Text this: Informed conditioning on clinical covariates increases power in case-control association studies