2004).

Hypothesis
• The human study should test a clearly defined hypothesis, using a detailed protocol that would
lead to quantitative information directly or indirectly relevant to assessment of health risk from
ELF and RF exposure and allow any other independent laboratory(s) to reconstruct the study
and replicate the findings for validation and allow the results be combined for meta analysis
where appropriate.

GLP
• The protocol should meet the letter and spirit of all relevant regulations concerning experiments
using human subjects, and have prior approval of all relevant review boards. Personnel working with
volunteers require special training and oversight.
• For instance, where radioactively labelled compounds could be injected into the human subject’s
bloodstream [such as in Positron Emission Tomography (PET) or Single Photon Emission Computed
Tomography (SPECT)] exploratory research in animals should first be considered to investigate brain
areas of interest with alternate techniques, such as quantitative autoradiography (QAR) to map
regional cerebral blood flow (RCBF). A RCBF study with rats exposed to RF would indicate if there is
something worth looking for in humans. (Society of Nuclear Medicine Brain Imaging Council, 1996;
Blodgett et al., 2005)
• Adherence to ethical rules must be indicated.
• A well-described criterion for inclusion and exclusion of volunteers is required.

EMF Exposure Systems


• It is essential for high quality research that accurate assessment of RF and ELF exposure is an integral part of all studies and that each
research team include scientists and engineers skilled in traceable EMF dosimetry, in sufficient detail for replication (Triages).

• Current ELF and RF future research should be applicable to electrical and mobile telephone systems in use. For RF research the focus
should be on signal protocols in use, e.g. second, third and fourth generation signals (2G, 3G, 4G) including their modulation
patterns. For radars, the frequency and pulsing regimes should be applicable to current and emerging systems. [derived from Kuster
and Balzano, 1996; Schönborn et al., 2000; IEEE Std C95.6-2002; ICNIRP, 2003; Foster and Repacholi, 2004; Kuster et al., 2004;
Nikoloski et al., 2005; Andersen, and Foster Presentations, COST 281 Zurich Feb 17-18, 2005; Reilly, 2005]

RF Exposure Metrics
• Specific Absorption Rate (SAR) in Watts per kilogram (W/kg) is the fundamental RF dosimetry
parameter.
• Since RF adverse effects are typically set on the basis of a 1°C temperature rise, it is essential in
the design of local RF exposures in human acute experiments that temperature rise as well as
SAR is measured and verified using numerical/thermal modelling of the exposures. There are
some recognized uncertainties indicated by the range of the modelling data relating
temperature rise with localized SAR (Quoted from Draft Std PC95.1-2005, Section C.2.2.2.1.2).
• In cognitive behavioural research the assessment of the metrological quality and uncertainties
of measurement (variability reliability and validity) of all the psychological tests (HCN, 2004) and
dosimetry measurements employed must be stated [Re: Electroencephalograms (EEGs) see
Angelone et al., 2004 and see General Research Priorities above for more detail); Re: Positron
Emission Tomography (PETs) see Blodgett et al., 2005; Re: Functional Magnetic Resonance
Imaging (fMRIs) see Cabeza and Nyberg, 2000)].
• Threshold responses derived from using at least 3 levels of dose and duration of exposure data
where possible are sought in addition to sham-exposed controls (‘Sham/Sham’). Appropriate
positive controls as well as non-RF heating controls can be very important to help assess the
metric of the human response and the potential effects of RF heating should be included.

Human RF Exposure in the Near Field

• It is necessary for replication and accurate interpretation of results that the experimental
exposure setup be defined in stereotaxic coordinates for the position of the antenna in relation
to the human head. It is necessary to measure, under the experimental conditions, the SAR
exposure pattern from the phone antenna in the stereotaxically defined position in relation to
the neuroanatomy of the (phantom) head and brain, taking into account the dielectric
properties of various tissues of the typical head and neck. It is important to numerically
calculate the same SAR values by the gram (1cc) through out the exposed head, brain and neck
and compare these values with the experimental measurement values and thermal modelling
for verification [IEEE, ICES, COST 281: Joint workshop 2004/09 Thermal Physiology Workshop,
Paris].
• It is also important to measure the emissions of the exposure system and ambient room
emissions in the experimental conditions. (Excell, 1996, 1998)
• Information about the internal magnetic field should be provided.

Human RF Exposure in the Far Field

• Improved assessment techniques are needed to analyse the range of RF field exposures and
absorption experienced by individuals. (Repacholi, Conference Nov ’96; Repacholi, 1998;
Neubauer et al., 2005).
• The IEEE C95.1-1999 Standard for RF safety calls for spatially averaged measurements of
incident power density to verify compliance with maximum permissible exposure limits. Human
exposure to RF power radiated by mobile base station antennas can be assessed by means of
the incident power density averaged over the body. The convenience of adopting this quantity
lies in the well-behaved decay away from the antenna. (Balzano and Faraone, 1999; Faraone et
al., 2000)
• Both numerical modelling and experimental measurement are important to verify whole body
human exposure [Balzano and Faraone, 1999; Faraone et al., 2000; Neubauer et al., 2005].

ELF Exposure Metrics

• In the low frequency range external magnetic fields are measured as flux density (B) in milliTesla
(mT) and magnetic field strength (H) in amps per meter (A/m) respectively (IEEE Std C95.6-
2002).

• In the low frequency range internal electric fields are measured as in situ (in tissue) electrical
forces (Volts per metre; V/m) and behaviourally as electrostimulation which is defined as
induction of a propagating action potential in excitable tissue by an applied electrical stimulus;
electrical polarization of presynaptic processes leading to a change in post synaptic cell activity
[Quoted from IEEE Std C95.6-2002].

Results Analysis
• Results should be analysed and interpreted on the basis of the prior stated hypothesis (Triages).
• No data should be discarded without valid reason (e.g. equipment failure, procedures not
followed, non-participation). Reasons for this should be recorded.
• Other tests of significance post hoc should be stated as such and an appropriate ‘multiple
comparisons’ correction (I.E. Bonferroni, Tukey, or Empirical Bayes Adjustments) be applied.
Post hoc results can be considered hypothesis generating for new research but cannot be
considered as the direct or main result of the presented research. (Keppel, 1982; Steenland et
al., 2000).
• The known reliability and validity of the tests employed and the normal variance of these tests
stated in the methods should be referred to in the analysis of the results. The variance in the
scores due to experimental exposure need to be stated in the interpretation of the results taking
into account the normal variance of the test scores without treatment, and variance of the test
scores under sham control, positive control, and negative control treatment conditions. [IEEE
Std C95.6-2002; Draft IEEE Std PC95.1-2005; HCN, 2004; Foster and Repacholi, 2004; Triages].

Conclusion
• The conclusion should be drawn on the basis of the prior stated hypothesis of the study
(Triages).
• Knowledgeable interpretation of the results must consider the established science (outside of
the EMF area) in the field of investigation such as biophysical mechanisms of interaction,
cognition, sleep, etc. [Repacholi, 1998; ICNIRP, 1998; Kandel et al., 2000; Kandel, 2001; IEEE Std
C95.6-2002; ICNIRP, 2003; Foster, 2003; D’Andrea et al., 2003; NRPB, 2004; Draft IEEE Std
PC95.1-2005]
• Results should be viewed with respect to previously accepted scientific principles before
ascribing the results to new theories. Research findings pointing to previously unidentified
relationships should be carefully evaluated and appropriate additional independent studies
should be conducted before the findings are accepted (Repacholi, 1998).
• Theories (e.g., for mechanisms of interaction) should make sufficiently concrete predictions that
they can be tested experimentally and be capable of being verified, if correct (Schwan, 1988;
Repacholi, 1998; Foster, 2000; Adair, 2003; Foster, 2003; Foster and Repacholi, 2004; Draft IEEE
Std PC95.1-2005).

Publication
• Well-designed and -conducted studies should be published regardless of the outcome, since
negative results are as useful as positive studies in the context of the EMF database for the
evaluation of human health risks and the setting of standards (Kheifets et al., 2003).

Scientific Responsibilities
Scientists are required to take responsibility for:

• the sensitive issue of data protection and confidentiality,


• the new opportunities, such as the increasing availability of computerized data,
• the incorporation of molecular methods to aid the investigation of mechanistic pathways and
gene-environment interactions,
• and the development and utilization of sophisticated statistical approaches.
[Quoted from Rushton and Elliott, 2003; see also Conrads et al., 2004; Dorman, 2005; Espina et al.,
2005; Gilham et al., 2005; LaPorte, 2005]

Replication
• Preferably, the experiments should be repeated and the data confirmed independently, or the
claimed effects should be consistent with results of similar experiments, for which the biological
systems involved are comparable (Repacholi, 1998).

Meta Studies
• Since the technologies of electricity and radio telecommunications serve billions of people it is
appropriate to address the remaining health concerns where possible in large multi-centred
international studies (Haarala et al., 2004).
• Where possible, feasibility studies (pilot studies) for multi-centred studies should be undertaken
with input from international experts to verify and validate every aspect of the methodology of
the EMF field study under question. The feasibility studies should be open for further scientific
comment and published in peer reviewed journals before these large studies are funded and
carried out.
• Simultaneous human experimental replications should use the same or similar methodologies
including standardized questionnaires and should report data to provide quantitative results
that can be compared and or combined for meta analysis of all the independently collected data
(Haarala et al., 2004).
• In meta studies the data collection and analyses should be done according to the hypothesis and
methodology of the published protocol.
In Vivo Study Design

Please refer to the‘General Experimental Design Criteria’ and the ‘General Research Priorities’ above
and the in vivo triage below for further details.
Introduction
• Animal studies provide information concerning the interaction of EMFs with living systems that
display the full repertoire of body functions, such as immune responses, cardiovascular changes
and behaviour.
• Individual animals in inbred strains are genetically identical, thus ensuring a relative consistency of
response to the agent in question reducing the variance and improving the sensitivity of the
study.
• Transgenic or gene knockout animal models of certain diseases have further increased the value
of animal studies to reveal potential adverse health effects. (NRPB, 2004)
• Animal studies can be carried out over the brief lifetime of the animal (~ 2 years) to answer health
questions about lifetime exposure effects.
• Animal studies can use exposures above guideline limits and thus report dose, duration and threshold
data.
• The number of subjects can be increased to improve the statistical power of the experiments.
• Extrapolation of this information to humans may be considered, if there are proven homologies
in structures and processes in animals and humans. For example, animal studies have been very
useful in helping unravel the sequence of genetic events in the development of a number of
human cancers (Balmain and Harris, 2000) (NRPB, 2004) and in the molecular mechanisms of learning
(Kandel et al., 2000, 2001).
• IARC (2002) noted that, with regard to cancer ‘in the absence of adequate data on humans, it is
biologically plausible and prudent to regard agents and mixtures for which there is sufficient
evidence of carcinogenicity in experimental animals as if they presented a carcinogenic risk to
humans’.
Hypothesis
• The project should test a clearly defined hypothesis, using a detailed protocol that would lead to
quantitative information directly or indirectly relevant to assessment of health risk from ELF and
RF exposure and allow any other independent laboratory(s) to reconstruct the study and
replicate the findings for validation.

Good Laboratory Practice


Good Laboratory Practice (GLP) should be used throughout the design and conduct of any study
where possible and practical, but especially with large and long-term animal studies (see, e.g., FDA,
1993, NTP 1992).
• A specific protocol, consistent with the GLP guidelines, should be established and documented.
Any changes instituted during the course of the study should also be documented.
• The protocol should include randomized, double blind, symmetric handling of animals and
specimens, except when precluded by the nature of the experiment or biological system.
• The protocol should include all appropriate controls (positive, negative, cage controls, sham-
exposed etc.).
• Investigators should be blind to whether they are working with exposed or control animals or
biological systems.
Quality assurance (QA) procedures should be included in the protocol, including traceable dosimetry
and monitoring of the programme by both a team from within the experimental staff and an
independent group, as required by GLP [Repacholi, 1998; Repacholi and Greenebaum, 1999;
Schönborn et al., 2004; ICNIRP, 2003; IEEE Std C95.6-2002; IEEE Std C95.1, 1991/1999; Draft IEEE Std
PC95.1-2005].

Methodology
• Well-characterized biological systems (animal species) should be used, preferably ones that are
well established (validated) from the scientific literature that can be used reliably by other
laboratories.
• The biological system used should be appropriate to the end-point(s) studied.
• Threshold responses derived from using at least 3 levels of dose and duration of exposure data
where possible are sought in addition to sham-exposed controls (‘Sham/Sham’). Appropriate
positive controls as well as non-RF heating controls can be very important to help assess the
metric of the response of the biological system and the potential effects of RF heating should be
included.
The a priori estimated statistical power of the experiment, based on prior knowledge and the
number of tests planned, should be sufficient to reliably detect the expected size of the effect (often
as small as 10-20%) (In vivo Triage).

Traceable Dosimetry

1. Current and RF future research should be applicable to electrical and mobile telephone
systems in use. For RF research the focus should be on signal protocols in use, e.g.
second, third and fourth generation signals (2G, 3G, 4G) including their modulation
patterns. For radars, the frequency and pulsing regimes should be applicable to current
and emerging systems. [Derived from Kuster and Balzano, 1996; Schönborn et al., 2000;
IEEE Std C95.6-2002; ICNIRP, 2003; Foster and Repacholi, 2004; Kuster et al., 2004;
Nikoloski et al., 2005; Andersen, and Foster Presentations, COST 281 Zurich Feb 17-18,
2005; Reilly, 2005]

2. Environmental conditions, such as temperature, humidity, light, vibration, sound, and


background EMFs, should be measured and recorded periodically. All experimental
conditions should be the same for all groups, except for EMF exposure [Schönborn et al.,
2004; Engineering and In vivo Triages].
2. Experimental EMF should be fully characterized and measured periodically. Waveform, pulse
shape and timing, frequency spectrum, harmonics and transients from both continuous sources
and from switching exposure systems on and off, should all be measured where appropriate.
Background fields, such as ambient, equipment-derived, and crossover fields from other
exposure systems, are also important and need to be characterized. Time-varying and static
components should be measured, as well as the polarization and directions of the fields. Field
modulation introduced by experimental factors such as motion of sample shakers should be
noted and measured whenever possible. Positioning of animals within exposure systems should
be noted and randomised where appropriate.
3. Experimental dosimetry should confirm any calculated values; measurements should be
taken at multiple points in any in vivo model; should be reported in SAR (W/kg) - not
simply as a field strength measure; should specify whether whole body or local SAR for
animal studies; and should detail the method of measurement. (Repacholi, Conference,
Nov ’96; Repacholi, 1998; International EMF Project/ ICNIRP, 1999)
4. In considering experimental studies more generally, a considerable problem in the
interpretation of experiments is that many of them have given insufficient detail
concerning exposure conditions. Moreover, in the case of pulsed fields when SAR values
are quoted it is often unclear whether these refer to the average SAR or to the peak SAR
during pulses. It is very important to make this distinction, since the peak SAR can be
1000 or more times the average value. Full details should be provided of experimental
conditions including maximum SAR per pulse for pulsed radiation.
5. Failure to adequately characterize and control for experimental conditions (i.e.
immobilization) with appropriate cage control animals, could significantly mask any
potential effects mediated by the RF field on stress-related parameters (Stagg et al.,
2001).
6. A comprehensive uncertainty, variability and artifact dosimetric analysis is required
to achieve consistent interpretation of the results (Quoted from Kuster et al., 2004; IEEE
Std 1528-2003)
Data collection and quality assurance

1. The full protocol, including QA, (including double blind conditions) should be followed
strictly, as should GLP provisions for monitoring this.

2. Data should be recorded contemporaneously and back-up copies kept of all electronic
data.

3. No data should be discarded without valid reason (e.g. equipment failure, procedures not
followed). Reasons for this should be recorded.

4. As part of the QA programme, at least one independent reassessment of all or an


appropriate sample of specimens should be made, when assays require an independent
judgment by the investigator (e.g., histological evaluations).
5. Where possible, samples should be stored for future reference.

Results Analysis
• The stored data set should contain all data, and if any data are excluded from an analysis, clear,
legitimate reasons for doing so should be recorded.
• Analysis techniques should be appropriate to the data and the hypothesis and results should be
analysed and interpreted on the basis of the prior stated hypothesis (In Vivo Triage).
• Other tests of significance post hoc should be stated as such and an appropriate ‘multiple
comparisons’ correction (I.E. Bonferroni, Tukey, or Empirical Bayes Adjustments) be applied.
Post hoc results can be considered hypothesis generating for new research but cannot be
considered as the direct or main result of the presented research. (Keppel, 1982; Steenland et
al., 2000).
• The known reliability and validity of the tests employed and the normal variance of these tests
stated in the methods should be referred to in the analysis of the results. The variance in the
scores due to experimental exposure needs to be stated in the interpretation of the results
taking into account the normal variance of the test scores without treatment, and variance of
the test scores under sham control, positive control, and negative control treatment conditions.
[IEEE Std C95.6-2002; Draft IEEE Std PC95.1-2005; HCN, 2004; Foster and Repacholi, 2004; In
Vivo Triage].
Conclusions
• Conclusions should be drawn on the basis of the hypothesis, be fully supported by the data and
include all-important implications of the data set.
• Reports should include enough data and information concerning materials and methods to allow
independent assessment of the conclusions and discussion.
• Knowledgeable interpretation of the results must consider the established science (outside of
the EMF area) in the field of investigation such as biophysical mechanisms of interaction,
genotoxicity, cancer, memory [Lee and Nurse, 1987; Repacholi, 1998; ICNIRP, 1998; Kandel et
al., 2000; Kandel, 2001; IEEE Std C95.6-2002; ICNIRP, 2003; D’Andrea et al., 2003; Foster, 2003;
Vijayalaxmi and Obe, 2004; Obe and Vijalalaxmi, 2005; Draft IEEE Std PC95.1-2005; Moulder et
al., 2005]
• Results should be viewed with respect to previously accepted scientific principles before
ascribing the results to new theories. Research findings pointing to previously unidentified
relationships should be carefully evaluated and appropriate additional independent studies
should be conducted before the findings are accepted (Repacholi, 1998; Adair, 2003; Foster,
2003).
• Theories (e.g., for mechanisms of interaction) should make sufficiently concrete predictions that
they can be tested experimentally and be capable of being verified, if correct (Schwan, 1988;
Repacholi, 1998; Adair, 2003; Foster, 2000; Foster, 2003; Foster and Repacholi, 2004; Draft IEEE
Std PC95.1-2005).
Publication
• Well-designed and -conducted studies should be published regardless of the outcome, since
negative results are as useful as positive studies in the context of the EMF database for health
risk evaluation (Kheifets et al., 2003).
• Timely peer reviewed publication is essential. (Repacholi, 1996; Repacholi, 1998; International
EMF Project/ICNIRP, 1999; Kheifets et al., 2003).

Scientific Responsibilities
Scientists are required to take responsibility for:

• the sensitive issue of data protection and confidentiality,


• the new opportunities, such as the increasing availability of computerized data,
• the incorporation of molecular methods to aid the investigation of mechanistic pathways and
gene-environment interactions,
• and the development and utilization of sophisticated statistical approaches.
[Quoted from Rushton and Elliott, 2003; see also Conrads et al., 2004; Dorman, 2005; Espina et al.,
2005; Gilham et al., 2005; LaPorte, 2005]

Replication
• Preferably, the experiments should be repeated and the data confirmed independently, or the
claimed effects should be consistent with results of similar experiments, for which the biological
systems involved are comparable (Repacholi, 1998).

Meta Studies
• Since the technologies of electricity and radio telecommunications serve billions of people it is
appropriate to address the remaining health concerns where possible in large multi-centred
international studies [for example Perform A EU 2000-2005; FDA study USA, 2004-2008]
• Where possible, feasibility studies (pilot studies) for multi-centred studies should be undertaken
with input from international experts to verify and validate every aspect of the methodology of
the EMF field study under question. The feasibility studies should be open for further scientific
comment and published in peer reviewed journals before these large studies are funded and
carried out
• Simultaneous experimental replications should use the same or similar methodologies and
should report data to provide quantitative results that can be compared and or combined for
meta analysis of all the independently collected data.
• In meta studies the data collection and analyses should be done according to the hypothesis and
methodology of the published protocol and the guidelines for complex statistical analyses
(Keppel, 1982).
In Vitro Study Design

Introduction
Please refer to ‘General Experimental Design Criteria’ and the ‘General Research Priorities’ above
and the in vitro triage below for details as well.
Traditionally, cellular research was considered as the possible source of a plausible mechanism for
any biological effect due to exposure to RF or ELF signals [Repacholi and Cardis, 1997; Repacholi,
1998].
Great advances in cellular research have given us a new understanding of the interplay of cellular
research of genes, proteins and cell signalling in single living cells, and the functioning of the whole
human body in health and disease [Lee and Nurse, 1987; Kandel et al., 2000; Kandel, 2001; Conrads
et al., 2004; Dorman, 2005; Espina et al., 2005; Laporte, 2005].
These molecular advances have an impact on EMF research priorities. Molecular profiling may become essential at all levels of research
namely in epidemiology, human acute studies, animals studies and tissue cultures. We now see the incorporation of molecular methods
to aid the investigation of mechanistic pathways and gene-environment interactions.

Hypothesis
The in vitro study should test a clearly defined hypothesis, using a detailed protocol that would lead
to quantitative information directly or indirectly relevant to assessment of health risk from ELF and
RF exposure and allow any other independent laboratory(s) to reconstruct the study and replicate
the findings for validation and allow the results be combined for meta analysis where appropriate.
Good Laboratory Practice
Good Laboratory Practice (GLP) should be used throughout the design and conduct of any study
where possible and practical (see, e.g., FDA, 1993, NTP 1992).
Quality assurance (QA) procedures should be included in the protocol, including traceable dosimetry
and monitoring of the programme by both a team from within the experimental staff and an
independent group, as required by GLP [Repacholi, 1998; Repacholi and Greenebaum, 1999;
Schönborn et al., 2000; ICNIRP, 2003; IEEE Std C95.6-2002; IEEE Std C95.1, 1991/1999; Draft IEEE Std
PC95.1-2005]
Methodology
• Well-characterized biological systems or assays should be used, preferably ones that are well
established (validated) from the scientific literature that can be used reliably by other
laboratories (Murchan et al., 2003; Vijayalaxmi and Obe, 2004; Obe and Vijayalaxmi 2005).
• The biological system used should be appropriate to the end-point(s) studied.
• Threshold responses derived from using at least 3 levels of dose and duration of exposure data
where possible are sought in addition to sham-exposed controls (‘Sham/Sham’). Appropriate
positive controls as well as non-RF heating controls can be very important to help assess the
metric of the response of the biological system and the potential effects of RF heating should be
included.
• Methods of staining and scoring for tests including DNA breakage [such as micronuclei (MN),
chromosome aberrations (CA) and sister chromatid exchange (SCE)] should be standardised, and
their variability, reliability, and validity stated.
• Procedures should be double-blinded and scored by more than 1 independent laboratory before
breaking the code.
• The a priori estimated statistical power of the experiment, based on prior knowledge and the
number of tests planned, should be sufficient to reliably detect the expected size of the effect
(often as small as 10-20%) (In vitro Triage).
• Sufficient numbers of each sample for staining and scoring for tests including DNA breakage [for
instance, MN, CA, SCA] should be taken to insure the power of the study is large enough to find
a significant effect of exposure if there is one. [COST Action 281 Recommendation on an
Internationally Co-ordinated Research on Genotoxic Effects of Electromagnetic Radiation from
Mobile Communication Systems, http://www.cost281.org/activities/Gentox-recomm-
090304AW.doc; Brüske-Hohlfeld et al., 2001; Vijayalaxmi and Obe, 2004; Obe and Vijayalaxmi,
2005].
• Further experiments should include techniques for genomics, proteomics and transcriptomics to
give us information about genes and the state of protein such as whether it is activated or
inactivated. The power of newer protein arrays is that they give us a window into the
microcircuitry of the cell, revealing which pathways are being used at a particular moment
[Conrads et al., 2004; Espina et al., 2005]. They may provide crucial information that could even
be due to (IR) radiation therapy [Sreekumar et al., 2001] or possibly EMF exposure.
Exposure Conditions
There should be well-defined exposure conditions with improved control of the exposure parameters, continuous monitoring and control
of the environmental parameters, support of a double blinded study protocol, carefully characterized dosimetry including the average
specific absorption rate (SAR), standard deviation of the SAR, monitoring of the temperature load, determination of the local SAR and
possible temperature hotspots. (Nikoloski et al., 2005)

1. Temperature, atmosphere in CO² incubators, vibration, and stray fields from incubator heaters
and fans are sources of asymmetry (differences between exposed and control samples) that are
often overlooked in cell and tissue culture experiments. These must be measured with
appropriate instrumentation and every effort made to ensure that any differences are
minimized, except for EMF exposure of the "exposed" samples.
• The environmental requirements (e.g., stabilized temperature, atmospheric control,
sterility) must be strictly fulfilled.
• All relevant technical and biological parameters must be monitored during the experiment.
• The most important technical data should be logged in order to track possible malfunctions
of the system.
• All controlling and monitoring devices should be rigorously checked for interference under
worst- case considerations.
• Non-disturbance of commercial services must be ensured [Burkhardt et al., 1996; Schönborn
et al., 2000; Schuderer and Kuster, 2003; Nikoloski et al., 2005].
2. Contemporaneous positive controls for the assay and for non-RF thermal effects are
recommended and negative controls, both maintained under identical circumstances to
exposed cultures, sham-sham comparisons of multiple exposure systems, randomized handling
of cultures, and blinding, forced cooling and repeatable positioning of the flasks, should form
part of the study, as appropriate. It is strongly recommended that any pathology or
histopathology be performed in a blinded fashion.
3. To characterize electric fields or induced currents in cultures, electrode geometry and materials
(including agar bridges, etc.), dish shape and dimensions, depth of medium and specimen
dimensions, conductivity (RF and ELF) and dielectric constant (RF only) of medium are
important.
4. Comparisons of E field, SAR and temperature measurements should result in good agreement.
5. Signal characteristics and field strengths inside the medium must be relevant to risk assessment.
Therefore, the system should allow for field strengths that are at least as high as the relevant
limits defined by the regulatory bodies [ANSI, 1992; CENELEC, 1998; ICNIRP, 1998] or the highest
exposure occurring in real life situations.
6. The field distribution at the location of the cell culture should be as homogeneous as possible.
The target value shall be better than ±30% [Kuster and Schönborn, 2000; Schönborn et al., 2000;
2001; Schuderer and Kuster, 2003; Nikoloski et al., 2005].
7. The signal characteristics must be well defined. In particular, this requires a signal source that is
well defined with respect to the frequency, modulation scheme, power stability, and noise level
(Schönborn et al., 2000]. Field values should be measured directly. Electrophoretic products
should be considered and measured, where possible, when electrodes are used.
8. ELF magnetic field experiments should consider the factors above as they apply to induced
current. The angle between applied field and medium, as well as the angle between applied ELF
fields and the local DC field, should be measured.
9. When using media, serum or other reagents that may have variation from batch to batch,
serious consideration should be given to purchasing and storing sufficient stocks in a single
batch for the duration of the experiment. Similarly, the characteristics of cell lines derived from
a standard source should not be allowed to diverge over time. There should be backup stocks
from the original source.
10. For experiments lasting more than a few days and in all cases where samples or stock cultures
are maintained for extended periods or data are gathered or stored electronically, backup
systems must be installed to protect the work against equipment or power supply failure.
11. The costs for design, construction, and maintenance of the system should be reasonable. The
system should be easy-to-use and as foolproof as possible. (Quoted from Schönborn et al.,
2000).

Results Analysis
• Results should be analysed and interpreted on the basis of the prior stated hypothesis (In vitro
Triage).
• Other tests of significance post hoc should be stated as such and an appropriate ‘multiple
comparisons’ correction (I.E. Bonferroni, Tukey, or Empirical Bayes Adjustments) be applied.
Post hoc results can be considered hypothesis generating for new research but cannot be
considered as the direct or main result of the presented research. (Keppel, 1982; Steenland et
al., 2000).
• The known reliability and validity of the tests employed and the normal variance of these tests
stated in the methods should be referred to in the analysis of the results. The variance in the
scores due to experimental exposure needs to be stated in the interpretation of the results
taking into account the normal variance of the test scores without treatment, and variance of
the test scores under sham control, positive control, and negative control treatment conditions.
[IEEE Std C95.6-2002; Draft IEEE Std PC95.1-2005; Foster and Repacholi, 2004; In vitro Triage].

Conclusion
• The conclusion should be drawn on the basis of the prior stated hypothesis of the study (In vitro
Triage).
• Knowledgeable interpretation of the results must consider the established science (outside of
the EMF area) in the field of investigation such as biophysical mechanisms of interaction,
genotoxicity, DNA, RNA and protein replication, cell cycle, reproduction, apoptosis, free radicals,
molecular mechanisms, and cell signalling [Repacholi, 1998; ICNIRP, 1998; IEEE Std C95.6-2002;
ICNIRP, 2003; Foster, 2003; Vijayalaxmi and Obe, 2004; Obe and Vijalalaxmi, 2005; Draft IEEE Std
PC95.1-2005].
• Results should be viewed with respect to previously accepted scientific principles before
ascribing the results to new theories. Research findings pointing to previously unidentified
relationships should be carefully evaluated and appropriate additional independent studies
should be conducted before the findings are accepted (Repacholi, 1998; Adair, 2003; Foster,
2003).
• Theories (e.g., for mechanisms of interaction) should make sufficiently concrete predictions that
they can be tested experimentally and be capable of being verified, if correct (Schwan, 1988;
Repacholi, 1998; Adair, 2003; Foster, 2000; Foster, 2003; Foster and Repacholi, 2004; Draft IEEE
Std PC95.1-2005).

Publication
• Well-designed and -conducted studies should be published regardless of the outcome, since
negative results are as useful as positive studies in the context of the EMF database for health
risk evaluation (Kheifets et al., 2003).

Scientific Responsibilities
Scientists are required to take responsibility for:

• the sensitive issue of data protection and confidentiality,


• the new opportunities, such as the increasing availability of computerized data,
• the incorporation of molecular epidemiological methods to aid the investigation of mechanistic
pathways and gene-environment interactions,
• and the development and utilization of sophisticated statistical approaches.
[Quoted from Rushton and Elliott, 2003; see also Conrads et al., 2004; Dorman, 2005; Espina et al.,
2005; Gilham et al., 2005; Keppel, 1982; LaPorte, 2005]
Replication
• Preferably, the experiments should be repeated and the data confirmed independently, or the
claimed effects should be consistent with results of similar experiments, for which the biological
systems involved are comparable (Repacholi, 1998).

Meta Studies
• Since the technologies of electricity and radio telecommunications serve billions of people it is
appropriate to address the remaining health concerns where possible in large multi-centred
international studies [COST 281, Directive, March 11, 2004].
• Where possible, feasibility studies (pilot studies) for multi-centred studies should be undertaken
with input from international experts to verify and validate every aspect of the methodology of
the EMF field study under question. The feasibility studies should be open for further scientific
comment and published in peer reviewed journals before these large studies are funded and
carried out (COST 281, Directive, March 11, 2004).
• Simultaneous in vitro replications should use the same or similar methodologies including
standardized tests and should report data to provide quantitative results that can be compared
and or combined for meta analysis of all the independently collected data.
• In meta studies the data collection and analyses should be done according to the hypothesis and
methodology of the published protocol.
This is a living document and research methods are updated continually as are the IEEE ICES
SC-4 Triages listed in Appendix A below.

REFERENCES

Adair ER, Berglund LG: On the thermoregulatory consequences of NMR imaging. Magn Reson Imaging 1986, 4:321-
33.

Adair ER, Berglund LG: Predicted thermophysiological responses of humans to MRI fields. Ann N Y Acad Sci 1992,
649:188-200.

Adair ER, Blick DW, Allen SJ, Mylacraine KS, Ziriax JM, Scholl DM. Thermophysiological responses of human
volunteers to whole body RF exposure at 220 MHz Bioelectromagnetics. 2005, DOI 10.1002/bem.20101
Adair ER, Cobb BL, Mylacraine KS, Kelleher SA: Human exposure at two radio frequencies (450 and 2450 MHz):
similarities and differences in physiological response, Bioelectromagnetics 1999, 20(Suppl 4):12-20.

Adair ER, Kelleher SA, Mack GW, Morocco TS: Thermophysiological responses of human volunteers during
controlled whole-body radio frequency exposure at 450 MHz, Bioelectromagnetics 1998, 19:232-245.

Adair ER, Mylacraine KS, Allen SJ, Thermophysiological consequences of whole-body resonant RF exposure (100
MHz) in human volunteers. Bioelectromagnetics 2003, 24:489-501.

Adair ER, Mylacraine KS, Cobb BL: Human exposure to 2450 MHz CW energy at levels outside the IEEE C95.1
Standard does not increase core temperature. Bioelectromagnetics 2001, 22:429-439.

Adair RK, Biophysical limits on athermal effects of RF and microwave radiation.


Bioelectromagnetics 2003, 24:39-48.

Ahlbom A (1996): Some fundamental aspects of epidemiology with reference to research on magnetic fields and
cancer. In: R Matthes (Ed): "Non-Ionising Radiation". Proc.3rd Non-Ionising Radiation Workshop, 22-26 April,
Baden, Austria: ICNIRP, International Commission on Non-Ionising Radiation Protection, D-85764 Oberschleißheim,
Germany, 17-27.

Ahlbom A, Green A, Kheifets L, Savitz D, Swerdlow A; ICNIRP (International Commission for Non-Ionizing Radiation
Protection) Standing Committee on Epidemiology, Epidemiology of health effects of radiofrequency exposure,
Environ Health Perspect, 2004 Dec;112(17):41-54. Review.

Ahrens W, Krickeberg K, Pigeot I, An introduction to epidemiology, in Handbook of Epidemiology, Ahrens W, Pigeot


I (Eds), Springer, Heidelberg, 2005. 3-36.

Ahrens W, Pigeot I (Eds) Handbook of Epidemiology, Springer, Heidelberg, 2005.


Andersen J-B, Signal forms in wireless applications, presentation in COST281 Workshop - Do sinusoidal versus non-
sinusoidal waveforms make a difference? and 8th MCM (Zurich) 2005/02/17-18.
http://www.cost281.org/documents.php?node=93&dir_session=

Angelone LM, Potthast A, Segonne F, Iwaki S, Belliveau JW, Bonmassar G, Metallic electrodes and leads in
simultaneous EEG-MRI: Specific absorption rate (SAR) simulation studies, Bioelectromagnetics. 2004
May;25(4):285-295.

Balmain A, Harris C C, Carcinogenesis in mouse and human cells: parallels and paradoxes, Carcinogenesis, (2000), 21(3),
371–7.

Balzano Q, Exposure metrics for RF Epidemiology: Cellular phone handsets. Radiation Prot. Dos, (1999) 83:1-2: 165-
170.

Balzano Q, Faraone A, Human exposure to cellular base station antennas, IEEE International Symposium on
Electromagnetic Compatibility, 1999, 2:924-928.

Beaglehole R, Bonita R, Kjellström T, Basic epidemiology. Geneva: World Health Organization,


(1993), 175.

Bernardi P, Cavagnaro M, Pisa S, Piuzzi E, Power absorption and temperature elevations induced in the human
head by a dual-band monopole-helix antenna phone, IEEE Trans. Microwave Theory Tech., 2001, 49: 2539 – 2546.

Bernardi P, Cavagnaro M, Pisa S, Piuzzi E, Specific absorption rate and temperature


increases in the head of a cellular-phone user, IEEE Trans. Microwave Theory
Tech., 2000, 48: 1118-1126.

Blettner & Schlattmann, Meta-analysis in epidemiology, in, Ahrens W, Pigeot I (Eds) Handbook of Epidemiology,
Springer, Heidelberg, 2005, 829-858.
Blodgett TM, Casagranda B, Townsend DW, Meltzer CC, Issues, Controversies, and Clinical Utility of Combined
PET/CT Imaging: What Is the Interpreting Physician Facing? AJR Am J Roentgenol, 2005 May;184(5 Suppl):S138-45.

Blundell T.L. (1996) Structure-based drug design. Nature 384 Suppl 7:23-26.

Bracken, T., Kheifets, L. and Sussman, S., (1993): Exposure assessment for power frequency electric and magnetic
fields (EMF) and its application to epidemiologic studies. Journal of Exposure Analysis and Environmental
Epidemiology, 3: 1-22.

Brodsky LM, Habash RW, Leiss W, Krewski D, Repacholi M. Health risks of electromagnetic fields. Part III: Risk
analysis. Crit Rev Biomed Eng. 2003;31(4):333-354

Burkhardt M, Pokovi K, Gnos M, Schmid T, Kuster N, Numerical and experimental dosimetry of Petri dish exposure
setups, Bioelectromagnetics (1996) 17:6 : 483 - 493

Brüske-Hohlfeld I, Scherb H, Bauchinger M, Schmid E, Fender H, Wolf G, Schmitz-Feuerhake I, Schröder H, Stephan


G, Csicsaky M, & Wichmann HE, A cluster of childhood leukaemias near two neighbouring nuclear installations in
Northern Germany: prevalence of chromosomal aberrations in peripheral blood lymphocytes, Int. J. Radiat. Biol.
(2001): 77: 1: 111-116.

Cabeza R,Nyberg, L, Imaging Cognition II: An Empirical Review of 275 PET and fMRI studies. Journal of Cognitive
Neuroscience (2000), 12,1: 1-47.

Cardis E., Kilkenny M, International case control study of adult brain head and neck tumours: Results of the
feasibility study. Radiation Protection Dosimetry 1999 83:1-2. 179-184.

Cardis E. Rice JM, Criteria for health risk assessment with the International EMF Project, 1997, 35-47. In: Non-
Thermal Effects of RF Electromagnetic Fields. Munich 20-21, November 1996. Editors J.H. Bernhardt, R. Matthes
and M.H Repacholi. ICNIRP 3/97, WHO.
Conrads TP, Fusaro VA, Ross S, Johann D, Rajapakse V, Hitt BA, Steinberg SM, Kohn EC,
Fishman DA, Whitely G, Barrett JC, Liotta LA, Petricoin EF 3rd, Veenstra TD. High-resolution
serum proteomic features for ovarian cancer detection. Endocr Relat Cancer. 2004
Jun;11(2):163-78.

COST Action 244bis, Chairman Veyret B, Study on an Action Plan on Possible Health Effects Related to the Use of
Radiotelephones. COST Action 244bis (Project COST-STY-98-4023) Project Final Report, 1999.

COST 244 EC Chairman, McKinlay AF: Possible health effects related to the use of radiotelephones. Proposals for a
research programme by a European Commission Expert Group. Directorate General XIII, Telecommunications,
Information Market and Exploitation of Research, Brussels: European Commission.,1996.

COST Action 281 Recommendation on an Internationally Co-ordinated Research on Genotoxic Effects of


Electromagnetic Radiation from Mobile Communication Systems http://www.cost281.org/activities/Gentox-
recomm-090304AW.doc

Chou C K, Basic problems of diversely reported biological effects of radio frequency fields, Radiats Biol Radioecol.
2003 Sep-Oct;43(5):512-518.

Chou C-K, D’Andrea JA, Reviews of effects of RF fields on various aspects of human health: Introduction
Bioelectromagnetics 14 Nov 2003, 24, S6:S5-S6.

D'Andrea JA, Chou C-K, Johnston SA, Adair ER, Review: Microwave effects on the nervous system,
Bioelectromagnetics Supplement 2003, 6:S107-S147.

Diem E, Schwarz C, Adlkofer F, Jahn O, Rudiger H, Non-thermal DNA breakage by mobile-phone radiation
(1800MHz) in human fibroblasts and in transformed GFSH-R17 rat granulosa cells in vitro, Mutat Res. 2005 Jun
6;583(2):178-183.
Dorman Janice S, Director, Collaborating Center for WHO Multinational Project for Childhood Diabetes (DiaMond),
U of Pittsburgh Molecular epidemiology and technology transfer, http://www.pitt.edu/~rlaporte/who.html

Elliott P, Jarup L, Ahlbom A, Cohort study of mobile phone users (pilot study) in press
http://www.mthr.org.uk/research_projects/mthr_funded_projects/elliot.htm

Elliott P, Wakefield JC, Bias and confounding in spatial epidemiology. In: Spatial Epidemiology: Methods and
Applications, Elliott P, Wakefield J, Best N, Briggs DJ, Eds, 2000 Oxford University Press, Oxford, 68–84.

Elliott P, Wartenberg D, Spatial epidemiology: current approaches and future challenges, Environ Health Perspect.
2004 Jun;112(9):998-1006. Review.

Elliott P, Westlake AJ, Kleinschmidt I, Hills M, Rodrigues L, McGale P, et al., The Small Area
Health Statistics Unit: a national facility for investigating health around point sources of
environmental pollution in the United Kingdom, J Epidemiol Community Health, 1992, 46:345–
349.

EPRI, Residential transient magnetic field research: Interim report, (1994), Project RP2966-07. Palo Alto, California:
Electric Power Research Institute, Report TR-103470

Espina V, Geho D, Mehta AI, Petricoin EF 3rd, Liotta LA, Rosenblatt KP, Pathology of the future: molecular profiling
for targeted therapy. Cancer Invest. 2005; 23(1):36-46.

Excell PS, Olley P, Jackson NN, Modelling of an arbitrarily-oriented mobile telephone handset in the Finite-
Difference Time-Domain field computation method, Applied Computational Electromagnetics Society Journal,
(1996) 11(2), 55-65.

Excell PS, Computer modelling of high frequency electromagnetic field penetration into the human head,
Measurement and Control 31(1998), (6), 170-175.
Exposure Metrics and Dosimetry for EMF Epidemiology: Proceedings of an International Workshop National
Radiological Protection Board, Chilton, UK September 7-9, 1998. Editor A.F. McKinlay & M.H. Repacholi. Radiation
Protection Dosimetry 1999, 83:1-2. 1-193.

Faraone, A., Tay, R, Y-S., Joyner, K.H., Balzano, Q. Estimation of the average power density on the vicinity of cellular
base-station collinear array antennas IEEE Transactions on Vehicular Technology May 2000, 49, 3:984-996.

Feychting M, Ahlbom A, Kheifets L, Emf and health, Annu Rev Public Health. 2005, 26:165-189.

Foster KR, Biological Effects of Radiofrequency Fields: Does Modulation Matter? presentation in COST281
Workshop - Do sinusoidal versus non-sinusoidal waveforms make a difference? and 8th MCM (Zurich) 2005/02/17-
18. http://www.cost281.org/documents.php?node=93&dir_session=

Foster KR, Electromagnetic Field Effects and Mechanisms, IEEE Engineering in Medicine and Biology Magazine
1996,15 (4): 50-56

Foster KR, Mechanisms of interaction of extremely low frequency electric fields and biological systems. Radiat Prot
Dosimetry, 2003, 106(4):301-10.

Foster KR, Adair ER, Modeling thermal responses in human subjects following extended exposure to
radiofrequency energy. Biomed Eng Online. Feb 2004 28;3(1):4. http://www.biomedical-engineering-
online.com/content/3/1/4

Foster KR, Repacholi MH, Biological Effects of Radiofrequency Fields: Does Modulation Matter? Radiation Research
(2004) 162:219-225.

FDA: Good laboratory practice for nonclinical laboratory studies. Food and Drug Administration, US Department of
Health and Human Services. (1993) Federal Register 21 CFR Ch. 1 (4-1-93 Edition) Part 58, pp 245-258.
Gajsek P, Pakhomov AG, Klauenberg BJ, Electromagnetic field standards in Central and Eastern European countries:
current state and stipulations for international harmonization, Health Phys. 2002 Apr;82(4):473-483.

Gandhi O, Li Q X, Kang G, Temperature rise for the human head for cellular telephones and for peak SARs
prescribed in safety guidelines, IEEE Trans. Microwave Theory Tech., 2001, 49: 1607-1613.

Gilham C, Peto J, Simpson J, Roman E, Eden TO, Greaves MF, Alexander FE; UKCCS Investigators Day care in
infancy and risk of childhood acute lymphoblastic leukaemia: findings from UK case-control study, BMJ, 2005 Jun
4;330(7503):1294-1299. http://bmj.bmjjournals.com/cgi/rapidpdf/bmj.38428.521042.8Fv1.

Greenland, S, The impact of prior distributions for uncontrolled confounding and response bias: a case study of the
relation of wire codes and magnetic fields to childhood leukemia, Journal of the American Statistical Association,
2003, 98, 47-54.

Haarala C, Ek M, Bjornberg L, Laine M, Revonsuo A, Koivisto M, Hamalainen H, 902 MHz mobile phone does not
affect short term memory in humans. Bioelectromagnetics 2004. 25:452-456.

Habash I, Health risks of electromagnetic fields. Part I: Evaluation and assessment of electric and magnetic fields,
Crit Rev Biomed Eng. 2003;31(3):141-95

Habash I, Health risks of electromagnetic fields. Part II: Evaluation and assessment of radio frequency
radiation.,Crit Rev Biomed Eng. 2003;31(3):197-254.

HCN, Health Council of the Netherlands, TNO study on the effects of GSM and UMTS signals on well-being and
cognition: Review and recommendations for further research, 28 June 2004 HCN report No U-886\EvR\RA\673-A2
publication no. 2004/13E. http://www.gr.nl/pdf.php?ID=1042

Hill A B, The environment and disease: association or causation? Proc. R. Soc. Med (1965), 58:295-300.
Hirata A, Morita M, Shiozawa T, Temperature increase in the human head due to a dipole antenna at microwave
frequencies, IEEE Trans. Elect. Compatibility, 2003. 45:109-116.

Hirata A, Shiozawa T, Correlation of maximum temperature increase and peak SAR in the human head due to
handset antennas, IEEE Trans. Microwave Theory Tech., 2003 51: 1831-1841.

Huber R, Schuderer J, Graf T, Jütz K, Borbely AA, Kuster N, Achermann P, Radio frequency electromagnetic field
exposure in humans: Estimation of SAR distribution in the brain, effects on sleep and heart rate,
Bioelectromagnetics (8 April 2003) 24:4:262-276.

IARC Monograph ‘Non-ionizing Radiation, Part 1: Static and Extremely Low Frequency Electric and Magnetic Fields’
Vol. 80, 2002.

ICNIRP, Exposure to Static and Low Frequency Electromagnetic fields, Biological Effects and Health Consequences
(0-100 kHz), Eds Matthes R. McKinlay AF, Bernhardt JH, Vecchia P, Veyret B, 2003, Märkl-Druck, München,
Germany. http://www.icnirp.de/.

ICNIRP, Guidelines for limiting exposure to time-varying electric, magnetic, and electromagnetic fields (up to 300

GHz), Health Physics, 1998; 74 (4):494-522.

ICNIRP, Non-thermal effects of RF electromagnetic fields. R Matthes, J Bernhardt and MH Repacholi, Eds,
Proceedings of Munich meeting, Nov. 1996. ICNIRP Pub. 3/97. From: ICNIRP C/- Bundesamt für Strahlenschutz,
Institut für Strahlenhygiene, Ingolstädter Landstraße 1, D-85764 Oberschleißheim, Germany.

IEEE Std 1528-2003. Recommended Practice for Determining the Peak Spatial-Average Specific Absorption Rate
(SAR) in the Human Head from Wireless Communications Devices: Measurement Techniques.
IEEE Std C95.6-2002, IEEE Standard for Safety Levels with Respect to Human Exposure to Electromagnetic Fields, 0
to 3 kHz, September 2002, IEEE, New York, USA.

IEEE Std C95.1-1991/1999, IEEE Standard for Safety Levels with Respect to Human Exposure to Radio Frequency
Electromagnetic Fields, 3 kHz to 300 GHz. 1999, IEEE, New York, USA.

IEEE, ICES, COST 281: Joint workshop 2004/09 Thermal Physiology Workshop (Paris)
http://www.cost281.org/documents.php?node=85&dir_session=

International EMF Project/ICNIRP, Working group draft report from Erice November 1999: Review of low-level
pulsed radiofrequency effects and research needs.

Ivancsits S, Pilger A, Diem E, Jahn O, Rudiger HW, Cell type-specific genotoxic effects of intermittent extremely low-
frequency electromagnetic fields, Mutat Res. 2005 Jun 6;583(2):184-188.

Johansen C, Boice J Jr, McLaughlin J, Olsen J, Cellular telephones and cancer--a nationwide cohort study in
Denmark, J Natl Cancer Inst., 2001 Feb 7;93(3):203-7.
http://jncicancerspectrum.oupjournals.org/cgi/content/full/jnci;93/3/203

Johansen C, Boice JD Jr, McLaughlin JK, Christensen HC, Olsen JH, Mobile phones and malignant melanoma of the
eye, Br J Cancer., 2002 Feb 1;86(3):348-349.

Johnston S, WHO criteria for EMF health risk assessment, in Biological Effects of Electromagnetic Fields. (Ed) P
Stavroulakis; Springer 2003. 733-777

Johnston SA, Scherb H. Mobile phone use and acoustic neuromas. Epidemiology. 2005 May;16(3):416-7; author
reply 417-8.
Kandel ER, The Molecular Biology of Memory Storage: A Dialogue Between Genes and Synapses Science (2001)
294: 5544: 1030-1038.

th
Kandel ER, Schwartz J, Jessell M, (eds) Principles of Neural Science 4 Edition, (2000), McGraw Hill, London, 1-
1414.

nd
Keppel G., Design and Analysis A Researcher’s Handbook 2 Edition 1982, Prentice Hall, New Jersey.

Kheifets L, Repacholi M, Saunders R, Thermal stress and radiation protection principles. International Journal of
Hyperthermia, May-June 2003, 19, (3): 215–224.

Kuster, N. Dosimetric needs for laboratory and epidemiological research, In, Biological effects of non-thermal
pulsed and amplitude modulated RF electromagnetic fields and related health risks, Eds: Bernhardt, J.H., Matthes
R., Repacholi M.H. ICNIRP; 3/1997: 27-34.

Kuster N, Balzano Q, Experimental and numerical dosimetry. In: N Kuster, Q Balzano, JC Lin (Eds) Mobile
communications safety, (1996),London: Chapman Hall pp 13-64.

Kuster N, Schönborn F, Recommended Minimal Requirements and Development Guidelines for Exposure Setups of
Bio-Experiments Addressing the Health Risk Concern of Wireless Communications, Bioelectromagnetics Oct 2000,
21:508-514.

Kuster N, Schuderer J, Christ A, Futter P, Ebert S, Guidance for exposure design of human studies addressing health
risk evaluations of mobile phones, Bioelectromagnetics, 2004 Oct;25(7):524-529.

LaPorte Ronald E, Telecommunications and Disease Monitoring Director: The Centre was established to assist with
the WHO global program in the area of non-communicable diseases. http://www.pitt.edu/~rlaporte/who.html
Lee MG, Nurse P. (1987) Complementation used to clone a human homologue of the fission
yeast cell cycle control gene cdc2. Nature 327: 31-35.

McKinlay A. Radiotelephones and human health: A European research initiative. Radiation Research Dosimetry.
1997, 72: 3-4: 313-320.

Moulder JE, Foster KR, Erdreich LS, McNamee JP, Mobile phones, mobile phone base stations and cancer: a review
Int JRadiat Biol, 2005, 81:3:189-203.

Murchan S, Kaufmann ME, Deplano A, de Ryck R, Struelens M, Zinn CE, Fussing V, Salmenlinna S, Vuopio-Varkila J,
El Solh N, Cuny C, Witte W, Tassios PT, Legakis N, van Leeuwen W, van Belkum A, Vindel A, Laconcha I, Garaizar J,
Haeggman S, Olsson-Liljequist B, Ransjo U, Coombes G, Cookson B, Harmonization of pulsed-field gel
electrophoresis protocols for epidemiological typing of strains of methicillin-resistant Staphylococcus aureus: a
single approach developed by consensus in 10 European laboratories and its application for tracing the spread of
related strains. J Clin Microbiol. 2003 Apr, 41(4):1574-1585.

NIEHS (1998). Assessment of health effects from exposure to power line frequency electric and magnetic fields.
NIEHS working group report. Research Triangle Park: National Institute of Environmental Health Sciences, 1998.

Neubauer G, Röösli M, Feychting M, Hamnerius Y, Kheifets L, Kuster N, Ruiz I, Schüz J, Überbacher R, Wiart J, Study
on the Feasibility of Epidemiological Studies on Health Effects of Mobile Telephone Base Stations-Final Report.
FSM- Project No. A2003-9, March 2005 Copy No. 1 ARC-IT-0124 http://www.mobile-
research.ethz.ch/english/projekte_e.htm; http://www.mobile-research.ethz.ch/var/pub_neubauer_pref14.pdf

Nikoloski N, Frohlich J, Samaras T, Schuderer J, Kuster N, Reevaluation and improved design of the TEM cell in vitro
exposure unit for replication studies. Bioelectromagnetics, 2005 Apr; 26(3):215-224.

NRPB: McKinlay AF, Allen SG, Cox R, Dimbylow PJ, Mann SM, Muirhead CR, Saunders RD,
Sienkiewicz ZJ, Stather JW, Wainwright PR, Review of the Scientific Evidence for Limiting
Exposure to Electromagnetic Fields (0–300 GHz), 30 April 2004 Documents of the NRPB: 15: 3.
http://www.hpa.org.uk/radiation/publications/documents_of_nrpb/abstracts/absd15-3.htm
NRPB Workshop Non-Ionising Issues, Dennis, J. A. Stather, J. Eds, Radiation Protection Dosimetry, 1997, 72; 3/4: 1-
336.

NTP: Specifications for the conduct of studies to evaluate the toxic and carcinogenic potential of chemical,
biological and physical agents in laboratory animals for the National Toxicology Program (NTP). Attachment 2.
August 1992 (Including modifications through 9/95). Available from: National Institute of Environmental Health
Sciences, Environmental Toxicology Program, PO Box 12233, Research Triangle Park, NC 27709, USA.

Obe G; Vijayalaxmi. Controversial cytogenetic observations in mammalian somatic cells exposed to extremely low
frequency electromagnetic radiation: A review and future research recommendations Bioelectromagnetics. 2005
May 10

Reilly JP, An analysis of differences in the low-frequency electric and magnetic field exposure standards of ICES and
ICNIRP, Health Physics, 2005, 89(1):71-80.

Repacholi MH, Low-level exposure to radiofrequency fields: Health effects and research needs.
Bioelectromagnetics (1998) 19: 1-19.

Repacholi MH, and Cardis E, Criteria for EMF health risk assessment. Radiat Prot Dosim, (1997):72: 305-312.

Repacholi MH, Greenebaum B, Interaction of static and extremely low frequency fields on living systems: Health
effects and research needs. Bioelectromagnetics, (1999) 20:3:133-161.

Rothman KJ, Chou CK, Morgan R, Balzano Q, Guy AW, Funch DP, Preston-Martin S, Mandel J, Steffens R, Carlo G,
Assessment of cellular telephone and other radiofrequency exposures for epidemiological research, Epidemiol,
(1996), 7:291-298.

nd
Rothman KJ, Greenland S, Modern Epidemiology. 2 Ed. Lippencott-Raven, 1998.
Rushton L, Elliott P, Evaluating evidence on environmental health risks, Br Med Bull, 2003; 68:113-128.

Scherb H, Methodological Issues in Epidemiology, presentation in COST281 -Are RF-fields able to raise the risk of
cancer? Schriesheim, 2004/11/23. http://www.cost281.org/documents.php?node=94&dir_session=

Schönborn F, Pokovi K, Kuster N, Dosimetric analysis of the carousel setup for the exposure of rats at 1.62 GHz,
Bioelectromagnetics 2004, 25:16-26.

Schönborn F, Pokovic K, Burkhardt M, Kuster N. Basis for optimization of in vitro exposure apparatus for health
hazard evaluations of mobile communications. Bioelectromagnetics. 2001 Dec, 22(8):547-59.

Schönborn F, Pokovic K, Wobus AM, Kuster N, Design, optimization, realization, and analysis of an in vitro system
for the exposure of embryonic stem cells at 1.71GHz, Bioelectromagnetics 2000, 21:372-384.

Schuderer J, Kuster N, Effect of the meniscus at the solid/liquid interface on the SAR distribution in Petri dishes and
flasks, Bioelectromagnetics, 2003 Feb, 24(2):103-108.

Schwan H.P. Biological effects on non-ionizing radiation: Cellular properties and interactions: Ann Biomed Eng.
1988, 16:245-263.

Society of Nuclear Medicine Brain Imaging Council, Ethical clinical practice of functional brain imaging, J Nucl Med.,
1996 Jul, 37(7):1256-1259.

Sreekumar A, Nyati MK, Varambally S, Barrette TR, Ghosh D, Lawrence TS, Chinnaiyan AM. Profiling of cancer cells
using protein microarrays: discovery of novel radiation-regulated proteins. Cancer Res. 2001 Oct 15;61(20):7585-
93.
Stagg, R0. B., Hawel, LH, III, Pastorian, K, Cain, C, Adey, WR, Byus CV, Effect of immobilization and concurrent
exposure to a pulse-modulated microwave field on core body temperature, plasma ACTH and corticosteroid, and
brain ornithine decarboxylase, Fos and Jun mRNA. Radiation Research, 2001, 155,4: 584–592.

Steenland K, Bray I, Greenland S, Boffetta P, Empirical Bayes adjustments for multiple results in hypothesis-
generating or surveillance studies, Cancer Epidemiol Biomarkers Prev. 2000 Sep, 9:895-903.

Stewart, Sir W., Challis, L., Barclay, L., Barton, M-N., Blakemore, C., Coggon, D., Cox, D., Fellows, J., Repacholi, M.,
Rugg, M., Swerdlow, A., Varma, T. Independent Expert Group on Mobile Phones, Mobile Phones and Health. May
11, 2000. http://www.iegmp.org.uk/ NRPB, Chilton, Didcot, Oxon.

Stolwijk JAJ, Hardy JD, Control of body temperature, In Handbook of Physiology, Section 9, Reactions to
environmental agents, Ed Douglas HK, Bethesda, MD, American Physiological Society, 1977, 45-69.

UK Childhood Cancer Study Investigators (UKCCS) The United Kingdom Childhood Cancer Study: objectives,
materials and methods, British Journal of Cancer (2000) 82, 1073-1102.

Van Leeuwen G M, Lagendijk J J, Van Leersum B J, Zwamborn A P, Hornsleth S N, Kotte A N, Calculation of change
in brain temperatures due to exposure to a mobile phone, Phys. Med. Biol., 1999, 44:2367-2379.

Vaul, J. A. Excell, P. S. The European `CEPHOS' (cellular phone standard) dosimetry project, IEE Conference
Publication 1999 ISSU 461: 65-68.

Vijayalaxmi, Obe G, Controversial cytogenetic observations in mammalian somatic cells exposed to radiofrequency
radiation, Review, Radiat Res. 2004 Nov, 162(5):481-96.

Wainwright P, Thermal effects of radiation from cellular telephones, Phys. Med. Biol., 2000, 45: 2363-2372.
Wang J, Fujiwara O, FDTD computation of temperature rise in the human head for portable telephones, IEEE Trans.
Microwave Theory Tech., 1999, 47:1528-1534.

WHO Environmental Health Criteria (EHC) Monograph 35 on ELF Fields, (1992)

Yioultsis T V, Kosmanis T I, Kosmidou E P, Zygiridis T T, Xenos T D, Tsiboukis T D, A comparative study of the


biological effects of various mobile phone and wireless LAN antennas, IEEE Trans. Magnetics, 2002, 38:777-780.
Appendix A

IEEE ICES SC-4 TRIAGES

1. In Vivo Working Group Triage, 16 screens


2. In Vitro Working Group Triage, 17 screens
3. Epidemiology Working Group, Triage, 13 screens
4. Engineering Working Group Triage, 17 screens
5. Chair Working Group Triage 7 screens
The 5 Triages (listed below) were developed by Martin Meltz and ICES SC-4 colleagues to regularise
and automate the compilation of anonymous peer review results of the literature identified by the
Literature Surveillance Working Group (See also ICES SC-4, page 7).

For Example IEEE SC-4: In Vivo Triage: 16 Screens Are Summarized Below:
Questions on the form include the date the paper is received by reviewer, the paper accession
number, and the reviewer code number. The rating scale on each criteria is: high 3; acceptable 2;
low 1; not acceptable= 0.
Briefly, there 9 criteria rating scales:
A. Reviewer rating for clarity of statement if objective, specific goals and/or hypothesis. 0-3
B. Reviewer rating for completeness of description of biological system exposed. 0-3
C. Reviewer rating for completeness of description of the time/duration of exposure and biological
response. 0-3
D. Reviewer rating for completeness of description of the organs systems studied and endpoints
examined. 0-3
E. Reviewer rating for confidence in the methodologies employed. 0-3
F. Reviewer rating for confidence in the completeness of the data reporting and the merit if the
data analysis. 0-3
G. Reviewer rating for confidence in the conclusion of the authors. 0-3 (give a statement of the author’s conclusions)

H. Conclusion Presentation (Does the title and the abstract accurately reflect what was measured?).
0-3
I. Overall technical merit rating of the in vivo bioeffect review. 5-1
I request that the chair have this paper reviewed by the statistics WG: yes/no
Overall rating, if it is 2 or 1 your are requested to give reasons for the score.
Category score A..B..C..D..E..F..G.. Overall Score. 0-3
Study will become important to the standards setting process if independently replicated. yes/no
Relevance for human standard setting. 0-3

All the 5 Working Group Triages are attached below for your use:

1. In Vivo, 2.In Vitro, 3. Epidemiology, 4. Engineering, 5. WG Chair.

[1] The IEEE is today the world’s largest technical professional society, with more than 365,000 members in over 150
countries.

[2] IEEE Std C95.6-2002, “IEEE Standard for Safety Levels with Respect to Human Exposure to Electromagnetic Fields, 0 to 3
kHz,” and IEEE Std C95.1-1991/1999, “IEEE Standard for Safety Levels with Respect to Human Exposure to Radio Frequency
Electromagnetic Fields, 3 kHz to 300 GHz.”

[3]Detailed guidelines on the conduct of high quality laboratory research can be found in the
good laboratory practice guidance of the US Food and Drug Administration (FDA, 1993) and in
the specifications of the US National Toxicology Program (NTP, 1992).

[4]General uncertainty in measurement is defined as: The estimated amount by which the
observed (measured) or calculated value of a quantity may depart from the true value.
Ordinarily taken as the sum of the random errors at the 95% confidence level and the estimated
upper limit of the systematic error. NOTE—The uncertainty is often expressed as the average
deviation, the probable error, or the standard deviation. IEEE Std 1528-2003
Use of magnetic field dosemeters for occupational exposure assessment

Philip Chadwick, EMF Dosimetry Handbook Project Co-ordinator

This summary should be read in conjunction with John Swanson's chapter on Power frequency
EMF measurements

In 1997 the UK Health and Safety Executive


published Contract Research Report CRR
1997:128 on the suitability and use of an
EMDEX II magnetic field dosemeter for the
assessment of the magnetic field exposure of
induction heating workers.

This is a report of good general usefulness as


the induction heating environment can give
rise to some of the highest occupational
exposures to magnetic fields at a range of
frequencies, from les than 10 Hz to greater
than 10 kHz. The waveforms of the magnetic
fields can be complex, with a high harmonic
content, and exposures can be transitory and
very dependent on the details of working
practices.

This brief summary is provided for the EMF


Dosimetry Handbook by permission of HSE

The HSE report contains some very useful general conclusions about the use of personal
dosemeters in industry:
• Dosemeters will give better estimates of true operator exposure than can be derived from
spot measurements and observations of working practices

• The use of dosemeters on multiple workers or for prolonged periods will minimise the
perturbation of working practices

• Exposures which occur infrequently and which would not be predicted from routine
observation may be captured by a dosemeter

• Variations in emitted magnetic flux density during a long production cycle will be
accounted for

Several aspects of magnetic field dosimetry in industrial environments were emphasised:

• The exposure environment must be characterised, to allow comparison with exposure


standards and to ensure that the magnetic flux densities encountered are within the
frequency and dynamic ranges of the dosemeter

• The dosemeter data should be examined carefully for evidence of bad data and the range
of recorded exposures compared with those predicted from spot measurements.
Discrepancies should be investigated further

• The functionality of the dosemeter should be checked immediately before it is issued and
after it has been returned
• It is a feature of many induction heating environments that the magnetic flux density
varies across the dimension of the human body. It must be recognised that the position at
which the dosemeter is worn may affect the maximum magnetic flux density that it
records.

Particular conclusions related to practical aspects of the use of dosemeters included:

• Most of the operators wore dosemeters for one or two days only, and it is conceivable that
their working practices may have been affected, consciously or otherwise.

• Some of the dosemeter data are not consistent with the spot measurements of magnetic
flux density made at each site and it is likely that the magnetic field emissions vary
markedly between different stages in the production process

• Reliable personal dosimetry would require the wearing of a dosemeter for many days.

• The exposure of induction heater operators is characterised by transient exposure maxima


as they approach the coils and the dosemeter should be set to the highest sample rate
possible to avoid problems of aliasing. Where aliasing does occur, the amplitude of the
maxima is likely to be underestimated but warning of this is given by the presence of the
bad data flag in the key byte of the EMDEX data file. Workers' exposure records should be
checked for the presence of this flag and if it is present, the data treated with caution.

• Sample rate is often limited by the available memory of the dosemeter, and memory
constraints also preclude routine measurement of the separate vector components of
magnetic flux density.

• It is vital that the battery not be allowed to run down before the data file is transferred to a
computer or the exposure information will be lost.
• If the dosemeter is dropped, one or more sensing coils may be damaged. This may not be
obvious to the user but could seriously affect the validity of the data gathered. It is
important that the integrity of each data channel is checked with a test source before and
after each exposure assessment is made.

Dosemeter record from operator of induction furnace


Power frequency EMF measurements

John Swanson, Technology. & Science Labs., National Grid Transco Research. & Development. Centre,
Leatherhead, UK

1 General principles
Measurements of electric and magnetic fields can be made for various purposes. Depending on the
purpose, different techniques will be appropriate. There is no single “correct” procedure; the correct
procedure will depend on the reason the measurements are being made and the use that will be made
of the results.

When embarking on a measurement programme, therefore, it is important first to identify clearly the
objective. This objective must be used to decide on the procedures to be followed. This chapter
discusses the choices and issues involved, and gives suggestions for appropriate procedures, but is not a
substitute for intelligent, informed choice on the part of the person specifying or performing the
measurement.

The main international standard relating to measurements at power frequencies is a 1998 IEC standard
IEC-61786, “Measurement of low-frequency magnetic and electric fields with regard to exposure of
human beings – Special requirements for instruments and guidance for measurements”. The provisions
of this standard are referred to where appropriate in the following material.

2 Types of measurement
The following are some of the main reasons for performing measurements. This is not, however, an
exhaustive list.

Simple characterisation of the field in a building

The objective is to characterise the field in a home, a work location, or similar, by a single number.
Depending on resources and time available, this can be done by a single measurement at a single
location; a single measurement at each of a series of locations; a sequence of measurements over time
at a single location; or a sequence of measurements over time at multiple locations.
Identification of sources

The objective is to make measurements of the field at a location, specifically how it varies over space or
time, to enable the source of the field to be identified.

Comprehensive characterisation of the field in a building

The objective is to collect more data on the field in a building than just a single number, so as to permit
extraction of desired information at a later date.

Characterisation of sources

The objective is to perform measurements that relate to a particular source of field (eg a power line, an
item of equipment) rather than to the field in a particular place, so as to characterise that source.

Personal exposure

The objective is to measure the exposure of a person over a period of time during which they are
exposed to fields from various sources.

Compliance with exposure limits

The objective is to assess whether a given set of EMF exposure limits are exceeded. This can be done
either by assessing the fields in an area in which people will be present, or by monitoring the exposure
of the people, or a combination.

Laboratory measurements

The objective is to measure the field within experimental apparatus in a laboratory, eg the field
produced by coils used to expose an experimental system to magnetic fields.
3 Choice of instrument for magnetic field
measurements

3.1 Sensor technology

There are three technologies for measuring magnetic fields: fluxgate magnetometers, Hall effect
devices, and search coils.

3.1.1. Fluxgate magnetometers

These comprise a ferromagnetic core which is driven in and out of magnetic saturation in opposite
directions by a high-frequency current generated within the instrument. The addition of an external
field creates an asymmetry, and the instrument uses this asymmetry to measure the field. Fluxgate
magnetometers are sensitive to all external fields up to a certain frequency, static fields as well as
alternating fields. The static and alternating components of the field can be separated and recorded
separately by the instrument circuits.

Fluxgate magnetometers are usually the preferred choice when it is desired to measure static fields as
well as alternating fields. The fluxgate sensor itself can be made reasonable small and is usually remote
from the bulkier remainder of the instrument, so they can have uses where it is necessary to probe the
field in difficult locations. Other than these specific applications, their use is limited, as they are more
expensive and have higher battery consumption than search coil instruments. Care should be taken if
using them in laboratory settings, as they produce a finite high-frequency field themselves as part of
their operation which could perturb an experimental setup.

3.1.2. Hall effect devices

These devices pass a current through a suitable semiconductor, and detect the field via the voltage
produced across the element perpendicular to the current. Like fluxgate magnetometers, they detect
the instantaneous total field, the sum of the static and any alternating fields. They are often used in
other applications because they can measure high fields at the Tesla level and above. However, they are
usually not very sensitive to low fields and suffer from zero-point drift, and the probes are often fragile,
which means they have few applications in EMF dosimetry.
3.1.3. Search coils

Search coils are simply coils of wire. An alternating magnetic field induces a voltage in the coil. Except
for the specialised instances where fluxgate magnetometers are appropriate, search coils are the
preferred technology for EMF dosimetry where it is not desired to measure the static field as well.

Characteristics of search-coil instruments

• Size of coil

To obtain a large enough signal, search-coil instruments tend either to have relatively large coils (10-20
cm), or small coils (less than 1 cm) with ferrous cores. Larger coils are perfectly acceptable as long as
the field itself does not vary significantly over the area of the coil. Thus they will often be acceptable
under power lines or in the middle of rooms, but will be less appropriate close to conductors or to
equipment. IEC specifies that the coil should have area 0.01 m2 or less.

It is sometimes recommended to use large coils so as to measure the average field over an area
comparable to part of the human body. Usually, however, where the field varies over space this rapidly,
it is preferable to measure the variation of the field with a small probe and then to apply any desired
averaging to the results of those measurements.

Small, ferrous-cored coils are often preferred and produce a smaller, more versatile instrument. The
one disadvantage is that the ferrous core produces non-linearities at high fields, greater than 1 mT. For
most purposes this is irrelevant as fields are rarely that high, but care should be exercised if high fields
are to be measured.

• Number of coils

Each coil measures the component of field in one direction. Instruments have either one or three
orthogonal coils. The choice depends on the purpose of the measurement. When identifying and
investigating different sources of field, a single coil can be useful, as the extra information on the
direction of the field can assist in identifying the source. For most other applications, however, three
coils are more useful.

The resultant field from multiple sources at the same frequency always traces out an ellipse. There are
two alternative ways of quantifying an elliptically polarised field, shown in figure 1.

Figure 1 Alternative measures of an elliptically polarised field. 1:

“maximum” field. 2: “resultant” field

Vector 1 gives the rms of the field along the major axis of the ellipse, which is the direction of maximum
field. This is known as the “maximum” of the field. Vector 2 gives the actual rms of the field, known as
the “resultant field”.

The “maximum” field, vector 1, can be measured by rotating a single coil until the maximum value is
obtained. The “resultant” field, vector 2, can be measured using three orthogonal coils. The rms of the
signals from the three coils are combined as root-sum-of-squares to give the resultant field. This applies
regardless of the orientation of the coils relative to the field.
Three orthogonal air-cored coils can be arranged so that, to a good approximation, their centres are
coincident and therefore they measure the field at the same point in space. Figure 2a shows an
example. Ferrous-cored coils are nearly always physically separate. This still gives satisfactory results as
long as the field does not vary over the distance scale of the separation of the coils. As this can be small
(of the order of a cm) this is usually acceptable. Figure 2b shows three such ferrous-cored coils.

Figure 2 Examples of instruments using

(a) three orthogonal coils arranged to be centred on a (b) three separate orthogonal coils
single point

If only a single-axis instrument is available, the resultant (total) field can be measured by performing
three successive orthogonal measurements. This can have limited accuracy, however, if the field varies
over the time taken to perform these measurements. It can also be difficult to locate the sensor at the
same point each time; if accuracy is required, a jig to locate the instrument should be used.

3.2 Type of display

Increasingly, most displays are digital. The main occasion when an analogue display is more useful is
when investigating a source, when variations in the field as the meter is moved need to be readily
observed, but even here a digital display can be used.
3.3 Range and resolution

Average background fields in homes (ie fields in the general volume of the home, not close to
equipment) range from a few tens of nanotesla in some European countries to over a hundred
nanotesla in North America. If performing measurements in a high-field country, and particularly if the
emphasis of the measurements is on high fields in that country (eg on identifying homes with fields
greater than 200 or 400 nT), a resolution of 10 nT will be adequate. In many other instances for
measurements in homes, however, a resolution of 1 nT is desirable.

The desired range of the instrument depends on the maximum field likely to be encountered.
Background fields in homes are rarely greater than 1 µT; fields in industrial settings or near domestic
equipment can be 1 mT or more. Occupational exposure limits are typically of the order of a mT, so a
range of at least this is necessary to assess compliance with such limits.

3.4 Storage abilities

If using measurements of field to investigate a source, no storage ability is necessary; the changes in
field as the instrument is moved are observed in real time. For some other purposes, storage within the
instrument is not needed; the reading can be written down or entered into a database. For many
purposes, however, it is helpful or necessary to store results within the instrument. The choice of
measurement interval and total duration is a compromise limited by the total storage capacity available
in the instrument.

3.5 Frequency response

The output of a search coil is proportional to frequency. Most instruments adapt this frequency
response within the instrument to give a final output which is either broadly flat between a lower cut-off
(typically 20-30 Hz) and an upper cut-off (typically 500 Hz – 5 kHz), or is sensitive only to the power
frequency, 50 or 60 Hz.

The choice of frequency response is determined by the characteristics of the field to be measured and
the purpose of the measurements. In many cases, the source of the field is the power system. Then,
the main component of field will be at 50 or 60 Hz, with smaller harmonics at multiples of this,
principally three times or “third harmonic”, 150 or 180 Hz. Most flat responses will be adequate for
this. However, if there is particular interest in harmonics, or if the sources is such that higher harmonics
(eg from some ac-dc power conversion processes) or lower frequencies (eg from some railways at one-
third the power frequency) are present, then an instrument with a particular frequency response may
be appropriate.
Note that search coils produce a signal when rotated in a static field. To avoid this being erroneously
recorded as a power-frequency signal, many instruments have filters to reduce the response at low
frequencies. If an instrument sensitive to lower frequencies is used (eg because railways are to be
measured), extra care must be taken to avoid this interference.

Exactly how flat a frequency response is depends on the sophistication of the instrument. However,
achieving a flatter frequency response also tends to increase the power consumption and the weight of
the instrument. Therefore, instruments designed to be small, light and with long battery life may have
poorer frequency responses. This is part of the choice that has to be made, but in many instances, the
flatness of the frequency response will not affect the result much.

3.6 rms and other measures

Increasingly, instruments are made with true rms detection of the field, and this is regarded as
preferable for most purposes.

Alternatives are rectified average and peak detection. There may be occasions when each of these is
desirable. However, in general, these are a legacy of less developed instrument technologies, and
should be avoided. They cause particular problems when harmonics are present.

3.7 Accuracy

The accuracy required of an instrument should, in principle, be determined by the measurement


purpose. However, IEC specifies an overall uncertainty in the measurement of a uniform field of ±(10%
of reading +20 nT), and most commercial instruments can be regarded as accurate to at least this 10%
value.

3.8 Other factors

IEC specifies that an instrument should function over a temperature range of 0 to 45 °C and from 5% to
95% relative humidity.

If measurements are to be done in the high electric fields close to high-voltage equipment, the
instrument should be immune to these fields. IEC specifies that fields of 20 kV m-1 should produce no
more than 20 nT change in the reading. Most instruments will have satisfactory EMC performance; IEC
specifies immunity to 3 V m-1 from 150 kHz to 1 GHz assessed in accordance with standard IEC 61000
(IEC 1995).

4 Choice of instruments for electric-field


measurements
Most practical electric-field measurements are made by measuring the voltage (or current) between
two parallel plates perpendicular to the electric field. Other technologies are available, but have few
advantages. Such a meter is a “free body” meter; the alternative is a “ground reference” meter, which
measures the voltage between a single electrode and ground, but this is less common.

Parallel-plate sensors for electric fields tend to be physically larger than magnetic-field sensors.
Therefore, electric-field instruments are more often single-axis. However, three-axis meters are
available, with the three separate parallel-plate assemblies either on three orthogonal faces of a cube or
spread over the surface of a sphere.

One common measurement scenario is near or under power lines. In this instance, the electric field is
essentially vertical for the first metre or two at least above ground level. Therefore, in this instance, a
single-axis meter, aligned to measure the vertical field, is perfectly satisfactory.

Many of the issues discussed under magnetic fields apply to measurements of electric fields as well:
frequency response, storage, type of display, etc. There is, however, one major issue with electric fields
that does not apply to magnetic fields.

4.1 Perturbed and unperturbed electric fields

Electric fields are perturbed by any conducting object. This includes the person making the
measurement and any conducting supports for the sensor.
Electric-field measurements should be either of the unperturbed field or of the perturbed field and it
should be clear which is being measured. Measurements made by a meter worn by a person are
unavoidably of the perturbed field. In most other cases, however, it is preferable to attempt to measure
the unperturbed field. This can sometimes be an ambiguous definition, where the field is perturbed by
an object not connected with the measurement process. The usual objective is to measure the field that
would be present in the absence of the person, but including any perturbation caused by fixed objects
that would still be present when the person is not.

To obtain an unperturbed measurement, the instrument sensors must be supported in a way that does
not perturb the field, and the person performing the measurement must be sufficiently distant.

The instrument can be supported either on an insulating pole held horizontally, or on a vertical
insulating support (a tripod or similar). In both cases, if readings in real time (as opposed to stored and
read only later) are desired, the sensors are connected to the rest of the instrument by a radio link or a
fibre-optic, or the instrument is self-contained with a display large enough to be read from a distance.

A vertical tripod or similar support is often mechanically easier and makes it easier for the operator to
be distant. However, in humid conditions, moisture can settle on both a tripod or a pole, making them
less good insulators and perturbing the reading. As most electric fields are vertical near ground level,
the vertical support, which is parallel to the field lines, leads to greater perturbation than the horizontal
pole, which runs perpendicular to the field lines, along an equipotential, and therefore is less sensitive
to small amounts of moisture. This is a reason for preferring horizontal poles when some degree of
perturbation is regarded as inevitable.

The person making the measurements should be 2 m distant from the instrument to ensure that
perturbation is negligible. Measurements at 1 m may be acceptable but less accurate. Hand-held
measurements should be avoided.

The perturbation caused by the person making the measurement is, simplistically, to increase the field
towards the top of their body, and to reduce it towards the bottom of their body. There is therefore a
height of measurement, around 1.5 m above ground and therefore higher than the 1 m above ground
which is recommended as a suitable measurement height for other reasons, where the perturbing effect
of the person is neutral. This means in practice, the distance of the person from the instrument may not
be so critical. However, this should not be relied on, and for good measurements, the person should
always be 2 m distant.

Measurements of the perturbed electric field are made by attaching a meter to a person’s body.
However, the reading obtained is extremely dependent on the amount of perturbation, which in turn is
extremely dependent on the exact location of the meter on the person’s body. For certain simple
geometries and for certain locations on the body (eg the top of the head of a person standing upright in
a vertical field, where the factor is approximately 20) it is possible to calculate the conversion factor
from the perturbed to the unperturbed field. In most situations, however, this conversion factor cannot
be calculated, and varies as the person moves anyway. Therefore, such perturbed measurements have
limited use, and should never be compared directly to unperturbed fields.

The need to measure unperturbed fields undoubtedly adds considerably to the effort and cost of making
electric-field measurements, but in most situations, it is the unperturbed rather than the perturbed field
which is the most useful.

5 Calibrations
It is necessary to be sure that any instrument used is giving an acceptably accurate reading; this is
achieved by calibration. IEC requires the overall uncertainty of the calibration process to be no greater
than ±(5% +10 nT).

Most instruments will be supplied calibrated by the manufacturer. Periodic re-calibration will be
needed. This can be performed either by the user or by sending the meter to a specialist calibration
service or to the manufacturer. Calibration systems are not simple, and probably only users with
multiple instruments will be able to justify maintaining their own calibration systems.

Magnetic fields are calibrated by placing them in a known field produced by a coil system of known
geometry. It is common to refer to these coils as a Helmholtz arrangement. In fact, a Helmholtz pair,
two equal coaxial circular coils separated axially by one radius, is a legacy of the time when fields had to
be calculated analytically; its main advantage is the ability to express analytically the field and its first
few derivatives at the centre. There are alternative systems which are more efficient at producing a
uniform field (for a review, see Kirschvink 1992). The simplest system is a single square coil; the best
results are achieved from three or four coaxial coils, of carefully chosen numbers of turns and axial
separation. The size of the coil is determined by the requirement that the field at the centre must be
reasonably uniform over the area of the search coil in the instrument being calibrated. IEC requires less
than 1% (or 1.5% for larger probes under certain circumstances) departure of the field anywhere over
the area of the search coil from the field at the centre. For a 0.1 m diameter search coil, this could be
achieved by a single square coil of side 1 m, or a Merrit four-coil system of side 0.4 m.

IEC requires the field in the calibration coil to be known to ±3%. In most practical situations, the
geometry of the coil system is known to rather better than this, so the limit on the accuracy of the
calibration system is the accuracy of the measurement of the current flowing in the coils.

Electric fields are calibrated by placing them between two parallel plates across which a known voltage
is applied. The uniformity of the field is improved by grading rings round the edges of the system, but to
avoid perturbation, the system of plates should still be placed a minimum distance away from objects
that might perturb the field. The limit on the size of the plates is then proximity effects between the
meter and the plates. IEC specifies that a meter of maximum dimension 0.23 m requires plates of 1.5 m
square separated by 0.75 m.

Calibration at high electric fields may involve applying voltages to the calibration plates sufficiently high
such that corona becomes an issue and corona guard rings become necessary.

For both electric and magnetic fields, when a calibration is performed, there are three alternatives. One
is simple to accept the instrument as passing the calibration provided it is within a certain margin of the
correct field. A second is to adjust the instrument so it reads the correct value, and a third is to record a
correction factor which should be applied to any readings taken to give the correct reading. IEC allows
all three, and the choice is determined by the ease of adjusting the meter and the accuracy required.
The option of recording and subsequently applying a correction factor gives the highest accuracy, but is
best avoided in many practical measurement situations where it increases the potential for mistakes.

All calibrations should be performed as part of a quality-controlled calibration system with traceable
records. Exactly how this is done and what is required depends on the quality regime in operation.
Calibrations should be performed at set intervals. It is common to make this yearly, but if documented
experience shows that the meter retains its calibration acceptably for longer than this, as will often be
the case with commercial instruments, then a longer calibration interval is acceptable.

6 Sample measurement procedures


This chapter has emphasised that there are no universal rights and wrongs in EMF measurements; the
choices made depend on the purpose of the measurements. The following are therefore suggestions for
the particular scenarios identified at the start of this chapter and should not be regarded as definitive.

• Simple characterisation of the field in a building

The objective is to characterise the field in a home, a work location, or similar, by a single number.
Depending on resources and time available, this can be done by a single measurement at a single
location; a single measurement at each of a series of locations; a sequence of measurements over time
at a single location; or a sequence of measurements over time at multiple locations.

Probable measurement procedure: use a battery-powered instrument with logging facilities and three
ferrous-cored coils. Leave it in a standard location for 24 hours or longer. The location should be clearly
specified in the study protocol, eg “1 m above floor level at the middle of the side of the child’s bed”,
and should be distant (at least 1 m) from any items of electrical equipment. This usually means the
centre of a room, or as near it as is compatible with the occupants’ use of the home. Standardisation
may be improved by providing a stand to hold the meter. The logging interval will be determined by the
need for the logging capacity to last the required time. An instrument with 1 nT resolution will be
preferable in homes but 10 nT resolution may be acceptable.

Similar measurements for electric fields are possible but will be more problematical because of
perturbations by people walking near to the meter.

• Identification of sources

The objective is to make measurements of the field at a location, specifically how it varies over space or
time, to enable the source of the field to be identified.
Probable measurement procedure: use a hand-held, battery-powered instrument with a clear read out
(analogue will be easier but digital is acceptable as well). Walk round the location observing the reading
so as to identify areas of high field, and track these to their source. It may be helpful to have a single-
axis readout as the direction of the field helps determine the direction of the current producing it.

• Comprehensive characterisation of the field in a building

The objective is to collect more data on the field in a building than just a single number, so as to permit
extraction of desired information at a later date.

Probable measurement procedure: use several instruments to measure further characteristics of the
field. For example: to characterise the field at more than one location over a house, one approach is to
leave several instruments logging for 24 hours at locations spread over the house. Another approach is
to measure a profile of field (using a single instrument but making the measurements as close together
as possible) from one side of the house to the other.

• Characterisation of sources

The objective is to perform measurements that relate to a particular source of field (eg a power line, an
item of equipment) rather than to the field in a particular place, so as to characterise that source.

Probable measurement procedure to characterise a power line: use a three-coil magnetic field meter.
The size of the coils is less important. Choose one or a few standardised locations, eg directly under the
centreline of the power line at mid-span, and possible some standard distances perpendicular to the
line, eg 10 m, 25 m, 50 m. Perform measurements at 1 m above ground. For electric fields, use a single-
axis meter, aligned to measure the vertical field, held on a horizontal insulating pole at 1 m above
ground, or a clean, dry vertical insulating stand. If extra detail is desired, measure the field at more than
one height: ground level, 1 m and 2 m above ground. Measurements at one point in time obviously give
the field only at that time. If the load on the line is available, the measured field at one load can be
scaled to other loads; or the measurements can be logging measurements over an extended period of
time.
• Personal exposure

The objective is to measure the exposure of a person over a period of time during which they are
exposed to fields from various sources.

Probable measurement procedure: use a small three-axis meter. The priority is to prolong battery life
and to reduce the bulk, so factors such as the flatness of the frequency response may be sacrificed. Get
the subjects to wear it at a standard position on their body, possibly in a pouch on a belt, or for children,
in a child-friendly back-pack or similar. Preferably arrange for any display to be blanked to reduce the
temptation for the subject to experiment with approaching high-field sources. Give clear instructions on
what to do at night time, ie where to place the meter when it is taken off.

There are some very specific issues related to personal exposure measurement (dosimetry) for the
assessment of occupational exposures to magnetic fields. They are discussed here.
References
Diplacido J, Shih C H and ware B J 1978 Analysis of the proximity effects in electric field measurements
IEEE Transactions on Power Apparatus and Systems PAS-97 2167-2177

IEC 1995 Electromagnetic compatibility (EMC) IEC 61000

IEC 1998 Measurement of low-frequency magnetic and electric fields with regard to exposure of human
beings – special requirements for instruments and guidance for measurements IEC 61786

Kirschvink J L 1992 Uniform magnetic fields and double-wrapped coil systems: improved techniques for
the design of bioelectromagnetic experiments Bioelectromagnetics 13 401-411
Guidelines for the RF exposure assessment of metallic
implants
This document available as a pdf here

November 2006

Dr Vitas Anderson Dr Robert McIntosh

L3, 170 Pacific Hwy 400 Burwood Road


St Leonards (Sydney) NSW 2065 Hawthorn (Melbourne) VIC 3122
AUSTRALIA AUSTRALIA
Vitas.Anderson@kordia.com.au robert.l.mcintosh@team.telstra.com
Abstract

This chapter of the EMF Dosimetry Handbook provides guidance for assessing whether persons
bearing metallic implants inside their bodies should be restricted from exposure to the upper tier
limits of the radiofrequency (RF) safety guidelines (1998) published by the International
Commission for Non-ionising Radiation Protection (ICNIRP) and the C95.1 (2005) standard of
the Institute of Electrical and Electronic Engineers (IEEE). The recommendations presented here
are based on original research by the authors, investigations of specific implant cases by the
Telstra Research Laboratories in Melbourne, Australia and various publications in the scientific
literature. Wherever possible, rules-of-thumb have been developed to provide simple and
practical ways for assessing implants and for some external body worn metallic objects.
Nonetheless, there remain some assessment scenarios that will still require detailed analysis.
Document history

Release
Date issued Details of revision
number
First released issue from Anderson and McIntosh,
R1 21 Nov 2006
reviewed by the ACRBR

Acknowledgements

This material is based on research sponsored by the Air Force Research Laboratory, under
agreement number FA4869-06-1-0115. The U.S. Government is authorized to reproduce and
distribute reprints for Governmental purposes notwithstanding any copyright notation thereon.

The authors also acknowledge the following organisations for their sponsorship and support of
the research and literature review that was required for the drafting this chapter:

• Telstra Corporation Ltd, Australia


• Kordia Pty Ltd, Australia ( http://www.kordiasolutions.com/ )

Lastly, the authors thank Mr Raymond McKenzie for his helpful reviews of this chapter on
behalf of the Australian Centre for Radiofrequency Bioeffects Research ( www.acrbr.org.au ).
Index

1 Background 4
1.1 Types of metal implants in the body 4
1.2 RF heating and human exposure limits 4
1.3 Safety targets for RF tissue temperature increases 4
1.4 RF safety assessments of metal implants 5
2 RF and thermal factors affecting implant assessments 6
2.1 RF factors 6
2.1.1 RF absorption in the body 6
2.1.2 RF absorption around the metal implant 6
2.2 Thermal factors 7
3 Canonical studies of implant rods and infinite plates 9
3.1 Introduction 9
3.2 Canonical modelling of plane waves travelling through layered tissues 9
3.3 Canonical modelling of plane waves reflected off metallic planar boundaries 13
3.4 Canonical modelling of rod implants in infinite medium 15
3.4.1 General approach 15
3.4.2 Use of VAR instead of SAR 15
3.4.3 Model setup 16
3.4.4 Peak VAR vs. frequency for E polarization exposure of a 40 mm rod in infinite bone
18
3.4.5 Peak VAR variation with rod length 20
3.4.6 Peak VAR variation with shape of rod tip 22
3.4.7 Peak VAR variation with field orientation 23
3.4.8 Peak VAR variation with dielectric value of medium 23
4 Specific implant assessments 25
4.1 Linear implants (e.g. pins, rods, and long narrow plates) 26
4.2 Screws 27
4.3 Arterial stents 27
4.4 Wide plates 28
4.5 Pacemakers 28
4.6 Loops 29
4.7 Cochlear implant systems and auditory brainstem implant (ABI) systems 30
4.8 Spectacles 31
4.9 Jewellery 31
4.10 Tooth fillings, caps, and orthodontic braces and plates 31
4.11 Implanted retinal stimulators 32
4.12 Implanted radiators 32
4.13 Spinal fusion systems and cervical fixation devices in MRI 33
5 General rules-of-thumb and observations for implant assessments 34
5.1 General observations 34
5.2 Thermal effects 34
5.3 Rods and other linear objects 34
5.4 Screws 35
5.5 Arterial stents 35
5.6 Wide Plates 35
5.7 Pacemakers 36
5.8 Loops 36
5.9 Cochlear Implant Systems 36
5.10 Spectacles 36
5.11 Jewellery 36
5.12 Tooth fillings, caps, and orthodontic braces and plates 36
5.13 Shrapnel and shotgun pellets 36
6 References 37
1Background
1.1Types of metal implants in the body
Many people carry pieces of metal implanted within their bodies, which vary in their origin.
These metal implants can for example be unwanted remnants of shrapnel, or more commonly,
screws, rods, wires, pins or plates that are implanted by orthopaedic surgeons to repair broken
bones and worn joints. Other metallic implant types include arterial stents and implanted
electronic devices such as cardiac pacemakers and cochlear implants. External body-worn
metallic objects, such as spectacles, jewellery, and the outer components of the cochlear implant
system, are also considered in this chapter.

1.2RF heating and human exposure limits


Tissue heating is a well established effect of exposure to electromagnetic radiofrequency (RF)
fields due to the absorption of RF power from fields induced inside the body. The common
metric for RF heating is the Specific energy Absorption Rate, or SAR in W/kg, which is simply
related to the internal RF electric field at any point by:

where Eint is the magnitude of the internal electric field (V/m), σ is the electrical conductivity of
the tissue (S/m), and ρ is the mass density of the tissue (kg/m³).

Metallic implants can sometimes concentrate the RF heating effect around them by the way they
scatter the incident RF field. This possibility has been recognised in various RF safety guidelines
and standards (ICNIRP, 1998, IEEE, 2005, ARPANSA, 2002) which caution that the potential
for exceeding allowable exposure limits for localised SAR around metal implants should be
assessed for persons exposed up to upper tier limits, i.e. the occupational limits in the ICNIRP
Guidelines for electromagnetic exposures (1998) and the controlled environment limits in the
IEEE C95.1 RF safety Standard (2006). The lower tier limits prescribed for general public
exposures in these documents incorporate substantial additional safety margins that are generally
regarded as providing sufficient protection for implant RF field enhancements.

The localised SAR limits in the ICNIRP Guidelines (1998) and the IEEE C95.1 standard (2006)
are assessed by averaging the point SAR over a mass of 10 g, usually in the shape of a cube, in
recognition of the thermal diffusion properties of tissues. Different upper tier limits apply to
different parts of the body as indicated in Table 1 below:
Table 1 Upper tier limits for localised SAR in the ICNIRP Guidelines (1998) and the IEEE
C95.1 (2006) standard for human exposure to RF fields.

ICNIRP Guidelines (1998) IEEE C95.1 standard (2006)

Head and torso 10 W/kg Head (except pinna), torso, 10 W/kg


upper arms, elbows, thighs
and knees

Arms and legs 20 W/kg The pinnas and limbs distal 20 W/kg
to the elbows and knees

The upper tier RF limits for exposure to ambient electric (E) and magnetic (H) fields in both the
ICNIRP Guidelines (1998) and the IEEE C95.1 Standard (2006) have been primarily formulated
to restrict whole body average (WBA) SAR absorption to less than 0.4 W/kg for standing
children and adults exposed to uniform plane wave fields. There is a general presumption that
these E and H-field limits will also ensure that the localised 10/20 W/kg SAR limits are not
exceeded for all circumstances.

1.3Safety targets for RF tissue temperature increases


Tissue temperature rise is a more fundamental indicator of RF heating hazard than localised SAR
as it includes the effect of the body’s capacity to dissipate RF heating. A conservative safety
target is to restrict RF tissue heating to no more than 1°C in the head and torso (ICNIRP 1998).
For other parts of the body that are more tolerant of temperature increases and have less critical
functions (i.e., where the higher 20 W/kg SAR limit of the ICNIRP Guidelines or the IEEE
C95.1 standard is applied), then a target temperature rise of 2°C would seem appropriate.

1.4RF safety assessments of metal implants


Detailed assessments of SAR concentrations and temperature rises around metal implants
generally require complex analyses and specialised skills that are beyond the reasonable
capabilities and resources of the great majority of affected persons and organisations.
Furthermore, due to enormous diversity in the size, shape and location of metal implants, as well
as the many different possible scenarios for RF exposure, it has been difficult to make
generalizations from one particular implant assessment to another.

As a result, and despite warnings from RF safety standards and guidelines, most persons bearing
personal metal objects and working in high RF fields have not been assessed for the potentially
adverse RF heating of those metal objects. To address this lack, this chapter offers practical and
accessible guidelines, or rules-of-thumb, for making such assessments for as many implant
scenarios as possible.

In devising these rules-of-thumb, the authors have drawn upon many sources including their own
research, previous assessments conducted by the Telstra Research Laboratories in Melbourne,
Australia, as well as published scientific papers on this topic, and have assumed that:
• The rules-of-thumb only apply to persons who are not exposed above the upper tier limits
of the ICNIRP Guidelines (1998) or the IEEE C95.1 standard (2006).
• A metallic implant in the head or torso can be considered safe if the localised SAR
(averaged over a 10 g cube) in tissue around the metal object does not exceed 10 W/kg or
if the RF induced temperature rise in tissue around the metal object does not exceed 1°C.
• A metallic implant in the limbs and pinna can be considered safe if the localised SAR
(averaged over a 10 g cube) in tissue around the metal object does not exceed 20 W/kg or
if the RF induced temperature rise in tissue around the metal object does not exceed 2°C.
• The rules should hold for all orientations of the metal objects with respect to the incident
field since an individual will generally move about in the field.
• There is no RF heating of the implant itself, i.e. the implant is assumed to be a perfect
electrical conductor and the only RF heating occurs in the tissue around the implant.

As this topic is still in a relatively early stage of development, it should be expected that at least
some of the rules-of-thumb offered in this chapter will require further amendment as more
research is accumulated. Nonetheless, it is hoped that the publication of even imperfect rules-of-
thumb will at least be a useful starting point and impetus for the development of better
guidelines, and preferable to making no assessments at all.
2RF and thermal factors affecting implant
assessments
2.1RF factors
2.1.1RF absorption in the body
The factors influencing the electromagnetic interaction between metal implants and the RF
exposure field external to the body are varied and complex. Firstly, one must consider the
general interaction of the body with the RF exposure field, as this affects the local incident field
exposure around the implant. The main factors that affect the efficiency and distribution of RF
absorption in the body are:

1. The frequency of the RF source


2. The polarisation of the incident RF field with respect to the body and its parts
3. The position of the RF source to the body which may lead to partial body exposures and
near field coupling effects
4. The size, shape and grounding of the body
5. The dielectric properties (permittivity and conductivity) of body tissues

The dependence of SAR distribution on the RF exposure frequency offers a number of avenues
for developing rules-of-thumb. For certain frequency ranges, the internal SAR may be too low to
exceed peak allowable limits in particular parts of the body, even with RF field concentrations
around the implants. Thus, it would be useful to identify those frequency ranges where metallic
implant assessments are not necessary in all or parts of the body. Conversely, in certain
frequency ranges, there may be parts of the body where localised SAR levels are relatively high,
and where the additional SAR enhancement effect of the implant is more critical. Frequency
ranges that are particularly worth noting include:

1. Frequencies above ~4-6 GHz


At frequencies above this range the small skin depth of absorption, δ, can provide
effective RF shielding of implants buried in the body. At one skin depth, the point SAR
will diminish by a factor of 0.14 relative to point SAR at the surface.
2. Frequencies up to the MF band (300 kHz – 3 MHz)
In this range RF coupling to the body is weak, and E-field limits in the ICNIRP
Guidelines (1998) and the IEEE C95.1 Standard (2006) are predicated on the more
stringent requirements of protecting against external shock and burns arising from contact
with passively charged conductors.
3. Frequencies in the HF (3–30 MHz) and VHF (30–300 MHz) bands
In this range, whole and partial body resonances occur. Induced RF currents in the ankle
and neck are of particular interest due to the concentration of RF current flows in these
narrowed conduction areas.
2.1.2RF absorption around the metal implant
Having established a base level of RF exposure in the body, the next step is to determine how the
metal object perturbs and possibly concentrates the SAR around it. This should include a
consideration of the following factors:

1. The size of the metal object


2. The shape of the metal object
3. Any gaps in the metal object
4. Location of the metal object within the body
5. Dielectric values of tissues around implanted metal objects
6. Whether the implant traverses local tissue boundaries
7. Orientation of the implant with respect to the local induced in vivo fields.
8. Distribution of the in vivo field around the implant (more important for large implants)

It should be noted that passive metal objects cannot of themselves generate any additional RF
energy in accordance with the thermodynamic law for conservation of energy. However, due to
RF field scattering, they can redistribute the incident RF energy around them, leading to SAR
concentrations at some points and corresponding SAR reductions in other areas.

There are at least four basic mechanisms of SAR enhancement around implants as displayed in
Figure 1:

1. SAR enhancement at the ends of implants, particularly when the long axis is parallel to
the in situ electric field
2. SAR enhancement in gaps of linear implants
3. SAR enhancement in the gaps of broken loops that are cut by changing magnetic flux
density (B)
4. Constructive interference in surface layers with underlying metallic plates

Figure 1 Four basic mechanisms of SAR enhancement around metallic implants

2.2Thermal factors
Thermal factors that can influence the local temperature rise around RF exposed implants
include:

1. The thermal conductivity and physical structure of the implant which influences the
ability of the implant to redistribute temperature variations around it through internal heat
transfer.
2. The size and specific heat capacity of the implant which alters the thermal mass of the
implant, and affects the transient response to heating.
3. The heat transfer environment around the implant including: the thermal conductivity of
surrounding tissues; the micro blood perfusion of surrounding tissues, and; the proximity
of implant to large blood vessels.
4. The proximity of the implant to the body surface, where heat transfer from the skin to the
ambient environment becomes important.

To a good first approximation, heat transfer inside the body can be numerically modeled using
the classic bioheat equation (Pennes, 1948):

where T is the tissue temperature (°C), c is the specific heat capacity (J/kg°C), K is the thermal
conductivity (W/m°C), A0 is the metabolic heat production (W/m³), b is the heat-sink strength
from each tissue volume by blood perfusion (W/m³ °C), and Tb is the temperature of the
perfusing blood. The desired solution for T is obtained when the system reaches steady-state
thermal equilibrium.

Heat transfer at the surface of the body can be modelled as a convective boundary:

where h is the convection coefficient (W/m²°C), Ta is the ambient temperature (°C), qe is the
evaporative heat loss (W/m²) and n is the direction of the unit normal to the surface. The
convection coefficient may include a linearised component for heat radiation.

A favoured method for computational analysis of the bioheat equation in human bodies is the
finite difference (FD) technique whereby complex heterogeneous models of the human body and
implants can be represented by voxels in a regular rectangular mesh. A particular advantage of
this approach is that the finite difference mesh can be made to coincide with the voxel mesh of
an RF model based on the Finite Difference Time Domain (FDTD) technique, thereby allowing
easy transfer of the calculated RF SAR data to the FD thermal analysis. For a detailed example
of this approach see, for example, McIntosh et al. (2005).
3Canonical studies of implant rods and
infinite plates
3.1Introduction
This chapter section is drawn from a project report by the authors (Anderson and McIntosh,
2004) for a study on metallic implants that was sponsored by the Asian Office for Aerospace
Research and Development (AOARD) of the United States Air Force Office of Scientific
Research (AFOSR). It also includes observations gathered from earlier implant modelling for
specific assessments that were conducted at the Telstra Research Laboratories in Melbourne,
Australia.

Results and conclusions about implants are provided from canonical studies of the following
areas:

• Calculations of SAR attenuation in planar multilayer skin/muscle/bone/metal models


exposed to a plane wave; and;
• A canonical assessment of rod implants exposed to a plane wave in infinite medium that
investigated the influence of size, tip shape, rod orientation and the dielectric properties
of the surrounding tissue medium.

3.2Canonical modelling of plane waves travelling through layered tissues


Radios waves are attenuated as they travel through lossy materials such as human tissues. At
high frequencies, above 4-6 GHz, the rate of attenuation with depth is so pronounced for human
exposures that most of the RF power is absorbed at the body surface. At these frequencies,
metallic objects located deeper in the body may be effectively shielded from RF exposures.

The extent of this shielding can be gauged by examining a simple canonical model of a plane
wave travelling through multiple planar layers representing skin, muscle and bone, as
represented in Figure 2 below. This scenario has been modelled in a commercial RF analysis
package, FEKO v4.1 (EMSS, 2003), using Greens functions for planar multilayered substrates.
The final bone layer extends infinitely in the z direction.

Figure 2 Model setup for examination of a plane wave incident on infinite multilayers of
skin, muscle and bone.

Skin and muscle thickness over bone varies in different locations of the body. In some places,
there is effectively no muscle layer at all, e.g. on the forehead and shins. To cover these different
scenarios, the following scenarios were analysed:
• 3 mm skin layer, infinite bone
• 5 mm skin layer, infinite bone
• 7 mm skin layer, infinite bone
• 5 mm skin layer, 10 mm muscle layer, infinite bone
• 5 mm skin layer, 30 mm muscle layer, infinite bone

The models were examined in the frequency range of 1-10 GHz using dielectric values for dry
skin, skeletal muscle and cortical bone from Gabriel (1996) as shown in Table 3.

Table 2 Tissue dielectric values for multilayer models

Frequency Dry skin Skeletal muscle Cortical bone


(GHz)
εr σ (S/m) εr σ (S/m) εr σ (S/m)

1 40.9 0.900 54.8 0.978 12.4 0.156

2 38.6 1.265 53.3 1.454 11.7 0.310

4 36.6 2.340 50.8 3.016 10.5 0.727

6 34.9 3.891 48.2 5.202 9.6 1.203

8 33.2 5.824 45.5 7.798 8.8 1.680

10 31.3 8.014 42.8 10.626 8.1 2.136

Results of the multi tissue layer model analyses are shown in Figures 3 to 7. The curves for each
frequency have been normalised so that the point SAR = 1 at the air skin surface.
Figure 3 Normalised point SAR in a multilayer tissue model (3 mm skin, infinite bone)
exposed to a plane wave over the frequency range 1-10 GHz.
Figure 4 Normalised point SAR in a multilayer tissue model (5 mm skin, infinite bone)
exposed to a plane wave over the frequency range 1-10 GHz.

Figure 5 Normalised point SAR in a multilayer tissue model (7 mm skin, infinite bone)
exposed to a plane wave over the frequency range 1-10 GHz.
Figure 6 Normalised point SAR in a multilayer tissue model (5 mm skin, 10 mm muscle,
infinite bone) exposed to a plane wave over the frequency range 1-10 GHz.
Figure 7 Normalised point SAR in a multilayer tissue model (5 mm skin, 30 mm muscle,
infinite bone) exposed to a plane wave over the frequency range 1-10 GHz.

A number of general trends are evident from these results:

1. SAR decays more rapidly with depth as the frequency of exposure increases. For all cases
studied, the point SAR at a depth of 10 mm had diminished by at least a factor of 10 for
exposures above 6 GHz. This observation lends support to the treatment of RF exposures
above 6 GHz as a surface heating phenomenon.
2. The large disparity in dielectric values between bone and skin or muscle causes a
reflected wave from the bone interface. This can lead to a significant standing wave
pattern in the skin and/or muscle exhibiting constructive or destructive interference
depending on the layer thickness and the exposure wavelength, λ, which is dependent on
frequency. A constructive interference pattern occurs when the skin/muscle layer is
approximately a quarter wavelength thick, resulting in enhanced SAR. This phenomenon
was most pronounced at 1-2 GHz for the models studied.
3. At the depth of the bone layer, the point SAR is substantially diminished compared to
SAR at the surface in all of the studied cases.

3.3Canonical modelling of plane waves reflected off metallic planar


boundaries
An obvious thread to follow up from the observation of standing waves described in the
preceding section is the constructive interference patterns that can result from RF waves
reflected off a planar metal surface underneath the skin. Classical transmission line theory
indicates that the maximal constructive interference occurs when the thickness of the skin
between the air and plate is a quarter of the RF wavelength, λ, in the skin. Using a FEKO model
as indicated in Figure 8, the calculated field pattern in a 3 mm layer of skin in front of a metal
boundary is shown in Figure 9. It shows the maximal SAR levels occur at 4 GHz where the
wavelength in skin is approximately 12 mm, in accordance with the λ/4 expectation.

Figure 8 Model setup for examination of a plane wave incident on an infinite layers of skin
overlaying a perfect electrical conducting (PEC) plane. The averaging cube for calculating
1 g or 10 g SAR is positioned against the air/skin surface and extends behind the metal
plane where SAR equals zero.
Figure 9 Point SAR distribution for a 1 mW/cm² plane wave normally incident on a 3 mm
thick layer of skin overlaying a metallic plane. Results were calculated in a similar FEKO
model as described in the previous section.

For other skin thicknesses, the frequency at which maximal quarter wave enhancement occurs is
indicated in Table 3 as gauged by the 10 g average SAR over the shape of a cube at the surface.
The point SAR in the skin layer for exposures at these frequencies at the upper tier ambient E-
field limit in the ICNIRP Guidelines (1998) is shown in Figure 9.

Table 3 10 g average SAR (in the shape of a cube) in a skin layer overlaying a metal plane
exposed to a normally incident plane wave at the upper tier limit for ambient E-field
exposure ICNIRP Guidelines (1998), as shown in Figure 10.

skin thickness 3 mm 4 mm 5 mm 6 mm 7 mm 8 mm

λtissue = 4 * skin thick 12 mm 16 mm 20 mm 24 mm 28 mm 32 mm

Freq (GHz) 4.10 3.04 2.41 1.99 1.70 1.47

ICNIRP E-field limit


(W/m²) 50 50 50 50 42.5 36.8

Max 10 g avg SAR (W/kg)


at ICNIRP E-field limit 1.126 1.130 1.126 1.116 0.944 0.804
Figure 10 Point SAR distribution in a layer of skin (3 – 8 mm thick) overlaying a metallic
plane. Exposure consisted of a normally incident plane wave of intensity equal to the
ICNIRP Guidelines (1998) upper tier (occupational) E or H field reference levels at the
quarter wave frequency for each skin thickness. SAR is zero behind the metal plane
interface.

Table 3 indicates that the ICNIRP 10 g localized SAR limits (10 W/kg for head and torso, 20
W/kg for the limbs) are not exceeded for ambient exposures at the occupational field limits. The
quarter wave enhancement effect appears to monotonically decrease for increasing skin thickness
greater than 4 mm (see figure 10).

3.4Canonical modelling of rod implants in infinite medium


3.4.1General approach
As depicted in Figure 1, SAR can be enhanced at the tips of linear metal structures, particularly
when the E-field is oriented parallel to the longest dimension of the implant. This phenomenon
has been investigated by the authors in a series of canonical models for rods exposed to a plane
wave in an infinite dielectric medium, and with particular regard to the following factors:

1. The length of the rod


2. The shape of the rod tip
3. The orientation of the rod with respect to the incident E-field exposure
4. The dielectric medium around the rod
3.4.2Use of VAR instead of SAR
Rather than calculate mass averaged SAR around the rod, it was decided that the Volumetric
Absorption Rate (VAR) in W/m³ averaged over a fixed sized cube was a more appropriate metric
for comparing the relative RF field enhancements. The VAR at any point is calculated as
VAR = σ|E|², c.f. SAR = σ|E|²/ρ. The RF power calculated by integrating point VAR over a 10
cm³ cube is equivalent to the RF power obtained by integrating SAR over a cube of 10 g mass if
the density of the medium is 1000 kg/m³, as is commonly assumed for most tissue types.

The decision to choose volume averaged VAR as the comparison metric was based on its greater
ease of calculation and because it is more directly related to tissue temperature rise. On the first
point, the density of metals (steel ~ 8000 kg/m³) is much higher that the density of tissues (~
1000 kg/m³) which can substantially affect the size of a constant mass averaging cube when it
intersects a metal implant and hence greatly complicates the calculation of mass averaged SAR
compared to a constant size VAR averaging cube.

Moreover, this variability in the size of a SAR averaging mass also introduces an arbitrary
variation in the level of RF power deposited in the cube which makes mass averaged SAR less
directly related to temperature rise than volume averaged VAR. The more direct coupling
between VAR and tissue temperature can be plainly seen in the bioheat equation for steady state
RF heating shown below:

ρSAR VAR

3.4.3Model setup
The analyses were performed using Method of Moment (MoM) analysis in the FEKO v4.1
software (EMMS, 2003). The 10 cm³ volume averaged VAR was calculated by averaging point
VAR in a 24 x 24 x 24 cubic array as depicted in Figure 13. The rod models consisted of a
perfect electrically conducting (PEC) round rod exposed to a plane wave in an infinite tissue
medium. The volume averaged VAR over a 10 cm³ cube was calculated along the length of the
rod as shown in Figure 14.
Figure 11 The 10 cm³ averaging cube for VAR was subdivided into a cubic array of 24 x 24
x 24 cuboids. The 10 cm³ VAR was obtained by averaging the point VAR at the centre of
each of the13,824 cuboids.

Figure 12 FEKO model of a PEC rod implant exposed to a plane wave in an infinite tissue
medium. The 10 cm³ VAR was evaluated along the length of the rod.

In these canonical analyses, the rod was immersed in an infinite tissue medium of either bone or
muscle. The analyses were conducted over a frequency range of 0.1 MHz to 10 GHz with
uniform logarithmic spacing of 5 points per decade (1, 1.6, 2.5, 4, 6.3, 10). The tissue dielectric
values were obtained from Gabriel et al. (1996) as shown in Table 4.

Table 4 Relative permittivity, εr, and conductivity, σ, of cortical bone and skeletal muscle
used in the canonical analyses of a rod exposed to a plane wave in infinite tissue medium.
The plane wave wavelength, λ, and the skin depth, δ, in the tissues are also shown.

freq Cortical bone Muscle


(MHz)
εr σ λ(mm) δ (mm) εr σ λ(mm) δ (mm)
(S/m) (S/m)
0.1 2.28 2.08 67274 11379 8.09 3.62 15624.2 2815.1
E+2 E-2 E+3 E-1

0.16 2.11 2.10 52218 9088 6.95 3.75 11893.4 2230.8


E+2 E-2 E+3 E-1

0.25 1.97 2.12 40694 7366 5.76 3.96 9092.6 1769.7


E+2 E-2 E+3 E-1

0.3 1.91 2.14 36633 6765 5.23 4.07 8137.3 1602.1


E+2 E-2 E+3 E-1

0.4 1.82 2.18 30889 5917 4.34 4.28 6835.0 1360.7


E+2 E-2 E+3 E-1

0.63 1.66 2.27 23298 4775 2.97 4.65 5227.7 1038.4


E+2 E-2 E+3 E-1

1 1.45 2.44 17230 3793 1.84 5.03 4032.0 785.2


E+2 E-2 E+3 E-1

1.6 1.19 2.70 12571 2935 1.07 5.35 3128.8 594.6


E+2 E-2 E+3 E-1

2.5 9.33 3.03 9331 2251 6.40 5.59 2470.6 460.7


E+1 E-2 E+2 E-1

3 8.32 3.19 8278 2012 5.22 5.68 2244.2 416.2


E+1 E-2 E+2 E-1

4 6.87 3.44 6870 1682 3.85 5.81 1927.7 355.4


E+1 E-2 E+2 E-1

6.3 5.01 3.86 5144 1272 2.49 5.99 1513.7 278.5


E+1 E-2 E+2 E-1

10 3.68 4.28 3837 969 1.71 6.17 1179.3 218.8


E+1 E-2 E+2 E-1

16 2.79 4.70 2831 748 1.25 6.34 909.6 172.4


E+1 E-2 E+2 E-1

25 2.25 5.09 2096 597 9.93 6.51 705.5 138.6


E+1 E-2 E+1 E-1

30 2.09 5.25 1844 548 9.18 6.58 634.2 127.1


E+1 E-2 E+1 E-1
40 1.89 5.51 1497 483 8.26 6.69 533.7 111.4
E+1 E-2 E+1 E-1

63 1.67 5.94 1057 403 7.25 6.88 401.0 91.6


E+1 E-2 E+1 E-1

100 1.53 6.43 722 343 6.60 7.08 293.1 76.7


E+1 E-2 E+1 E-1

160 1.43 7.05 479 295 6.17 7.31 206.7 65.9


E+1 E-2 E+1 E-1

250 1.37 7.84 318 255 5.90 7.57 143.7 58.5


E+1 E-2 E+1 E-1

300 1.34 8.27 268 239 5.82 7.71 122.8 56.1


E+1 E-2 E+1 E-1

400 1.31 9.13 204 213 5.71 7.96 95.0 52.6


E+1 E-2 E+1 E-1

630 1.28 1.13 132 169 5.58 8.58 62.3 47.3


E+1 E-1 E+1 E-1

1000 1.24 1.56 84.7 121 5.48 9.78 40.0 40.7


E+1 E-1 E+1 E-1

1600 1.19 2.42 53.9 76.2 5.38 1.24 25.3 31.7


E+1 E-1 E+1 E+0

2500 1.14 4.04 35.3 44.6 5.27 1.77 16.4 21.9


E+1 E-1 E+1 E+0

3000 1.11 5.06 29.8 35.2 5.21 2.14 13.7 18.0


E+1 E-1 E+1 E+0

4000 1.05 7.27 22.8 24.0 5.08 3.02 10.4 12.7


E+1 E-1 E+1 E+0

6300 9.46 1.27 15.2 13.03 4.78 5.57 6.79 6.68


E+0 E+0 E+1 E+0

10000 8.12 2.14 10.3 7.27 4.28 1.06 4.48 3.34


E+0 E+0 E+1 E+1
3.4.4Peak VAR vs. frequency for E polarization exposure of a 40 mm rod in infinite bone
In an initial exploration of linear resonance mechanisms, a model of a round rod (40 mm long, 4
mm diameter) was exposed to a plane wave in an infinite bone medium. The incident E-field was
set to 1 V/m at the centre of the rod, with E parallel to the rod’s axis as depicted in Figure 12.
The 10 cm³ VAR along the length of the rod was normalized with respect to the unperturbed 10
cm³ VAR at the same location when the rod is not present. The peak relative VAR enhancement
along the rod is plotted in Figure 13 over the frequency range 0.1 MHz to 10 GHz.

Figure 13 Peak relative VAR enhancement caused by the presence of a 40 cm long rod
exposed to an E polarised plane wave in infinite bone medium.

The graph depicted in Figure 13 displays some interesting features. Firstly, a resonant response is
clearly evident at around 630 MHz with a peak VAR enhancement of around 4.2. Below this
frequency, there is a lower and constant enhancement of 2.1. Above the resonance, the VAR
enhancement drops to unity, i.e., there is no enhancement of 10 cm³ average VAR.

Figure 14 displays field plots of point VAR at frequencies in each of the three frequency regions
just described. At resonance (630 MHz), the peak VAR enhancement is clearly seen at the tips of
the rods. A relative reduction in VAR can also be observed around the middle of the implant,
illustrating the important general principle of power conservation. In other words, the rod cannot
create additional VAR, but simply redistributes the available RF power around it provided by the
incident exposure. At sub-resonance, (1 MHz in Figure 16), a similar pattern of field
enhancement/reduction around the rod occurs as at resonance. This similarity was evident for all
frequencies below resonance. However, above resonance, there is a clear qualitative shift in the
pattern of VAR distribution. At 6.3 GHz, where λ is 15.2 mm and small compared to the rod
length, the rod acts as an electrically large scatterer, producing a pattern of standing waves
immediately in front of it (note also the attenuation of the E-field to the left of the picture as the
field propagates towards the implant). Over the 21.5 mm side length of the 10 cm³ averaging
cube (which is large relative to λ/4), the constructive and destructive peaks of these standing
waves average out to unity. The logarithmic attenuation of the plane wave as it travels through
the medium is also quite evident at these high frequencies.

Figure 14 Field plots of point VAR around the 40 cm long cylinder implant at sub-
resonance (1 MHz), resonance (630 MHz) and supra-resonance (6.3 GHz) for an incident
plane wave E-field of 1 V/m at the centre of the rod. All three plots use the same linear
scale for point VAR.
3.4.5Peak VAR variation with rod length
The influence of rod length on 10 cm³ VAR enhancement was studied by analysing 4 mm
diameter round rods of varying lengths (20, 30, 40, 80, 160 and 320 mm long) exposed to a plane
wave in an infinite bone medium with E parallel to the rod axis. The relative enhancement of 10
cm³ VAR is shown in Figure 15. The plots for each rod length display the same general resonant
features as described previously, though with some interesting differences.

Firstly, the frequency of the resonant peak changes with rod length. In particular, the resonant
peak appears to occur when the rod length is around one third of the exposure wavelength, as
previously reported by Fleming et al. (1999).

Secondly, rod length influences the magnitude of the relative 10 cm³ VAR enhancement at the
resonant peak and at sub-resonance. Figure 16 shows the peak 10 cm³ VAR enhancement at
resonance. Only a modest enhancement is seen for small rods (×1.57 for a 20 mm length), but the
enhancement is quite substantial for larger rods (×37.4 for 160 mm length). The increase in 10
cm³ VAR enhancement appears to be fairly linear for rod lengths greater than 40 mm. Figure 17
shows 10 cm³ VAR enhancement in the plateau sub-resonance region, with similar trends as
described for the resonant VAR enhancement.

Figure 15 Variation in the peak relative enhancement of 10 cm³ VAR for 4 mm diameter
round rods of varying length exposed to an E polarised plane wave in infinite bone
medium.
Figure 16 Variation in the peak enhancement of 10 cm³ VAR with rod length at resonance
in an infinite bone medium.

Figure 17 Variation in the peak enhancement of 10 cm³ VAR with rod length for plane
wave exposure at sub-resonance frequencies in an infinite bone medium.

3.4.6Peak VAR variation with shape of rod tip


The influence of rod tip shape on the peak relative enhancement of 10 cm³ VAR is shown in
Figure 18 for 80 mm long round rods (4 mm diameter) exposed to an E polarized plane wave in
infinite bone medium with either flat or conical tips. They indicate that the flat tip rod exhibited a
very similar 10 cm³ VAR enhancement response compared to the rod with conical tips,
indicating that the rod tip shape is not a significant cause of variation.

80 mm

Figure 18 Variation in the peak enhancement of 10 cm³ VAR with frequency for flat and
pointed end tips of an 80 cm round rod exposed to an E polarised plane wave in an infinite
bone medium..

3.4.7Peak VAR variation with field orientation


An important determinant of RF field enhancement around long linear structures is the
orientation of the E-field with respect to the implant. Maximum coupling at frequencies around
or below resonance occurs when the E-field is parallel to the long axis of the structure. This is
clear in Figure 19 which illustrates the peak relative VAR enhancement around an 80 mm long
round rod (4 mm diameter, flat ends) exposed to a plane wave in infinite bone medium. In
contrast to the maximum enhancements when E is parallel to the rod, virtually no enhancement is
seen when E is perpendicular to the rod. Halfway between these orientations at 45°, the peak
sub-resonant enhancement is around 59% of the parallel enhancements, but the resonant
response is much more subdued being only 32% of the parallel enhancement. A second smaller
resonant peak is also evident.
Figure 19 Variation in the peak enhancement of 10 cm³ VAR with frequency for different
polarisations of a plane wave E-field incident on an 80 cm round rod in an infinite bone
medium.

3.4.8Peak VAR variation with dielectric value of medium


In the final series of canonical rod implant models, the influence of the tissue dielectric
properties was examined. In these analyses, flat tip rods of various lengths were exposed to E
polarized plane waves in both infinite bone and infinite muscle medium. The results of these
analyses are shown in Figure 20 with comparison to the results obtained in the bone medium.
Figure 20 Variation in the peak enhancement of 10 cm³ VAR with frequency for different
rod lengths exposed to an E polarised plane wave in infinite bone and muscle medium.

These results reveal some very obvious and interesting trends that are consistent for all of the
different length rods. Firstly, the 10 cm³ peak VAR enhancement in the sub-resonance frequency
range is exactly the same for rods immersed in muscle as for rods in bone. Likewise, in the
supra-resonance region, the 10 cm³ peak VAR enhancement trends to unity for rods in both
mediums. However, in the resonant region there are clear differences wherein the enhanced
resonant response seen in bone, is completely damped out when the rod is placed in muscle. It
would seem reasonable to speculate that this damping effect is due to the increased RF power
losses arising from the higher conductivity of muscle, noting that VAR (and SAR) is directly
proportional to conductivity. Note also that the actual value of VAR and SAR may be higher in
muscle than in bone, even though the relative enhancement as shown in Fig. 20 is less. For rods
of all lengths immersed in muscle, the peak 10 cm³ VAR enhancement is negligible (i.e., < ×1.4)
for frequencies above 500 MHz.
4Specific implant assessments
In this section we describe specific implant assessments that have been published in the literature
as well as unpublished assessments conducted by the Telstra Research Laboratories for Telstra
RF workers.

The majority of these studies employed numerical modelling (primarily FDTD and MoM), with
metal objects placed in or near human body models or canonical models to perform the
assessments. In some cases this was followed by the use of thermal modelling (primarily FD) to
assess the resultant thermal changes.

For a separate review of the literature and for a general discussion on metallic implants see
Virtanen et al. (2006).
4.1Linear implants (e.g. pins, rods, and long narrow plates)
Type Dimensions Placement Exposure Comments Source

Pin/rod 7 - 28 mm Different 250 mW, Resonance Virtanen et al.


long and orientations 900/1800 found around (2005)
0.5 - 8 mm of pin/rod MHz λ/3 (14 mm
diameter located on mobile for
the skin or phone 900 MHz).
in muscle, type Enhancement
fat or bone exposure of average
of a simulated SAR found to
cylinder by a be 2 to 3
model of monopole times but
the body on metal study
handset concludes
box that the “…
10 mm enhancement
from skin is unlikely to
surface be
problematic.”
Long narrow 220 mm Attached to RF plane Resonance Telstra
plate long and ~ humerus wave around 50– individual
20 mm with six exposure 100 MHz. assessment
thickness screws and Peak 10 g (2005)
surrounded SAR = 1.03
by muscle W/kg at
in whole bottom of rod
body for 10 W/m2
FDTD input and
model hence
complies
with ICNIRP
Guidelines
(1998)

Intramedullary 440 mm Inside 900 MHz Negligible Telstra


nail long and femur plane effect individual
15 mm (attached wave assessment
diameter with screw exposure (2001)
65 mm
from top)
in whole
body
FDTD
model

Pin Wire 5 - In a 250 mW, Peak 10 g Cooper and


25 mm long spherical 900 MHz SAR Hombach (1996)
(peak at model of a dipole increases no
~λ/2 = head, 15 mm more than
20 mm). parallel to from head 30%
Diameter dipole and (GSM900 compared to
either 0 10 mm mobile tissue without
(filament) from phone implant. Peak
or 1 mm surface type 1 g SAR is
(pin). exposure) around
3.9 W/kg for
filament 20-
25 mm in
diameter

Summary: The assessments conducted at 900 MHz indicate that short (5-28 mm long) and
long (440 mm) implants do not cause excessive SAR, which is in line with the results of the
canonical modelling of rods (see figure 20). The assessment of the long narrow plate (220
mm) in the humerus (upper arm) provides some limited evidence that implant assessments in
that location are not likely to be problematic at any frequency.

4.2Screws
Type Dimensions Placement Exposure Comments Source

Screws ~ 15 mm In elbow in RF plane Resonance Telstra individual


long and ~ whole body wave around 900 assessment (2003)
4 mm wide FDTD exposure MHz but
model effect
negligible

Summary: This assessment lends support to the notion that small implants do not cause
excessive SAR concentration.

4.3Arterial stents
Type Dimensions Placement Exposure Comments Source

Coronary 34 mm long Top of the 100-900 At 100 MHz Telstra individual


artery and 2 mm heart in the MHz the peak 1 g assessment
stent diameter middle artery plane SAR near the (2001)
(modelled as wave stent was 0.35
cylinder) in exposure W/kg for 10
whole body W/m2
FDTD model exposure.

Coronary 6 mm and The stent is Heating Conclude that Foster, Goldberg


artery 25 mm modelled as a furnaces - the and Bonsignore
stent diameter cylinder in 6.25 and ANSI/IEEE (1999)
stents infinite muscle 92.6 kHz C95.1-1992
tissue with RF and standard
blood inside. MRI limits provide
Measurements considered adequate
were made in protection.
egg-white. Paper also
confirms
convective
cooling
effects of
blood vessel.

Summary: Studies show no cause for concern. The potential for adverse heating around
stents is substantially mitigated by the convective cooling from blood flow within the artery.
4.4Wide plates
Type Dimensions Placement Exposure Comments Source

Circular 50 mm Implanted 100 MHz- Resonant response McIntosh et al.


cranioplasty diameter, on the 3 GHz occurred at around (2005)
plate curved front of the plane- 200-300 MHz
around cranium wave (peak 10 g SAR is
forehead, around 5- from the around 0.8 W/kg
and 1.5 mm 6 mm front to for 10 W/m2 input
thick under the the back power at the
surface of of the occupational
the body limit) and
forehead in cumulative
whole interference in the
body scalp at around
FDTD 2100-2800 MHz
model (4.9 W/kg for 50
W/m2) (where the
scalp thickness
was λ/4). The
resultant
temperature
increase is less
than 1 °C.

Circular 15 - 22 mm In a 250 mW, Peak 1 g SAR is Cooper and


Disk diameter spherical 900 MHz around 3.5 W/kg Hombach
model of a dipole for disk around 18 (1996)
head, (15 mm mm in diameter
10 mm from
from the head)
surface mobile
phone
type
exposure

Summary: In comparison to the linear enhancements for rod tips, the peak linear SAR
enhancement is generally more spread around the circumference of a plate which reduces the
peak 10 g average SAR. The cumulative interference effect (with fields in the skin-plate
interface giving rise to increased SAR in front of the plate) has been shown to not lead to
SAR limits being exceeded (in line with the canonical modelling of section 3.3).
4.5Pacemakers
Summary: There have been many studies in the area of the interaction between
electromagnetic fields and pacemakers but the main topic of interest has been interference
issues (see, for example, Hrabar et al. (2001)).
4.6Loops
Type Dimensions Placement Exposure Comments Source

Loops 15 - 50 mm Different 250 mW, Resonance found Virtanen et al.


diameter orientations 900/1800 when loop is λ/3 (2005)
of pin/rod MHz to λ/2 in diameter.
located on mobile Relative
the skin or in phone type enhancement for
muscle, fat exposure 1 g SAR up to
or bone of a (from 2.7× at 900 MHz
cylinder monopole), for 30 mm
model of the 10 mm diameter ring,
body from although at typical
surface mobile phone
power levels “…
enhancement is
unlikely to be
problematic.”
Averaged SAR
highest when loop
is in muscle.

Wire ~ 20 - 25 Ties around Plane wave Temperature rise Hocking, Joyner


loop mm the sternum. exposure at estimated to be and Fleming
and diameter Heat transfer 3, 9.5, 1.3 °C for 50 (1991)
tie with ties analysed in 80,1650- W/m2 at 80 MHz.
around 7-12 finite 3000 MHz The occupational
mm long element (FE) limit at this
model. frequency in
ICNIRP (1988) is
10 W/m2 so all
max temperature
increases < 1 °C.

Summary: Assessments here for a variety of dimensions and exposure conditions show no
cause for concern. There are no published studies of SAR enhancement in the gaps of loops
– further research is needed.
4.7Cochlear implant systems and auditory brainstem implant (ABI) systems
Type Placement Exposure Comments Source

Cochlear Models of 900 MHz mobile Local increases Franzoni et al.


implant cochlea phone about electrode (2006)
system (assessed using arrays of SAR and
the Finite temperature
Integration
Algorithm
(FIT))

Cochlear Standard Mobile phone At 900 MHz (1800 McIntosh et al.


implant placement for type exposure MHz) peak 10 g (2006[1,2])
system internal and (from dipole) at averaged SAR
(with metal external 900 MHz/250 reached near the
hook over components, in mW and 1800 implant was
the ear) whole body MHz/125 mW. calculated to be
FDTD model Dipole placed 10 1.31 W/kg (0.93
mm from ear. W/kg) and the
Different peak temperature
orientations increase was
considered. 0.33 °C (0.16 °C).

Cochlear Standard 50 Hz, 50 kHz, 5 Satisfied ICNIRP Bahr and Boltz


implant placement for MHz and mobile exposure limits (2006)
system (no internal phone
external component, in frequencies
component) whole body 900 MHz,
FDTD model 1750 MHz and
1950 MHz

ABI and Receiver- MRI imaging No heating due to Chou, McDougall


cochlear stimulator (RF component devices observed and Chan (1995)
implant placed on at 63.8 MHz)
system (no outside of head
external and leads
components placed across
in either) central slice of
head in
phantom

Summary: These studies are fairly comprehensive for cochlear implant systems covering a
wide variety of exposure conditions and show no cause for concern.
4.8Spectacles
Dimensions Placement Exposure Comments Source

Several About 1.5 – 3 SAR averaged over eye Edwards and


common frame human GHz plane increases by up to 160% Whittow (2005)
shapes head wave at 50 (compared to when
FDTD W/m2 and spectacles not present)
model 1.8 GHz and decreases up to 80%
dipole depending upon type of
spectacles. Safety
standards satisfied.

Frame 166 mm About 450 MHz, Minor decrease in 10 g Troulis (2003)


wide, lens human 1W average SAR with
aperture 68 x head monopole spectacles (compared to
40 mm FDTD placed in when spectacles not
model front of present)
face

About Mobile SAR levels comply with Yelkenci and


human phones Austrian safety standard Magerl (2000)
head 915 MHz ÖNORM S1120 (4 W/kg
FDTD GSM and for 1 g mass)
model 1.9 GHz
DECT

About a 835 MHz, Unaveraged SAR in eye Anderson and


phantom 600 mW, increased up to 29% with Joyner (1995)
handset spectacles present but still
with pull- below RF safety standard
out limits
antenna

120 mm wide, About a 2 – 4 GHz Spectacles give “… either Griffin (1983) and
lens phantom TEM a shielding or Griffin and
circumference waves enhancement effect …” Davias (1983)
150 mm, but “No serious human
wings 150 mm health effect is
long conclusively revealed
…”.

Summary: , The fairly thorough consideration here has been driven by the concern of
possible enhancement of RF energy in the eyes by the spectacles. In particular, the study by
Edwards and Whittow (2005) was quite exhaustive in the number of frames and exposure
scenarios examined.

4.9Jewellery
Type Dimensions Placement Exposure Comments Source

Ear- 1. Band 1. Band along 900 MHz Increased point Fayos-Fernández


rings: 86 mm back side of dipole SAR values et al. (2006)
1. long, 5 mm pinna. simulating were observed
Band thick, and 2. Studs along mobile but no
2. 10 mm back side of phone significant
Three wide pinna differences were
studs 2. One 25 found when
mm long considering the
and two Placed in and 10 g volume
around 10 about sphere averaged SAR.
mm and head
diameter models
(assessed using
FIT and FDTD
techniques) and
inside and
outside of a
phantom shell.

Summary: The study presented raises no concerns, even with the jewellery in close
proximity to the RF source, and one item of significant size (86 mm long). Even though
jewellery can be worn on the body surface the dimensions of jewellery is usually small (< 20
mm).

4.10Tooth fillings, caps, and orthodontic braces and plates


Type Placement Exposure Comments Source

Teeth In human Mobile phones 915 SAR levels comply with Yelkenci and
caps head FDTD MHz GSM and Austrian safety standard Magerl (2000)
model 1.9 GHz DECT at ÖNORM S1120 (4 W/kg
maximum levels for 1 g mass)

Summary: The Yelkenci et al. (2000) study raises no concerns, as could be expected for very
small metal implants. Moreover, given that the oral cavity is routinely subjected and adapted
to substantial heat loads (e.g. from a hot cup of coffee), then RF heating at upper tier limits
would seem to be comparatively trivial.

4.11Implanted retinal stimulators


Type Placement Exposure Comments Source

Implanted In eye, 46 mW, 2 MHz Peak temperature DeMarco et al.


retinal adjacent to extraocular rise of 0.6 °C, (2003), Lazzi et al.
stimulator retina, in a 2- transmitting without blood flow, (2003)
D FDTD head multiturn coil and 0.4 °C, with
model around 20 mm blood flow assumed
from eye

Implanted In eye, 50 mW, 1.45 and Peak 1 g SAR is Gosalia K and


retinal adjacent to 2.44 MHz 1.59 W/kg at Lazzi G (2003)
stimulator retina, in extraocular 2.44 GHz and
human head transmitting 0.83 W/kg at
FDTD model multiturn coil 1.45 GHz

Summary: These studies raise little concern. Would need to consider each case with
improvements and changes in the technology.

4.12Implanted radiators
Type Dimensions Placement Exposure Comments Source

Implanted Loop 5 mm Normal to chest 403 MHz, Peak 10 g Scanlon (2004)


radiator x 10 mm wall with hip- 25 µW SAR is
mounted monitor, radiated 1.79 W/kg
in human body power
FDTD model

Summary: Particular study gives example that satisfies safety standards. Such devices can
be expected to become more prevalent in society with the increase in medical monitoring
systems and would need to be reviewed for each new technology.
4.13Spinal fusion systems and cervical fixation devices in MRI
Type Dimensions Placement Exposure Comments Source

Cervical In MRI Study Chou, Hover,


fixation standard imaging confirmed McDougall and Ren
devices position in (RF pain to (2004)
phantom component patient caused
at 63.8 by significant
MHz) heating

Spinal Not stated In MRI Temperature Chou, McDougall


fusion standard imaging increase less and Chan (1997)
stimulator position in (RF than 2 °C
(two wire phantom component unless broken
leads and at 63.8 electrode
flat metal MHz) present which
case) gave rise to
around the 14 °C
lumbar increase
vertebrae

Spinal Wires 11.8 In central MRI Highly Buechler, Durney


fusion cm long. position in switched- localised and Christensen
implant Case 3.67 box gradient increase in (1997)
(two wires cm by 2.3 shaped magnetic the E-field up
and a flat cm. torso fields. to 197 times
metal Exposure compared to
case) 600 Hz when implant
magnetic not present
field

Summary: Studies show situations of significant concern for MRI patients, but are difficult
to generalise to assessments of persons exposed to the upper tier limits.
5General rules-of-thumb and observations
for implant assessments
Listed below are rules of thumb and general observations which may be useful in determining
whether an implant requires detailed assessment for a person exposed up to the upper tier limits
of the ICNIRP Guidelines (1998) or the IEEE C95.1 standard (2006). These recommendations
are based on the canonical modelling in section 3 and the specific implant assessments reviewed
in section 4.

5.1General observations

1. RF field enhancements around an implant are affected by the frequency of exposure, the
shape and size of the implant, its orientation with respect to the polarization of the in situ
field and the dielectric properties of the surrounding tissue medium.
2. The absolute level of the SAR around an implant will also depend on the incident RF
field levels in the body area around the implant. Thus, the potential for excessive
localized SARs around an implant is only likely in parts of the body where in situ fields
are already relatively high. Conversely, implants located in parts of the body which are
relatively well shielded would not generally require assessment, especially in in low
conductivity tissues like bone.
3. A metallic implant is a passive re-radiator, and of itself cannot create additional RF
power absorption in the body. Thus the overall RF heating in the general vicinity of the
implant will remain about the same. One possible exception to this rule is the case of a
large implant in one leg (e.g., a metal rod in the tibia), which by providing a lower
impedance conductance path preferentially diverts additional current flow to that leg for
exposure frequencies around and below whole body resonance frequencies.
4. Constructive and destructive interference effects can enhance or diminish the RF field
level in a skin or skin/muscle layer above bone depending on the thickness of the layers
and the frequency of exposure, thereby affecting the incident exposure of an implant
located there. See Figures 3-7 for point SAR plots vs frequency for various tissue layer
thicknesses.
5. SAR attenuation at the skin surface is very substantial at frequencies above 6 GHz and
thereby provides RF shielding protection against metallic implant enhancements in the
body (see Figures 3-7).
6. 10 cm³ average Volumetric Absorption Rate (VAR) is a better metric for assessing RF
heating effects around implants than 10 g average SAR as it is not affected by mass
density changes between the metal implant and the surrounding tissue and is more closely
related to temperature rise.

5.2Thermal effects
7. Some implants are located in a thermal environment where efficient heat transfer
mechanisms will greatly mitigate any localized heating around parts of the implant. For
example, the temperature of an arterial stent is strongly controlled by the convective heat
transfer of the arterial blood flow passing through it. Metal plates located close to the skin
(e.g., plates on the outside of the cranium) are another example, as are all forms of body-
worn metallic objects (e.g. spectacles, jewellery, and the external component of a
cochlear implant system).

5.3Rods and other linear objects

8. In low loss tissues such as bone, a maximal resonant response for E parallel rods occurs
when the rod length is equal to one third of the exposure wavelength (see Figure 15).
This resonant enhancement increases linearly with the length of the rod (see Figure 16).
Long rods can cause very substantial field enhancement at their tips, and hence are more
likely candidates for detailed assessments. This resonance effect is damped out in tissues
with higher electrical conductivity such as muscle (see Figure 20).
9. A common mechanism for RF enhancement around a metallic implant is the field
concentrations that appear at the opposite ends of an implant where the projected length
of the implant against the incident RF electric field vector is the longest. For rods and
other linear structures, the enhancement mostly occurs at the end tips (see Figure 14). The
level of enhancement depends on the size of the implant with respect to the wavelength
of the exposure, which in turn is inversely proportional to the exposure frequency.
10. Short rods less than 20 mm in length do not cause any significant field enhancement
around the implant, which may in part be due to the averaging effect of the 10 cm³ VAR
volume (or 10 g SAR mass). It is probably reasonable to infer from this that all objects
with a maximum dimension of 20 mm or less will not require assessment.
11. The diminished enhancement of SAR at the tips of implants for supra-resonance removes
the need to assess a large class of implants of certain lengths above certain frequencies
which are located at certain distances below the skin (use Figure 20 in combination with
Figures 3-7 for assessment guidance).
12. For rods of all length immersed in muscle, the 10 cm³ VAR enhancement is low (< ×1.4)
for frequencies above 500 MHz (see Figure 20). Similar observations may apply in other
high loss tissues. The re-radiated fields around an implant tend to decay very quickly in a
lossy dielectric tissue environment.
13. The RF field enhancement at the ends of an implant is constant for frequencies below
resonance. The level of this enhancement increases with the rod length (see Figure 17)
and is independent of the dielectric properties of the surrounding tissue (see Figure 20).
14. There is no significant RF enhancement at the ends of an implant in the supra resonant
frequency range. The supra resonant range occurs at lower frequencies for longer
implants (see Figure 20).
15. The 10 cm³ VAR enhancement at the tips of linear implants diminishes substantially for
non parallel E polarizations. No field enhancement is seen for E polarizations that are
perpendicular to rods. Hence, a person moving with respect to the exposure source would
likely reduce implant SAR enhancements when averaged over time.
16. The tip shapes on rods have negligible impact on localised RF heating. This is probably a
consequence of the small size of tips relative to the 10 cm³ VAR averaging volume.
Likewise, RF field enhancements that occur around any sharp point in the implant would
be so localized that their influence would not be noticeable in a 10 cm³ VAR of 10 g SAR
averaging mass.

5.4Screws

17. No assessment required for screws up to 20 mm in length. Longer screws may require
assessment depending on the frequency of exposure and the level of RF shielding at the
implant body location. The pointed end of a screw (where the localised E-field can be
elevated) does not require specific assessment when determining 10 g mass averaged
SAR (see Figure 18).

5.5Arterial stents

18. Detailed studies published so far show no cause for concern for stents up to 34 mm in
length (see section 4.3). In addition, the potential for adverse heating around stents is
substantially mitigated by the convective cooling from blood flow within the artery.
Hence no assessments are required for all stents.

5.6Wide Plates

19. Metal plates that lie directly beneath the skin may enhance SAR in the skin at microwave
frequencies due to constructive interference which is maximized when the thickness of
the skin is equal to a quarter wavelength of the RF exposure in that tissue. For skin
thicknesses between 3 to 8 mm, the quarter wave resonance for a normally incident
exposure ranges from 4.1 to 1.5 GHz (see Table 3). This enhancement however does not
appear to cause the 10 g average SAR to exceed the 10 W/kg upper tier limit for ambient
field exposures below the upper tier reference levels for E or H of the ICNIRP Guidelines
(1998).

20. The peak linear SAR enhancement is generally more spread around the circumference of
a wide plate compared to a rod which reduces the peak 10 g average SAR. As seen in
figure 20, the level of enhancement rises with increasing implant length. For exposure
frequencies less than 200 MHz, plates larger than 50 mm in diameter may need to be
assessed for linear enhancement effects if situated close to the skin where shielding is low
(see sections 3.2 and 4.4).
5.7Pacemakers
21. The body of the pacemaker should not be of concern based upon the studies listed above
for wide plates, particularly as they are usually located deep inside the body. Further
studies are required in regards to the influence of the pacemaker leads.

5.8Loops

22. A loop shaped metal implant which is oriented normal to the in situ H-field may produce
enhanced SAR in any gap in the loop. This phenomenon requires further investigation.

5.9Cochlear Implant Systems

23. No assessment required. See section 4.7 for further details.

5.10Spectacles

24. No assessment required. See section 4.8 for further details.

5.11Jewellery

25. Even though there have been limited assessments performed on jewellery (see section
4.9) there should be no requirement for an assessment. Jewellery is not usually positioned
near vital body tissues, the exception being piercings near the eye but these are typically
small (< 20 mm) and do not require assessment. Heat transfer mechanisms will easily
dissipate any localised heating.

5.12Tooth fillings, caps, and orthodontic braces and plates

26. No assessment required. The oral cavity is naturally adapted to higher heat loads (e.g.
from a hot cup of coffee) which in part is due to the convective and evaporative cooling
from respiration air flow. This may reasonably be expected to provide a larger margin of
safety from RF heating around metallic implants in the mouth. Tooth fillings and caps are
also small in size.

5.13Shrapnel and shotgun pellets


27. No assessment required for pieces up to 20 mm in length. Longer shrapnel pieces may
require assessment depending on the frequency of exposure and the level of RF shielding
at the implant body location.
6References
Anderson V, McIntosh RL. 2004. First stage report on AOARD metallic implant project. Tokyo:
Research Contractor Report for the Asian Office of Aerospace Research and Development
(AOARD) of the United States Air Force Office of Scientific Research (AFOSR). 50 p.
Australian Radiation Protection and Nuclear Safety Agency. 2002. Radiation Protection
Standard: Maximum exposure levels to radiofrequency fields – 3 kHz to 300 GHz.
ARPANSA Radiation Protection Series No 3.
Bahr A, Boltz T. 2005. Schutz von Personen mit Cochleaimplantaten in elektromagnetischen
Feldern. Report prepared for the Federal Ministry for Economics and Labour, Germany.
Project 37/02. See abstract at
http://www.bmwi.de/BMWi/Redaktion/PDF/A/abschlussbericht-der-studie-zu-
cochleaimplantaten-kurzfassung,property=pdf,bereich=bmwi,sprache=de,rwb=true.pdf and
document at http://www.bmwi.de/BMWi/Redaktion/PDF/A/abschlussbericht-der-studie-zu-
cochleaimplantaten,property=pdf,bereich=bmwi,sprache=de,rwb=true.pdf [Both last accessed
1 Sep 2006]
Buechler DN, Durney CH, Christensen DA. 1997. Calculation of electric fields induced near
metal implants by magnetic resonance imaging switched-gradient magnetic fields. Magnetic
Resonance Imaging, 15(10): 1157-1165.
Chou CK, Hover A, McDougall JA, and Ren RL. 2004. Radio frequency heating due to cervical
fixation devices during magnetic resonance imaging. Presentation abstract for 21st
Bioelectromagnetics Annual Meeting, June 2-24, Long Beach, USA.
Chou CK, McDougall JA, Chan KW. 1995. Absence of radiofrequency heating from auditory
implants during magnetic resonance imaging. Bioelectromagnetics 16:307-316.
Chou CK, McDougall JA, Chan KW. 1997. RF heating of implanted spinal fusion stimulator
during magnetic resonance imaging. IEEE Transactions on Biomedical Engineering 44:367-
373.
Cooper J, Hombach V. 1996. Increase in specific absorption rate in human heads arising from
implantations. Electronics Letters 32:2217-2219.
DeMarco SC, Lazzi G, Liu W, Weiland JD, and Humayun MS. 2003. Computed SAR and
thermal elevation in a 0.25 mm 2d model of the human eye and head in response to an
implanted retinal stimulator – Part I: models and methods. IEEE Trans on Ant and Prop.
51:2274-2285.
Edwards RM, Whittow WG. 2005. Applications of a genetic algorithm for identification of
maxima in specific absorption rates in the human eye close to perfectly conducting spectacles.
IEE Proc-Sci Meas Technol 152:89-96.
EMSS. 2003. FEKO version 4.1. Stellenbosch, South Africa: EM Software & Systems-S.A.
(Pty) Ltd.
Fayos-Fernández J, Arranz-Faz C, Martínez-González A and Sánchez-Hernández D. 2006.
Effect of pierced metallic objects on SAR distributions at 900 MHz. Bioelectromagnetics
27:337-353.
Fleming AHJ, Anderson V, Rowley JR. 1999. Computational estimate of the frequency response
of metallic implants in biological tissues exposed to RF fields. Proceedings 1999 ACES
Annual Review of Progress in Applied Computational Electromagnetics.
Foster KR, Goldberg R, Bonsignore C. 1999. Heating of cardiovascular stents in intense
radiofrequency magnetic fields. Bioelectromagnetics 20:112-116.
Franzoni C, Parazzini M, Tognola G, Ravazzani P. 2006. Analysis of the internal fields
distribution and SAR evaluation in a model of implanted cochlea exposed to 900 MHZ.
Presentation abstract for 28th Bioelectromagnetics Annual Meeting, June 11-15, Cancun,
Mexico.
Gabriel S, Lau RW, Gabriel C. 1996. The dielectric properties of biological tissues: III.
Parametric models for the dielectric spectrum of tissues. Physics in Medicine & Biology
41(11):2271-93.
Gosalia K, Lazzi G. 2003. SAR distribution and thermal elevation in a human head model due to
the operation of the data telemetry link and implanted chip in a retinal prosthesis. IEEE Int
Antennas and Propagation Symposium and USNC/CNC/URSI Nth American Radio Science
Meeting, June 22-27. 3:1075-1078.
Griffin DW. 1983. A microwave antenna method of measuring the effect of meta-framed
spectacles on microwaves near the eye. Proc Antennas and Propagation Society International
Symp. 21:253-256.
Griffin DW, Davias N. 1983. An investigation of the effect of metal-framed spectacles on
electromagnetic radiation near the eyes. Department of Electrical Engineering, The University
of Adelaide, South Australia, report Dec 1983.
Hocking B, Joyner KH, Fleming HJ. 1991. Implanted medical devices in workers exposed to
radio-frequency radiation. Scand J Work Environ Health 17:1-6.
Hrabar S, Balachandran W, and Bartolić. 2001. A simple model of electromagnetic coupling
between mobile telephone and implanted pacemaker. Proceedings Medicon 2001, 12-15 June
2001, Pula, Croatia.
International Commission on Non-Ionizing Radiation Protection. 1998. Guidelines for limiting
exposure to time-varying electric, magnetic, and electromagnetic fields (up to 300 GHz).
Health Phys 74: 494-522.
Institute of Electronic and Electrical Engineers. 2005. IEEE Standard for Safety Levels with
Respect to Human Exposure to Radio Frequency Electromagnetic Fields, 3 kHz to 300 GHz.
New York: IEEE International Committee on Electromagnetic Safety (SCC39). IEEE std
C95.1.
Lazzi G, DeMarco SC, Liu W, Weiland JD, and Humayun MS. 2003. Computed SAR and
thermal elevation in a 0.25 mm 2-d model of the human eye and head in response to an
implanted retinal stimulator – Part II: results. IEEE Trans on Ant and Prop. 51:2286-2295.
McIntosh RL, Anderson V, and McKenzie RJ. 2005. A numerical evaluation of SAR distribution
and temperature changes around a metallic plate in the head of a RF exposed worker,
Bioelectromagnetics 26:377-388.
McIntosh RL, Iskra S, McKenzie RJ, Chambers J, Metzenthen B, Anderson V. 2006[1].
Assessment of SAR and thermal changes near a cochlear implant for mobile phone type
exposures, in preparation.
McIntosh RL, Iskra S, McKenzie RJ, Chambers J, Metzenthen B, Anderson V. 2006[2].
Assessment of SAR and thermal changes near a cochlear implant system for mobile phone
type exposures, Bioelectromagnetics Society 28th Annual Meeting Abstract Book, Cancun,
Mexico, 11-15 June 2006.
Pennes HH. 1948. Analysis of tissue and arterial blood temperature in the resting human
forearm. Journal of Applied Physiology 1:93-122.
Scanlon WG. 2004. RF safety considerations for bodyworn and body-implanted radiators.
Available from: http://grouper.ieee.org/groups/scc28/scanlon_pres_ICES0602.ppt. [Last
accessed 1 Apr 2006]
Troulis SE, Evans NE, Scanlon WG, Trombino G. 2003. Influence of wire-framed spectacles on
specific absorption tare within human head for 450 MHz personal radio handsets. Electronics
Letters 39:1679-1680.
Virtanen H, Huttunen J, Toropainen A, Lappalainen R. 2005. Interaction of mobile phones with
superficial passive metallic implants. Physics in Medicine and Biology. 50:2689-2700
Virtanen H, Keshvari J, Lappalainen R. 2006. Interaction of radio frequency electromagnetic
fields and passive metallic implants - A brief review. To appear in Bioelectromagnetics.
Yelkenci T, Magerl G. 2000. Die Berechnung der von Mobiltelefonen im menschlichen Kopf
hervorgerufenen spezifischen Absorptionsraten. Elektrotechnik und Informationstechnik
117:744-749.

Version R1 38/38
Radiofrequency/Microwave Safety Standards
R. C. Petersen

Introduction

Radiofrequency (RF)/microwave safety standards generally refer to standards, regulations, recommendations and
guidelines that specify basic restrictions and exposure limits for the purpose of protecting human health.
Contemporary standards are based on the results of critical evaluation and interpretation of the relevant scientific
research – ideally, all laboratory and epidemiology research that relates any biological response, from short-term and
long-term exposure, would be included. From this evaluation, a threshold is established for the most sensitive
confirmed response that could be considered harmful to humans. To account for uncertainties in the data and to
increase confidence that the limits are well below the levels at which an adverse effect could occur, the resulting
threshold is lowered by a somewhat arbitrary safety factor. RF safety standards have evolved over several decades
from a simple single value that is applicable over a broad band of frequencies e.g., 10 MHz to 100 GHz, to
sophisticated frequency and time dependent limits that cover a much greater frequency range, e.g., 3 kHz to 300
GHz. The early single-value limits were usually expressed in terms of incident power density and were based on
simple models predicting whole-body heating; contemporary standards address effects associated with
electrostimulation at low frequencies, effects associated with whole-body heating, effects associated with surface
heating and usually include limits on induced and contact, exposure to pulses of high peak but low average power,
and localized exposure. The evolution of the development of the standards and guidelines developed by committees
of the American National Standards Institute (ANSI), the International Commission on Non-Ionizing Radiation
Protection (ICNIRP), the Institute of Electrical and Electrical Engineers (IEEE) and the National Council on
Radiation Protection and Measurements (NCRP) is described below.

The early years

Although the interest in the potential effect on humans exposed to RF energy goes back almost a century, it was only
toward the end of World War II that a concerted effort was made to try to understand the interaction of RF energy
with biological systems and, from this understanding, establish criteria to protect against effects that could be
considered harmful. The effort in the United States mainly stemmed from anecdotal reports of various effects by
radar technicians and others who came in contact with various military radars, e.g., temporary male sterility from
exposure to radar beams, the induction of opacities in the lens of the eye. Although by this time the heating effects
of RF energy was well-understood and the technology had been applied in medicine for decades, e.g., RF diathermy,
the anecdotal reports in conjunction with studies reporting lens opacities in the eyes of subject animals exposed to
microwave energy (e.g., Richardson, et el., , Clark et al. , Daily, et al. ) and a report of cataracts in a radar technician
(Hirsch and Parker ), resulted in a coordinated effort to understand the interaction of RF/microwave energy with
biological systems and to establish safety limits. As clearly evidenced by the several orders of magnitude differences
between the protection guides initially adopted by different organizations worldwide during the mid to late 1950’s,
there was little agreement as to suitable protection criteria or an appropriate rationale for establishing these criteria.

Organized efforts to seek an understanding of the possible interaction mechanisms and the effects on human beings
of exposure to electromagnetic energy at RF/microwave frequencies began in the United States with a number of
meetings and symposia. Such meeting included the “Symposium on Physiologic and Pathologic Effects of
Microwaves” held at the Mayo Clinic in 1955, the “First Annual Tri-Service Conference on Biological Hazards of
Microwave Radiation” and the “Second Annual Tri-Service Conference on Biological Hazards of Microwave
Radiation,” in 1957 and 1958, respectively . The purpose of these meetings was to bring together key researchers in
the radiation hazards area in order to discuss ongoing research and identify needed research and, ultimately,
establish science-based safety limits. As pointed out by Mumford , during the time period of these symposia, and
even before, a number of widely different exposure limits were recommended and used by different organizations in
the US. These recommendations, expressed in terms of incident power density, ranged from 100 µW/cm2 to
100 mW/cm2. The upper level was based on an apparent threshold for opacities in the lens of the eye (cataracts),
which was estimated by Hirsch and Parker to be of the order of 100 mW/cm2. Others, e.g., Williams, et al. and Ely
et al. , found higher thresholds but there was general agreement that 100 mW/cm2 should be considered an
approximate threshold for biological damage, i.e., levels above this value were considered hazardous and should be
avoided. In 1953, one Department at Bell Telephone Laboratories added a 30 dB safety factor to the level
considered hazardous and adopted 100 µW/cm2 as a safe level. In 1954 General Electric adopted 1 mW/cm2 as a
safe level, some organizations informally adopted 10 mW/cm2 as a potentially hazardous level, and still others
merely adopted 100 mW/cm2 as a hazardous level without specifying a safe limit. Based on a number of animal
studies and discussions at the symposia noted above, it became apparent by the late 1950’s that 100 µW/cm2 was too
conservative, 100 mW/cm2 was probably not conservative enough and most organizations adopted an exposure limit
of 10 mW/cm2 as recommended by Schwan and Li . This value was based on a simple thermal model that limited
the rise in core temperature of an exposed individual to less than 1ο C, assuming that about half of the incident
energy was absorbed. The frequency range was 10 MHz to 100 GHz.

USAS C95.1-1966 and ANSI C95.1-1974

In 1960, the first formal RF safety standards project was approved in the US when the American Standards
Association (ASA)1 approved the initiation of Radiation Hazards Standards Project C95 and the establishment of a
committee (C95), which was charged with developing standards through an open consensus process. The scope of
the committee was “Hazards to mankind, volatile materials, and explosive devices which are created by man-made
sources of electromagnetic radiation. The frequency range of interest extends presently from 10 kHz to 100 GHz. It
is not intended to include infrared, X-rays or other ionizing radiation.” The C95 Committee, co-sponsored by the
Department of the Navy (Bureau of Ships), and the American Institute of Electrical Engineers,2 was chaired by
Schwan; there were six members on the committee, including the chairman. The committee deliberated for
approximately six years and in 1966 the first C95.1 standard, USAS C95.1-1966, was published . The exposure limit
was presented as a “Radiation Protection Guide” (RPG), defined as the radiation level which should not be exceeded
without careful consideration of the reasons for doing so. The RPG for whole-body exposure was 10 mW/cm2 across
the frequency spectrum from 10 MHz to 100 GHz. Included were an averaging time of 6 minutes and a
corresponding energy density limit of 1 mWh/cm2. The 6 min averaging time appears to have come from the
diathermy literature, although this is not stated in the standard. It is noted that the RPG is applicable in moderate
thermal environments and that under conditions of moderate to sever heat stress the RPG should be reduced
accordingly. The entire standard is less than one and one-half pages in length. Although some at the time considered
the RPGs only applicable in the occupational environment, nowhere in the standard is this stated or implied.

A revision of USAS C95.1-1966 was published in 1974 by the American National Standards Institute (ANSI) as
ANSI C95.1-1974 . The normative part of the standard was still less than two pages in length; the RPGs for
continuous whole body exposure, expressed in terms of incident power density and energy density, remained at
10 mW/cm2 and 1 mWh/cm2, respectively. Because it was then recognized that important exposures could occur in
the near field, particularly in the workplace, limits were also given separately in terms of mean squared electric and
magnetic field strengths. Although the mean-squared electric and magnetic field strengths were each based on an
equivalent power density of 10 mW/cm2, it was considered important to assess each independently, at least at
frequencies below 300 MHz. It is noted in the standard that the RPGs were based on the currently available
literature, it was the consensus of the committee that effects associated with tissue heating remain dominant, and the
RPGs should protect against such effects. It is also noted that at the time, sufficient information concerning
modulation effects, peak power effects and frequency dependent effects was not adequate to substantiate
adjustments to the RPGs to account for these effects . The frequency range over which the RPGs applied remained
10 MHz to 100 GHz. During the eight year development of the 1974 revision the working group (Subcommittee 4)3
had grown considerably in size, totaling almost 70 members.

ANSI C95.1-1982 Standard and the National Council on Radiation Protection and
Measurements (NCRP) Recommendations (1986)
Each revision of the C95.1 standard was more scientifically sound, albeit more complex than its predecessor, with
major changes appearing in ANSI C95.1-1982 (the revision of ANSI C95.1-1974). These changes were related to
the significant advances that occurred in the 1970s in instrumentation and the techniques for measuring complex
electromagnetic fields and, most important, in the field of RF dosimetry. Advances in dosimetry included the use of
numerical techniques to study energy absorption patterns in simple spheroidal and block models of humans and
animals and the use of thermography to study the absorption characteristics of complex realistic models of animals
and anthropomorphic models of humans. These studies led to a clearer understanding of the frequency-dependent
absorption by objects in an RF field, in particular the pronounced resonance over a narrow range of frequencies, the
extent of which depended on the geometry and orientation of the object in the field. Under optimal exposure
conditions, i.e., conditions yielding maximal absorption, it was found that the absorption cross section at resonance
could be 2-3 times greater than the geometrical cross section (see Figure 1). From this understanding, it became
apparent that realistic future protection guides should be frequency-dependent, something that was studied by Soviet
scientists in the early 1960s, cf. Pressman . Thus, while the RPGs in the 1966 and 1974 C95.1 standards were
independent of frequency, by the mid 1970’s it was recognized that the amount of RF energy absorbed by an object
in the field would be frequency dependent—as should the RPGs.

RF dosimetry studies , i.e., the study of RF absorption in models of humans and animals, were carried out by Guy , ,
Guy et al. , Gandhi, et al. , , Durney , Durney et al. , Hagmann and Gandhi , and others, using thermographic
techniques and numerical modeling. These studies led to an understanding of how the incident and internal
electromagnetic fields are related as a function frequency, field polarization, and size, geometry, orientation and
composition of the object in the field. They also led to the recognition of the need for a dosimetric quantity to relate
the incident fields to the internal fields, a quantity that would be more directly related to a biological effect than the
incident fields alone. This need became very apparent during the literature evaluation that led to the 1982 standard
where the general criticism of the growing body of literature was a complete lack of consistency in reported results,
particularly with respect to the field parameters necessary for determining the internal field distributions or energy
absorbed from the field. In many cases only the incident power density was reported without mention of other
parameters necessary for estimating these quantities and, hence, made the comparison of studies difficult at best.
This lack of consistency and completeness also helped explain large differences in the incident power density
reported for the same biological effect in different animal species, and in the same animal species under different
exposure conditions. It was agreed that an appropriate quantity for establishing meaningful thresholds and allowing
comparison across frequency and animal species should be analogous to “dose,” and “dose rate” used by the
ionizing radiation community. This then would be the basic parameter that should be reported so that the results of
studies at different frequencies, using different animal species and widely different exposure conditions e.g., plane
wave, TEM cell, cylindrical cavity, could be compared. Once a threshold for an adverse effect is determined in
terms of the “dose rate,” i.e., the rate at which energy is absorbed from the field, the growing understanding of RF
dosimetry would provide the means for relating this threshold to the incident fields and, with a suitable safety factor,
to realistic frequency-dependent RPGs.
Figure 1—Calculated whole-body average SAR versus frequency for simple
models of the average man for three standard polarizations. The incident
power density is 1 mW/cm2. Curves E and H refer to exposure geometries
where the major axis of the body is aligned with the electric field (E) and
the magnetic field (H), respectively; K refers to the geometry where the
direction of propagation is in the direction of the major axis of the body.
(From Durney, et al. )

Various quantities and terms were proposed for an appropriate dosimetric quantity including “absorbed power
density” expressed in units of W/cm3 or W/kg, and “dose, and “dose rate,” i.e., the energy imparted to a unit mass of
biological material (dose) and the rate at which energy is imparted to unit mass (dose rate). After considerable
discussion and debate within C95 Subcommittee 4 (SC4) during the 1970’s, there was consensus that dose and dose-
rate were appropriate. To avoid confusion and the connotation associated with terms traditionally used in ionizing
radiation protection, “dose” and “dose rate” were named “specific absorption” (SA), defined as the incremental
energy absorbed by (dissipated in) an incremental mass, and “specific absorption rate” (SAR), defined as the time
rate of incremental energy absorbed in (dissipated in) an incremental mass—specific meaning that it is unique to
RF/microwave frequencies.. The units of SA and SAR are J/kg and W/kg respectively. Although SA and SAR first
appear as the defining RF dosimetric quantities in the 1981 NCRP Scientific Committee 39 Report (No. 67) , it had
already been accepted by C95 SC4 in the late 1970s during the development of C95.1-1982 and was used effectively
to compare the results of studies in the database in order to determine an SAR threshold for effects considered
adverse. From this threshold and the results of the increasing number of dosimetry studies, frequency-dependent
limits expressed in terms of the incident fields were derived. These limits were called Radiofrequency Protection
Guides (RFPG) in order to mitigate confusion with the term RPG used by the ionizing radiation community.
Compliance with the RFPGs, ensures that the SAR remains below the threshold (with an adequate margin of safety)
under various exposure conditions and for various size humans from infants to adults.
As indicated above, the 1966 and 1974 C95.1 standards were based on the assumption that effects to protect against
are related to gross thermal effects associated with elevations in core temperature. By 1980, however, a number of
studies reporting effects that occurred at levels where significant temperature increases were not observed or
expected (i.e., “athermal effects”) began to appear in the scientific literature. These studies warranted careful
examination and were included in the list of citations considered by SC4 during the development of the 1982 C95.1
standard. (It is pointed out in the 1982 standard that “classification and judgment of findings were made without
prejudgment of mechanisms of effects,” i.e., the intent of the subcommittee was to protect exposed humans against
harm from adverse effects associated with any interaction mechanism, including effects associated with an elevation
in body temperature” ). During the literature selection process that led to ANSI C95.1-1982, several hundred
experimental studies reporting effects associated with RF energy were reviewed and a select list of 32 studies was
compiled in accordance with the following criteria: demonstrability (positive effects), relevance, reproducibility and
dosimetric quantifiability (i.e., was the SAR reported or was there enough information in the report regarding the
exposure setup to allow determination of the SAR).

Studies that demonstrated general evidence of morbidity or debilitation, chronic or acute, were emphasized . The
bias toward positive findings added a degree of worst-case conservatism to the resulting exposure limits. When
positive results were demonstrated for a specific biological endpoint by several laboratories, those studies that
demonstrated the effect at the lowest SAR and longest exposure duration were selected. Biological endpoints were
grouped in the 15 categories shown in Table 1 along with the number of studies that met the selection criteria in
each category. Reports of specific effects induced by low frequency amplitude-modulated RF carriers, e.g., calcium
efflux from chick brain tissue, were included but were not considered adverse for the following reasons: inability of
the SC4 members to relate the effect to human health; the narrow range of effective modulation frequencies; the
study author’s finding that the effect is reversible. The studies were reviewed by the biologists on SC4 and also by
the physically trained scientists and engineers with emphasis on reliability, evidence of adverse effects, and whether
the study had been independently replicated in another laboratory. The engineers also determined the SAR for each
of the studies.

Following the critical evaluation of the selected studies, the subcommittee agreed that the most sensitive, reliable
confirmed biological response that could be considered potentially harmful to humans is disruption of food-
motivated learned behavior. Even though this effect is, modest, transient and represents an adaptive response, it
serves to identify a threshold for potentially harmful effects . It was also assumed that while behavioral disruption
was demonstrated to be transient and reversible after acute exposure, chronic exposure could lead to irreversible
injury. The threshold for behavioral disruption was found to reliably occur within a narrow range of whole-body-
averaged SARs between approximately 4 to 8 W/kg, across animal species from rodents to primates, frequencies
from 600 MHz to 2450 MHz, and incident power densities that ranged from 10 to 50 mW/cm2. Thus it was agreed
by SC4 that the appropriate biological endpoint for acute exposures should be disruption of behavior, and the
corresponding threshold, in terms of whole-body-average SAR, should be set at 4 W/kg. That is, SAR values above
4 W/kg could produce adverse effects while SARs below 4 W/kg were not shown to result in effects that could be
considered hazardous.
Table 1—Category of exemplary reports selected from the experimental literature by SC4 for the
development of ANSI C95.1-1982

Environmental factors (effects of temperature on the specific endpoint – 3 studies

Behavior and physiology – 6 studies

Immunology – 4 studies

Teratology – 1 study

Central nervous system/blood-brain barrier – 4 studies

Cataracts – no reliable studies were found reporting cataracts at levels ≤ 10 mW/cm2

Genetics (no reliable studies were found reporting genetic effects at levels ≤ 10 mW/cm2

Human studies – no reliable human studies were found

Thermoregulation and metabolism – 5 studies

Biorhythms – 1 study

Endocrinology – 3 studies

Development – 3 studies

Evoked auditory response (RF hearing) – no studies

Hematology – 2 studies

Cardiovascular – 1 study

There was considerable deliberation during the development of ANSI C95.2-1982 as to an appropriate margin of
safety and whether a single frequency-dependent RFPG should apply to exposures of the public and the worker. In
order to ensure an adequate margin of safety, a safety factor of 10 was incorporated, which was considered adequate
to protect members of the public and the worker because of the conservatism already built into the 4 W/kg threshold.
Thus, a whole-body-averaged SAR value of 0.4 W/kg was adopted as the basis of the standard (basic restriction)
from which the frequency-dependent RFPGs (also called derived limits, investigation levels, reference levels) would
be derived.

By 1980 the field of RF dosimetry had advanced to the point where reliable techniques were available to determine
the incident power density that would limit the whole-body-averaged SAR to a specific value. Theoretical analyses
were carried out to determine the magnitude of the incident fields that would limit the whole-body-averaged SAR to
0.4 W/kg under worst-case exposure conditions, i.e., conditions that would maximize energy absorption. The results
of these analyses demonstrated that under plane wave exposure conditions, energy absorption in models of humans,
ellipsoids, animals, etc., is generally maximal when the major axis of the exposed object is aligned with E-field
vector of the incident field and, under these exposure conditions, absorption increases with the square of frequency,
reaches a maximum (resonance), then decreases linearly with increasing frequency over a limited range of
frequencies, and then remains relatively constant. Work by Gandhi and others showed that under these conditions,
maximum absorption (resonance) occurs when the length of the long axis of the exposed object is approximately
0.36 to 0.4 wavelengths. For example, the resonant frequency varies from about 79 MHz to 54 MHz, respectively,
when the height of the body ranges from 1.52 m to 1.98 m. Moreover, it was found that when the object is in contact
with a ground plane, the resonant frequency is about one-half the value found when it was not in contact. Data were
complied from a number of dosimetry studies and a family of resonance curves and plotted as a function of whole-
body-average SAR versus frequency for humans ranging in size from infants to tall adults, both in and not in
conductive contact with a ground plane and with the long axis of the body parallel to the E-field vector of the
incident field. The result of this exercise, normalized to an incident power density of 1 mW/cm2 (which limits the
maximum SAR at resonance to 0.4 W/kg) is shown in the Appendix of C95.1-1982, . The RFPGs were obtained by
rearranging these data to determine the maximum incident power density that would limit the whole-body-averaged
SAR to 0.4 /kg across the frequency range of interest. The results showed that the incident power density required to
maintain an essentially constant SAR in humans of all sizes could be approximated by a broad resonance curve that
decreased as 900/f2 up to 30 MHz, above which it remained constant up to 300 MHz, then rose as f/300 to 1500
MHz, above which it remained relatively constant at 5 mW/cm2. Although the limiting incident power density
continues to increase with decreasing frequency for frequencies below 30 MHz, the RFPG was limited to 100
mW/cm2 for frequencies below 3 MHz to prevent reactions at the body surface caused by the relatively high E-fields
(> 600 V/m), e.g., perception and electric shock. The averaging time remained 6 min over the entire frequency
range. There was some concern by members of the subcommittee about the 6 min averaging time as it applies to
pulses of high peak power but low average power because time averaging single pulses of extremely short duration,
e.g., a few microseconds, leads to unrealistically high exposure limits. No agreement was reached on how to treat
this situation and, therefore, there are no explicit peak power limitations in the 1982 standard.

In addition to RFPGs for whole-body exposures, the 1982 standard contained the following exclusions: 1) The
RFPG for whole body exposure at frequencies between 300 kHz and 100 GHz could be exceeded if it could be
shown using laboratory techniques that the resulting SAR averaged over the whole body would not exceed 0.4 W/kg
and the peak spatial average SAR could not exceed 8 W/kg as averaged over any one gram of tissue. The 8 W/kg
was based on the peak to average SAR values reported in a number of animal studies where it was found that
typically the peak to average SAR ratio was 20 to 1; 2) At frequencies between 300 kHz and 1 GHz, the RFPGs
could be exceeded if the RF input power to the devices is 7 W or less, which is based on limiting the peak spatial-
average SAR to 8 W/kg.

Finally, the standard contained the following caveat: “Because of the limitations of the biological effects database,
these guides are offered as upper limits of exposure, particular for the population at large. Where exposure
conditions are not precisely known or controlled, exposure reduction should be accomplished by reliable means to
values as low as reasonably achievable [ALARA] .” This last sentence often has been quoted out of context by
applying it to RF exposure in general.

NCRP Report No. 86

A number of important events occurred during the interval between approval of ANSI C95.1-1982 and IEEE C95.1-
1991, including comprehensive reviews of the extant RF bioeffects literature by a scientific committee of the
National Council on Radiation Protection and Measurements (NCRP).4 Although NCRP is concerned mostly with
ionizing radiation, in 1973 Scientific Committee 53 (SC53 – now SC89-5) was convened to carry out a
comprehensive review the scientific literature and make recommendations for limiting exposures to RF energy.
SC53 consisted of 6 members, 5 advisory members and 5 consultants (NCRP )—8 of whom were at the time also
members of SC4 of the ANSI C95 committee. Whereas SC4 adopted criteria for selecting studies specifically
relevant to standard setting (e.g., demonstration of positive effects, relevance, reproducibility, dosimetric
quantifiability), and consequently reviewed in detail a relatively small number of reports, SC53 carried out complete
review of the literature and close to 1000 studies were included in the NCRP literature evaluation (the literature
cutoff date was 1982 but a few 1983 references are included). With the quality of the selected reports ranging from
excellent to poor, including some which appear to be nothing more than anecdotal reports, value judgments had to
be made in interpreting and assessing the quality of each of the studies. Reports were divided roughly by biological
endpoint into the categories shown in Table 2.
Table 2—Category of reports reviewed NCRP SC53 SC4 during the development of NCRP Report 86

Macromolecular and cellular effects

Chromosomal and mutagenic effects

Carcinogenesis

Effects on growth, reproduction and development

Effects on the hematopoietic and immune systems

Effects on the endocrine system

Effects on cardiovascular function

Interaction with the blood-brain-barrier

Interactions with the central nervous system

Behavioral effects

Cataractogenesis

Human studies

Thermoregulatory response in human beings

Mechanisms of interactions

As did SC4 of the C95 committee, the members of SC53 also concluded that the most sensitive and statistically
significant biological endpoint was behavioral disruption. Although the carrier frequencies for behavioral disruption
ranged from 225 to 5800 MHz, across animal species from laboratory rats to rhesus monkeys (see Table 3), the
incident power densities ranged from 8 to 140 mW/cm2 and the exposure conditions included near field, far field,
planewave, multipath, CW and modulated RF, the SAR threshold for behavioral disruption narrowly ranged from 3
to 9 W/g, which is in fair agreement with the threshold reported in ANSI C95.1-1982.

Table 3—Comparison of power density and SAR thresholds for behavioral disruption in trained
laboratory animals (from Osepchuk and Petersen )

Species and 225 MHz (CW) 1.3 GHz 2.45 GHz (CW) 5.8 GHz
Conditions (Pulsed) (Pulsed)

Norwegian Rat
Power Density
— 10 mW/cm2 28 mW/cm2 20 mW/cm2
SAR
2.5 W/kg 5.0 W/kg 4.9 W/kg

Squirrel Monkey
Power Density
— — 45 mW/cm2 40 mW/cm2
SAR
4.5 W/kg 7.2 W/kg
— —

Rhesus Monkey
Power Density
8 mW/cm2 57 mW/cm2 67 mW/cm2 140 mW/cm2
SAR
3.2 W/kg 4.5 W/kg 4.7 W/kg 8.4 W/kg

For the frequency range where surface effects predominate, SC53 went further than the C95 committee and
recommended lowering the RFPG if there is a likelihood of coming into contact with grounded metallic objects. To
prevent RF burns at the point of contact, the recommendation was to lower the RFPG such that the induced RF
current does not exceed 200 mA. This is to be done on a case by case basis.

Recommendations, based on the low-frequency modulation-specific effects literature, e.g., calcium efflux studies,
were also included. It was pointed out that it is not known whether these affects lead to a risk to human health, but
the reliability of the studies and their independent confirmation in avian and mammalian species dictates the need
for caution . The recommendation is as follows: “If a carrier frequency is modulated at a depth of 50% or greater at
frequencies between 3 and 100 Hz, the exposure criteria for the general population shall also apply to occupational
exposures.” The incorporation of this caveat, which was based on reported frequency and intensity “windows,” was
extremely controversial and has not been accepted by other standard-setting bodies or incorporated into
contemporary science-based standards and guidelines.

Although the RFPGs in the 1982 C95.1 standard and the NCRP Report are far more realistic and sophisticated than
those used before 1982, both suffer serious shortcomings, including the following: 1) There are no limitations on
peak power for pulses of high intensity but low average power—the 6 min averaging time allows exposure to short
pulses in excess of those known to cause burns at frequencies where the energy is deposited superficially, i.e.,
frequencies above a few GHz; 2) There is no explicit guidance to limit induced current at the lower frequencies, i.e.,
frequencies below a few MHz, to prevent electric shock and RF burns; 3) The magnetic field strength limits, which
correspond to the equivalent plane wave power density of the RFPG, is unrealistic at the low frequencies where the
magnetic field is inefficiently coupled to the body; 4) No clear distinction is made between whole-body and partial-
body exposure (Petersen ). Many of these issues were addressed in the next revision of the 1982 C95.1 standard, i.e.,
IEEE C95.1-1991.5 Table 4 is a comparison of the rationale between NCRP Report 86 and ANSI C95.1-1982.
Table 4—Comparison of rationale: ANSI/NCRP (from Petersen )

Parameter ANSI C95.1-1982 NCRP Report No. 86

Recognition of whole-body resonance Yes Yes

Incorporation of dosimetry(SAR) Yes Yes

Database of experimental literature Relatively small (32 citations) Large

Most significant biological endpoint Behavioral disruption Behavioral disruption

Whole-body-averaged SAR associated 4-8 W/kg 3-9 W/kg


with behavioral disruption

Limiting whole-body-averaged SAR 0.4 W/kg 0.4 W/kg


0.08 W/kg*

Averaging time 6 min 6 min


30 min*

Criterion for limits below 3 MHz Surface effects, e.g., perception, electric Surface effects, e.g., perception, electric
shock (E field)) shock (E field).

Criterion for localized exposure Whole-body-averaged SAR < 0.4 W/kg


Peak spatial average SAR < 8 W/kg
Peak spatial average SAR < 8 W/kg
Peak spatial average SAR < 2 W/kg*

Special criterion for modulated fields No Yes (for occupational exposure)

Specific limits for high peak, low No No


average power pulses

*General population

Institute of Electrical and Electronics Engineers (IEEE) C95.1-19916


Almost immediately after ANSI C95.1-1982 was published, Subcommittee 4 of the ANSI C95 Committee began
work on the next revision with emphasis on addressing some of the recognized shortcomings mentioned above. As
with the earlier revisions, the literature was culled for relevant studies and a total of 321 papers were identified by
the Literature Surveillance Working Group. (See Figure 2 for a graphical depiction of the literature evaluation
process.) Although most of the selected reports were published before 1985, several reports, particularly those
relating to shock, burns and peak-power effects, were published after 1985. Those peer-reviewed studies that
reported effects at whole-body averaged SARs less than 10 W/kg, and which also met the criteria of the
Engineering, Biological and Statistical Evaluation Working Groups, were sent to the Risk Assessment Working
Group, whose charge was to determine the threshold SAR above which potentially deleterious effects are likely to
occur in humans, even if the effects are reversible . As in the case of ANSI C95.1-1982 and the NCRP Report, the
working group concluded that a threshold SAR of 4 W/kg is appropriate to protect against behavioral disruption,
which was once again found to be the most sensitive and reliable biological endpoint. Without saying that
behavioral disruption is a “thermal” effect, it was noted that behavioral disruption in laboratory animals was
accompanied by a core temperature increase of approximately 1 °C and the effect, regardless of the interaction
mechanism, was reversible. Effects reported to be non-thermal, e.g., modulation specific effects such as changes in
calcium efflux from chick brain tissue, were again considered but it was the consensus of the Risk Assessment
Working Group that such effects were inconsistent, could not be related to human health, and, therefore, not useful
for standard setting. It was also the consensus of the Risk Assessment Working Group that a safety factor of 10 to
account for dosimetric, biological and other uncertainties would provide an adequate margin of safety, thereby
yielding a basic restriction of 0.4 W/kg in terms of whole-body-averaged SAR.

Figure 2—Graphical depiction of the ANSI/IEEE literature evaluation process

Unlike C95.1-1982, however, which consisted of a single tier that was considered protective of all, the SC4 Societal
Implications Working Group recommended following NCRP and including a separate lower tier for exposures that
take place in uncontrolled environments. There recommendation was based on the following argument : “To some,
it would appear attractive and logical to apply a larger, or different, safety factor to arrive at the guide for the general
public. Supportive arguments claim subgroups of greater sensitivity (infants, the aged, the ill, and the disabled),
potentially greater exposure durations (24 hours/day vs. 8 hours/day), adverse environmental conditions (excessive
heat and/or humidity), voluntary versus involuntary exposure, and psychological/emotional factors that can range
from anxiety to ignorance. Non-thermal effects, such as efflux of calcium ions from brain tissues, are also
mentioned as potential health hazards .” However, this is followed by “The members of Subcommittee 4 believe the
recommended exposure levels should be safe for all, and submit as support for this conclusion the observation that
no reliable scientific data exist indicating that

1. Certain subgroups of the population are more at risk than others;


2. Exposure duration at ANSI C95.1-1982 levels is a significant risk;
3. Damage from exposure to electromagnetic fields is cumulative; or
4. Non-thermal (other than shock) or modulation-specific sequelae of exposure may be meaningfully related
to human health.”
and “No verified reports exist of injury to human beings or of adverse effects on the health of human beings who
have been exposed to electromagnetic fields within the limits of frequency and SAR specified by previous ANSI
standards, including ANSI C95.1-1982.” Thus, any scientific justification for the lower tier is tenuous at best.
However, the C95 standards are developed through an open consensus process and the majority of the voting
members agreed that a lower tier is appropriate.

The lower tier was derived by reducing the upper tier value by a factor of 5, at least in the resonance region where
SAR is important resulting in a whole-body-averaged SAR of 0.08 W/kg. However, unlike other standards and
guidelines that set limits based on population groups, i.e., an upper tier for occupational exposure and a lower tier
for exposure of the general population, the committee concluded that it would be more meaningful to address the
exposure environment rather than the exposed population to help clarify the assignment of an appropriate set of
limits to personnel, particularly in the workplace. Thus, the derived limits of the upper tier, now referred to as
maximum permissible exposure values (MPE) to be consistent with the use of the term in other standards relating to
non-ionizing radiation protection, e.g., the laser safety standards ANSI Z136.1 and IEC 60825, apply to exposures in
controlled environments; the MPEs of the lower tier apply to exposures in uncontrolled environments. Controlled
environments are considered locations where exposures may be incurred by individuals who are aware of and have
control of their potential for exposure, e.g., as a concomitant of their employment, or by other cognizant persons;
uncontrolled environments are locations where there is exposure of individuals who have no knowledge or control
of their exposure (in living quarters, offices or in workplaces where there are no expectations that exposure levels
may exceed the MPE recommended for lower tier). (See Figure 3 for graphical representation of the IEEE C95.1-
1991 MPEs.)
F

igure 3—Graphical representation of the C95.1-1991 MPEs expressed in term of the E-field equivalent plane
wave power density

In addition to a lower tier and the use of exposure environments rather that exposed populations, a number of other
significant changes appear in the 1991 revision of ANSI C95.1-1982. These include the following:

1. Increased frequency range: The frequency range of the 1982 C95.1 standard is 300 kHz to 300 GHz; the
frequency range of the 1991 standard is 3 kHz to 300 GHz.
2. Magnetic field limits: The magnetic field limits, which correspond to the equivalent free-space power
density of the RFPGs in the 1982 standard, were relaxed in the 1991 standard at frequencies below 3 MHz
in order to more realistically reflect the contribution of the magnetic field to the SAR.
3. Power density limits at quasi-optical frequencies: The MPEs in terms of incident power density were
relaxed from 5 mW/cm2, the value in the 1982 standard for frequencies above 1.5 GHz, to 10 mW/cm2 for
frequencies above 3 GHz (exposures in controlled environments) and for frequencies above 15 GHz
(exposures in uncontrolled environments). This more realistically reflects biological effects associated with
surface heating where the penetration depth is comparable to that of infrared radiation (IR). This is
consistent with the corresponding MPEs at IR wavelengths found in the laser safety standards, e.g., ANSI
Z136.1 and IEC 60825, for exposure to large area beams (greater than 1,000 cm2) , .
4. Averaging time: More realistic averaging times are incorporated in the 1991 standard in order to address a
number of issues including exposure to short high peak power pulses. The averaging time is as follows: In
the frequency region where surface heating predominates, the averaging times decrease with increasing
frequency. For exposures in controlled environments, the averaging time now decreases from a value of 0.1
h at 15 GHz to 10 s at 300 GHz and decreases from 0.5 h at 15 GHz to 10 s at 300 GHz for exposures in
uncontrolled environments. The shorter averaging time mitigates against conditions where skin burns could
occur from short but intense exposures to small areas of the skin, which would be permitted with the longer
averaging times found in the NCRP recommendations and the 1982 ANSI C95.1 standard. For example, an
averaging time of 0.1 h at wavelengths where the penetration depth is comparable to that of the far-IR
would allow a 0.5 s exposure to small areas of the skin that exceeds the 1.2 - 2.4×104 mW/cm2 skin burn
threshold reported by Evans et al. . At 300 GHz, the 10 s averaging time is consistent with the
corresponding averaging time at 300 GHz found in the laser safety standards and guidelines.
5. Peak power limits: Peak power limitations have been incorporated to preclude high specific absorption
(SA) that could result from exposure to increasingly short, high-amplitude pulses. Specifically, for
exposures to pulsed RF fields of pulse durations less than 100 ms and frequencies in the range of 100 kHz
to 300 GHz, the MPE in terms of peak power density for a single pulse is limited to the MPE (under normal
averaging time conditions) multiplied by the averaging time in seconds, divided by five times the pulse
width in seconds, i.e.

If more than five pulses occur during the averaging time, normal time averaging will further reduce the
permissible peak power. In addition, a peak E-field limit of 100 kV/m is included and takes precedence over the
SA limits above. The peak power limits are based on the literature on the evoked auditory response in humans
(microwave hearing) and RF energy induced unconsciousness (stun effect) in rodents. The SA limits are
conservative with respect to the stun effect but the peak power density limits are above the threshold for
microwave hearing, which while annoying, is not considered harmful.
6. Partial-body exposure: Most situations, particularly in the workplace, exposures are to non-uniform fields
over portions of the body and not to uniform plane-wave fields. It was therefore decided that it is
appropriate to address such situations with criteria that would allow relaxation of the MPEs under partial-
body exposure conditions. Specifically, the spatial peak mean squared field strengths and the equivalent
power density permitted under partial-body exposure conditions are allowed to exceed the spatial average
(as averaged over the projected area of the body), as a function of frequency, up to a factor of 20 times.
This relaxation is based on animal studies and dosimetric studies which show that under uniform plane-
wave exposure conditions, the spatial peak SAR exceeds the whole-body-averaged SAR by a factor of
about 20 times. The use of the partial-body relaxation provision is limited because of the accompanying
caveat “The following relaxation of power density limits is allowed for exposure of all parts of the body
except the eyes and the testes.” This precludes practical implementation in many exposure scenarios. The
reasoning behind inclusion of the caveat was concern that at frequencies where the penetration depth was
comparable to that in the IR portion of the spectrum, the relaxation would allow exposures to the eye that
would exceed the IR MPEs for the eye and skin in the laser safety standards—even with the reduced
averaging time.
7. Induced and contact current limits: Induced and contact current limits are incorporated to protect against
surface effects (e.g., shocks and burns) associated with electric-field induced currents which predominate at
frequencies below a few MHz. For the controlled environment, the maximum contact current and the
induced RF current through each foot is limited to 1000 f mA (0.003 < f ≤ 0.1 MHz) and 100 mA (0.1 MHz
< f < 100 MHz) and to 450 f mA (0.003 < f ≤ 0.1 MHz) and 45 mA (0.1 MHz < f < 100 MHz) for the
uncontrolled environment. The averaging time is 1 s. Guidance is also included on how measurements of
foot and contact current should be performed.
8. Minimum measurement distance: In order to minimize the problem of proximity effects, i.e., erroneous
measurement results associated with coupling between the sensor (antenna) elements in the instrument and
the reactive fields from re-radiating structures, a minimum separation distance of 20 cm from any object is
recommended.
9. Low power device exclusion: This exclusion pertains to devices that emit RF energy without control or
knowledge of the user. It is generally applied to hand-held devices such as two-way radios. Specifically, at
frequencies between 100 kHz and 450 MHz, the MPE may be exceeded if the radiated power is 7 W or less
for the controlled environment and less than 1.4 W for the uncontrolled environment. At frequencies
between 450 and 1500 MHz, the MPE may be exceeded if the radiated power is 7(450/f ) W or less for the
controlled environment and less than 1.4(450 /f ) W for the uncontrolled environment. This exclusion does
not apply to devices with the radiating structure maintained within 2.5 cm of the body, e.g., personal
wireless communication devices such as mobile telephones.
10. SAR Exclusions: As in the 1982 standard, the SAR exclusion allows exposures in excess of the MPEs if it
can be shown by reliable means (e.g., laboratory studies) that the whole-body-averaged and peak spatial-
average SAR (basic restrictions) are not exceeded. Unlike the SAR exclusions in the 1982 C95.1 standard,
which were applicable over the entire frequency range of 300 kHz to 300 GHz, and did not specify a
geometric shape for the 1-g averaging volume for localized exposure, the 1991 standard specifies a realistic
frequency range of 100 kHz to 6 GHz and an averaging volume in the shape of a cube to eliminate the
problem of grossly overestimating the peak spatial-average SAR at frequencies where the depth of
penetration is superficial, i.e., at millimeter-wave frequencies. Limiting the frequency range to that where
SAR is meaningful and assigning a cubic geometry to the averaging volume more accurately represents the
potential for hazard. The recommended whole-body-average SAR exclusion for exposures in controlled
environments remains the same as that for the single-tier exclusion in the 1982 C95.1 standard, i.e., 0.4
W/kg (but applicable over the narrower frequency range indicated above). The corresponding value for
exposures in uncontrolled environments is 0.08 W/kg. The peak spatial-average SAR is 8 W/kg and 1.6
W/kg for the uncontrolled and uncontrolled environments, respectively (but applicable over a narrower
frequency range and averaged over any 1-g of tissue in the shape of a cube). The following additional SAR
exclusion is included: for exposure of the extremities, i.e., the hands wrists, feet and ankles, the MPEs can
be exceeded provided the peak spatial-average SAR of 20 W/kg (controlled environment) and 4 W/kg
(uncontrolled environment) in any 10-g of tissues in the shape of a cube is not exceeded (and the induced
current and contact current limits are not exceeded).

Compared with the RFPGs found in the 1982 C95.1 standard and the NCRP recommendations, the 1991 MPEs are
far more complex and sophisticated. The complexity in application and measurement is more than offset by having
scientifically defensible limits that realistically address known RF hazards by ensuring that the thresholds for
adverse effects are not exceeded.

The 1991 standard was approved by the IEEE Standards Board in 1991 and published in 1992. It was also approved
for use as an American National Standard by ANSI in 1992. At the time IEEE C95.1-1991 was approved, SC4 had
125 members; approximately 72% from the research community (including university, military and public health
service laboratories), the rest from industry (~10%), industry (consulting ~3%), government (administration ~4%),
and general public and independent consultants (~11%). At the same time C95.1-1991 was approved, IEEE C95.3-
1991 was also approved . C95.3-1991 was developed by SC1 (Techniques, Procedures, and Instrumentation) and
replaces ANSI C95.3-1973 and ANSI C95.5-1981 . This recommended practice describes instrumentation,
measurement techniques and computational techniques that can be employed to assess compliance with the basic
restrictions and MPEs of the C95.1 standards.

In 1997 C95.1-1991 was reaffirmed (without change); in 1999 Supplement 1 was approved to address certain
ambiguities in the 1991 standard. A definition of spatial average and recommendations on how spatial average
should be measured, i.e., by scanning (with a suitable measurement probe) a planar area equivalent to the area
occupied by a standing adult human (projected area), is included in Supplement 1. Also, the averaging time for
induced and contact current was increased from 1 s to 6 min for frequencies where heating predominates, i.e.,
above 100 kHz, and rms ceiling values of 500 and 220 mA for the controlled and uncontrolled environments,
respectively, were added as were E-field limits (expressed as a percentage of the MPEs) below which induced
current measurements are not required. A detailed description of how induced and contact current should be
measured was also added. There were also a number of other changes including the clarification of averaging
volume as it applies to average spatial-peak SAR, clarification of the term radiated power as it applies to low-
power hand-held devices, clarification of the measurement distance requirements for certain direct radiators (the
separation distance for measurements made in proximity to any directly radiating structure or any of its
attachments was reduced to 5 cm but remained at 20 cm for indirect radiators and reflectors).

In 2004, a request from IEEE SCC347 led to the development of an amendment (C95.1b-2004) that helped clarify
issues relating to the determination of the peak spatial-average SAR associated with the use of hand-held mobile
transceivers intended to be operated placed against the side of the head. This amendment, which was approved in
2004, assigns the same basic restrictions to the pinna as those applicable to the extremities, i.e., peak spatial-
average SAR values of 20 and 4 W/kg, for the controlled and uncontrolled environments, respectively, averaged
over any 10 g of tissue in the shape of a cubical volume surrounding an evaluation point. For this purpose, the
evaluation point is defined as “either the geometric center of the electric field probe sensors at a site used for
experimental SAR measurement, or the location of the incremental volume (voxel) in a numerical computation.”

International Commission on Non-Ionizing Radiation Protection (ICNIRP) Guidelines


The most commonly used standards throughout the world are based on the IEEE C95 standards, the
recommendations of the National Council on Radiation Protection and Measurements (NCRP), and the guidelines of
the International Radiation Protection Association’s (IRPA) International Commission on Non-Ionizing Radiation
Protection (ICNIRP).8 Like IEEE and NCRP, ICNIRP is an organization with established scientific committees that
review the literature and make recommendations regarding exposure to RF/microwave energy. The most recent
ICNIRP guidelines were approved in November 1997 and published in 1998 . At the time the guidelines were
developed, the Commission included the participation of 17 scientists and 11 external experts from 12 different
countries, including Sweden, Australia, Great Britain, Germany, Poland, and the US. Like IEEE and NCRP, ICNIRP
carried out an extensive review and interpretation of the literature, from which exposure guidelines were developed.
As in the case of the ANSI, IEEE and NCRP committees, the ICNIRP guidelines are based on studies reporting
established effects. In agreement with the rationale of C95.1-1991, ICNIRP also found that the established effects
that should be used for developing exposure criteria were surface effects at the lower frequencies, e.g.,
electrostimulation, shocks and burns, and effects associated with tissue heating at the higher frequencies. Although a
number of in vitro studies were reviewed, the focus was on in-vivo studies. A number of epidemiological studies of
reproductive outcome and cancer were reviewed but because of the lack of adequate exposure assessment and
inconsistency of results these studies were found to be of little use for establishing science-based exposure criteria.
Studies reporting athermal effects, including “window effects,” e.g., effects associated with ELF amplitude
modulated (AM) RF fields, were also considered but ICNIRP concluded “Overall, the literature on athermal effects
of AM electromagnetic fields is so complex, the validity of reported effects so poorly established, and the relevance
of the effects to human health is so uncertain, that it is impossible to use this body of information as a basis for
setting limits on human exposure to these fields” .

Like the ANSI/IEEE and NCRP committees, ICNIRP determined that SAR is the valid dosimetric parameter over
the broad whole-body resonance region and also found that the most reliable and sensitive indicator of potential
harm was behavioral disruption, with a threshold SAR of 4 W/kg. A safety factor of 10 was incorporated for
exposure in the workplace, and an additional factor of 5 for exposure of the general public yielding maximum
whole-body-average SAR values of 0.4 and 0.08 W/kg, respectively (called basic restrictions). In addition, basic
restrictions in terms of peak spatial-average SAR of 10 and 2 W/kg averaged over any 10 g contiguous tissue are
recommended for localized exposure. The somewhat less arbitrary ICNIRP peak spatial-average SAR limits are
thought to be based on effects to the eye. Specifically, the threshold associated with the induction of lens opacities in
the eyes of rabbits has been shown to be greater than 100 W/kg. The mass of the eye is about 10 g – by
incorporating safety factors of 10 and 50 times, the resulting peak spatial-average values are 10 and 2 W/kg
averaged over any 10 g of contiguous tissue for occupational exposure and exposure of the public, respectively.

There are also a number of differences between the ICNIRP derived limits (called reference levels) and the MPEs
found in the 1991 IEEE standard but these differences are mostly related to engineering issues, e.g., models used to
relate the incident fields to the basic restrictions, and differences in philosophy of determining safety factors, and not
with any specific biological response or its threshold. Differences between the ICNIRP guidelines and the C95.1-
1991 standard include a broader frequency range for the ICNIRP guidelines (0 to 300 kHz9 compared with 3 kHz to
300 GHz for C95.1-1991), different values for the induced and contact current limits, a slightly higher basic
restriction for localized exposure, (10 and 2 W/kg for the upper and lower tiers, respectively, compared with 1.6 and
8 W/kg in C95.1-1991), a different averaging volume for the localized exposure basic restriction (“over any 10 g of
contiguous tissue” compared with “over any 1 g of tissue in the shape of a cube” in C95.1-1991, a broader resonance
region (10 to 400 MHz compared with 30 to 300 MHz in C95.1-1991), a broader frequency range over which SAR
applies (100 kHz to 10 GHz compared with 100 kHz to 6 GHz in C95.1-1991), and lower peak-power limits. The
ICNIRP peak power limits are based on the evoked auditory response (microwave hearing) whereas the C95.1-1991
limits are based on the stun-effect in small animals (with a suitable margin of safety). That is, while ICNIRP
considers “microwave hearing” a harmful effect, it is considered a possible annoyance in the C95.1-1991 standard—
but not a hazard. There are a number of other minor differences.

IEEE process

Compared with other committees that develop recommendations and guidelines for exposure to RF/microwave
energy, e.g., ICNIRP and NCRP, C95 committees are by far the largest, most innovative, and had the greatest
influence on RF/microwave safety standards worldwide . The subcommittees are open to anyone with a direct
material interest and the standards development process has always been open, formal and transparent at every level,
i.e., the process is different from that of other committees, such as ICNIRP and NCRP, which tend to be closed,
informal and somewhat non-transparent. While the committee operated as an ANSI committee during the
development of the 1991 C95.1 standard, then as an ANSI Accredited Standards Committee, then, during the last
two years a committee sponsored by the IEEE SASB, in each instance it was subject to the formal rules of the
sponsoring organization to ensure due process at every level. In order to understand how the IEEE committees
function, the process will be described briefly before discussing the latest revision of the C95.1 standard (IEEE
C95.1-2005).

The standards coordinating committees that operate under the sponsorship of the IEEE SASB must rigidly adhere to
the policies and procedures of the IEEE, IEEE SASB and the approved polices of the committees. In general, the
process begins with the submittal of a Project Authorization Request (PAR) to the New Standards Committee
(NesCom), a standing committee of the IEEE SASB. (See Figure 4 for a flowchart that depicts the process.) The
PAR outlines the scope and purpose of the proposed standard, the reasons for developing the standard, the number
of members of the working group, when the draft will be ready for sponsor ballot, potential conflicts with the scope
of other standards or standards projects, plus a number of other questions that must be answered by the submitter.
Once deemed complete and accurate by NesCom, a recommendation can be made to the SASB for approval.
Following SASB approval, the working group (in this case SC4) can move forward with the development of drafts.
In accordance with IEEE SASB and ICES procedures, the membership of SC4 consists of volunteers representing
all stakeholders (membership is open to all parties with a direct material interest – IEEE membership is not
required). The membership of SC4 consists of volunteers in engineering, physics, statistics, epidemiology, life
sciences, medicine, and the public with a balance of representatives from government, industry, academia, and the
general public. This wide-ranging participation, including thorough discussions and open decision making, is the
hallmark of the process that led to C95.1-2005 .
When a draft is finally approved by the working group, following the same process mandated for sponsor balloting
described below, (except that balloting is carried out by SC4—not the IEEE SA Balloting Center), the draft is
submitted to the IEEE SA Balloting Center for sponsor ballot (i.e., by the parent committee – SCC28). Sponsor
balloting begins when the IEEE Balloting Center notifies members of the balloting pool that a standard is ready for
sponsor ballot and invites members to join the Ballot Group for that standard. The balloting pool consists of the
parent committee members (SCC28 – now Technical Committee 95 of the IEEE International Committee on
Electromagnetic Safety—ICES) 10 plus all interested parties that may have joined the balloting pool. The balloting
pool is open to any IEEE Standards Association (IEEE SA) member, or any non-member who elects to pay a
nominal fee to vote and receive drafts. Members of the parent committee who wish to vote, but are not members of
the IEEE SA, first have to be approved by the SASB. Requests from the sponsor chair to the SASB secretary
outlining why these individuals should be permitted to vote, what they bring to the committee, etc., are usually
placed on the consent agenda of the next quarterly SASB meeting and, unless pulled off for discussion, are approved
with the agenda. During this time the standard usually undergoes a mandatory editorial review by IEEE Standards
Department project editors, a review by SCC10 (Terms and Definitions – to ensure that all terms and definitions are
in accord with IEEE definitions where such definitions exist), SCC14 (Quantities, Units, and Letter Symbols – to
ensure consistent use of units and letter symbols), and, in some cases, a legal review.

Approval at the sponsor level requires a 75% response from the members of the Ballot Group (including
abstentions) and 75% affirmative votes (the ratio of positive to positive plus negative votes) after ballot resolution.
Attempts must be made to resolve every negative ballot and every substantive comment that accompanied a ballot,
and their resolution (by an ad hoc ballot resolution working group) must be circulated to the Ballot Group to allow
each member to confirm, change his or her vote or comment (only on issues raised during the initial ballot or
previous recirculation ballot). Once a consensus is achieved the standard, ballot results, copies of the PAR, copies of
the recirculation ballots, ballot resolution, and other relevant material are submitted to the SASB Review Committee
(RevCom), also a standing committee of the IEEE SASB. RevCom reviews the scope of the standard to ensure that
it is in accord with the scope of the PAR, that the draft has gone though legal review (when necessary), editorial
review, review by SCC10, and SCC14 and that the Policies and Procedures of the sponsor and those of the IEEE
SASB have been meticulously followed to ensure that the process was open, transparent, and due process was
afforded at every level. When these conditions are met, RevCom can deem the ballot valid and recommend approval
by the IEEE SASB. (RevCom deals only with procedural issues—not technical issues.) Once approved, the draft
standard becomes an IEEE standard and is forwarded to ANSI for public comment and recognition as an American
National Standard. Because of the potential sensitivity of the C95 standards, ICES Policies and Procedures require
formal balloting at the working group level adhering strictly to IEEE SASB procedures, i.e., all members of SC4 are
invited to join the ballot group, all comments submitted with each ballot are addressed, each revised draft resulting
from ballot resolution and all comments and their disposition are circulated to all members of the ballot group to
allow them to reaffirm or change their original vote.

IEEE C95.1-2005
C95.1-2005 is far more detailed and inclusive than its predecessor C95.1-1991. The 2005 standard is divided into
two major parts—normative and informative. The normative part contains the scope and purpose, the normative
references, definitions, recommendations (basic restrictions and MPEs), rules for assessing compliance and the role
of an RF safety program. The informative part contains 7 Annexes. The first three explain the revision process,
summaries of the literature by biological endpoint, and the rationale for the revision. Examples of practical
applications of the standard to typical exposure situations are also included as is a glossary of commonly used terms,
the bibliography of seminal papers from the International EMF Project (IEEE/WHO) database that are cited in
establishing the basic restrictions and thresholds, and a bibliography of other cited publications.

At the time C95.1-2005 was approved, SC4 had 132 members, 42% from outside the US representing 23 countries.
Of the 132 members, 36 were from academia, 56 from laboratories and administrative branches of federal agencies
and the Department of Defense, 22 were from industry, 26 were independent consultants, and 2 represented the
general public. Of these, 73 participated in the balloting, 57 approved, 5 disapproved with comments and 11
abstained, resulting in 92% approval. During sponsor balloting, the Ballot Group had 59 members, 58 returned
ballots, 51 approved, 2 disapproved with comments, 1 disapproved without comments and 4 abstained, resulting in
96% approval. The standard has grown in length from less than one and one-half pages (C95.1-1966) to more than
250 pages—the majority of which addresses the literature reviews and evaluations and the rationale, particularly as
it applies to changes.

As with the earlier C95.1 standards, the revision of the 1991 standard began with the identification of relevant
papers by the SC4 Literature Surveillance Working Group. The focus was on the identification of reliable studies
reporting biological responses – from reversible effects and responses of adaptation to irreversible and biologically
harmful effects. At the literature evaluation cutoff date, 31 December 2003, the Literature Surveillance Working
Group identified over 2200 papers from a number of databases and inputs from federal agencies and other
organizations that were regularly polled. Findings of studies published between 1950 and December 2003 were
considered, including a number of studies reviewed for C95.1-1991. Although the literature cutoff date was
December 2003, a few papers published in 2004 and 2005 were included. New insights gained from improved
experimental and numerical methods and a better understanding of the effects of acute and chronic RF
electromagnetic field exposures of animals and humans were considered during the evaluation process. Every
attempt was made to include and to evaluate all of the relevant literature in the database, with emphasis on studies
carried out under low level exposure conditions where increases in temperature could not be measured or were not
expected. SC4 agreed that only peer-reviewed papers and technical reports of original research would constitute the
primary database on which any risk analysis would be based. Abstracts and presentations at scientific meetings or
technical conferences were expressly excluded. A list of all 1143 papers that were evaluated during the development
of C95.1-2005 can be found in Annex E of the standard.11

The literature evaluation was carried out by the Engineering, Epidemiology, In vivo, and In vitro Working Groups.
In addition, a Mechanisms Working Group was established to evaluate the technical significance of particular
interaction mechanisms with regard to standard-setting. The Engineering WG was tasked with assessing of the
exposure systems, field characteristics and measurements, dosimetry, specific absorption rates, induced currents and
fields, and temperature/humidity measurements and whether or not the information provided was sufficient to allow
a full understanding of how the experiment was performed.

The Epidemiology WG was originally tasked with the evaluation of each paper for study design and population
segments, quality of the methods and implementation, merit of data acquisition and analysis for specific endpoints,
and presence or absence of positive statistical associations. Similarly, the In Vivo and In Vitro WGs examined the
technological methodologies employed in each published paper, including the exposure conditions, specific organ
systems and/or biological endpoints, the engineering and statistical methodologies employed, and the relevance of
each study for standard-setting. The in vitro papers typically emphasized possible effects at the cellular level,
including those on cell viability and proliferation, genotoxicity, cell transformation, molecular synthesis, and cell
function; the in vivo papers typically examined possible effects of exposure on the whole organism or on specific
organ systems, including effects on the embryo/fetus, reproductive ability, immunological system, functional
alterations of the metabolic or thermoregulatory system, various histological endpoints, and behavioral changes.

Many of the evaluations went through a formal process beginning with the chair of each WG providing copies of
each paper to two independent reviewers, together with specially designed and approved review forms. These forms
were in a computer format that required numerical scoring by individual reviewers for entry into a computerized
database. When a review was completed, the reviewer gave the paper an overall technical merit rating on a 5-point
scale. The rating scale was: Very High = 5; Moderately High = 4; Acceptable = 3; Low = 2; and Very Low = 1. For
ratings of 1 or 2, a request was made for justification for the low score in writing by the reviewer. Strong
discordance between the two reviews of a given paper required a third independent review. Periodically, the chair of
each WG would submit a summary of the completed evaluations to the Chair of the Risk Assessment WG (RAWG)
whose charge was to evaluate the implied risk for human beings of exposure to RF electromagnetic fields.

After several years it became clear that the literature evaluation process would not be completed on time following
the formal protocol described above. While the engineering WG evaluated nearly all of the papers in the database
and the In Vivo WG evaluated more than 90% of their assigned papers, few epidemiology and in vitro papers were
evaluated by members of their respective WGs because of a lack of qualified reviewers. Rather than try to evaluate
every paper in the database following the protocol described above, certain individuals with considerable expertise
in specific areas volunteered or were asked to prepare review papers and summarize their findings in specific topic
areas. These included, for example, cancer induction or promotion, teratologic effects, ocular effects, epidemiology,
thermoregulation, and animal behavior. In each topic area, one of the goals was to search for definable hazards. The
texts and conclusions of the various review papers were made available to the RAWG; the summaries and
conclusions from each review paper, which appear in Annex B of C95.1-2005, were further enhanced by 12 review
papers published in Supplement 6, 2003 of Bioelectromagnetics . These included reviews of the epidemiology and
in vitro literature. The evaluation process took advantage of all completed evaluations in the computerized database
plus the review papers.

The overall results of the literature evaluation and review process were used to determine the thresholds of
individual responses and dose response functions, i.e., the lowest level at which a potential harmful effect occurs and
the function that relates dose rate, e.g., SAR, to response magnitude. The weight of evidence approach was used
throughout to develop the thresholds and dose response functions, i.e., the same approach used to develop guidance
for assessment of risk from chemical and other physical agents known to be hazardous.12 SC4 agreed that the
recommendations (basic restrictions and MPEs) should protect against “established adverse health effects in human
beings associated with exposure to electric, magnetic and electromagnetic fields in the frequency range of 3 kHz to
300 GHz” . The term adverse health effect is defined as “A biological effect characterized by a harmful change in
health.” Notes to the definition point out that 1) adverse effects do not include biological effects without a harmful
health effect, changes in subjective feelings of well-being that are a result of anxiety about RF effects or impacts of
RF infrastructure that are not physically related to RF emissions, or indirect effects caused by electromagnetic
interference with electronic devices, and 2) sensations (perceptions by human sense organs) per se are not
considered adverse effects. Thus a sensation of warmth at millimeter and other wavelengths and the microwave
auditory effect under the underlying special conditions are not recognized as effects to be protected against by this
standard. Painful or aversive electrostimulation resulting from exposure at frequencies below 0.1 MHz is treated as
an adverse effect” . This definition, though somewhat narrower than the WHO definition of adverse effect, i.e., “A
biological effect that has a detrimental effect on mental, physical and/or general well being of exposed people either
in the short-term, or long term” (cf. ), was chosen to eliminate some of the ambiguity and subjectivity associated
with the broader definition.

Once the hazard threshold was identified and enough supporting information was available, a safety factor was
applied to the threshold to derive the basic restrictions and MPEs based on the best available scientific information
using the conservative approach common in standard setting. The safety factor, which is influenced by the
uncertainty in the knowledge of the degree of hazard associated with the hazard threshold, is selected to prevent
exceeding the hazard threshold value with a sufficiently wide margin. The magnitude of a safety factors in the 2005
standard ranges from unity at low frequencies, where effects associated with electrostimulation occur, e.g.,
sensation, to significantly greater values at frequencies where heating effects occur, i.e., above 100 kHz. In all cases,
however, the selection of the appropriate safety factor was based on informed expert opinion after considering the
underlying biological and engineering uncertainties applicable to the exposed population for a broad range of
exposure conditions.

The results of the literature evaluation and review process by the SC4 working groups did not provide evidence that
would warrant a change in the scientific basis for the adverse effect level for frequencies between 100 kHz and
3 GHz. The threshold value for whole-body average SAR was again found to be 4 W/kg, and again the most reliable
reproducible biological endpoint was found to be behavioral disruption of food-motivated behavior in laboratory
animals, including non-human primates. Although this conclusion was based on the results of animal studies, it was
agreed that whole-body-averaged SARs above 4 W/kg could be potentially harmful in humans. This is the same
threshold SAR and endpoint found during the development of C95.1-1982, C95.1-1991, the 1986 NCRP report and
the 1998 ICNIRP guidelines. The upper boundary of the frequency range over which whole-body-average SAR is
considered the appropriate basic restriction metric was reduced from 6 GHz in the 1991 standard to 3 GHz based on
RF penetration depth calculations. Also, peak spatial-average SAR values were changed from 1.6 W/kg and 8 W/kg
for the lower and upper tiers to 2 W/kg and 10 W/kg, respectively, and the corresponding tissue averaging mass was
changed from 1g to 10 g. This change is based partially on the biologically based rationale of ICNIRP related to
exposure of the eyes, the extensive theoretical biophysical research quantifying RF energy penetration in biological
tissue, and the desire to harmonize with the ICNIRP guidelines where scientifically justified.

The rationale to set exposure limits for effects associated with electrostimulation at the lower frequencies and
temperature-related effects at higher frequencies is explained thoroughly in the standard. Improved numerical and
measurement methods in RF dosimetry have increased knowledge about the SAR-temperature relationship
following RF energy deposition in human tissue, which is essential when assessing potential biological and health
effects of RF exposures. A number of special considerations have been reviewed and are explained in detail in the
annexes of the standard.

The 2005 standard incorporates a reasonably large margin of safety and, unlike the earlier standards, an RF safety
program is required to provide part of the margin of safety for those exposed above the lower tier, now called an
“action level,” rather than exposures in uncontrolled environments. The choice of the term “action level” for the
lower tier, rather than limits for the “general public” or “uncontrolled environment,” stems from the fact that the
committee concluded that the weight of scientific evidence supports the conclusion that there is no measurable risk
associated with RF exposures below the basic restrictions of the upper tier of this standard. The lower tier, with an
additional safety factor, recognizes public concerns and also supports the process of harmonization with other
recommendations and guidelines, e.g., the NCRP recommendations and the ICNIRP guidelines, and defines the
level above which implementation of an RF safety program is recommended. The purpose of the action level is to
initiate measures, i.e., implementation of an RF safety program as defined in IEEE C95.7-2005 , to prevent
exposures above the upper tier. (The basic restrictions and MPEs of the lower tier can be used for the general
population.) The standard is especially conservative, since the safety factors are applied against perception
phenomena (electrostimulation and behavioral disruption), which are far less serious effects than any permanent
pathology or even reversible tissue damage that could occur at much higher exposure levels than those for
perception phenomena .

This revision of IEEE Std C95.1 maintains many of the characteristics of the previous standard but also contains a
number of differences from earlier editions that address new dosimetry findings and that simplify the use and
application of the standard. These similarities and differences are described in Annex C of C95.1-2005 and are
summarized below.

Similarities:
• All relevant reported biological effects at either low (“non-thermal”) or high (“thermal”) levels were
evaluated. Research on the effects of chronic exposure and speculations on the biological significance of
low-level interactions have not changed the scientific basis of the adverse effect level.
• Whole-body-average and peak spatial-average SAR remain the basic restrictions over much of the RF
spectrum and remain the same as in the earlier standards and guidelines, i.e., 0.4 and 0.08 W/kg.
• The MPE for exposures in controlled environments remain the same as in C95.1-1991.

• The averaging time remains 6 min for frequencies below 3 GHz for effects associated with tissue heating;
but the averaging time for effects associated with electrostimulation is now 0.2 s for an rms measurement
(not 1 s as in the 1999 Supplement to C95.1-1991 ).

Differences:

• Both C95.1-1991 and C95.1-2005 contain two tiers. While the weight of scientific evidence supports the
conclusion that no measurable risk is associated with RF exposures below the limits of the upper tier, it is
impossible to scientifically prove absolute safety and, hence, a lower tier has been set with an extra margin
of safety. The lower tier recognizes public concerns, takes into account uncertainties in laboratory data and
in exposure assessment, and supports the process of harmonization with other standards, e.g., the NCRP
recommendations and the ICNIRP guidelines. While the basic restrictions and MPEs of the upper tier in
both standards apply to exposures in controlled environments; the lower tier of the 2005 standard is now an
action level, rather than specific limits for exposures in uncontrolled environments. This action level, above
which an RF safety program shall be implemented to protect against exposures that exceed the upper tier, is
tied to C95.7-2005 (RF safety programs) . For practical purposes, however, the lower tier may also be used
for the general public.
• The upper frequency boundary over which the whole-body-averaged SAR is deemed to be the basic
restriction (i.e., the “resonance” region) has been reduced from 6 GHz to 3 GHz.
• The lower tier MPEs for long-term exposure are different from those in C95.1-1991 and are in general
more restrictive between 300 MHz and 300 GHz.
• The peak spatial-average SAR values have been changed from 1.6 W/kg and 8 W/kg for lower and upper
tiers to 2 W/kg and 10 W/kg, respectively.
• The averaging mass for determining the peak spatial-average SAR has been changed from 1 g of tissue in
the shape of a cube to 10 g of tissue in the shape of a cube.
• Although implicit in previous versions of the C95.1 standard, e.g., as an SAR exclusion, the present
standard explicitly relies on “basic restrictions.”
• The C95.1-2005 requires the development and implementation of an RF safety program in controlled
environments.
• A more realistic averaging time (based on thermal modeling by Riu and Foster ) for both the upper and
lower tiers has been incorporated for frequencies above 3 GHz to take into account penetration depth,
which decays rapidly above 5 GHz. (This resolves the need for the caveat “except for the eyes and
testes” associated with the partial-body relaxation criteria found in the 1991 standard.)
• The upper frequency at which maximum induced and contact currents are specified is now 110 MHz
compared with 100 MHz in the previous standard.
• The frequency at which the upward ramp begins for the relaxation of the power density limits for localized
exposure has been changed from 6 GHz to 3 GHz.

In recognition of the differing impact of exposure to particular frequencies, the standard provides sections devoted to
three frequency bands: 3 kHz to 5 MHz, 100 kHz to 3 GHz and 3 GHz to 300 GHz. The limits in the first band,
which protects against adverse effects associated with electrostimulation, overlaps the second band where the limits
also protect against effects associated with heating. The limits in the third band protect against effects associated
with heating, particularly superficial heating. Differences between C95.1-1991 and C95.1-2005 within each of these
bands are as follows:
• 3 kHz to 5 MHz: The basic restrictions, based on effects associated with electrostimulation, are now
provided in terms of the in situ electric fields for different regions of the body. Magnetic field MPEs are
specified for the arms and legs and for the head and torso. The electric field MPE for exposure of the whole
body has been increased for exposures in controlled environments; the corresponding magnetic field MPE,
with separate requirements for different regions of the body, has been increased for exposures in controlled
environments and for the lower tier (action level), and have been made frequency dependent. Formulas for
determining maximum permitted peak electric fields for both in situ and environmental conditions are
included.
• 100 kHz to 3 GHz: The peak spatial-average SAR criteria for localized exposure of any tissue excluding
the hands, wrists, forearms, feet, ankles, lower legs and pinnae, have been relaxed from 8 W/kg and 1.6
W/kg to 10 W/kg and 2 W/kg for the upper and lower tiers, respectively. The corresponding averaging
mass has been increased from 1 g to 10 g of tissue in the shape of a cube.13 The peak spatial-average SAR
for the hands, wrists, forearms, feet, ankles, lower legs and pinnae, remains 20 W/kg for the upper tier and
4 W/kg for the lower tier. The contact current limits for the frequency range of 100 kHz to 110 MHz have
been subdivided into touch and grasping conditions, with the grasping condition confined to the controlled
environment. The permissible touch contact current has been reduced for both the controlled environment
and the lower tier (action level). In the transition region where effects associated with electrostimulation
and tissue-heating occur (100 kHz to 5 MHz), the basic restrictions and MPEs for both must be met. In
order to harmonize with ICNIRP, the frequency dependence of the MPEs for frequencies between 300
MHz and 300 GHz has changed and the values made more stringent, but only for the lower tier.
• 3 GHz to 300 GHz: The upper frequency of the SAR region has been reduced from 6 GHz to 3 GHz to
better reflect the quasi-optical nature of tissue interactions. As indicated above, the principal change has
been in the values and frequency dependence of the MPEs above 300 MHz for the lower tier, but the MPE
at 300 GHz and the corresponding averaging time remains the same as in C95.1-1991.

As in the earlier standards, the recommendations are expressed in terms of basic restrictions MPEs, i.e., reference
levels, investigation levels. The basic restrictions are limits on the in situ electric field strength for the for the brain,
heart, extremities and other tissues (different limits for each) for frequencies between 3 kHz and 100 kHz, whole-
body-averaged SAR and peak spatial-average SAR (with relaxed limits for the extremities and the pinna) for
frequencies between 100 kHz and 3 GHz, and incident power density for frequencies between 3 GHz and 300 GHz.
The MPEs, which are derived from the basic restrictions, are limits on external fields and induced and contact
current. In the region where effects associated with electrostimulation predominate, i.e., between 3 kHz and 100 kHz
(up to 5 MHz for certain pulsed fields), the MPEs are expressed in terms of the external electric and magnetic field
strengths for the head and torso with separate values for the limbs. MPEs for the undisturbed electric field strength
(absent a person) are also provided for frequencies between 3 kHz and 100 kHz. In the region where whole-body
heating predominates (100 kHz to 3 GHz), the MPEs are expressed in terms of the incident electric and magnetic
field strengths for frequencies up to 300 MHz for the upper tier (exposures in controlled environments) and up to
400 MHz for the lower tier (action level – general public) above which the MPEs for both tiers are expressed in
terms of the plane-wave equivalent power density. In the transition region of 0.1 to 5 MHz, each of the two sets of
MPEs and basic restrictions apply. In this transition region the MPEs and basic restrictions based on heating will be
more restrictive for long-term exposures to CW fields, while the MPEs and basic restrictions based on the effects of
electrostimulation will be more restrictive for short-term exposure, e.g., short isolated pulses of low duty factor. For
frequencies greater than 3 GHz, the MPEs and basic restrictions are expressed in terms of incident power density
and are equivalent. Figure 5 and Figure 6 are graphical representations of the C95.1-2005 MPEs for the upper tier
(exposures in controlled environments) and the lower tier (action levels), respectively. Figure 7 shows a comparison
of the C95.1-2005 MPEs (in terms of the E-field equivalent power density) with the corresponding MPE of the 1998
ICNIRP guidelines. Table 5 is a comparison of features of the ICNIRP guidelines with the corresponding features of
IEEE C95.1-2005.
Figure 5—Graphical representation of the C95.1-2005 MPEs for the upper tier in the frequency
region where effects associated with heating predominate (from )
Figure 6—Graphical representation of the C95.1-2005 MPEs for the lower tier in the frequency
region where effects associated with heating predominate )
Figure 7—Comparison of the C95.1-2005 MPEs (lower tier – expressed in terms of E-field
equivalent power density) with the ICNIRP MPEs for the general public. The upper tier MPEs of
C95.1-2005 are the same as the C95.1-1991 MPEs.
Table 5—Comparison of 1998 ICNIRP guidelines and C95.1-2005: Region where the predominant interaction
mechanism is tissue heating

Parameter ICNIRP IEEE C95.1-2005

Frequency range ~ 100 kHz to 300 GHz ~ 100 kHz to 300 GHz

Recognition of whole-body resonance Yes Yes

Incorporation of dosimetry(SAR) Yes Yes

Database of experimental literature Large Large (~1100 citations)

Most significant biological endpoint Behavioral disruption (associated with ~ Behavioral disruption (associated with ~
1°C temperature rise) 1°C temperature rise)

Whole-body-averaged SAR associated 1-4 W/kg ~ 4W/kg


with behavioral disruption

Limiting whole-body-averaged SAR 0.4 W/kg (Occupational) 0.4 W/kg (Controlled Environment)

– Applicable frequency range 0.08 W/kg (General Public) 0.08 W/kg (Action Level)

100 kHz to 10 GHz 100 kHz to 3 GHz

Peak spatial-average SAR (localized 10 W/kg (Occupational) 10 W/kg (Controlled Environment)


exposure)
2 W/kg (General Public) 2 W/kg (Action Level)
–Averaging volume 10-g of contiguous tissue
10 g of tissue in the shape of a cube
–Averaging time 6 min (Occupational)
6 min (Controlled Environments)
6 min (General Public)
30 min (Action Level)

Limits for extremities

–Upper tier 20 W/kg (limbs) 20 W/kg (extremities including pinnae)

–Lower tier 4 W/kg (limbs) 4 W/kg (extremities including pinnae)

100 kHz < f ≤ 10 GHz 100 kHz < f ≤ 3 GHz


–Applicable frequency range

Averaging time (f > 100 kHz)

–Upper tier 6 min (f ≤ 10 GHz) decreasing to 10 s at 6 min (f ≤ 3 GHz) then decreasing to 10 s at


300 GHz
–Lower tier 300 GHz)
6 min (f ≤ 10 GHz) decreasing to 10 s at
300 GHz
6 min (3 kHz ≤ f ≤ 1.34 MHz). E 2 and H 2
have different averaging times for

1.34 MHz < f ≤100 MHz but both are equal


to 30 min at 100 MHz. For 100 MHz < f ≤
5 GHz the averaging time is 30 min and
then decreases to 10 s at 300 GHz.

Induced and contact current limits

–Upper tier 40 mA (limb currents) 90 mA (each foot)

–Lower tier 20 mA (limb current) 45 mA (each foot)

100 kHz ≤ f ≤ 110 MHz 100 kHz ≤ f ≤ 110 MHz


–Applicable frequency range

Special criterion for modulated fields No No

Specific limits for high peak, low Yes—Based on evoked auditory response Yes—Based on the stun-effect
average power pulses (“microwave hearing”)

RF safety program Not specifically Yes – IEEE C95.7-2005. The BRs and
MPEs of the lower tier (action level) are
linked to an RF safety program to mitigate
against exposures that could exceed the
BRs and MPEs of the upper tier.
Other RF standards

There are a number of other standards available for assessing compliance with the C95.1 standards and the ICNIRP
guidelines. These include measurement standards, such as IEEE C95.3-2002 , which describes measurement and
computational techniques for assessing human exposure to electric, magnetic and electromagnetic fields, instrument
types and limitations, measurement uncertainties, plus a number of other instrument and measurement issues.
Although the scope of the standard covers the frequency range of 100 kHz to 300 GHz, its practical range of
applicability is closer to 100 kHz to few GHz. There are also a number of product standards available for assessing
compliance of specific products, such as hand-held mobile telephones intended to be operated while placed next to
the head, with the basic restrictions (peak spatial-average SAR) found in C95.1-2005 and the ICNIRP guidelines.
These include IEEE 1528-2003 , IEEE 1528a-2005 and IEC 62209 . These standards describe in detail the
procedures for determining the peak spatial-average SAR in an anthropomorphic model of the human head (filled
with liquid head-tissue simulant) by means of a robotically-controlled miniature electric field probe. Calibration
techniques, measurement uncertainties, recipes for the tissue-equivalent liquids and techniques for measuring their
electric properties are just a few of the issues addressed in these standards. These IEEE and IEC standards are in
complete harmony. Neither of these standards specifies a specific basic restriction—they can be used for conformity
assessment against any commonly used value and averaging mass. Similar standards are now under development to
carry out the same assessments using numerical simulations, e.g., FDTD techniques. An IEC project team is now in
the process of developing a standard specific to assessing human exposure to mobile telephone base stations.

IEEE C95.7-2005 (RF safety programs) presents guidelines and procedures that can form the basis of a an RF safety
program1 for controlling hazards associated with RF sources that operate in the frequency range of 3 kHz to 300
GHz . C95.7 is a general-purpose standard with the goal of preventing potentially hazardous exposures to
electromagnetic fields, currents, and/or contact voltages. The standard is modeled somewhat after the laser safety
standards, e.g., ANSI Z136.1 and IEC 60825 , where areas in which exposure may be possible are characterized into
one of four exposure categories according to the potential risk for exposure in excess of prescribed limits and then
specifying the appropriate controls to reduce the likelihood of over-exposure. This standard is designed to
complement the IEEE C95 family of standards but may find use in the development of effective programs to ensure
conformance with other guidelines, standards, or regulations. Warning signs and labels, which generally are part of
any safety program, can be found in IEEE C95.2-1999 .

Summary

Contemporary science-based RF/microwave safety standards and guidelines are based on in-depth evaluations and
interpretations of the extant scientific literature. This is an on-going process and, as such, the recommendations in
terms of safe limits of exposure evolve as the research becomes more focused and the quality of the research
improves. The simple single value frequency-independent limit proposed more than five decades ago has evolved
into the sophisticated rather complex standards and guidelines now used throughout the world. The biggest influence
on the direction of the standards was the better understanding of dosimetry issues gained through the thermographic
studies and numerical modeling that began in the 1970’s. This provided the means for establishing meaningful
parameters for relating the external fields to the internal fields and provided the means for readily comparing study
results. The concept of SAR, first proposed in 1981 for use as a basic restriction over a limited frequency range, led
to the current whole-body-averaged SAR 4 W/kg threshold for adverse effects (behavioral disruption), which has
not changed even though the literature database has grown tremendously during that time. Although the MPEs of
IEEE C95.1-2005 may differ slightly from those of the 1997 ICNIRP guidelines, the differences are minor and have
to do with the engineering aspects of relating the incident fields to the internal fields and the assigned margins of
safety—not with philosophical differences in the interpretation of the biology. With each revision, the basic
restrictions and derived limits (reference levels, MPEs) of the two most often cited standards and guidelines, i.e., the
C95.1 standards and the ICNIRP guidelines, converge.

Bothe the ICNIRP guidelines and C95.1 standards are living documents. If any new adverse effect is established
which would require a change in the standard, for example, the standard can be promptly revised by amendments.
The IEEE committee continues the literature surveillance and evaluation of the bioeffects research for the next
revision. The future replacement of peak SAR by temperature or temperature increase was discussed as a possibility
during development of the 2005 standard; dosimetry studies are now in progress to identify the relationship between
temperature rise and peak spatial average SAR for future consideration.

References

1. Richardson, A. M., Duane, T. D. and Hines, H. M. “Experimental lenticular opacities produced by


microwave irradiation,” Arch. Phys. Med., vol. 29, pp. 765-769, December, 1948.
2. Clark, J. W., Hines, H. M. and Salisbury, W. W. “Exposure to microwaves,” Electronics, vol. 22, pp. 66-
67, May, 1949.
3. Daily, L. Jr., Wakim, K. G., Herrick, J. F., Parkhill, E. M. and Benedict, W. L., “The effects of microwave
diathermy on the eye of the rabbit,” Am. J. Ophthalmol., vol. 35, pp. 1001-1017, July, 1952.
4. Hirsch, F. T. and Parker, J. T., “Bilateral lenticular opacities occurring in a technician operating a
microwave generator,” A. M. A Arch. Ind. Health, vol. 6, pp. 512-517, December, 1952.
5. Mumford, W. W., “Some technical aspects of microwave radiation hazards,” Proc. IRE, pp. 427–447, Feb.
1960.
6. Williams, D. B., Mohahan, J. P., Nicholson W. J., and Aldrich, J. J., “Biologic effects studies on
microwave radiation, time and power thresholds for the production of 12.3 cm microwaves,” IRE Trans. on
Medical Electronics, vol. ME-4, pp. 17-22, February, 1956.
7. Ely, T. S., Goldman, D. E., Hearon, J. Z., Williams, R. B. and Carpenter, H. M., Heating Characteristics of
Laboratory Animals Exposed to Ten-Centimeter Microwaves. Naval Med. Res. Report. Project NM001
056.13.02, March 21, 1957.
8. Schwan, H. P. and Li, K., “The mechanism of absorption of ultrahigh frequency electromagnetic energy in
tissues as related to the problem of tolerance dosage,” IRE Trans. on Medical Electronics, vol. ME-
4.February, 1956.
9. USAS C95.1-1966, Safety Level of Electromagnetic Radiation With Respect to Personnel, United States of
America Standards Institute, New York, NY.
10. ANSI C95.1-1974, Safety Level of Electromagnetic Radiation With Respect to Personnel, American
National Standards Institute, New York, NY.
11. Durney, C. H., Johnson, C. C., Barber, P. W., Masoudi, H., Iskander, M. F., Lords, J., Ryser, D. K., Allen,
S. L. and Mitchell,. J. C. Radiofrequency Dosimetry Handbook, Second Edition. Report SAM-TR-78-22,
USAF School of Aerospace Medicine, Brooks Air Force Base, Texas, 1978.
12. Pressman. S., Electromagnetic Fields and Life, (Translation by Sinclair, F. L.,), Plenum Press, New York,
1970.
13. Guy, A. W., “Analyses of electromagnetic fields induced in biological tissues by thermographic studies on
equivalent phantom models,” IEEE Trans. Microwave Theory and Tech., vol. MTT-19, pp. 205-214, 1975.
14. Guy, A. W., “Quantitation of induced electromagnetic field patterns in tissue, and associated biological
effects,” In Biological Effects and Health Hazards of Microwave Radiation, Czerski, P. (Ed), pp. 203-216,
Polish Medical Publishers, Warsaw, 1974.
15. Guy, A. W., Weber, M. D. and Sorensen, C. C., “Determination of power absorption in man exposed to
high-frequency electromagnetic fields by thermographic measurements on scale models,” IEEE Trans. on
Medical Electronics, vol. 23, pp. 361-371, 1976.
16. Gandhi, O. P., Hunt, D. L. and D’Andrea, J. A., “Deposition of electromagnetic energy in animals and in
models of man with and without grounding and reflector effects,” Radio Science, vol. 12, no. 6S, pp. 39-48,
1977.
17. Gandhi, O. P., Sedigh, K., Beck, G. S. And Hunt, E. L., Distribution of electromagnetic energy deposition
in models of man with frequencies near resonance. In Biological Effects of Electromagnetic Waves,
Johnson, C. C. and Shore, M. L., (Eds), DHEW Publications (FDA) 77-8011, vol. 2, pp. 44-67, 1976.
18. Durney, C. H., “Electromagnetic dosimetry for models of humans and animals: A review of theoretical and
numerical techniques,” Proceedings of the IEEE, vol. 68, 1980.
19. Hagmann, M. J. and Gandhi, O. P., Numerical calculations of electromagnetic energy deposition in models
of man with grounding and reflector effects,” Radio Science, vol. 14, no. 6S, pp. 23-29, 1979.
20. NCRP, Radiofrequency Electromagnetic Fields – Properties, Quantities and Units, Biophysical Interaction,
and Measurements, NCRP Report no. 67, National Council on Radiation Protection and Measurements,
Bethesda, MD, 1981.
21. ANSI C95.1-1982, American National Standard Safety Levels with respect to Human Exposure to Radio
Frequency Electromagnetic Fields, 300 kHz to 100 GHz, American National Standards Institute, New
York, NY.
22. NCRP, Biological Effects and Exposure Criteria for Radiofrequency Electromagnetic Fields. Bethesda:
NCRP Report no. 86, National Council on Radiation Protection and Measurements, Bethesda, MD, 1986
23. Osepchuk, J. M. and Petersen, R. C. “Safety Standards for Exposure to Electromagnetic Fields,” IEEE
Microwave Magazine, Vol. 2, No. 2, pp. 57-69, June 2001.
24. Gandhi, O. P., “State of knowledge for electromagnetic absorbed dose in man and animals,” Proc. IEEE,
vol. 68, pp. 24-32, 1980.
25. Petersen, R. C., "Radiofrequency/Microwave Protection Guides," Health Physics, Vol. 61, No. 1, July
1991.
26. IEEE C95.1-1991, IEEE Standard for Safety Levels with Respect to Human Exposure to Radio Frequency
Electromagnetic Fields, 3 kHz to 300 GHz, IEEE New York, NY.
27. ANSI Z136.1, American National Standard for the Safe Use of Lasers, American National Standards
Institute, New York, NY.
28. IEC 60825-1, IEC Standard Safety of Laser Products - Part 1: Equipment Classification, Requirements and
User's Guide, International Electrotechnical Commission, Geneva, (2001).
29. Evans, E., Brooks, J., Schmidt, F, Williams, R., Ham, W.T. Jr., “Flash burn studies on human volunteers,”
Surgery, vol. 37, pp. 280-297, 1955.
30. International Commission on Non-Ionizing Radiation Protection (ICNIRP), Internet site
http://www.icnirp.org/aim.htm
31. ICNIRP (International Commission on Non-Ionizing Radiation Protection), “Guidelines for limiting
exposure to time-varying electric, magnetic, and electromagnetic fields (up to 300 GHz),” Health Physics,
vol. 74, pp. 494 - 522, 1998.
32. IEEE C95.3-1991, IEEE Recommended Practice for the Measurement of Potentially Hazardous
Electromagnetic Fields - RF and Microwave. (Replaces ANSI C95.3-1973 and ANSI C95.5-1981.)
33. ANSI C95.3-1973, American National Standard Techniques and Instrumentation for the Measurement of
Potentially Hazardous Electromagnetic Radiation at Microwave Frequencies, American National Standards
Institute, New York, NY.
34. ANSI C95.5-1981, American National Standard Recommended Practice for the Measurement of Hazardous
Electromagnetic Fields - RF and Microwave, American National Standards Institute, New York, NY.
35. IEEE C95.1-1991: 1999 edition, IEEE Standard for Safety Levels with Respect to Human Exposure to
Radio Frequency Electromagnetic Fields, 3 kHz to 300 GHz, IEEE New York, NY.
36. IEEE C95.1b-2004, IEEE Standard for Safety Levels with Respect to Human Exposure to Radio Frequency
Electromagnetic Fields, 3 kHz to 300 GHz - Amendment 2: Specific Absorption Rate (SAR) Limits for the
Pinna, IEEE New York, NY.
37. IEEE C95.1-2005, IEEE Standard for Safety Levels with Respect to Human Exposure to Radio Frequency
Electromagnetic Fields, 3 kHz to 300 GHz, IEEE New York, NY.
38. Petersen, R.C., “Radiofrequency safety standards-setting in the United States,” in Bersani (ed.), Electricity
and Magnetism in Biology and Medicine, Plenum Press, New York, 1999, pp. 761-764.
39. Bioelectromagnetics, Supplement 6, Wiley-Liss, 2003.
40. WHO, “Model legislation for electromagnetic fields protection,” Internet site http://www.who.int/peh-
emf/standards/EMF_model_legislation%5b1%5d.pdf, 2006.
41. IEEE C95.7-2005, IEEE Recommended Practice for Radiofrequency Safety Programs,. IEEE New York,
NY.
42. Riu P. J., Foster K. R., “Heating of tissue by near-field exposure to a dipole: a model analysis,” IEEE
Trans. Biomed Eng., vol. 46, pp. 911 - 917, 1999
43. IEEE C95.3-2002, Recommended Practice for Measurements and Computations of Radio Frequency
Electromagnetic Fields With Respect to Human Exposure to Such Fields, 100 kHz–300 GHz, IEEE New
York, NY.
44. IEEE 1528-2003, IEEE Recommended Practice for Determining the Peak Spatial-Average Specific
Absorption Rate (SAR) in the Human Head from Wireless Communications Devices: Experimental
Techniques, IEEE New York, NY.
45. IEEE 1528a-2005, IEEE Recommended Practice for Determining the Peak Spatial-Average Specific
Absorption Rate (SAR) in the Human Head from Wireless Communications Devices: Measurement
Techniques Amendment 1: CAD File for Human Head Model (SAM Phantom), IEEE New York, NY.
46. IEC 62209-1, Human Exposure to Radio Frequency Fields from Hand-held and Body-mounted Wireless
Communication Devices – Human Models, Instrumentation, and Procedures –Part 1: Procedure to
Determine the Specific Absorption Rate (SAR) for Hand-held Devices used in Close Proximity to the Ear
(Frequency Range of 300 MHz to 3 GHz), International Electrotechnical Commission, Geneva, (2005).
47. IEEE C95.2-1999, IEEE Standard for Radio-Frequency Energy and Current-Flow Symbols, IEEE New
York, NY.
1 The American Standards Association later became the American National Standards Institute (ANSI) that now serves as a
clearing house for standards developed through an open consensus process.

2 In 1963 the American Institute of Electrical Engineers merged with the Institute of Radio Engineers to form a new professional
society, the Institute of Electrical and Electronics Engineers (IEEE).

3 At the time (1974), the C95 Committee consisted of the following seven subcommittees: SC1 (Techniques, Procedures and
Instrumentation); SC2/3 Terminology and Units of Measurements); SC4 (Safety Levels and/or Tolerances with Respect to
Personnel); SC5 (Safety Levels and/or Tolerances with Respect to Electro-Explosive Devices); SC6 (Safety Levels and/or
Tolerances with Respect to Flammable Materials); SC7 (Medical Surveillance).

4 NCRP is a non-profit corporation chartered by the U.S. Congress. The Charter of the NCRP includes as one of its objectives
“To collect, analyze, develop and disseminate in the public interest information and recommendations about (a) protection
against radiation (referred to herein as radiation protection) and (b) radiation measurements, quantities and units, particularly
those concerned with radiation protection.” Although more focused on “ionizing radiation,” e.g., X-rays, gamma-rays, nuclear
radiation, NCRP has developed several reports that address radiofrequency issues.

5 During the period the revision of ANSI C95.1-1982 was developed (1982-1990), ANSI ceased sponsoring standards
committees and instead became a clearing house for standards developed by committees accredited by ANSI. Although the
C95 committee became an ANSI Accredited Standards Committee (ANSI ASC C95), there was consensus of the membership
that it would be beneficial to explore the possibility of operating under the sponsorship of the Institute of Electrical and
Electrical Engineers (IEEE), which is also an ANSI accredited standards developer. After several meetings with IEEE staff, in
1989 the C95 committee began operating as a Standards Coordinating Committee (SCC28 – now the International Committee
on Electromagnetic Safety – ICES) under the sponsorship and subject to the rigid rules, procedures, and oversight of the IEEE
Standards Board (now the IEEE Standards Association Standards Board – SASB).
6 IEEE is a non-profit technical professional society with more than 365,000 members in 150 countries. Within IEEE are 39
societies, including the Consumer Electronics Society, Education Society, Electromagnetic Compatibility Society, Engineering
in Medicine and Biology Society, Information Theory Society, Neural Networks Society, Society on Social Implications of
Technology. While many IEEE societies sponsor standards committees, when the scope of a proposed standard overlaps the
scope of several societies, “Standards Coordinating Committees” (SCC) are established to develop such standards. IEEE
membership is not a requirement for participation on an IEEE SCC or on any of its subcommittees.

7 IEEE SCC34 was a standards coordinating committee established in 1995 to develop product standards relative to the safe
use of electromagnetic energy. At the time, SCC34 SC2 was developing a standard (IEEE 1528) for measuring the peak spatial-
average SAR from RF-emitting devices intended to be operated while placed next to the head (mobile telephones).

8 As indicated on their website “ICNIRP’s beginnings go back to 1973 when, during the 3rd International Congress of the
International Radiation Protection Association (IRPA), for the first time, a session on non-ionizing radiation protection was
organized.” This was followed in 1974 by the formation of a working group on non-ionizing radiation, in 1975 by a study group
to review the field of non-ionizing radiation, in 1977 by the creation of the International Non-Ionizing Radiation Committee
(INIRC), and in 1992 ICNIRP was chartered as an independent non-profit scientific body. ICNIRP is also a formally
recognized non-governmental organization in non-ionizing radiation for the World Health Organization and the International
Labour Office. The work of ICNIRP is carried out by the main Commission, with support of consulting members and four
Standing Committees; Epidemiology, Biology, Physics and Engineering and Optical Radiation.

9 Basic restrictions are provided over the frequency range extending from “up to 1 Hz” to 300 GHz.

10 In order to provide a better description of the international aspects of its activities, the name “IEEE International Committee
on Electromagnetic Safety” (ICES) was approved for use by the IEEE SASB in 2000. Then, in March 2005, the IEEE SASB
approved a new committee, still called ICES, which includes two technical committees: TC34 (formerly SCC34) and TC95
(formerly SCC28). This new committee operates as a standards coordinating committee (SCC39) under the sponsorship of the
IEEE SASB. Currently the membership of the two technical committees (not including subcommittee members) stands at
approximately 150 professionals, with a balance of disciplines and a balanced representation from the medical, scientific,
engineering, industrial, government, and military communities, representing 26 countries. ICES is now international and
influence of the C95 standards is now global in scope. Through the World Health Organization’s standards harmonization
effort, ICES is working closely with other expert groups toward the development a single science-based global standard.

11 The complete list of papers in the IEEE/WHO database is available online at Internet site http://www10.who.int/peh-
emf/emfstudies/IEEEdatabase.cfm

12 For purposes of this standard, the weight of scientific evidence is defined as “the outcome of assessing the published
information about the biological and health effects from exposure to RF energy. This process includes evaluation of the quality
of test methods, the size and power of the study designs, the consistency of results across studies, and the biological
plausibility of dose-response relationships and statistical associations.”

13 The rationale for changing the peak spatial-average SAR and averaging volume was in part the desire to harmonize with the
ICNIRP guidelines where scientifically justified and in part based on recent theoretical biophysical research and
thermophysiological data showing the inability of RF energy to cause significant local temperature increases in small tissue
volumes for inducing adverse health effects .

16
The BioElectric Shield Company has been dedicated to helping create a more balanced and peaceful
world one person at a time since 1990.

In the 1980’s, when Dr. Charles Brown, DABCN, (Diplomate American College of Chiropractic
Neurologists), the inventor of the Shield, became aware that a certain group of his patients exhibited
consistent symptoms of stress and a slower rate of healing that the rest of his patient population. This
group of patients all worked long hours in front of CRT computer screens for many hours a day, and
usually 6 days a week. He began researching the effects of electromagnetic radiation in the literature, and
found there were many associated health effects. He wanted to help these patients, and hoped that he
could come up with a low-tech, high effect product.

In 1989, he had a series of waking dream that


showed him a specific pattern of crystals. Each
of 3 dreams clarified the placement of the
crystals. He showed the patterns to an individual
who can see energy and she confirmed that the
pattern produced several positive effects. She
explained that the Shield interacts with a
person’s energy field (aura) to strengthen and
balance it. Effectively it created a cocoon of
energy that deflects away energies that are
not compatible. In addition, the Shield acts to
balance the physical, mental, emotional and spiritual bodies of the aura.

A series of studies was conducted to investigate the possible protection from EMF's wearing this kind of
device. Happily the studies were consistent in showing that people remained strong when exposed to
these frequencies. Without the shield, most people showed measurable weakening in the presence of both
EMF’s and stress. Of interest to us was that these same effects were noted when people IMAGINED
stress in their lives. It seems obvious that how we think and what we are exposed to physically both have
an energy impact on us. The Shield addresses energy issues-stabilizing a person's energy in adverse
conditions. See “How the Shield Works” for more information.

Since that time, we have sold tens of thousands of Shields and had feedback from more people than we
could possibly list. Here are just a few of the testimonials we have gotten back from Shield wearers.

Dr. David Getoff was one of the earliest practitioners to begin wearing a Shield and doing his own testing
with patients with very good results (video).

BioElectricShield.com | About the Shield | Shield Products | In the Media | Contact


OUR MISSION

Our mission is to make the BioElectric Shield available worldwide. In doing so, we feel we are part of the
solution to the health crisis that is, in part, caused by exposure to electromagnetic radiation and well as
exposure to massive amounts of stress, from situations and other people’s energy.

We also want to bring more peace, balance and joy to the world - and the Shield offers a vibration of
peace, love, and balance in a world filled with fear and uncertainty. Selling a Shield may seem like a
small thing in the scheme of things, but each Shield helps one more person find a greater sense of ease,
balance and protection, allowing them to focus on living their dreams

To enhance your sense of well-being, (In addition to the Shield, ) we offer other products that provide
health and wellness benefits on many levels.

By working together we can, and are, accomplishing miracles.

Charles W. Brown, D.C., D.A.B.C.N.

Dr. Brown graduated in 1979 w ith honors from Palmer College of Chiropractic. He
is a Di plomate of the National Board of Chiropractic Examiners and a Di plomate of
the A merican B oard o f Chiropractic Neu rologists. He al so i s cer tified i n Ap plied
Kinesiology. Dr. Brown has had his own radio show "Health Tips". Additionally, he
has t aught an atomy at B oston Un iversity an d t he New E ngland I nstitute o f M assage
Therapy.

He invented the BioElectric Shield, Conditioning Yourself for Peak Performance (a DVD of series of
Peak Performance Postures with Declarations) and Dr. Brown’s Dust and Allergy Air Filters, as well as
Dr. Brown’s Dust and Allergy Anti-Microbial, Anti-Viral Spray. He is presently working on other
inventions.

Dr. Brown’s experience of the Shield is that it has helped him move deeper into spiritual realms, quantum
energy, and creative meditative spaces. It has always been his desire to help others, and he is grateful that
the Shield is helping so many people worldwide.

Virginia Bonta Brown, M.S., O.T.R.

As child, I always wanted others feel better. As a teenager, I volunteered as a


Candy Striper at the local hospital, wheeling around a cart of gifts to patients’
rooms. The hospital setting didn’t really draw me, so summers were spend
teaching tennis to kids at a wonderful camp in Vermont. With the idea of
becoming a psychologist, I received a B.S. degree from Hollins College in

BioElectricShield.com | About the Shield | Shield Products | In the Media | Contact


psychology and worked with drug addicts for a year. Called by the practicality of Occupational Therapy, I
received an M.S. degree in Occupational Therapy from Boston University in 1974

For the next 16 years, working with ADD, ADHD, autistic and other special needs children was my
passion. Because of my specialty in Sensory Integration Dysfunction (a technique based on neurology), I
met Anne Shumway Cook, RPT, PhD, a brilliant PT, with a PhD in neurophysiology. We created special
therapy techniques for children with vestibular (balance and position in space) dysfunction while she
worked with the Vestibular Treatment Center at Good Samaritan, and while I managed the therapy
services of the Children's Program at this same hospital in Portland, Oregon. A fun project at that time
also included collaborating with a team of other therapists to create a therapy in the public schools manual
for OT, PT and Adapted PT procedures. It included goals and treatment plans which has served as a
model for nearly every school district in the United States. There was nothing quite so satisfying as
seeing a child move from frustration to joy as they began to master their coordination and perceptual
skills.

For the next seven years, I shifted my focus. Married to Dr. Charles Brown, we decided that I’d begin to
work with him in his Pain and Allergy Clinic, first in Boston and then in Billings, Montana. During this
time I began to hear people talk about how thoroughly stressed out they were by their job environment.
Their neck and shoulders hurt from sitting in front of computer screens. They were fatigued and
overloaded dealing with deadlines and other stressed out people! They wanted to be sheltered from the
“storm” of life. Though conversation, myofascial deep tissue and cranio-sacral therapy helped them, the
stress never disappeared. It was our patients who really let us know that something that managed their
environment and their energy would be a wonderful miracle in their lives.

What could we do to help them? I became an OT so I could help children and adults accomplish whatever
it was that they wanted to do. When my husband, Dr. Brown, invented the Shield, initially I felt I was
abandoning my patients. Running the company meant I didn’t spend as much time in the clinic. But then
I saw what the Shield was accomplishing with people. They got Shields and their lives began to improve.
People told me they felt less overwhelmed, didn’t get the headaches in front of the computer, were less
affected by other people’s energy and enjoyed life more. I began noticing the same thing!

In 2000, we received a request for a customized shield for a child with ADD/ADHD. After it was
designed, our consultant told us that she could create a special shield that would help any person with
these symptoms. Read more about the ADD/ADHD Shield.

When we started the company in 1990, I was still seeing patients nearly full time. I was wearing the
Shield and began to notice something different about my own life. At the clinic, I noticed my energy was
very steady all day. Instead of being exhausted at the end of the day, particularly when I had treated
particularly needy patients, I was pleasantly tired and content. I noticed I was more detached from the
patient’s problem. In other words, I didn’t allow it to tire me. Instead I became more compassionate and
intuitive about what they needed to help them. I was able to hear my Guides more clearly as they helped
me help them. As I wore it during meditation, I felt myself go deeper into a space of Unity of all things,
from people to mountains to stars.

BioElectricShield.com | About the Shield | Shield Products | In the Media | Contact


Over the years, I’ve spoken with many, many people, from all walks of life. Because they consistently
tell me how much it’s helped them, I become more committed each year to offer this to as many people
as possible. It is my belief that the Shield is a gift from the Divine, and that those who wear it will be
helped on earth to accomplish their own mission, with greater health and greater compassion. For this
reason, it is my desire to provide the blessing of the BioElectric Shield to as many people as possible.

Carolyn (Workinger) Nau:

I joined the BioElectric Shield Company in January 1994 when the shipping and order
department consisted of one computer and a card table. With my help, the company
grew to what it is today. From 1994 to 2000 I traveled and did approximately 100 trade
shows, talking to people, muscle testing and really finding out how much difference the
Shield makes in people’s lives.

An empath and natural intuitive, I have personally found the Shield to be one of my most important and
valued possessions, as it assists me in not taking on everyone else’s stuff. That ability has also been
invaluable when I talk to and connect with clients in person, over the phone or even via email. I am
frequently able to “tune in” and help advise on the best Shield choice for an individual.

I felt a strong pull to move to California and reluctantly left the company in 2000. While in California I
met the love of my life, David Nau. After being married on the pier in Capitola, we relocated to
Milwaukee, Wisconsin where he’d accepted a job as design director of an award winning exhibit firm.
David is an artist and designer, and has taken all the newest photos of the Shields. They are the most
beautiful and accurate images we have ever had!
Through the magic of the internet I was able to return to working with the company in January 2008. I
love how things have changed to allow me to live where I want and work from home. I am fully involved
and even more excited about the Shield’s benefits and the need for people to be strengthened and
protected. I am thrilled to be back and loving connecting with old and new customers. It’s great to pick up
the phone and have someone say, “Wow, I remember you. You sold me a Shield in Vegas in 1999”

How did I get started making Energy Necklaces? It's not every day that going to a trade show can totally
change your life. It did mine. I must have been ready for a drastic change. I just didn't know it. I guess
I’ve just always been a natural Quester.

Quite by chance, I went to the Bead and Button Show in Milwaukee. The show is an entire convention
center filled with beads, baubles and semi-precious stones. I looked over my purchases at the end of the
first day and realized I didn't have enough of some for earrings. So I went back with a friend who
normally is the voice of reason. I thought if I got carried away she’d help me stop. Joke was on me.

I was unable to resist all those incredible goodies. My friend turned out to be a very bad influence, she’d
find fabulous things and hold semi-precious and even precious stones in front of me saying "Have you
seen this?". How can a woman resist all that beauty? I can’t! I couldn't. I walked out with a suitcase full
of beads and stones. The only problem was, I didn’t even know how to make jewelry.

BioElectricShield.com | About the Shield | Shield Products | In the Media | Contact


I spent the summer taking classes, reading books, practicing jewelry making. Immediately people were
stopping me in the street asking about the jewelry I was wearing. It finally dawned on me that just maybe
I was meant to design and share my creations. Thus Bold Bodacious Jewelry was born.

I still laugh about this whole process. Obviously the Universe or someone was guiding me. Looking back
it should have been obvious that I was buying enough to start a business. But at the time, it just felt like
the right thing to do. Not a conscious plan. Sometimes following your gut can change your life.

In the fall of 2008, I felt a pull to examine how various gemstones could enhance the protective and
healing effects of the BioElectric Shield. I also wanted to wear great jewelry and my gold and diamond
Shied at the same time, so I created something new so I could do that. After making a few “Shield energy
necklaces”, I was convinced that not only was my jewelry beautiful and fun to wear, it had additional
healing qualities as well. Since then I’ve been immersed in studying stones and their properties, paying
particular attention to the magical transformation that happens when stones are combined. Much like the
Shield, the combined properties of the stones in my jewelry are more powerful than the same combination
of stones loose in your hand. To view gem properties and styles to complement your shield, please visit
Shield Energy necklaces .

David Nau:

We’re pleased to have added David to our team. David is an award winning creative
designer who readily calls on the wide variety of experience he has gained in a
design career spanning over thirty years. His familiarity with the business allows
him to create a stunning design, but also one that works for the needs of the client.
The design has impact, and functions as needed for a successful event. Having
owned his own business, David maintains awareness of cost as he designs, assuring the most value
achieved within a budget.

A Graduate of Pratt Institute, Brooklyn, NY, David’s career has included positions as Senior Exhibit
Designer, Owner of an exhibit design company, Design Director, and Salesman. This variety of positions
has provided experience in all phases of the exhibit business; designing, quoting, selling, directly working
with clients, interfacing with builders and manufacturers, staging and supervising set-up.

David has worked closely with many key clients in the branding of their products and themselves in all
phases of marketing, both within and outside the tradeshow realm. He has designed tradeshow exhibits,
museum environments and showrooms for many large accounts including Kodak, Commerce One,
Candela Laser, The Holmes Group, Kendell Hospital Products, Enterasys, Stratus, Pfizer, Ligand
Medical, Polaroid, Welch Allyn, and Nortel. He has also designed museum and visitor centers for
Charlottesville, NASA Goddard, Hartford and Boston children’s museums.

David’s artistic eye has added to other aspects of our BioElectric Shield site and we appreciate his
ongoing contributions. David is currently unemployed and so has started going to trade shows with
Carolyn. For someone who has been designing trade shows for 35 years actually being in the booth he
designed is a whole new experience for him.

BioElectricShield.com | About the Shield | Shield Products | In the Media | Contact


Sam Sokol

Sam is our Internet consultant, bringing expertise and wisdom to this area of
communication for our company. Sam works with a wide variety of companies
in many industries to build, market and maintain their online presence. He has
helped both small and big companies to increase their online sales and build their
businesses. He has helped us to grow BioElectric Shield by giving us direct
access to great tools to make changes to our web site.

Dedicated to helping create a more balanced and peaceful world one person at a time Let's change our
lives and our worlds one thought, one action at a time.

BioElectricShield.com | About the Shield | Shield Products | In the Media | Contact