British Journal of Anaesthesia 1992; 68: 48-53
SIGHS AND THEIR EFFECT ON THE BREATHING OF PATIENTS
ANAESTHETIZED WITH INFUSIONS OF PROPOFOL
N. W. GOODMAN AND I. G. KESTIN
SUMMARY
Twenty-one spontaneous sighs were analysed from
records of the breathing of 10 patients anaesthetized
for 22-42 (mean 28.7) min with propofol infusions.
Sighs occurred in eight patients, the rate varying
between once in 29 min and four times in 26 min.
There was no pattern in breaths preceding sighs, but
the succeeding breaths were altered. On the first
succeeding breath, tidal volume was reduced by
a mean of 32% (95% confidence limits 19-44%;
P < 0.01) and inspirator/ time by a mean of 15%
(95% confidence limits 8-22%; P < 0.01) of the
means of the preceding breaths. These effects lasted
on average at least 10 breaths. Expiratory time was
usually slightly prolonged after a sigh, but this
effect was less clear, less consistent and less
prolonged. Sighs in patients anaesthetized with
propofol reduce the ventilatory drive (in terms of
mean inspirator/ flow), and alter the timing, of
succeeding breaths.
KEY WORDS
Anaesthetics, intravenous propofol. Ventilation: spontaneous
sighing.
Bendixen, Smith and Mead [1] described a type of
deep breath in which a sudden reinforcement occurs
close to the peak of an otherwise normal inspiration:
an "augmented breath" [2]. Bartlett [3] studied rats
and described these biphasic breaths as "a vagally
mediated mechanoreflex, which requires and is
regulated by afferent information from peripheral
chemoreceptors ", although intact pulmonary vagal
innervation is not essential either in cats [4] or in
humans; for instance, people with heart-lung trans-
plants show augmented breaths [5]. Augmented
breaths affect the breaths that follow, and Khoo and
Marmarelis [6] suggested that the disturbance might
be used to calculate the gain of the peripheral
chemoreflex.
We have studied the occurrence of augmented
breaths, which for convenience we refer to as
"sighs" [7], in patients anaesthetized with infusions
of propofol. The two main points of interest were the
mechanisms of sighs and the effects of the sighs on
the subsequent pattern of breathing. We have
compared these patterns with those reported pre-
viously in humans and animals, and comment on
how the ideas of Khoo and Marmarelis [6] might
apply to anaesthetized man.
PATIENTS AND METHODS
The observations reported here were made during a
study of steady-state breathing before and during
infusions of propofol. The study was approved by
Southmead Hospital's Ethics Committee. We have
included here only the methods relevant to the
description of sighs.
Ten patients (nine female) gave consent to take
part in the study. We chose only those patients in
whom we considered the airway would be easy to
manage under general anaesthesia. They were aged
24-^16 yr and weighed 55-76 kg, and were about to
undergo gynaecological or dental surgery.
The patients were unpremedicated. A vein was
cannulated in each forearm, one for the infusion and
one for sampling. A probe for pulse oximetry was
attached to a finger and an arterial pressure cuff was
placed on the other arm for a non-invasive device.
Breathing was recorded by respiratory inductance
plethysmography (Respitrace model 10.9000) and
stored and analysed by a computer program (BBC
B + ) as reported previously [8]. In addition, for the
later patients, the output from the Respitrace was
recorded on magnetic tape (Racal Thermionic Store
4).
For the first part of the study, patients lay quietly
while resting breathing was recorded. Propofol was
given according to a scheme based on the work of
Roberts' group [9] and designed to give 20 min of
quasi-stable sedation followed by 20 min of quasi-
stable anaesthesia. An i.v. injection of propofol
0.5 mg kg"1 was given and the "sedating" infusion
started at 5 mg kg"1 h"1. After 10 min, the rate was
altered to 4 mg kg"1 h"1 and after 10 min at this rate
a second injection (1 mg kg"1) was given and the
"anaesthetizing" infusion started at 13 mg kg"1 h"1.
The rate was reduced after 10 min to 11 mg kg"1 h"1,
and after another 10 min to 9 mg kg"1 h"1. This final
rate was continued until the anaesthetist in clinical
charge of the patient for the operation took over
responsibility.
Patients breathed air unless the pulse oximeter
indicated a saturation of 90% or less. This happened
sometimes if apnoea occurred at the onset of true
anaesthesia. Most patients had minor degrees of
NEVILLE W. GOODMAN, M.A., D.PHIL., B.M., B.CH., F.C.ANAES. ; IAN
G. KESTIN, B.A., M.B., B.S., F.C.ANAES. ; University Department of
Anaesthesia, Medical School Unit, Southmead Hospital, Bristol
BS10 5NB. Accepted for Publication: June 14, 1991.
SIGHS UNDER PROPOFOL ANAESTHESIA
FIG. 1. A sigh in a patient anaesthetized with propofol. Output (uncalibrated, with inspiration upwards) from
respiratory inductance plethysmogrsphy (Respitrace): rib cage above; abdomen below. Time trace (top) shows
seconds. Note that the sigh develops from peak inspiration of what until then seems a normal breath.
airway obstruction intermittently during sedation,
but needed no intervention. Most patients required
some support of the airway when anaesthetized,
although no patient needed more than a finger on the
point of the chin.
Analysis
We looked at all breaths having the typical biphasic
inspiratory pattern of a sigh. We calculated the mean
tidal volume, inspiratory time and expiratory time of
the 10 breaths preceding each sigh, and used these
baseline values to normalize the sigh. Normalized
values from all analysed sighs were pooled to give
normalized means (and 95 % confidence limits (CL)
on those means) for each of the 10 preceding breaths,
the sighs and a number of breaths after the sighs.
Bendixen's group [1] suggested that, in man, any
disturbance lasted no more than 10 breaths; from
our initial analysis taking 10 succeeding breaths, the
variables had not always clearly returned to their
baseline values by then, so we analysed instead the
20 succeeding breaths.
For each of tidal volume, inspiratory time and
expiratory time, we calculated the 95 % CL on the
differences between the breaths immediately pre-
ceding and succeeding the sighs. We applied paired
Student's t test to determine how long the effect of
the sigh lasted. Spearman rank correlation was used
to test if the tidal volume of the sigh affected the
response to it.
Frequency and mean inspiratory flow were derived
from the measured variables.
The effect of sighs on the end-expiratory position
(measured as the "volume" registered at end-
expiration) and of any change in the relative
contribution of rib cage and abdomen to the tidal
volume were tested by Wilcoxon paired rank sum
test.
49
Numerical and statistical calculations were made
within a spreadsheet (ViewSheet B1.0: Acorn Com-
puters Ltd) or using Unistat-II (Unisoft Ltd). All
CL values are of the means. P < 0.05 was taken as
significant.
RESULTS
All patients remained rousable by quiet voice during
the sedating infusion; all were unrousable during the
anaesthetizing infusion, although some occasionally
made slight movements of the arms.
Sighs when anaesthetized
Anaesthetized breathing was recorded for between
22 and 42 min (mean 28.7 min). During this time
there were 25 sighs. Two occurred within a few
breaths of the second, anaesthetizing bolus. Dis-
counting these, the overall rate of sighing was about
once every 12 min, although this varied: one patient
did not sigh at all during 20 min when anaesthetized;
one patient sighed four times in 26 min and another
four times in 33 min. We saw no sighs less than
6 min apart. All sighs were typical augmented
breaths (fig. 1), although the variability of baseline
tidal volume was greater in some patients than
others.
Four sighs could not be analysed: the two
occurring after induction; one from a noisy section
of record; and another that was followed by a long
apnoea. This left 21 sighs from eight patients. These
21 sighs were registered by the Respitrace as having
tidal volumes 2.1 to 6.6 (mean 3.9) times as large as
baseline, with inspiratory times 1.4 to 2.1 (mean 1.6)
times as long and expiratory times 1.1 to 2.8 (mean
1.7) times as long.
The general effect of a sigh on the succeeding
breaths was to reduce tidal volume, shorten in-
50
1.2 -i
_3
O
0.6 -
1.2 -i
o
0.8 H
0.6 -
1.2 -i
o
D.
X Time (s)
0.8 -
-10 -5 sigh +5 +10 +15 +20
Breath number
FIG. 2. Effect of sighs under propofol anaesthesia on tidal volume,
inspiratory time and expiratory time of succeeding breaths. The
21 sighs have each been normalized to the means of each variable
for the preceding 10 breaths. Each point and error bar represent
the means and 95% CL on those means calculated from the
summed normalized sighs. Points — 10 to +14 are from 21 sighs;
+ 15 and +16 from 19 sighs; +17 and +18 from 18; and +19
and +20 from 17. The actual sighs are omitted.
spiratory time and lengthen expiratory time. The
greatest effect was on tidal volume; the least on
expiratory time.
Effect on tidal volume. The greatest effect was
generally on the tidal volume of the first succeeding
breath, which was less than any individual tidal
volume of the preceding 10 breaths for all but three
sighs. The smallest first succeeding breath was 27 %
of baseline. Of the three exceptions one, in which
tidal volume was unaffected, was recorded while the
patient was breathing 100 % oxygen, one was difficult
to assess because the sigh occurred at a time when
the tidal volume was increasing with time and there
was one sigh for which the greatest effect occurred at
the third succeeding breath.
There was no consistent relation between relative
size of sigh and effect on tidal volume, either
comparing by eye different sighs from the same
BRITISH JOURNAL OF ANAESTHESIA
750 - Q
1 500 -
250 -
3. 1.0 -
•= 0.5 -
4 "
2 -
500 -
250 -
£ 100 -
750
„ "
1 500 -
ffi 250 -\
30 60 90
FIG. 3. Breath-by-breath ventilatory variables around a sigh in a
patient anaesthetized with propofol. This is an analysis of the sigh
shown in figure 1. From above downwards: tidal volume (KT),
inspiratory rime (Tl), expiratory time (TE), mean inspiratory flow
( F T / T O , abdominal contribution (Abd.), and end-expiratory
position (EEP). The sigh is circled.
patient, or when all 21 sighs were grouped (Spear-
man's p = -0.02; P = 0.47).
The pooled, normalized tidal volumes are shown
in figure 2 (upper panel). The reduction, by 31.7%
(95% CL 19.4-^14.0%; P < 0.05) to about 70% of
baseline for the first breath after the sigh, returned
gradually towards baseline over 10-15 breaths. The
pooled 14th breath was the last to be significantly
different from baseline (P = 0.034).
Effect on inspiratory time. Inspiratory time was
decreased after all sighs, including the three with no
or unclear effect on tidal volume. The overall effect
on inspiratory time was less than on tidal volume (a
reduction by a mean of 15.1 %; 95 % CL 7.9-22.3%;
P < 0.01), but had about the same time course and
with the same proviso of remaining below the
baseline even at 15-20 breaths (fig. 2, middle panel).
Effect on expiratory time. For eight sighs, the
expiratory time of the first succeeding breath was
greater than any of the preceding 10 breaths. For the
other 13, eight were greater than the baseline mean,
but five were less. The mean increase in expiratory
time between the preceding and succeeding breath
 SIGHS UNDER PROPOFOL ANAESTHESIA 51
1500 -| - 3 9 % to +3%) (Wilcoxon P < 0.01). The usual
pattern was then for a gradual return to the relative
_ 1000 -
contributions of abdomen and rib cage seen in the
breaths preceding each sigh.
^ 500 - Examples of individual sighs. Figure 3 shows die
analysed variables from the sigh illustrated in figure
1, for this study the "typical" sigh and causing a
reduction in tidal volume, a decreased inspiratory
2 -i time, an increased expiratory time, a decrease in
inspiratory flow, a reduction in relative abdominal
-2 1 - contribution and no change in end-expiratory pos-
ition.
Figure 4 represents an artificial sigh, managed by
applying a bag and mask briefly to the face and
6 - o synchronizing a hand-squeezed breath; there was no
4 - response [10] at that breath, but the subsequent
breaths were affected in a manner similar to those
2 - after a spontaneous sigh, except that the end-
expiratory position was increased 136 ml. This
artificial sigh was not included in the analysis.
600 - Sighs when awake or sedated
» 400 - Patients sighed in both these stages. Unfortun-
E ately, sighs often came too soon before or after
200 - periods of arousal, or occurred when the breathing
was too irregular, for formal analysis.

DISCUSSION

© The mechanism of sighing can be separated from die

50 -
1500 -
1000 ^
30 60 90
Time (s)
FIG. 4. Breath-by-breath ventilatory variables around an artificial
sigh in a patient anaesthetized with propofol (same patient as
figure 3). From above downwards: tidal volume (KT), inspiratory
time (7*1), expiratory time (71B), mean inspiratory flow (KT/7I),
abdominal contribution (Abd.), and end-expiratory position
(EEP). The sigh is circled.
was 7.7% (95 % CL - 3 . 3 % to + 18.7%). Any over-
all effect lasted at most three breaths (fig. 2, lower
panel): the diree breaths after the sigh, as a group,
were 11.0% longer (95% CL 2.9-19.1%) than the
three before, but this grouping was post-hoc.
Effect on other variables. The opposing effects of a
decreased inspiratory time and an increased and
more variable expiratory time implied there was no
discernible effect on frequency after the sigh: 17.2 (SD
2.8) b.p.m. for the preceding breath and 17.3 (SD
3.6) b.p.m. for the succeeding breath. Inspiratory
flow was reduced by 20 % (95 % CL 9-31 %) on the
first breath after a sigh.
For the actual augmented breath, the abdominal
contribution to the sigh was usually decreased relative
to the contribution from the rib cage, sometimes
quite markedly (mean — 9.7 %; median — 6 %; range
subsequent effects of sighs on the patterns of
breathing. Mechanism includes the initiating factors,
the frequency of sighing and the shape (that is, die
relation of flow with time) of the sighs. Our
observations do not allow any inferences about the
initiation of sighs.
The frequency and shape were similar to those
described previously. The healthy young adults
studied by Bendixen's group [1] sighed about 10
times per hour. Patients with heart-lung transplants
but otherwise well sighed seven to eight times per
hour [5], and their breathing patterns were the same
as a control group. Patients anaesthetized with
isoflurane sighed six times per hour [11]. Our
anaesthetized subjects sighed on average five times
per hour, but in all studies on sighs the range of
frequency of sighing is large; we can make no
assertions about the effect of anaesthesia or of
propofol on die frequency of sighing.
The sighs were typically augmented breaths [1,2],
so presumably the motor response giving rise to
sighs remains intact during propofol anaesthesia.
Anaesthesia may alter die response to the sigh.
Figure 5 in Bendixen's paper [1] clearly shows a sigh
followed by breaths that are slightly larger and
markedly slower than preceding breadis. The investi-
gators made no comment about tidal volume, but
described "an average slowing of the order of two
breaths per minute... limited to the first ten breaths
following the sigh". Shea and colleagues' [5] subjects
showed "no systematic differences in either [fre-
quency] or [tidal volume] in the 3 breadis following
the augmented breath". The typical response to a
sigh for our anaesthetized subjects was closer to
52
Cherniack and co-workers' [4] description in anaes-
thetized cats—" breaths... always smaller in ampli-
tude, and... of shorter duration than the preceding
control breaths"—a description that they reported
as being "in general agreement with previous
workers", all of whom had studied anaesthetized
cats.
The tidal volume was reduced by almost every
sigh in our study although, if we had looked only at
frequency rather than separately at inspiratory and
expiratory times, we would not have detected the
effect of sighs on timing of a decrease in inspiratory
time, which occurred with every sigh, and an increase
in expiratory time, which was less consistent. These
effects have not been reported before. Effects were
quite prolonged: 10-12 breaths represents about
30-40 s at the breathing frequencies of our subjects.
The decreased inspiratory time was not short
enough to account completely for the lesser tidal
volume, so the sighs in our anaesthetized subjects
had two main effects: ventilatory drive (in terms of
mean inspiratory flow) was reduced and "the
inspiratory 'off-switch' threshold [was] temporarily
lowered" [4]. Some dislike the use of the terms "on-
switch" and "off-switch" [12], but anaesthetics
[13], including propofol [14], alter ventilatory
timing. The alteration of timing that follows a sigh
could be inherent to the process of sighing or a reflex
consequence of it. The inspiratory time was reduced
after our one artificial sigh, suggesting a reflex
consequence.
The mechanism of the reduced tidal volume is
more easily explained, and may also be more useful.
Propofol reduces the slope of the ventilatory re-
sponse to carbon dioxide [14]. When a sigh causes a
sudden decrease in the alveolar, and hence arterial,
partial pressure of carbon dioxide, the peripheral
chemoreceptors are easily able to respond to this
change and cause a reflex reduction in ventilatory
drive, the subject moving to a lower point on the
carbon dioxide response curve. This was the basis of
Khoo and Marmarelis's [6] suggestion of using the
effect of a sigh to estimate the gain of the peripheral
chemoreflex. They tested a mathematical model of
their estimation on anaesthetized dogs and in a
simulation.
As the estimation can be made from measures of
ventilation and end-tidal carbon dioxide concen-
tration, without the need for giving the patient
carbon dioxide to breathe, the method may be
feasible for monitoring ventilatory responsiveness
during anaesthesia; but there are problems. In
common with most models of ventilation, that of
Khoo and Marmarelis [6] ignored timing; output is
given as minute ventilation. Normally, anaesthetized
patients respond to carbon dioxide by a change in
tidal volume without a change in frequency [13], and
propofol fits this pattern [15]. If the frequency does
not change, minute ventilation and tidal volume are
equivalent; it does not matter which is measured.
After our sighs, tidal volume decreased partly
because inspiratory time was shorter. This additional
effect of the sigh on the inspiratory off-switch,
through vagal or other mechanisms [4], implies that
estimation of peripheral chemoreflex gain would
BRITISH JOURNAL OF ANAESTHESIA
differ according to whether minute ventilation or
tidal volume was taken as the output.
Khoo and Marmarelis [6] predicted that the
greatest effect of the sigh is on the second or third
succeeding breath, although the peripheral chemo-
receptors may cause changes in breathing more
rapidly than this. This discrepancy may be because
their model ignores timing. A further complication is
that, although anaesthetized patients do not show a
frequency response to inhaled carbon dioxide, the
chemoreceptors are important in the alteration of
timing of the sigh itself, at least in dogs with blocked
vagus nerves [16], in which the expiratory time of the
sigh was four times longer than normal.
In premature neonates, sighing may cause the
breathing to become unstable and to oscillate; this
instability decreases in the first few months of life
[17]. Cleave and co-workers [18] constructed a
mathematical model based on prolonged recordings
of breathing from these infants and analysis of the
breathing after sighs. As did Khoo and Marmarelis
[6], these workers described the output of the
respiratory centre in terms of tidal volume only.
Their model and techniques may have limited
application to anaesthetized adults, at least when
breathing is regular and not unduly disturbed after
sighs because "... to produce [an] unstable equi-
librium point it is necessary to depart from the
normal adult values" [18]. Instability generally
arises when the gain of a system is increased but
anaesthesia reduces gain.
Ponte and Sadler [19] suggested that the apnoea of
induction of anaesthesia with propofol occurred
because propofol depressed chemoreceptor dis-
charge. Our observations suggest that, at clinical
rates of infusion, the peripheral chemoreflex is active.
To measure the gain of the peripheral chemoreflex
from sighs requires an accurate measure of the end-
tidal partial pressure of carbon dioxide, which is not
easy on lightly anaesthetized patients breathing air,
particularly if tidal volumes are small. In animal
work, gain may be increased by hypoxia; inferences
about the gain in anaesthetized humans may come
from comparing sighs in patients breathing air and
then oxygen, because hyperoxia decreases the gain.
REFERENCES
1. Bendixen HH, Smith GM, Mead J. Pattern of ventilation in
young adults. Journal of Applied Physiology 1964; 19:
195-198.
2. Glogowska M, Richardson PS, Widdicorabe JG, Winning
AJ. The role of the vagus nerves, peripheral chemoreceptors
and other afferent pathways in the genesis of augmented
breaths in cats and rabbits. Respiration Physiology 1972; 16:
179-196.
3. Bartlett D. Origin and regulation of spontaneous deep
breaths. Respiration Physiology 1971; 12: 230-238.
4. Chemiack NS, von Euler C, Glogowska M, Homma I.
Characteristics and rate of occurrence of spontaneous and
provoked augmented breaths. Acta Physiologica Scandinavica
1981; 111: 349-360.
5. Shea SA, Homer RL, Banner NR, McKenzie E, Heaton R,
Yacoub MH, Guz A. The effect of human heart—lung
transplantation upon breathing at rest and during sleep.
Respiration Physiology 1988; 72: 131-150.
6. Khoo MCK, Marmarelis VZ. Estimates of peripheral chemo-
reflex gain from spontaneous sigh responses. Annals of
Bwmedicai Engineering 1989; 17: 557-570.
SIGHS UNDER PROPOFOL ANAESTHESIA
7. Cherniack NS, Longobardo GS. Abnormalities in respiratory
rhythms. In: Cherniack NS, Widdicombe JG, eds. Control of
Breathing, Chapter 22 (Vol. II of Fishman AP, ed. The
Respiratory System; Section 3 of Handbook of Phystology.)
Bethesda: American Physiological Society, 1986.
8. Goodman NW, Vanner RG, Wade JA. Effects of incremental
doses of alfentanil and propofol on the breathing of anaes-
thetized patients. British Journal of Anaesthesia 1989; 63:
548-553.
9. Roberts FL, Dixon J, Lewis GTR, Tackley RM, Prys-
Roberts C. Induction and maintenance of propofol an-
aesthesia. Anaesthesia 1988; 43 (Suppl.): 14-17.
10. Reynolds LB. Characteristics of an inspiration-augmenting
reflex in anesthetized cats. Journal of Applied Physiology
1962; 17: 683-688.
11. Grim PS, Frcund PR, Cheney FW. Effect of spontaneous
sighs on arterial oxygenation during isoflurane anesthesia in
humans. Anesthesia and Analgesia 1987; 66: 839-842.
12. Ponte J. Breathing pattern and propofol. British Journal of
Anaesthesia 1990; 64: 528.
13. Pavlin EG, Hornbein TF. Anesthesia and the control of
ventilation. In: Chemiack NS, Widdicome JG, eds. Control
of Breathing, Chapter 25. (Volume II of Fishman AP, ed. The
53
Respiratory System; Section 3 of Handbook of Physiology.)
Bethesda: American Physiological Society, 1986.
14. Goodman NW, Black AMS, Carter JA. Some ventilatory
effects of propofol as a sole anaesthetic agent. British Journal
of Anaesthesia 1987; 59: 1497-1503.
15. Goodman NW, Black AMS. Inter-relations of the volume
and timing components of ventilation during carbon dioxide
rebreathing in awake and anaesthetized subjects. British
Journal of Anaesthesia 1987; 59: 1504-1513.
16. Bowes G, Andrey SM, Kozar LF. Carotid chemoreceptor
regulation of expiratory duration. Journal of Applied Physi-
ology : Respiration, Environmental and Exercise Physiology
1983; 54: 1195-1201.
17. Fleming PJ, Goncalves AL, Lcvine MR, Woollard S. The
development of stability of respiration in human infants:
changes in ventilatory responses to spontaneous sighs. Journal
of Physiology (London) 1984; 347: 1-16.
18. Cleave JP, Levine MR, Fleming PJ, Long AM. Hopf
bifurcations and the stability of the respiratory control system.
Journal of Theoretical Biology 1986; 119: 299-318.
19. Ponte J, Sadler CL. Effect of thiopentone, etomidate and
propofol on carotid body chemoreceptor activity in the rabbit
and the cat. British Journal of Anaesthesia 1989; 62: 41-45.