University Airlangga

Faculty of Dental Medicine

ENVIRONMENTAL AND
OCCUPATIONAL EPIDEMIOLOGY

Prof.Dr.Titiek Berniyanti drg.M.Kes.

Contents

► A primer on epidemiology
► Kinds of epidemiological studies
► Bias
► Data analysis
► Environmental and occupational epidemiology
► Understanding clusters
► Measuring exposure
► Epidemiology and risk assessment
► Future directions
► Summary
A primer on epidemiology

► Epidemiology is the study of the Frequention, distribution

and determinants of health and disease in human
populations.
► If we can show that an exposure causes disease, we have a
chance to intervene and prevent disease occurrence.
► Epidemiology has provided evidence that many
environmental and occupational exposures are associated
with diseases.
A primer on epidemiology

► Epidemiology can give us the tools, the techniques of

study design and analysis, to determine whether a given
exposure is associated with a given disease. How do we
judge that an association is causal (a process sometimes
called causal inference )
► A general philosophical framework for judging causality
stems from the writings of the philosopher Karl Popper.
This framework posits that observations (especially
repeated observations) that one event (A) is followed by
another (B) enable the epidemiologist to form a
hypothesis, that is, a proposition that A causes B.
The key to Popperian philosophy is
• that all hypotheses (or theories of causation)
are tentative and may be disproved by
further testing.
• Hypotheses that are tested many times and
hold up tend to become accepted as
scientific facts (for example, we accept that
cigarettes cause lung cancer), but
• over the course of time many accepted
hypotheses are overthrown by new scientific
insights (we now know that miasma or foul
air does not cause cholera).
On a practical level, a famous set of criteria
set out by Austin Bradford Hill ( 1965 ) is
commonly used by epidemiologists to judge
whether a particular causal hypothesis is plausible,
whether the observed association between A and B makes it likely that
in fact A causes B.
Only one — the proper temporal relationship — is absolutely required:
the exposure must precede the disease. Although it seems this should always
be easy to know, sometimes it is not clear.
Other commonly used Hill criteria that favor causality are
Consistency (the association is repeated in many studies),
a large effect size (the exposed have much more disease than the
nonexposed),
a positive dose - response relationship (more exposure causes more
disease), and
biological plausibility (some biological explanation makes it reasonable
that A causes B).
Regulators and risk assessors must conclude from the weight of the
epidemiological
evidence, applying criteria such as these, whether an association is likely
to be causal.
A number of agencies, such as
• the International Agency for Research on Cancer (IARC),
• the National Toxicology Program (NTP),
• The Institute of Medicine (IOM, a part of the National Academy of
Sciences), and
• the Environmental Protection Agency (EPA),
regularly review epidemiological evidence and publish summaries in
which they evaluate whether associations are likely to be causal.
Epidemiology has provided evidence judged as causal that
many environmental and occupational exposures are
associated with diseases, including evidence
• associating lead with cognitive impairment in children,
• Trihalomethanes (in water) with bladder cancer,
particulate air pollution with cardiorespiratory disease,
• radon gas with cancer, and
• ergonomic stress with low back pain, to name just a few.
 Kinds of Epidemiological Studies
1. Descriptive Studies
► At the simplest level there are descriptive studies , which
► characterize a disease by factors such as age, sex, time, and geographical
region.
► These studies do not formally test a hypothesis that a specific exposure (or
risk factor ) is associated with a disease but rather describe patterns in
disease occurrence in terms of broad demographic and other variables.
► These studies are often first steps and may provide clues about factors that
cause disease.
► For example,
► the fact that malaria occurs mainly in tropical areas provides a clue that
warm climate may play a role in its transmission.
► The fact that heart disease occurs at a later age in women than men may
provide a clue that endogenous estrogen plays a protective role.
2. Correlational, or Ecological, Studies

► Descriptive studies are a close cousin to correlational studies , or

ecological studies , which study the correlation between disease rates
and some specific exposure, but at the level of groups rather than
individuals
► For example, one can correlate breast cancer rates in countries around
the world with degree of socioeconomic development; breast cancer
incidence is higher in richer, more urbanized countries.
► provide clues about possible risk factors for disease, factors that can
then be examined further in studies of individuals.
► ecological studies are viewed as weaker than studies of individuals,
because across a population, individuals with the risk factors are not
necessarily the same individuals who contract the disease.
3. Etiologic, or Analytical, Studies
► Generally studies of individuals in which the investigators seek to test a
specific hypothesis about exposure and disease.
► for example, whether pesticide exposure is associated with Parkinson’s
disease.
► Analytical studies can in turn be divided into two types, clinical trials and
observational studies.

a) Clinical Trials(randomized clinical trials)

► A sense the model for rigorous epidemiological studies. often done to compare
one medication or treatment to another.
► They are controlled experiments, because they assign treatment (or exposure)
randomly to one group and not another. The treated and untreated groups are
therefore likely to be comparable with regard to other variables
► Both treated and untreated groups are followed prospectively over time.
► Randomized clinical trials are generally impractical for studying
environmental and workplace exposures, because one cannot ethically
administer a suspected toxin to a human population.
Observational studies (uncontrolled studies,
or natural experiments)

► Observational studies less definitive than clinical trials.

► The three principal designs for observational studies are cohort, case-control, and
cross-sectional.
❖ Cohort studies start with an exposed group and a nonexposed group, both disease free,
and follow them forward in time to observe disease incidence or mortality rates.
► Disease rates in the exposed and nonexposed can be then compared using a rate ratio or
a rate difference. The observation period in cohort studies may start in the past and
move forward to the present (retrospective studies), or start in the present and move into
the future (prospective studies).
❖ Case-control studies are useful for rare diseases and common exposures, the opposite
of cohort studies. They can be carried out in the general population or in hospitals or
can be nested within cohorts. They are more subject to bias than cohort studies because
it is sometimes difficult to choose cases and controls who are representative of the
overall diseased and non diseased populations.

❖ Cross-sectional studies tend to measure exposure and disease at the same time. They
are often done when the outcome of interest is subclinical or asymptomatic disease.
Cross- sectional studies are seen as a weaker design than cohort and case-control
studies. However, they are often the only possible design and can pro- vide valid
results.
Data Analysis

► Methods of analysis in epidemiology typically depend on whether the exposure variable

and the disease variable are continuous variables or categorical variables. Most of the
approaches described previously consider disease to be a categorical (yes/no) variable
(often called a dichotomous variable). This is typically true of a specific disease: you
either get the disease or you don’t.
► Exposure variables may also be continuous (for example, cadmium in the urine) or
categorical (welder or nonwelder).
► When both exposure and disease variables are dichotomous, then one usually calculates
the measures referred to previously, such as a rate ratio or an odds ratio.
► When both the disease and the exposure are continuous variables, typically a
regression analysis is conducted (for example, linear regression), in which the outcome
is disease and the predictors include exposure and any other confounder variables about
which the investigator has data
► Large sample sizes confer greater statistical power to detect associations, and lead to
high precision. Precision is often presented by a confidence interval, which represents a
range of plausible values for the measure of effect.
► Precision is related to statistical significance. Statistically significant usually means that
the estimate of effect is different from the null value and that the difference is unlikely
to have occurred by chance.
► Epidemiologists now prefer to express the precision of study results with confidence
intervals rather than with p values and tests of statistical significance, partly because a
range of plau- sible values is more informative than a single test of statistical
significance.
The study of diseases and health conditions (occurring in the population) that are linked
to environmental factors.
• These exposures usually are involuntary.

Epidemiology’s Contributions to Environmental Health

Concern with populations • Use of observational data • Methodology for study designs
• Descriptive and analytic studies
Occupational and environmental epidemiology
Two sub-disciplines of epidemiology that focus on
studying the potential health risks of exposures to
chemicals, particulates, metals, physical factors,
infectious disease agents, and psychosocial factors
in the workplace and general environment.

Diseases of concern in occupational and

environmental epidemiology can include the entire
spectrum of health outcomes, such as cancer,
cardiovascular, neurological, respiratory,
immunological, and dermal disease as well as
injuries, reproductive and mental health outcom
ENVIRONMENTAL AND OCCUPATIONAL
EPIDEMIOLOGY
► Environmental epidemiology concerns environmental agents to
which large numbers of people are exposed involuntarily. This area of
concern usually excludes voluntary exposures such as alcohol,
cigarettes, and medications.
► However, it usually includes environmental (“secondhand”) tobacco
smoke and infectious agents in water supplies. Although this definition
is sometimes a bit
► Environmental exposures can be thought of as contributing either to
epidemics or to endemic diseases.
► Epidemics are unusual outbreaks of disease clearly above a normal
level and often caused by known agents, although some- times the
agent is initially unknown.
► In contrast, endemic diseases exist at constant, low (or background)
levels and may or may not have an environmental cause.
What is Environmental Epidemiology?
▪ Environmental epidemiology The study of diseases and health
conditions (occurring in the population) that are linked to
environmental factors.
▪ concerns environmental agents to which large numbers of people
are exposed involuntarily.
▪ This area of concern usually excludes voluntary exposures such as
alcohol, cigarettes, and medications.
▪ However, it usually includes environmental (“secondhand”)
tobacco smoke and infectious agents in water supplies.
▪ Although this definition is sometimes a bit, Environmental
exposures can be thought of as contributing either to epidemics or
to endemic diseases.
• Epidemics are unusual outbreaks of disease clearly above a normal level and
often caused by known agents, although some- times the agent is initially
unknown.
• In contrast, endemic diseases exist at constant, low (or background) levels and
may or may not have an environmental cause.
Epidemiology’s Contributions to
Environmental Health

• Concern with populations

• Use of observational data
• Methodology for study designs
• Descriptive and analytic studies
Concern with Populations
Environmental epidemiology studies a population in relation to morbidity and
mortality. – Example: Is lung cancer mortality higher in areas with higher
concentrations of “smokestack” industries?

Use of Observational Data

Epidemiology is primarily an observational science that takes advantage of
naturally occurring situations in order to study the occurrence of disease.

Methodology for Study Designs

Characteristic study designs used frequently in environmental
epidemiology: – Cross-sectional – Ecologic – Case-Control – Cohort

Two Classes of Epidemiologic Studies

• Descriptive
– Depiction of the occurrence of disease in populations according to
classification by person, place, and time variables.
• Analytic
– Examines causal (etiologic) hypotheses regarding the association
between exposures and health conditions.
Occupational epidemiology
► is the epidemiological study of illness or injury associated with
workplace exposures.
► Examples include the association of stressful repetitive motion
and carpal tunnel syndrome, welding and lung cancer, silica
exposure and kidney disease, and poor office ventilation and
respiratory illness.
► Occupational epidemiology often involves relatively high
exposures in relatively small numbers of people, often
geographically isolated at a worksite.
► Historically, occupational studies were carried out in the context
of very high exposures.
► Today, workplace exposures to suspected toxins are much lower
than in the past, at least in industrialized countries, and they are
less often the focus of occupational epidemiology.
Measuring Exposure
► Measuring exposure with as much accuracy as possible is key to
valid epidemiological studies.
► Accurate exposure assessment is essential to detecting and
quantifying a dose - response relationship, forexample, which is
one of the key elements supporting a causal relationship .
► Mismeasured exposure (as a continuous variable) usually leads to
flattening, or attenuating, a true dose-response.
► Misclassification of dichotomous exposure status (exposed versus
nonexposed) can severely bias results toward the null.
► It is often difficult to assess exposure accurately when exposure
must be estimated in the past, as in case-control studies, in
retrospective cohort studies, and in cross-sectional studies of the
impact of past exposures on current outcomes.
Measures of Disease Frequency

•Prevalence
• Point prevalence
• Incidence
• Incidence rate
• Case fatality rate
 Prevalence
Refers to the number of existing cases of a disease, health condition, or deaths
in a population at some designated time

Point Prevalence
Refers to all cases of a disease, health condition, or deaths that exist at a particular
point in time relative to a specific population from which the cases are derived.
Incidence

The occurrence of new disease or mortality

within a defined period of observation (e.g.,
week, month, year, or other time period) in
a specific population.
Formula for Incidence Rate
Case Fatality Rate (CFR)
Provides a measure of the lethality of a disease.

Case Fatality Rate (CFR) (continued)

Major Historical Figure:
John Snow
• An English anesthesiologist who linked a cholera outbreak in
London to contaminated water from the Thames River in the
mid1800s.
• Snow employed a “natural experiment,” a methodology used
currently in studies of environmental health problems.
Study Designs Used in
Environmental Epidemiology

• Experimental
• Case Series
• Cross-Sectional
• Ecologic
• Case-Control
• Cohort
Odds Ratio (OR)
• A measure of association for case- control studies.
• Exposure-odds ratio: – Refers to “… the ratio of
odds in favor of exposure among the cases [A/C] to
the odds in favor of exposure among the non-cases
[the controls, B/D].”
Odds Ratio Table

Odds Ratio Equation

Note that an OR >1 (when statistically significant) suggests a

positive association between exposure and disease or health
outcome.
Relative Risk (RR)
• The ratio of the incidence rate of a disease or health
outcome in an exposed group to the incidence rate of the
disease or condition in a non-exposed group.

Relative Risk (RR) Table

RR Equation

Notes: When an association is statistically significant: RR

>1 indicates that the risk of disease is greater in the
exposed group than in the nonexposed group. RR
Study Endpoints

• Self-reported symptom rates

• Physiologic or clinical examinations
• Mortality
The epidemiologic triangle.
Environment in the “Triangle”
• The term environment is defined as the domain in which disease-causing agents
may exist, survive, or originate; it consists of “All that which is external to the
individual human host.”

Host in the “Triangle”

• A host is “a person or other living animal, including birds and arthropods, that
affords subsistence or lodgment to an infectious agent under natural conditions.”
Agent in the “Triangle”
• Agent refers to “A factor, such as a microorganism, chemical substance, or
form of radiation, whose presence, excessive presence, or (in deficiency
diseases) relative absence is essential for the occurrence of a disease.”

Causality
• Certain criteria need to be taken into account in the assessment of a
causal association between an agent factor (A) and a disease (B).
Hill’s Criteria of Causality

• Strength
• Consistency
• Specificity
• Temporality
• Biological gradient
• Plausibility
• Coherence
 Bias
► “Systematic deviation of results or inferences from the truth.
Processes leading to such deviation. An error in the conception
and design of a study—or in the collection, analysis,
interpretation, reporting, publication, or review of data—leading
to results or conclusions that are systematically (as opposed to
randomly) different from the truth.” – Porta M. A Dictionary of
Epidemiology. 5th ed. New York, NY: Oxford University Press;
2008.
► Bias refers to the distortion of the true relationship between
exposure and disease.
► The most important sources of bias are selection bias,
confounding, and information bias.
1. Selection bias occurs when the relationship between exposure and disease in
the study population is not representative of the true relation between exposure
and disease in the general population because the investigator has selected the
study population in a nonrepresentative way.
2. Confounding refers to the distortion of the exposure-disease relationship by
a third variable that is associated both with exposure and with disease.
3. Information bias can occur when information obtained about either
exposure or disease is incorrect. One of the main sources of information bias
in epidemiological studies is mis- measurement or misclassification of
exposure.

Healthy Worker Effect

• Refers to the observation that employed populations tend to have a lower mortality
experience than the general population. • The healthy worker effect could introduce
selection bias into occupational mortality studies.

Confounding
• Denotes “… the distortion of a measure of the effect of an exposure on an outcome due
to the association of the exposure with other factors that influence the occurrence of the
outcome.” – Porta M. A Dictionary of Epidemiology. 5th ed. New York, NY: Oxford
University Press; 2008.
Limitations of Epidemiologic Studies

• Long latency periods

• Low incidence and prevalence
• Difficulties in exposure assessment
• Nonspecific effects
Biomarker of exposure

An alternative to estimating external exposure is to use a

biomarker of exposure .
Examples of such biomarkers are dioxin in blood, cotinine (a
metabolite of nicotine) in blood, and lead in bone.
Such biomarkers can be useful because they measure internal
dose rather than external exposure.
They may therefore take into account variation in absorption
and metabolism of the external dose, possibly providing a more
accurate estimate of the biologically relevant dose that can
cause disease.
Biomarker of exposure
However, there are many problems that may make a measure of
the internal dose less desirable than a measure of the external
exposure, including wide individual variation, diffi culty in
obtaining accurate laboratory measurements of the biomarker,
and possibly choosing the wrong biomarker in a metabolic
pathway that features several candidate toxins.
Perhaps more important in the case of retrospective exposure
assessment, few biomarkers of exposure persist long enough to
be useful for such a study.
Understanding Clusters
► A cluster is an apparently elevated number of cases of disease in a
limited area over a limited period of time.
► Typically the number of cases in the cluster is small, on the order of ten
or twenty rather than hundreds.
► Clusters typically come to the attention of public health authorities, who
must first determine whether a cluster in fact represents an unusually
high occurrence of disease.
► Studies of clusters have more chance of leading to the discovery of a
specific cause when the disease in question is extremely rare.
► Occupational clusters have somewhat more of a chance than
environmental clus- ters of representing a common cause because they
have a natural boundary (the worksite) and therefore avoid the
boundary problem inherent in environmental clusters.
In cross - sectional or prospective studies current exposure can be
measured more or less easily, depending on the agent of interest.
However,
when exposure must be estimated in the past, as in case - control
studies, in retrospective cohort studies, and in cross - sectional
studies of the impact of past exposures on current outcomes it is
often difficult to assess exposure accurately .
Therefore we focus here on the problem of retrospective exposure
assessment.
In case - control studies of bladder cancer and drinking water, for
example, subjects may be trying to remember their pattern of
drinking - water consumption over the past fifty years.
SUMMARY

► In each case, epidemiologists work to define and measure

exposures, to define and measure the health outcomes of interest,
and to define and measure other factors that may bear on the
association of inter- est. They also work to eliminate or control
sources of bias that may skew their findings,
► including confounding, selection bias, and information bias.
► Epidemiological data are invaluable in risk assessment, in
standard setting and other policymaking, and in dispute resolution
in environmental and occupational health.
In cross -sectional studies of lead and neurologic deficits in children, one
may wish not only to measure current lead levels via the blood but also
to assess prior exposure to lead via its measurement in bone.
In retrospective cohort studies, investigators may be estimating past
silica exposure for workers in a specifi c plant.
As can be seen in these examples, in some instances investigators
attempt to measure external exposure (water drinking patterns, silica in
workers ’ breathing zone) and in others they seek a biomarker of internal
exposures (blood and bone lead).
Below we discuss both these scenarios.
First, let us consider more thoroughly the example of assessment of past
exposure to silica among workers in a retrospective cohort study.
Suppose there are some existing silica exposure measurements made
during the past twenty years for some workers in some jobs. Such a
relatively short record is typically the case, as exposure measurements
were not often made until somewhat recently. However, the cohort may
have been employed over the past forty or fifty years, and because
investigators seek to conduct an exposure – response analysis, they
require an estimate of past exposure for all workers across all jobs at all
points in time
EPIDEMIOLOGY AND RISK ASSESSMENT

► The results of occupational and environmental epidemiological

studies can affect public health by alerting policymakers to new
hazards and possibly by triggering regulations about permissible
levels of exposure.
► When a number of studies point in the same direction, public
authorities are more likely to act.
► Quantitative meta- analysis originally used for clinical trials but
have been used extensively for observational studies in the last
decade. They can combine results from different study designs,
such as rate ratios from cohort studies and odds ratios from
case-control studies.
► A variant method to summarize data across studies is a pooled
analysis, in which the raw data for each study are obtained and
the combined data then reanalyzed. Pooled analyses are much
more time consuming but have the advantage of providing more
flexibility in the analysis. Meta-analyses are most often done to
determine a common ratio measure of disease rates (for example,
a rate ratio) in the exposed versus the nonexposed. However, they
may also be done to determine a common exposure-response
coefficient across a number of exposure-response analyses.
► Exposure-response analyses are of particular interest to public health authorities who
seek to determine a permissible exposure level for the public or for workers.
► The determination of a permissible exposure level is based on risk assessment.
► Risk assessment may be based on animal data or human data. The former requires
extrapolation from animals to humans and hence involves a considerable amount of
uncertainty. For this reason, human (epidemiological) data are preferred, but they may
not exist for the agent in question.
FUTURE DIRECTIONS

► Occupational epidemiology is becoming less and less concerned

with exposures to toxins, which are becoming less and less
prevalent in the workplace. Instead, inter- est is now focusing
more on other types of exposures that affect a large number of
workers. One such exposure is job stress, which is difficult to
measure but which may have large consequences via increasing
blood pressure or cardiovascular disease, or both.