Using covariate-specific disease prevalence information to increase the power of case-control studies