Home
[size=12]Case Control Study
[/size] In a case control study design, persons having the disease of interest are selected and the history of exposure to risk factor(s) is obtained from them. These subjects are known as cases. Similarly a set of persons is selected who do not have the disease. They are known as controls. The history of exposure to the risk factor(s) is obtained from them as well. A comparison is drawn between cases and controls to establish association between the risk factor(s) and the disease.
Since we start with people who already have the disease and go back in time to obtain history of exposure, such studies are also known as retrospective studies.(However one should know that cohort study may also be retrospective).
(A) Selection of Controls
Considerable attention should be given to selection of controls. Proper selection of control is even more critical than that of study subjects. As far as possible, control group should resemble the study group in some attributes except that control are not affected by the disease. It may be difficult to obtain exactly matching controls in the community, especially when diagnosis of the disease is difficult to establish. In such Situations, it may be difficult to exclude the presence of minor forms of disease among the control subjects. Controls are matched on those variables, which may be confounders. The predictor variables are measured in both cases and controls.
Controls may be obtained from a variety of sources such as (i) hospital patients, (ii) relatives,(iii)neighbors, and (iv)general population. When hospital patients are selected as controls, selection criteria should be defined before starting the study.
Selection of controls from among relatives is generally easy and practical because they have similar genetic and environmental background. They are accessible and it is easy to obtain their consent. When patients and controls are almost equally exposed to environmental factors, some people get the disease and others do not. In such cases, protection depends on the degree of exposure and inherent immunity in the individual. It is inappropriate to get relatives as controls in such setting. Since neighbors share the same environment, they should be preferred for investigation of genetic disorders.
…………Disease………….. | Total | |||
Present | Absent | |||
Exposure | Present | a | b | Exposed (a+b) |
Absent | c | d | Not exposed (c+d) | |
Total | Cases (a+c) | Controls (b+d) | a+b+c+d | |
.a indicates subjects with both risk factors and the disease
.b indicates subjects with the risk factors but not the disease
. c indicates the subjects with the disease but not with the risk factors
. d indicates subjects with neither the risk factors nor with the disease
Odds ratio = (a/b) / (c/d) = ad / bc
In Case control Study, The above is 2x2 contingency table to compute odds ratio
Just like the selection of study subjects, the control subjects should also be selected by random sampling. As already stated, distribution of attributes to be studied in the controls should also be stratified. Controls could be selected and paired in 1:1 or grater ratio.
(B) Odds Ration of Exposure.
Among the cases and controls, we will find some who did have the exposure to the risk factors and others who did not have the exposure. The findings can be, at the basic level, organized into a two by two contingency table as shown above. The strength of association is measured as a ratio of odds of developing the disease. Vis a Vis having the risk and not developing the disease. This measurement is known as exposure odds ratio,
The following is the criteria that need to be satisfied before planning a case control studies.
1. The primary aim is to select cases and matching controls, collect information on the extent of exposure to the risk factor(s) in both groups and then compare the quantum of exposure with respect to presence or absence of disease and provide a measurement of association by means of exposure odds ratio.
2. A clear definition of what constitutes a case should be available. Inconsistency in diagnosis criteria or lack of clear definition may introduce misclassification bias, which will be described in data analysis stage.
3. A Clear definition for inclusion as control should be available. This is to be decided on the basis of known confounders, bias and effect modifiers.
4. Number of cases and controls (Sample Size) should be adequate to have valid results.
5. The ratio of cases and controls should be at least 1:1 or 1:2. More than 4 controls per cases do not add significantly to the outcome of the study.
6. Controls and Cases should preferably be age and sex matched to eliminate the confounding effect of age and gender.
7. A case control study design is appropriate for studying diseases with very low incidence and prevailing rates.
To be Continued...................
» Video for our MPH colleagues. Must watch
» Salam
» Feeling Sad
» Look here. Its 2020 and this is what we found
» Sad News
» Pakistan Demographic Profile 2018
» Good evening all fellows
» Urdu Poetry