FletcherRobertH 2014 Chapter2Frequency ClinicalEpidemiologyT

Chapter 2
All rights reserved. May not be reproduced in any form without permission from the publisher, except fair uses permitted under U.S. or applicable copyright law.
Frequency
Here, it is necessary to count.
—P.C.A. Louis†
1787–1872a
KEY WORDS
Numerator Cohort studies
Example
Denominator Cumulative A 72-year-old man presents with slowly pro-
Prevalence incidence gressive urinary frequency, hesitancy, and drib-
Point prevalence Incidence density bling. A digital rectal examination reveals a
Period prevalence Person-time symmetrically enlarged prostate gland and no
Incidence Dynamic population nodules. Urinary flow measurements show a
Duration of disease Population at risk reduction in flow rate, and his serum prostate-
Case fatality rate Random sample specific antigen (PSA) is not elevated. The cli-
Survival rate Probability sample nician diagnoses benign prostatic hyperplasia
Complication rate Sampling fraction (BPH). In deciding on treatment, the clinician
Infant mortality rate Oversample and patient must weigh the benefits and haz-
Perinatal mortality Convenience samples ards of various therapeutic options. To simplify,
rate Grab samples let us say the options are medical therapy with
Prevalence studies Epidemic drugs or surgery. The patient might choose
Cross-sectional studies Pandemic medical treatment but runs the risk of worsening
Surveys Epidemic curve symptoms or obstructive renal disease because
Cohort Endemic the treatment is less immediately effective than
surgery. Or he might choose surgery, gaining
immediate relief of symptoms but at the risk
Chapter 1 outlined the questions that clinicians need
of operative mortality and long-term urinary
to answer as they care for patients. Answers are usu-
incontinence and impotence.
ally in the form of probabilities and only rarely as cer-
tainties. Frequencies obtained from clinical research
are the basis for probability estimates for the purposes
of patient care. This chapter describes basic expres- Decisions such as the one this patient and clinician
sions of frequency, how they are obtained from clini- face have traditionally relied on clinical judgment
cal research, and how to recognize threats to their based on experience at the bedside and in the clinics.
Copyright @ 2014. LWW.
validity. In modern times, clinical research has become suffi-

ciently strong and extensive that it is possible to ground
clinical judgment in research-based probabilities—
†
A 19th Century physician and proponent of the “numerical frequencies. Probabilities of disease, improvement,
method” (relying on counts, not impressions) to understand the deterioration, cure, side effects, and death are the
natural history of diseases such as typhoid fever. basis for answering most clinical questions. For this
17
EBSCO : eBook Collection (EBSCOhost) - printed on 5/10/2018 9:42 PM via UNIVERSIDAD DE LA FRONTERA
AN: 1473039 ; Fletcher, Robert H., Fletcher, Suzanne W., Fletcher, Grant S..; Clinical Epidemiology : The
Essentials
Account: s9456359.main.ehost
18 Clinical Epidemiology: The Essentials
patient, sound clinical decision making requires accu- event could have occurred (population). The two basic
rate estimates of how his symptoms and complica- measures of frequency are prevalence and incidence.
tions of treatment will change over time according to
which treatment is chosen. Prevalence
Prevalence is the fraction (proportion or percent) of

ARE WORDS SUITABLE a group of people possessing a clinical condition or
SUBSTITUTES FOR NUMBERS? outcome at a given point in time. Prevalence is mea-
sured by surveying a defined population and counting
Clinicians often communicate probabilities as words the number of people with and without the condition
(e.g., usually, sometimes, rarely) rather than as num- of interest. Point prevalence is measured at a sin-
bers. Substituting words for numbers is convenient gle point in time for each person (although actual
and avoids making a precise statement when one is measurements need not necessarily be made at the
uncertain about a probability. However, words are same point in calendar time for all the people in the
a poor substitute for numbers because there is little population). Period prevalence describes cases that
agreement about the meanings of commonly used were present at any time during a specified period
adjectives describing probabilities. of time.
Incidence
Example Incidence is the fraction or proportion of a group of
people initially free of the outcome of interest that devel-
Physicians were asked to assign percentage val- ops the condition over a given period of time. Incidence
ues to 13 expressions of probability (1). These refers then to new cases of disease occurring in a popula-
physicians generally agreed on probabilities tion initially free of the disease or new outcomes such as
corresponding to adjectives such as “always” or symptoms or complications occurring in patients with a
“never” describing very likely or very unlikely disease who are initially free of these problems.
events but not on expressions associated with Figure 2.1 illustrates the differences between inci-
less extreme probabilities. For example, the dence and prevalence. It shows the occurrence of
range of probabilities (from the top to the bot-
tom tenth of attending physicians) was 60% to
2010 2011 2012
90% for “usually,” 5% to 45% for sometimes,
and 1% to 30% for “seldom.” This suggests (as
authors of an earlier study had asserted) that
“difference of opinion among physicians regard-
ing the management of a problem may reflect
differences in the meaning ascribed to words
used to define probability” (2).
Patients also assign widely varying probabilities to

word descriptions. In another study, highly skilled
and professional workers outside of medicine thought
“usually” referred to probabilities of 35% to 100%;
“rarely” meant to them a probability of 0% to 15% (3).
Thus, substituting words for numbers diminishes
the information conveyed. We advocate using num-
bers whenever possible.
Onset
PREVALENCE AND INCIDENCE Duration
In general, clinically relevant measures of frequency are

expressed as proportions, in which the numerator is the
number of patients experiencing an event (cases) and Figure 2.1 ■ Incidence and prevalence. Occurrence of
the denominator is the number of people in whom the disease in 10,000 people at risk for lung cancer, 2010 to 2012.
Essentials
Chapter 2: Frequency 19
lung cancer in a population of 10,000 people over Prevalence and Incidence

the course of 3 years (2010–2012). As time passes, in Relation to Time
individuals in the population develop the disease.
They remain in this state until they either recover Every measure of disease frequency necessarily con-
or die—in the case of lung cancer, they usually die. tains some indication of time. With measures of prev-
Four people already had lung cancer before 2010, and alence, time is assumed to be instantaneous, as in a
16 people developed it during the 3 years of obser- single frame from a motion picture film. Prevalence
vation. The rest of the original 10,000 people have depicts the situation at that point in time for each
not had lung cancer during these 3 years and do not patient, even though it may, in reality, have taken sev-
appear in the figure. eral months to collect observations on the various peo-
To calculate prevalence of lung cancer at the ple in the population. However, for incidence, time
beginning of 2010, four cases already existed, so is the interval during which susceptible people were
the prevalence at that point in time is 4/10,000. observed for the emergence of the event of interest.
If all surviving people are examined at the begin- Table 2.1 summarizes the characteristics of incidence
ning of each year, one can compute the prevalence and prevalence.
at those points in time. At the beginning of 2011, Why is it important to know the difference between
the prevalence is 5/9,996 because two of the pre- prevalence and incidence? Because they answer two
2010 patients are still alive, as are three other people entirely different questions: on the one hand, “What
who developed lung cancer in 2010; the denomi- proportion of a group of people has a condition?”;
nator is reduced by the 4 patients who died before and on the other, “At what rate do new cases arise in a
2011. Prevalence can be computed for each of the defined population as time passes?” The answer to one
other two annual examinations and is 7/9,992 at question cannot be obtained directly from the answer
the beginning of 2011 and 5/9,986 at the beginning to the other.
of 2012.
To calculate the incidence of new cases develop- RELATIONSHIPS AMONG
ing in the population, we consider only the 9,996 PREVALENCE, INCIDENCE,
people free of the disease at the beginning of 2010 AND DURATION OF DISEASE
and what happens to them over the next 3 years. Five
new lung cancers developed in 2010, six developed Anything that increases the duration of disease
in 2011, and five additional lung cancers developed increases the chances that the patient will be identi-
in 2012. The 3-year incidence of the disease is all fied in a prevalence study. Another look at Figure 2.1
new cases developing in the 3 years (16) divided by will confirm this. Prevalent cases are those that remain
the number of susceptible individuals at the begin- affected, to the extent that patients are cured, die of
ning of the follow-up period (9,996), or 16/9,996 their disease, or leave the population under study,
in 3 years. What are the annual incidences for 2010, they are no longer a case in a prevalence survey. As a
2011, and 2012? Remembering to remove the previ- result, diseases of brief duration will be more likely to
ous cases from the denominator (they are no longer be missed by a prevalence study. For example, 15%
at risk of developing lung cancer), we would calculate of all deaths from coronary heart disease occur out-
the annual incidences as 5/9,996 in 2010, 6/9,991 in side the hospital within an hour of onset and with-
2011, and 5/9,985 in 2012. out prior symptoms of heart disease. A prevalence
Table 2.1
Characteristics of Incidence and Prevalence
a. Characteristic b. Incidence c. Prevalence

Numerator New cases occurring during a period of time among Existing cases at a point or period of
a group initially free of disease time

Denominator All susceptible people without disease at the All people examined, including cases
beginning of the period and non-cases
Time Duration of the period Single point or period
How measured Cohort study (see Chapter 5) Prevalence (cross-sectional) study
Essentials
study would, therefore, miss nearly all these events Similarly, the prevalence of prostate cancer on
and underestimate the true burden of coronary heart autopsy is so much higher than its incidence that the
disease in the community. In contrast, diseases of majority of these cancers must never become symp-
long duration are well represented in prevalence sur- tomatic enough to be diagnosed during life.
veys, even when their incidence is low. The incidence

of inflammatory bowel disease in North America is
only about 2 to 14 per 100,000/year, but its preva- SOME OTHER RATES
lence is much higher, 37 to 246/100,000, reflecting
Table 2.2 summarizes some rates used in health care.
the chronic nature of the disease (4).
Most of them are expressions of events over time. For
The relationship among incidence, prevalence
example, a case fatality rate (or alternatively, the
and duration of disease in a steady state, in which
survival rate) is the proportion of people having a
none of the variables is changing much over time, is
disease who die of it (or who survive it). For acute dis-
approximated by the following expression:
eases such as Ebola virus infection, follow-up time may
Prevalence = Incidence × Average be implicit, assuming that deaths are counted over a
duration of the disease long enough period of time (in this case, a few weeks)
Alternatively, to account for all of them that might have occurred.
For chronic diseases such as cardiovascular disease or
Prevalence/Incidence = Duration cancer, it is more usual to specify the period of obser-
vation (e.g., the 5-year survival rate). Similarly, com-
plication rate, the proportion of people with a disease
or treatment who experience complications, assumes
Example that enough time has passed for the complications to
have occurred. These kinds of measures can be under-
The incidence and prevalence of ulcerative estimations if follow-up is not really long enough.
colitis were measured in Olmstead County, For example, surgical site infection rates have been
Minnesota, from 1984 to 1993 (5). Incidence underreported because they have been counted up to
was 8.3/100,000 person-years and prevalence the time of hospital discharge, whereas some wound
was 229/10,000 persons. The average dura- infections are first apparent after discharge (6).
tion of this disease can then be estimated as Other rates, such as infant mortality rate and
229/100,000 divided by 8.3/100,000 = 28 years. perinatal mortality rate (defined in Table 2.2) are
Thus, ulcerative colitis is a chronic disease con- approximations of incidence because the children
sistent with a long life expectancy. The assump- in the numerator are not necessarily those in the
tion of steady state was met because data from denominator. In the case of infant mortality rate for a
this same study showed that incidence changed given year, some of the children who die in that year
little during the interval of study. Although were born in the previous year; similarly, the last chil-
rates are different in different parts of the dren to be born in that year may die in the following
world and are changing over longer periods of year. These rates are constructed in this way to make
time, all reflect a chronic disease. measurement more feasible, while providing a useful
approximation of a true rate in a given year.
Table 2.2
Some Commonly Used Rates
Case fatality rates Proportion of patients who die of a disease

Complication rate Proportions of patients who suffer a complication of a disease or its treatment
Infant mortality rate Number of deaths in a year of children <1 year of age
Number of live births in the same year

Perinatal mortality rate (World Number of stillbirths and deaths in the first week of life per 1,000 live births
Health Organization definition)
Maternal mortality rate Number of maternal deaths related to childbirth in a given year
Number of live births in the same population during the same year
Essentials
Defined Representative Disease/outcome

population sample present?
Population
at risk
No
Sample
Yes
Figure 2.2 ■ The design of a prevalence study.
STUDIES OF PREVALENCE period prevalence but a good estimate of point

AND INCIDENCE prevalence because of the narrow time win-
dow) ranged from a high of 4.6% in the United
Prevalence and incidence are measured by entirely
States to a low of 0.9% in Japan. Period preva-
different kinds of studies.
lence was higher; for example, in the United
Prevalence Studies States, the 12-month prevalence was 10.0% and
the lifetime prevalence was 16.9%. The authors
In prevalence studies, people in a population are
concluded that “major depressive episodes are
examined for the presence of the condition of interest.
a commonly occurring disorder that usually has
Some members of the population have the condition
a chronic-intermittent course” (7).
at that point in time, whereas others do not (Fig. 2.2).
The fraction or proportion of the population that has
the condition (i.e., cases) constitutes the prevalence
of the disease. Incidence Studies
Another term for prevalence studies is cross-
The population under examination in an incidence
sectional studies because people are studied at a
study is a cohort, which is defined as a group of peo-
“cross-section” of time. Prevalence studies are also called
ple having something in common when they are first
surveys if the main measurement is a questionnaire.
assembled and are then followed over time for the devel-
The following is an example of a typical prevalence
opment of outcome events. For this reason, incidence
study.
studies are also called cohort studies. A sample of
people free of the outcome of interest is identified and
observed over time to see whether an outcome event
Example occurs. Members of the cohort may be healthy at first
and then followed forward in time for the emergence of
The World Health Organization created a re- disease—for example, from being cancer-free until the
search consortium to study the cross-national onset (or not) of pancreatic cancer. Or, all of them may
prevalence of depression. More than 37,000 have a recently diagnosed disease (such as pancreatic
people, randomly selected from the general cancer) and then be followed forward in time to out-
population, were interviewed in 10 countries comes such as recurrence or death. Incidence studies
in North America, Latin America, Europe, and
will be discussed in greater detail in Chapters 5 and 7.

Asia. Major depressive episodes were diagnosed
by face-to-face interviews, using an instrument Cumulative Incidence
designed to give consistent results in different
languages and cultures. Response rates were To this point, the term “incidence” has been used to
57% to 90%. Thirty-day prevalence (actually a describe the rate of new events in a group of people
of fixed size, all members of which are observed over
Essentials
a period of time. This is called cumulative incidence

because new cases are accumulated over time. all care provided to county residents was pro-
vided within the county and most residents had
Incidence Density (Person-Years) agreed to let their records be used for research.
The population of the county was estimated

Another approach to studying incidence is to mea-
from census data at approximately 175,000.
sure the number of new cases emerging in an ever-
Incidence of herpes zoster, adjusted to the
changing population, one in which individuals are
age and sex of the U.S. adult population, was
under study and susceptible for varying lengths of
3.6 per 1,000 person-years and rose with age.
time. The incidence derived from studies of this type
Pain after the herpes attack occurred in 18%
is called incidence density because it is, figuratively
of these patients.
speaking, the density of new cases in time and place.
Clinical trials often use the incidence density
approach. Eligible patients are enrolled over a period
of time so that early enrollees are treated and followed
up for longer periods than late enrollees. In an effort
to keep the contribution of individual patients com- Incidence of herpes zoster infection was described
mensurate with their follow-up interval, the denomi- in person-years in a dynamic population, whereas pain
nator of an incidence density measure is not persons after infection was a cumulative incidence in which all
at risk for a specific time period but person-time at patients with herpes zoster were followed up.
risk for the outcome event. A patient followed for A disadvantage of the person-years approach is
10 years without an outcome event contributes 10 that it lumps together different lengths of follow-up.
person-years, whereas one followed for 1 year con- A small number of patients followed for a long time
tributes only 1 person-year to the denominator. Inci- can contribute as many person-years as a large num-
dence density is expressed as the number of new cases ber of patients followed for a short time. If patients
per total number of person-years at risk. with long follow-up are systematically different from
The person-years approach is especially useful those with short follow-up—perhaps because out-
for estimating the incidence of disease in dynamic come events take a long time to develop or because
populations, those in which some individuals in
patients with especially bad risk tend to leave the
the population are entering and others leaving it as population—the resulting incidence density will
time passes. Incidence studies in large populations depend on the particular combination of number
typically have an accurate count of new cases in the of patients and follow-up times. For example, the
population (e.g., from hospital records or disease reg- latency period between exposure to carcinogen and
istries), but the size and characteristics of the popula- onset of cancer is at least 10 years for most cancers.
tion at risk can only be estimated (from census and It might be possible to see an increase in cancer rates
other records) because the people in it are entering
and leaving the region continually. This approach
works because the proportion of people who enter or
leave is small, relative to the population as a whole Move into
(Fig. 2.3), so the population is likely to be relatively community
stable over short periods of time.
Born in
community
Example Community
Population
A study of the incidence of herpes zoster in-
fections (“shingles”) and its complications pro-
vides and example of both incidence density

and cumulative incidence. Investigators in Ol-
Die
mstead County, Minnesota reviewed medical
records of adult residents of the county (8).
Move out
Other studies showed that more than 98% of
Figure 2.3 ■ A dynamic population.
Essentials
in a study of 10,000 people exposed to a carcinogen

and followed up for 20 years. However, a study of
100,000 people followed for 2 years would not show Example
an increase, even though it involves the same num-
The world is in an obesity epidemic. What

ber of person-years (200,000), because the follow-up
is the prevalence of abnormal weight in the
time is too short.
United States? It depends on how abnormal is
defined. Figure 2.4 shows the distribution of
body mass index (BMI, one way of measuring
BASIC ELEMENTS OF obesity that takes into account both weight
FREQUENCY STUDIES and height) in U.S. men and women aged 40 to
59 years in 2007 through 2008 (9). The U.S.
To make sense of a study reporting prevalence, one
National Institutes of Health and World Health
needs careful definition of both the numerator and
Organization have recommended a classifica-
the denominator.
tion of BMI (Table 2.3). According to this clas-
sification, prevalence of obesity was 33.8%,
What Is a Case? Defining whereas prevalence of overweight (which in-
the Numerator cludes people with obesity and more) was 68%.
A substantial proportion of the population,
Cases might be people in the general population who about 5%, was in the most extreme (Class III,
develop a disease or patients in clinical settings with “morbid obesity”) weight class.
disease who develop an outcome event such as recur-
rence, complication of treatment, or death. In either
situation, the way in which a case is defined affects Rates may also be affected by how aggressively
rates. one looks for cases. For example, aspirin can induce
Most clinical phenomena (serum cholesterol, asthma in some people. How often does this occur?
serum calcium, thyroid hormone levels, etc.) exist It depends on the definition of a case. When peo-
on a continuum from low to high. The cutoff point ple are simply asked whether they have a breath-
defining a case can be placed at various points and ing problem after taking aspirin, rates are relatively
this can have large effects on the resulting prevalence. low, about 3% in adults. When a case is defined
We will discuss some of the reasons why one would more rigorously, by giving aspirin and measuring
place a cutoff at one or another point in Chapter 3 whether this was followed by bronchoconstriction,
and the consequences for a diagnostic test perfor- the prevalence of aspirin-induced asthma is much
mance in Chapter 8. higher, about 21% in adults (10). The lower rate
10
8
Population (%)
Overweight
6
Class I
4
Normal
weight Class II
2
Under- Class III
weight
0 Obese
10 15 20 25 30 35 40 45 50 55
Body mass index
Figure 2.4 ■ The prevalence of overweight and obesity in men, 2007 to
2008. (Data from Flegal KM, Carroll MD, Ogden CL, et al. Prevalence and trends in
obesity among US adults, 1999–2008. JAMA 2010;303(3):235–241.)
Essentials
Table 2.3
Classification of Obesity According to Example
the U.S. National Institutes of Health
and World Health Organization
Many cases of prostate cancer remain indolent

2 and are not detected during life. With use of
Classification Body Mass Index (kg/m )
prostate-specific antigen (PSA), a blood test for
Underweight <18.5 prostate cancer, more of these indolent cases
Normal weight 18.5–24.9 are now found. The test is relatively sensitive
Overweight 25.0–29.9 and leads to prostate biopsies that discover oth-
erwise undetected cancers. The result of wide-
Obesity ≥30
spread PSA testing in the United States has been
Obesity Class I 30.0–34.9 a rapid rise in the reported incidence of prostate
Obesity Class II 35.0–39.9 cancer, more than double in a few years (11)
Obesity Class III ≥40
(Fig. 2.5). The rise in incidence probably does not
(“severe,” “extreme,” reflect an increase in the true incidence in the
or “morbid”) population because it has occurred so fast and
because similar increases have not been seen
Data from Flegal KM, Carroll MD, Ogden CL et al. Prevalence
and trends in obesity among US adults, 1999–2008. JAMA
in countries where PSA testing is less common.
2010;303:235–241. Incidence has subsequently fallen somewhat,
presumably because the reservoir of prevalent
cases, brought to attention by this new test,
has been exhausted. However, incidence has
pertains to clinical situations, whereas the higher not fallen to pre-PSA levels and seems to have
rate tells us something about the biology of this reached a higher plateau, suggesting that new
disease. (incident) cases are being diagnosed more fre-
Incidence can also change if a more sensitive ways quently since PSA testing was introduced.
of detecting disease is introduced.
250
Age-adjusted Incidence / 100,000
200
150
100
PSA Approval
50
0
1975 1980 1985 1990 1995 2000 2005 2007
Year of diagnosis
Figure 2.5 ■ Incidence depends on the intensity of efforts to find cases. Incidence of
prostate cancer in the United States during the widespread use of screening with prostate-
specific antigen (PSA). (Redrawn with permission from Wolf AMD, Wender RC, Etzioni RB
et al. American Cancer Society guideline for the early detection of prostate cancer: Update
2010. CA Cancer Journal for Clinicians 2010;60:70–98.)
Essentials
What Is the Population? being selected. Probability samples are useful because
Defining the Denominator it is often more informative to include in the sample
a sufficient number of people in particular subgroups
A rate is useful only to the extent that the popula- of interest, such as ethnic minorities or the elderly. If
tion in which it is measured—the denominator of the members of these subgroups comprise only a small
rate—is clearly defined and right for the question. proportion of the population, a simple random sam-
Three characteristics of the denominator are espe- ple of the entire population might not include enough
cially important. of them. To remedy this, investigators can vary the
First, all members of the population should be sampling fraction, the fraction of all members of
susceptible to the outcome of interest; that is, they each subgroup included in the sample. Investigators
should comprise a population at risk. If members of can oversample low-frequency groups relative to the
the population cannot experience the event or condi- rest, that is, randomly select a larger fraction of them.
tion counted in the numerator, they do not belong in The final sample will still be representative of the
the denominator. For example, rates of cervical can- entire population if the different sampling fractions
cer should be assessed in women who still have a cer- are taken into account in the analysis.
vix; to the extent that cervical cancer rates are based On average, the characteristics of people in prob-
on populations that include women who have had ability samples are similar to those of the population
hysterectomies (or for that matter, men), true rates from which they were selected, particularly when the
will be underestimated. sample is large. To the extent that the sample differs
Second, the population should be relevant to the from the parent population, it is by chance and not
question being asked. For example, if we wanted to because of systematic error.
know the prevalence of HIV infection in the commu- Non-random samples are common in clinical
nity, we would study a random sample of all people research for practical reasons. They are called con-
in a region. But if we wanted to know the prevalence venience samples (because their main virtue is that
of HIV infection among people who use street drugs, they were convenient to obtain, such as samples of
we would study them. patients who are visiting a medical facility, are coop-
Third, the population should be described in erative, and are articulate) or grab samples (because
sufficient detail so that it is a useful basis for judg- the investigators just grabbed patients wherever they
ing to whom the results of the prevalence study could find them).
applies. What is at issue here is the generalizability Most patients described in the medical literature
of rates—deciding whether a reported rate applies and encountered by clinicians are biased samples
to the kind of patients that you are interested in. of their parent population. Typically, patients are
A huge gradient in rates of disease (e.g., for HIV included in research because they are under care in
infection) exists across practice settings from the an academic institution, are available, are willing to
general population to primary care practice to refer- be studied, are not afflicted with diseases other than
ral centers. Clinicians need to locate the reported the one under study, and perhaps also are particularly
rates on that spectrum if they are to use the infor- interesting, severely affected, or both. There is nothing
mation effectively. wrong with this practice as long as it is understood
to whom the results do (or do not) apply. However,
Does the Study Sample because of biased samples, the results of clinical
Represent the Population? research often leave thoughtful clinicians with a large
generalizability problem, from the research setting to
As mentioned in Chapter 1, it is rarely possible to their practice.
study all the people who have or might develop the
condition of interest. Usually, one takes a sample so
that the number studied is of manageable size. This DISTRIBUTION OF DISEASE BY
leads to a central question: Does the sample accu- TIME, PLACE, AND PERSON
rately represent the parent population?
Random samples are intended to produce repre- Epidemiology has been described as the study of the
sentative samples of the population. In a simple ran- determinants of the distribution of disease in popula-
dom sample, every individual in the population has tions. Major determinants are time, place, and person.
an equal probability of being selected. A more general Distribution according to these factors can provide
term, probability sample, is used when every per- strong clues to the causes and control of disease, as
son has a known (not necessarily equal) probability of well as to the need for health services.
Essentials
Time
and signs of a febrile respiratory illness, chest
An epidemic is a concentration of new cases in radiograph changes, lack of response to anti-
time. The term pandemic is used when a disease is biotics, and normal or decreased white blood
especially widespread, such as a global epidemic of cell count. Later, as more became known
particularly severe influenza (e.g., the one in 1918– about this new disease, laboratory testing
1919) and the more slowly developing but world- for the responsible coronavirus could be used
wide rise in HIV infection/AIDS. The existence of an to define a case. Cases were called “reported”
epidemic is recognized by an epidemic curve that to make clear that there was no assurance
shows the rise and fall of cases of a disease over time that all cases in the Beijing community were
in a population. detected.
Figure 2.6 also indicates when major con-
trol measures were instituted. The epidemic
declined in relation to aggressive quarantine
Example measures involving the closing of public gath-
ering places, identifying new cases early in their
Figure 2.6 shows the epidemic curve for Se- course, removing cases from the community,
vere Acute Respiratory Syndrome (SARS), in and isolating cases in facilities specifically for
Beijing, People’s Republic of China, during SARS. It is possible that the epidemic abated for
the spring of 2003 (12). In all, 2,521 proba- reasons other than these control measures, but
ble cases were reported during the epidemic. it is unlikely given that similar control measures
Cases were called “probable” because, at the in other places were also followed by a resolu-
time, there were no hard and fast criteria for tion of the epidemic. Whatever the cause, the
diagnosis. The working definition of a case decline in new cases allowed the World Health
included combinations of the following epi- Organization to lift its advisory against travel
demiologic and clinical phenomena: contact to Beijing so that the city could reopen public
with a patient with SARS or living or visiting places and resume normal international busi-
an area where SARS was active, symptoms ness and tourism.
Universities and schools closed Libraries, bars, theaters closed
200 Fever checks at airports begins Start to group patients with

Quarantine of close contacts SARS in designated wards
Number of probable cases
150 Training in management Designated fever clinics

of patients with SARS
100
SARS made reportable All patients with SARS
Contact tracing begins in designated hospitals
50
0
Mar 7 Mar 14 Mar 21 Mar 28 Apr 4 Apr 11 Apr 18 Apr 25 May 2 May 9 May 15 May 23 May 30
Date of hospitalization
Figure 2.6 ■ An epidemic curve. Probable cases of severe acute respiratory syndrome in Beijing March 2003 through
May 2003, in relation to control measures. (Adapted with permission from Pang X, Zhu Z, Xu F, et al. Evaluation of
control measures implemented in the severe acute respiratory syndrome outbreak in Beijing, 2003. JAMA 2003;290:
3215–3221.)
Essentials
Knowledge of a local epidemic helps clinicians get

the right diagnosis. For example, while serving as a in North America, Europe, and Australia and
primary care physician on a military base in Germany, low in Africa and Asia (Fig. 2.7) (13). This obser-
one author saw a child with a fever and rash on the vation has led to the hypothesis that environ-
hands and feet. With only a hospital-based clerk- mental factors may play a large part in the de-
ship in pediatrics to rely on, he was perplexed. But velopment of this disease. This hypothesis has
when he and his colleagues began seeing many such been supported by other studies showing that
children in a short time span, they recognized (with people moving from countries of low incidence
the help of a pediatric consultant) that they were in to those of high incidence acquire higher rates
the midst of an outbreak of coxsackievirus infection of colorectal cancer during their lifetime.
(“hand, foot, and mouth syndrome”), which is a dis-
tinctive but mild infectious disease of children.
When a disease such as iodine deficiency goiter or
Place polio (after global efforts to eradicate it) is limited to
The geographic distribution of cases indicates where a certain places, the disease is called endemic.
disease causes a greater or less burden of suffering and
provides clues to its causes. Person
When disease affects certain kinds of persons at the
same time and in the same places as other people who
Example are not affected, this provides clues to causes and guid-
ance on how health care efforts should be deployed. At
The incidence of colorectal cancer is very differ- the beginning of the AIDS pandemic, most cases were
ent in different parts of the world. Rates, even seen in homosexual men who had multiple sexual
when adjusted for differences in age, are high partners as well as among intravenous drug users. This
led to the early hypothesis that the disease was caused
5.0 cases/10,000/year
Figure 2.7 ■ Colorectal cancer incidence for men according to area of the globe. (Data from Center MM, Jemal A,
Smith RA, et al. Worldwide variations in colorectal cancer. CA Cancer J Clin 2009;59:366–378.)
Essentials
by an infectious agent transmitted in semen and Hodgkin disease, aplastic anemia, or systemic lupus ery-
blood. Laboratory studies confirmed this hypothesis thematosus. In contrast, some referral hospitals are well
and discovered the human immunodeficiency virus. prepared for just these diseases, and appropriately so.
Identification of the kinds of people most affected also
led to special efforts to prevent spread of the disease in What Are Prevalence Studies Not
them—for example, by targeting education about safe Particularly Good For?
sex to those communities, closing public bathhouses,
and instituting safe-needle programs. Prevalence studies provide only weak evidence of
cause and effect. Causal questions are inherently about
new events arising over time; that is, they are about
USES OF PREVALENCE STUDIES incidence. One of the other limitations of prevalence
Properly performed prevalence studies are the very studies, for this purpose, is that it may be difficult to
best ways of answering some important questions and know whether the purported cause actually preceded
are a weak way of answering others. or followed the effect because the two are measured at
the same point in time. For example, if inpatients with
What Are Prevalence hyperglycemia are more often infected, is it because
hyperglycemia impairs immune function leading to
Studies Good For? infection or has the infection caused the hyperglyce-
Prevalence studies provide valuable information about mia? If a risk factor (e.g., family history or a genetic
what to expect in different clinical situations. marker) is certain to have preceded the onset of dis-
ease or outcome, interpretation of the cause-and-effect
sequence is less worrisome.
Another limitation is that prevalence may be the
Example result of incidence of disease, the main consideration
The approach to cervical lymphadenopathy
in causal questions, or it may be related to duration
depends on where and in whom it is seen.
of disease, an altogether different issue. With only
Children with persistent cervical adenopathy
information about prevalence, one cannot determine
seen in primary care practice have only a 0.4%
how much each of the two, incidence and duration,
probability that the node represents cancer,
contributes. Nevertheless, cross-sectional studies can
mainly lymphoma. Clinicians should have a
provide compelling hypotheses about cause and effect
high threshold for biopsy, the definitive way of
to be tested by stronger studies.
determining whether or not cancer is present.
The underlying message is that a well-performed
However, adults seen in primary care practice
cross-sectional study, or any other research design, is
have a 4% chance of having an underlying can-
not inherently strong or weak but is only in relation
cer of the head and neck. For them, clinicians
to the question it is intended to answer.
should have a lower threshold for lymph node
biopsy. Rates of malignancy in referral centers
are much higher, about 60% in adults with cer-
vical adenopathy, and biopsies are usually done.
Example
The situation is different in different parts of Children living on farms are less likely to have
the world. In resource-poor countries, myco- asthma than children who live in the same re-
bacterial infections are a more common cause gion but not on farms. Farm and urban environ-
of lymphadenopathy than cancer. Thus, knowl- ments differ in exposure to microbes, suggesting
edge of prevalence helps clinicians prioritize that exposure to microbes may protect against
the diagnostic possibilities in their particular asthma. But the environments differ in many
setting and for the particular patient at hand. other ways. To strengthen this hypothesis, in-
vestigators in Germany examined mattress dust
and also settled dust from children’s bedrooms
Prevalence of disease strongly affects the interpre- and found that greater microbial diversity was
tation of diagnostic test results, as will be described in inversely related to asthma, independent of
greater detail in Chapter 8. farming (14). Together, these observations sug-
Finally, prevalence is an important guide to planning gest that exposure to a wider range of microbes
health services. In primary care practice, being prepared may protect against asthma, a causal hypothesis
for diabetes, obesity, hypertension, and lipid disorders to be tested by stronger research.
should demand more attention than planning for
Essentials
Review Questions
Read the following statements and mark the about 1/100 persons. On average, how many
best answer. years does the disease persist?

A. 10
2.1. Cancer registries report 40 new cases of
B. 25
bladder cancer per 100,000 men per year.
C. 33
Cases were from a complete count of all
D. 40
patients who developed bladder cancer in
E. 50
several regions of the United States, and the
number of men at risk was estimated from 2.6. Which of the following studies is not a cohort
the census data in those regions. Which rate study?
is this an example of?
A. The proportion of patients with stomach
A. Point prevalence cancer who survive 5 years
B. Period prevalence B. The risk of developing diabetes mellitus in
C. Incidence density children according to their weight
D. Cumulative incidence C. Complications of influenza vaccine
E. Complication rate among children vaccinated in 2011
D. The earlier course of disease in a group of
2.2. Sixty percent of adults in the U.S. population patients now under care in a clinic
have a serum cholesterol >200mg/dL (5.2 E. Patients admitted to an intensive care unit
mmol/L). Which rate is this an example of? and followed up for whether they are still
A. Point prevalence alive at the time of hospital discharge
B. Complication rate 2.7. A sample for a study of incidence of
C. Incidence density medication errors is obtained by enrolling
D. Cumulative Incidence every 10th patient admitted to a hospital.
E. Period prevalence What kind of sample is this?
2.3. You are reading a study of the prevalence of A. Stratified sample
uterine cervix infections and want to decide if B. Probability sample
the study is scientifically sound. Which of the C. Convenience sample
following is not important? D. Random sample
E. Oversample
A. Participants are followed up for a sufficient
period of time for anemia to occur. 2.8. Cohort studies of children with a first febrile
B. The study is done on a representative seizure have shown that they have a one in
sample of the population. three chance of having another seizure during
C. All members of the population are childhood. What kind of rate is this?
women.
D. Cervical infection is clearly defined. A. Point prevalence
E. The study is done on a sample from a B. Complication rate
defined population. C. Cumulative Incidence
D. Period prevalence
2.4. A probability sample of a defined population: E. Incidence density
A. Is invalidated by oversampling. 2.9. Which of the following would not increase

B. Is inferior to a random sample. the observed incidence of disease?
C. Is not representative of the population. A. More aggressive efforts to detect the disease
D. Results in a representative sample of B. A true increase in incidence
the population only if there are enough C. A more sensitive way of detecting the disease
people in the sample. D. A lowering of the threshold for diagnosis
of disease
2.5. The incidence of rheumatoid arthritis is E. Studying a larger sample of the
about 40/100,000/year and the prevalence is population
Essentials
2.10. Infection with a fungus, coccidioidomycosis, apparent after birth, most often in adolescence),
is common in the deserts of the southwestern which would be an appropriate cohort?
United States and in Mexico, but uncommon
A. Children born in North Carolina in 2012
elsewhere. Which of the following best
and examined for scoliosis until they are
describes this infection?

adults
A. Endemic B. Children who were referred to an
B. Pandemic orthopedic surgeon for treatment
C. Incident C. Children who were found to have scoliosis
D. Epidemic in a survey of children in North Carolina
E. Prevalent D. Children who have scoliosis and are
available for study
2.11. Twenty-six percent of adults report having E. Children who were randomly sampled
experienced back pain lasting at least a day in from the population of North Carolina
the prior 3 months. Which of the following in the spring of 2012
best describes this rate?
2.14. Which of the following are prevalence studies
A. Cumulative Incidence especially useful for?
B. Incidence density
C. Point prevalence A. Describing the incidence of disease
D. Complication rate B. Studying diseases that resolve rapidly
E. Period prevalence C. Estimating the duration of disease
D. Describing the proportion of people in a
2.12. Which of the following best describes a defined population with the condition of
“dynamic population”? interest
E. Establishing cause and effect
A. It is rapidly increasing in size.
B. It is uniquely suited for cohort studies. 2.15. Last year, 800,000 Americans died of heart
C. People are continually entering and disease or stoke. Which of the following best
leaving the population. describes this statistic?
D. It is the basis for measuring cumulative
incidence. A. Incidence density
E. It is the best kind of population for a B. Point prevalence
random sample. C. Cumulative Incidence
D. Period prevalence
2.13. For a study of the incidence of idiopathic E. None of the above
scoliosis (a deformity of the spine that becomes Answers are in Appendix A.
REFERENCES
1. Roberts DE, Gupta G. Letter to the editor. New Engl J Med before zoster vaccine introduction. Mayo Clin Proc 2007;82:
1987;316:550. 1341–1349.
2. Bryant GD, Norman GR. Expressions of probability: words 9. Flegal KM, Carroll MD, Ogden CL, et al. Prevalence and
and numbers. N Engl J Med 1980;302:411. trends in obesity among US adults, 1999-2008. JAMA 2010;
3. Toogood JH. What do we mean by “usually”? Lancet 1980;1:1094 303:235–241.
4. Loftus EV Jr. Clinical epidemiology of inflammatory bowel 10. Jenkins C, Costello J, Hodge L. Systematic review of preva-
disease: Incidence, prevalence, and environmental influences. lence of aspirin induced asthma and its implications for clini-
Gastroenterology 2004;126:1504–1517. cal practice. BMJ 2004;328:434–437.
5. Loftus EV Jr, Silverstein MD, Sandborn WJ, et al. Ulcerative 11. Wolf AMD, Wender RC, Etzioni RB, et al. American Cancer
colitis in Olmstead County, Minnesota, 1940-1993: inci- Society Guideline for the early detection of prostate cancer:
dence, prevalence, and survival. Gut 2000;46:336–343. update 2010. CA Cancer J Clin 2010;60:70–98.
6. Sands K, Vineyard G, Platt R. Surgical site infections occur- 12. Pang X, Zhu Z, Xu F, et al. Evaluation of control measures
ring after hospital discharge. J Infect Dis 1996;173:963–970. implemented in the severe acute respiratory syndrome out-
7. Andrade L, Caraveo-Anduaga JJ, Berglund P, et al. The epide- break in Beijing, 2003. JAMA 2003;290:3215–3221.
miology of major depressive episodes: results from the Inter- 13. Center MM, Jemal A, Smith RA, et al. Worldwide variations
national Consortium of Psychiatric Epidemiology (ICPE) in colorectal cancer. CA Cancer J Clin 2009;59:366–378.
surveys. Int J Methods Psychiatr Res 2003;12:3–21. 14. Ege MJ, Mayer M, Normand AC, et al. Exposure to environmen-
8. Yawn BP, Saddier P, Wollan PC, et al. A population-based tal microorganisms in childhood asthma. N Engl J Med 2011;
study of the incidence and complication rates of herpes zoster 364:701–709.
Essentials

FletcherRobertH 2014 Chapter2Frequency ClinicalEpidemiologyT

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

FletcherRobertH 2014 Chapter2Frequency ClinicalEpidemiologyT

Hochgeladen von

Copyright:

Verfügbare Formate

Chapter 2

validity. In modern times, clinical research has become suffi-

Prevalence is the fraction (proportion or percent) of

Patients also assign widely varying probabilities to

In general, clinically relevant measures of frequency are

lung cancer in a population of 10,000 people over Prevalence and Incidence

a. Characteristic b. Incidence c. Prevalence

a group initially free of disease time

veys, even when their incidence is low. The incidence

Case fatality rates Proportion of patients who die of a disease

Number of live births in the same year

Defined Representative Disease/outcome

Figure 2.2 ■ The design of a prevalence study.

STUDIES OF PREVALENCE period prevalence but a good estimate of point

will be discussed in greater detail in Chapters 5 and 7.

a period of time. This is called cumulative incidence

The population of the county was estimated

vides and example of both incidence density

in a study of 10,000 people exposed to a carcinogen

The world is in an obesity epidemic. What

Many cases of prostate cancer remain indolent

Universities and schools closed Libraries, bars, theaters closed

200 Fever checks at airports begins Start to group patients with

150 Training in management Designated fever clinics

Knowledge of a local epidemic helps clinicians get

best answer. years does the disease persist?

A. Is invalidated by oversampling. 2.9. Which of the following would not increase

describes this infection?

Das könnte Ihnen auch gefallen