Selection Bias - Cohort Study or Randomized Clinical Trial

Selection Bias in Cohort Studies Less Likely

Among initially selected subjects, selection bias in a cohort study is less likely to occur compared to case-control or CrossSectionalStudy cross-sectional studies.

The Reason: study participants (exposed or unexposed; treatment vs placebo) are selected (theoretically) before the outcome occurs.

As you can see in the 2x2 table below, since disease has not occurred yet among initially selected subjects, there is typically no opportunity for disproportionate sampling with respect to exposure and disease. (We cannot draw the 4 arrows from the Source Population to the Study Sample.)

In other words, among participants in a cohort study or randomized trial Because disease has, by definition, not yet occurred, there is no opportunity, at the beginning, to over or under sample any of the 4 cells in the 2x2 table.


Schematic of Cohort Study Sampling

As you can see in the schematic below all that is sampled is exposure status. Even if disproportionate sampling of exposed or unexposed groups occurs, it will not result in selection bias when forming measures of association.

All you do have at the beginning of a cohort study or RCT is knowledge of who is exposed (or treated) and unexposed (or untreated). We call these the margins.

Front-End Bias in Cohort Studies

Selection bias can occur in cohort study at the front end by unknowingly entering diseased individuals into the study.

Cohort studies are not foolproof, however, in terms of no-bias on the front end of choosing the cohort because sometimes truly diseased persons are unknowingly entered into the cohort.

Example of Front-End Bias in Cohort Studies: Exercise and Survival

Consider a study of the effects of exercise on all-cause mortality in persons who are thought to be completely healthy at baseline.

Schematic of Front-End Bias

As shown in the schematic below, selection bias will lead to spurious protective effect of exercise (assuming truly no effect).

At the end of the study, how well will the study sample represent the source population? It is the "no exercise" group where trouble exists.


Front-End Bias in Clinical Trials

Loss to Follow-up

Selection bias can occur in cohort study at the end in loss-to-follow-up.

Selection bias among initially selected subjects in longitudinal studies, however, is not the primary source of selection bias in these cohort studies/ RCTs. Selection bias occurs when those persons lost to follow up have a different probability of the outcome than those persons who remain in the analysis, that is informative censoring in at least one of the exposure groups, AND when the degree of informative censoring differs across exposure groups.

Note that this all starts with those who are lost having a different incidence of the outcome than those who remain. The problem, of course, is that we rarely know the outcome experience of those who leave, other than in studies of all cause mortality where we can use the National Death Index.

See Also

Other causes of selection bias

Greenland, S. (1977). Response and follow-up bias in cohort studies. Am J Epidemiol, 106 (3), 184-187.

Jeff Martin, MD

-- MaryB - 08 May 2009