Figure 1-3 illustrates the structure of case-control studies. Case-control studies investigate a hypothesis about an association between exposures or potential risk factors (eg, smoking, medical conditions, therapies) and outcomes of interest (eg, loss of visual acuity, development of glaucoma, corneal graft failure, complications of cataract surgery). Case-control studies select a group of participants with the disease of interest (cases) and a group of comparable individuals who are free of disease (controls). Each study compares the past exposures and characteristics of the 2 groups to determine whether differences exist between the groups. If so, the study will conclude that the exposures or characteristics that differ are associated with the disease.
Researchers select cases and controls from a current database and obtain the history of exposures through patient surveys and/or review of medical records. Thus, researchers can perform case-control studies more quickly and inexpensively than cohort studies (discussed later in the chapter), because cohort studies require extra time and money to follow participants. During record reviews or patient interviews, case-control studies can collect data on many potential risk factors simultaneously.
Figure 1-3 Simplified schematics of observational study designs.
However, exposure data may be less accurate in case-control studies than in cohort studies. For example, patients with retinal vein occlusion (cases) may be more likely than control patients to recall taking medications (eg, aspirin) in the past because control patients, who do not have the disease, may be less motivated to scrutinize their past behavior. Therefore, a higher proportion of cases than controls might report use of aspirin in the past 6 months, even if in truth the proportion of aspirin users was the same in both cases and controls. This recall bias in cases may strengthen the association between aspirin use and vein occlusion.
Case-control studies may be subject to selection bias if they do not have an appropriate control group. For example, a study may show that myopia offers a protective effect on retinal vein occlusion if the study collects its cases from a retinal group but collects its controls from a general ophthalmology practice that offers refractive surgery for myopia. The proportion of individuals with myopia would be smaller among the cases (from the retinal group) than among the controls (from the general practice). The study may conclude that myopia is protective against retinal vein occlusion, but the apparent association would actually be attributable to selection bias. To learn more about other sources of bias, an interested reader may consult general epidemiology textbooks, such as the following.
Rothman KJ, Greenland S, Lash TL. Modern Epidemiology. 3rd, mid-cycle revision ed. Philadelphia: Lippincott Williams & Wilkins; 2012.
Excerpted from BCSC 2020-2021 series: Section 1 - Update on General Medicine. For more information and to purchase the entire series, please visit https://www.aao.org/bcsc.