The effect of case-control ratio in a case-control association study

Abstract

The numbers of case and control subjects do affect statistical power of the association results in a case-control association study. The more the subjects, the more reliable results we get. Does the ratio of case/control also affect the results? In this study, we designed a series of case-control association studies based on a SNP chip as below in order to see the effect of case/control ratio. The association tests were done with 24 ASO stroke patients, 24 lacunar stroke patients, and 96 healthy controls for each stroke (age and sex matched).

case control case : control 24 stroke patients 24 controls 1:1 24 stroke patients 48 controls 1:2 24 stroke patients 72 controls 1:3 24 stroke patients 96 controls 1:4

We tested if there is any change in the association results as unmatched case-control studies. Undoubtedly, equal numbers of cases and controls are ideal. However, sometimes the number of cases is small and cannot be increased for a variety of reasons. An equally small number of controls would provide little ability to find associations. Increasing the number of controls up to a certain ratio improves the power of the study. A little power improvement results from increasing the number of controls.