Effect and Correction of Measurement Error in Pharmacoepidemiologic Studies with Covariates