Sensitivity Analyses in cohort studies are used to?

• A. Test uncertainty in sample sizes
• B. Test uncertainty in assumptions
• C. Test study feasibility
• D. Test data completeness

Answer: (B)

Sensitivity analyses in cohort studies probe how robust the findings are to uncertainty in the assumptions that underlie the analysis. In practice, the estimated association rests on several key assumptions: that all important confounders have been measured and properly adjusted for; that exposure and outcome are measured accurately; that the chosen model correctly represents the data; and that censoring or loss to follow-up does not bias results. Since real-world data rarely meet all these assumptions perfectly, sensitivity analyses explore how results would look under plausible alternative scenarios.

For example, you might test how the association changes if there is some unmeasured confounding, if exposure is misclassified, or if different methods are used to handle missing data (such as complete-case analysis versus multiple imputation). You could also vary the exposure definition, the timing of exposure, or the functional form of the model. If the main result remains similar across a range of reasonable scenarios, you gain confidence that the finding is robust. If the results shift substantially under plausible changes, the conclusion is more tentative and depends on those assumptions.

This approach is distinct from concerns about sample size uncertainty or study feasibility. It specifically targets how dependent the conclusions are on the assumptions built into the analysis.