In the previous article, I defined the case-control study and discussed issues regarding the selection of controls. In this second article devoted to case-control studies, I will discuss bias and the advantages and disadvantage of this clinical study design.
Bias in case-control studies
Selection bias: As mentioned previously, the selection of controls is the most important decision in designing a case-control study design; inappropriate selection of controls could bias the results of the study ( Table I ).
|Temporomandibular joint pain group||Control group|
|Selection bias||Select patients from high-income area||Select patients from low-income area|
|Recall bias||More likely to correctly recall potential exposure to trauma||More likely to incorrectly recall potential exposure to trauma|
|Observer bias||Might incorrectly ascertain the exposure if the status is known (ie, overestimate)||Might incorrectly ascertain the exposure if the status is known (ie, underestimate)|
Recall bias: In case-control studies, information on past exposures collected by interviewing study participants can be prone to recall bias. Patients with temporomandibular joint pain might be more likely to think about previous trauma that might have led to the outcome compared with participants in the control group. Recall bias usually (but not always) results in underreporting of past exposures in the control group or overreporting in the case group, causing overestimation of any association between exposure and outcome. If for some reason underreporting of the exposure occurs only among the cases (ie, reporting in the controls is accurate), this will result in underestimation of the effect. In other scenarios, such as the association between orthodontic therapy and periodontal disease, recall bias can be a smaller problem, since ascertainment of the exposure is likely to be accurate.
Observer bias: This bias occurs when the assessment of the exposure status of a subject in the study is affected by knowledge of whether that person is a case or a control. If the investigator collecting the information on the trauma knows the status of the participants (case or control), he or she might be inclined to underreport the exposure among the controls if he or she believes that the trauma is associated with temporomandibular joint dysfunction, and vice versa. In this situation, it would be beneficial if the investigator is unaware of the case-control status, but this is often difficult.
Advantages and disadvantages of case-control studies
The advantages are the following.
They are relatively easy and inexpensive to conduct compared with cohort studies.
Rare diseases or diseases with long latent periods can be examined, since the subjects are selected on the basis of their outcome, not their exposure status.
We can study multiple exposures for 1 outcome.
Losses to follow-up are not a problem.
The disadvantages are the following.
They are prone to selection bias and recall bias.
It is difficult to establish temporal causality. Is the outcome a consequence of the exposure or is the exposure a consequence of the outcome?
Case-control studies are not suitable for studying rare exposures because it would be difficult to recruit the required sample size.
Only the prevalence of the exposure among those with and without the outcome can be measured. Calculating prevalence or incidence in the population is not possible.
In Table II , we have 800 cases and 1700 controls, but this does not mean that the prevalence of temporomandibular joint dysfunction is 800 1700 + 800 = 0.32, or 32%, because the investigator decides how many cases and how many controls to select, thus varying the proportion of cases. Table III shows an example of when case-control studies are more appropriate.