Gravar-mail: Proper Analysis of Secondary Phenotype Data in Case-Control Association Studies