Abstract
Statistical power in matched case-control studies depends on both the correlation coefficient between cases and their matched controls (phi) and the prevalence of exposure among controls (P0). To examine the hypothesis that the value of increasing the control-to-case ratio beyond 5 varies with both phi and P0, the authors estimated statistical power for a hypothetical case-control study under different assumptions. The effect of increasing the control-to-case ratio depended on phi and, to a lesser extent, on P0. The results suggest that investigators consider including more than five controls per case when either phi is greater than about 0.2 or P0 is less than about 0.15. (Author/publisher)