We used the data of a randomized, placebo-controlled trial of a cholera vaccine to compare the results of case–control analyses with those of cohort analyses. Cases of cholera were selected from the trial population followed for three years following dosing. For each case, we selected 4 age-matched controls who had not developed cholera. For each case and control, GIS was used to calculate vaccine coverage of individuals in a surrounding “virtual” cluster. Specific selection strategies were used to evaluate the vaccine protective effects.
66,900 out of 108,389 individuals received two doses of the assigned regimen. For direct protection among subjects in low vaccine coverage clusters, we observed 78% (95% CI: 47–91%) protection in a cohort analysis and 84% (95% CI: 60–94%) in case–control analysis after adjusting for confounding factors. Using our GIS-based approach, estimated indirect protection was 52% (95% CI: 10–74%) in cohort and 76% (95% CI: 47–89%) in case control analysis. Estimates of total and overall effectiveness were similar for cohort and case–control analyses.
The findings show that case–control analyses of individually randomized vaccine trials may be used to evaluate direct as well as population-level vaccine protection.