Search @

Custom Search

Sunday, June 1, 2008

Analytical Epidemiology ( Community Dentistry )


What is epidemiology?
Englander (1962) defined epidemiology as “ the study of those factors which influence the occurrence and distribution of health , disease , defect , disability and death in populations .
Epidimiology as a science is organized into three distinct divisions which complement one another . they are as follows :
Descriptive Epidemiology
Analytical Epidemiology
Experimental Epidemiology
The descriptive and analytical studies are often called as “ Observational Studies . “

In analytical epidemiology , the subject of interest is the individual within the population. The object is not to formulate , but to test hypothesis. Nevertheless , the inference is not to the population from which they are selected.
Analytical studies compromise 2 distinct types of observational studies :
Case control studies
Cohort studies
From each of these study designs, one can determine,
Whether or not a statistical association exists betmeen a disease and a suspected factor
If one exists the strength of the association
Factors present Individuals with particular disease CASES

Absent Individuals without particular disease CONTROLS

Individuals exposed to Presence
Particular factor(s) or
Individuals unexposed to particular
Particular factor(s) diseases


In recent years , the case control approach has emerged as a permanent method of epidemiological investigation . The case control method has 2 distinct features.
Both exposure and outcome(disease) have occurred before the start of the study
The study proceeds backwards from effect to cause
it uses a control or comparison group to support or refute an inference
By definition , a case control study involves 2 populations – cases and controls . The unit is the individual rather than the group . The controls are generally selected at an equal rate with the number of cases and should have matching characteristics with respest to general variables such as age ,sex , etc. if these variables themselves are not the suggested exposure characteristics.
Framework of a Case Control Study
( The 2 x 2 contingency table )
Suspected or Cases ( Disease Control (disease
Risk Factors Present ) absent )
Present a b
Absent c d
a + c b + d
Basic Steps
There are 4 basic steps in conducting a case control study :
Selection of cases and controls
Measurement of exposure
Selection of cases and controls
Selection of cases
The following 2 specifications are crucial :
Diagnostic Criteria
Eligibility criteria
Sources of cases
general population
Selection of controls
The controls must be as similar to the cases as possible , except for the absence of the disease under study.
Sources of controls
Hospital controls
Neighborhood controls
General population

It is defined as the process by which we select controls in such a way that they are similar to case with regard to certain pertinent selected variables ( e.g. age ) which are known to influence the outcome of disease and which if not adequately matched for compatibility , could distort or confound the results.

3.Measurement of exposure
Definitions and criteria about exposure are just as important as those used to define cases and controls. Thus information about exposure should be obtained.

The final step is analysis , to find out
Exposure rates among cases and controls to suspected factors
Estimation of disease risk associated with exposure ( odds ratio )

Biases in case control studies
A bias is a systematic error in design , conduct or analysis of a study which leads us to an erroneous conclusion.
Biases in selection of cases
This is called selection bias or diagnostic bias where in diagnosis itself is more unlikely if the exposure is present in history.
Biases in investigating controls
Apart from the fact that the controls are less likely to recall exposure variables than the cases themselves , there are other sources of biases that can creep into case control studies. The investigations/tests/interviews etc. may lack depth in controls whereas the cases are thoroughly worked up. This obviously introduces a bias by inflating the exposure factor in cases.

3.Confounding Bias
This is the distortion of study effect mixed with another effect because of variables extraneous to the exposure disease association predictive of the disease . Specially when one has multiple isk factors which are related to each other , the confounding effect may appear. To reduce this effect one usually comes at matching between cases and controls .

4. Problems due to overmatching
Confounding factors are actually extraneous factors unequally distributed between exposure subgroups. One tries to reduce the confounding by matching the common ones such as age,sex,etc. At times investigations land up in table by overmatching where in potential confounder is matched among cases and controls. The study thus loses the power of proving an obvious association .

5.Bias in Analysis
The obvious result of presence of a confounder is at the time of analysis .The association observed entirely may be due to non-uniform distribution of the confounder among cases / controls. The remedy is the use of strtification but it is only possible in a larger study.

Relatively easy to carry out.
Rapid and inexpensive ( compared with cohort studies )
Requires comparatively fewer subjects
Particularly suitable to investigate rare diseases or diseases about which little is known. But a disease whish is rare in general population may not be rare in special exposure groups
No risks to subjects
Allows the study of several different aetiological factors
Risk factors can be identified . Rational revention and control programmes can be identified.
Ethical problems minimal
No attrition problems , because case control studies do not require follow up of individuals into then future.

Problems of bias relies on memory or past records, the accuracy of which may be uncertain, validation of information obtained is difficult or sometimes impossible.
Selection of an appropriate control group may be difficult .
We cannot measure incidence , and can only estimate the relative risk.
Do not distinguish between cases and associated factors.
Not suited to the evaluation of therapy or prophylaxis of disease.
Another major concern is the representativeness of cases and controls.

Cohort study is another type of analytical study which is usually undertaken to obtain additional evidence to refute or support the existence of an association between suspected cause and disease.
It is also known as
Prospective study
Longitudinal study
Incidence study
Forward looking study
The distinguishing features of cohort studies are
The cohorts are identified prior to the appearance of the disease under investigation
The study groups , so defined , are observed over a period of time to determine the frequency of disease among them
The study proceeds forward from cause to effect
Concept of cohort
In epidemiology, the term ‘cohort’ is defined as a group of people who share a common characteristic or experience within a defined time period.
e.g. age,occupation,exposure to a drug or vaccine, pregnancy , insured persons, etc.
The comparison group may be the general population from which the cohort is drawn , or it may be another cohort of persons thought to have had little or no exposure to the substance in question , but otherwise normal.
Indications of cohort studies
When there is good evidence of an association between exposure and disease , as deriver from clinical observations and supported by descriptive and case control studies.
When exposure is rare, but the incidence of disease is high among exposed e.g. special exposure groups like those in industries , exposure to x rays , etc.
When criterion of study population can be minimized e.g. follow up is easy, cohort is stable, co-operative and easily accessible.
When ample funds are available.
Framework of a cohort study
In contrast to case control studies which proceed from effect to cause , the basic approach in cohort studies is to work from cause to effect.
In cohort study, the exposure has occurred but the disease has not.
Exposed to putative a b a+b
Aetiological factor
Not exposed to putative c d c+d
Aetiological factor
In assembling cohorts , the following general considerations are taken into account:
The cohorts must be free from the disease under study. Thus if the disease under study is coronary heart disease , the cohort members are first examined and those who already have evidence of the disease under investigation are excluded.
Insofar as the knowledge of the disease permits, both the groups should be equally susceptible to the disease under study , or effectively reflect any difference in disease occurrence.
Both the groups should be comparable in respect of all the possible variable, which may influence the frequency opf the disease.
The diagnostic and eligibility criteria of the disease must be defined before hand, this will depend upon the availability of reliable methods for recognizing the disease when it develops .
A well designed cohort is considered the most reliable means of showing an association between a suspected risk factor and subsequent disease because it eliminates many of the problems of the case control study and approximates the experimental model of the physical sciences.
Types of cohort study
Prospective cohort study
Retrospective cohort study
Combination of retrospective and prospective cohort study
Prospective cohort study
A prospective cohort study or “current” cohort study is one in which the outcome has not yet occurred at the time the investigation begins. Most prospective studies begin in the present and continue into future.
Retrospective cohort study
A Retrospective cohort study is one in which the outcomes have all occurred before the start of the investigation.
It is also known as
Historical cohort study
Prospective study in Retrospective
Non-current prospective study
Combination of retrospective and prospective cohort study
In this type of study, both the retrospective and prospective elements are combined. The cohort is identified from past records and is assessed of date for the outcome. The same cohort is followed up prospectively into the future for further assessment of outcome.
Elements of a cohort study
Selection of study subjects
Obtaining data on exposure
Selection of comparison groups
Selection of study subjects
The subject of a cohort study are usually assembled in one of 2 ways –
General population
When the exposure or cause of death is fairly frequent in the population , cohorts may be assembled from the general population, residing in well defined geographical, political and administrative areas.
Special groups
These may be special groups or exposure groups that can readily be studied

1.Select Groups
These may be preformed groups (e.g.doctors,nurses,lawyers,teachers,civil servants), insured persons, obstetric population, college alumni, govt. employees, volunteers, etc. These groups are usually a homogenous population.

2.Exposure Groups
If the exposure is rare, a more economical procedure is to select a cohort of persons known to have experienced the exposure.
Obtaining data on exposure
Information about exposure may be obtained directly from the
Cohort members – through personal interviews or mailed questionnaires
Review of records – certain kinds of information can be established only from medical records
Medical examination or special tests – Some types of information can only be obtained by Medical examination or special tests
Environmental surveys – This is the best source for obtaining information on exposure levels of the suspected factor in the environment where the cohort lived or worked.
Information about exposure should be collected in a manner that will allow classification of cohort members
According to whether or not they have been exposed to the suspected factor
According to the level or degree of exposure in the case of special exposure groups
Selection of comparison groups
Internal Comparisons
In some cohort studies , no outside comparison group is required. The comparisons are in built .
External comparisons
When information on degree of exposure is not available , it is necessary to put an external control to evaluate the experience of the exposed group.

C. Comparison with general population rates
If none is available , the mortality experience of the exposed group is compared with the mortality experience of the general population in the same geographic area as the exposed people.
The limiting factors in using general population rates for comparison are :
Non-availability of population rates for the outcome required
The difficulties of selecting the study and comparison groups which are representative of the exposed and non exposed segments of the general population
Follow up
One of the problems in cohort studies is the regular follow-up of all the participants . The procedures required compromise:
Periodic medical examination of each member of the cohort
Reviewing physician and hospital records
Routine surveillance of death records
Mailed questionnaires , telephone calls , periodic home visits – preferably all 3 on an annual basis .
The data are analysed in terms of
Incidence rates of outcome among exposed and non-exposed
Estimation of risk
1.Incidence Rates

In a cohort study , we can determine incidence rates directly in those exposed and those not exposed.
2. Estimation of risk
Having calculated the incidence rates , the next step is to estimate the risk of outcome in the exposed and non exposed cohorts .
Relative risk
Attributable risk
Relative Risk
Relative risk ( RR ) is the ratio of the incidence of the disease among exposed and the incidence among non exposed.
RR = Incidence of disease ( or death ) among exposed
Incidence of disease ( or death ) among non exposed
Estimation of relative risk is important in aetiological enquiries . It is a direct measure of the strength of association between suspected cause and effect .
Attributable risk
Attributable risk ( AR ) is the difference in incidence rates of disease ( or death ) between an exposed group and non exposed group.
It is often expressed as a parent .
AR = Incidence of disease rate among exposed minus incidence x 100 of disease rate among non exposed
Incidence rate among exposed
Attributable risk indicates to what extent the disease under study can be attributed to the experience.
Population attributable risk
It is the incidence of the diseases ( or death ) in the total po9pulation minus the incidence of disease ( or death ) among those who were not exposed to the suspected causal factor .
The concept of population attributable risk is useful in that it provides an estimate of the amount by which the disease could be reduced in that population if the suspected factor is eliminated or modified .
Relative Risk vs Attributable Risk
Relative risk is important in etiological enquiries . Its size is a better index than is attributable risk for assessing the aetiological role of a factor in disease . The larger the relative risk , the strength the association between cause and effect . But relative risk does not reflect the potential public health importance as does the attributable risk . Attributable risk gives a better idea than does relative risk of the impact of successful preventive or public health programme might have in reducing the problem .

Bias in cohort studies
There are 5 different bias types that are operative in a cohort study.
Selection bias
Follow up bias
Information bias
Confounding bias
Post Hoc bias
Selection Bias
Selection bias occurs when the group actively studied does not reflect the same distribution of characteristics like age , sex , etc. occurring in the general population .
Follow up bias
If the rate of disease is different among those lost to follow up , then internal validity of the study may be affected , that is , the relationship between exposure and outcome may be changed .
Information bias
This occurs when there is an error in the classification of 9ndividuals with respect to the outcome variable .
This may result measyrement errors , imprecise measurements and misdiagnosis os cases .
Confounding bias
It occurs when other factors that are associated with the outcome and exposure variables do not have the same distribution in the exposed anhd unexposed groups . The 2 common confounders in cohort studies are the factors of smoking and age . The risk of disease varies with age for almost all diseases .
Post hoc bias
It arised due to the use of data from o cohort study to make observations which were not part of the original study . Thus , interesting relationships are often observer in cohort studies which were not originally anticipated .

Incidence can be calculated
Several possible outcomes related to exposure can be studied simultaneously i.e. we can study the association of the suspected factor with many other diseases in addition to the one under study .
Cohort studies provide a direct estimate of relative risk.
Dose-response ratios can also be calculated
Since comparison groups are formed before disease develops , certain forms of bias can be minimized like misclassifications of individuals into exposed and unexposed groups .

Cohort studies involve a large number of people . They are generally unsuitable for investigating uncommon diseases or diseased with low incidence in the population .
It takes a long time to complete the study and obtain results by which time the investigators may have died or the participants may have changed their classification .
Administrative problems - loss of experienced staff , loss of funding,etc
The cohorts may migrate , lode interest or simply refuse to provide required information.
Those who volunteer may not be representative of all the individual with the characteristic of interest.
There may be changes in the standard methods or diagnostic criteria of the disease over prolonged follow up
Cohort studies are expensive
The study itself may alter peoples behaviour
Ethical problems
Practical considerations dictate that we must concentrate on a limited number of factors possible related to disease outcome
The main differences between case control and cohort studies are as follows :
Case control study Cohort Study
Retrospective Prospective
Disease has already occurred Disease is expected to occur in
the future
Presence of exposure in cases Development of disease in
And controls compared exposed and non exposed
Relatively easy to carry out Time consuming and difficult to
Carry out
Useful for rare cases with Suitable for common diseases
Smaller numbers with common exposure
Can have one outcome, but Can have multiple exposures
Can have multiple exposures
Only derives odds ratio Derives relative risk , attributable
Substantial biases can occur Biases are generally lower
Relatively less costly and no Expensive and dropout rate higher

Soben Peter
K Park

No comments: