The Impact of Restrictive Drug Coverage Policies on Pharmacoepidemiologic Methods and Health Outcomes.

摘要

The unintended consequences of restrictive drug coverage policies on epidemiologic research methods and population health outcomes have been understudied. The primary objective of this program of research was to study the impact of these policies on the magnitude and direction of potential bias within administrative database studies. This was achieved through three related studies: 1) an observational study that estimated the magnitude of drug exposure misclassification in administrative data across seven therapeutic classes; 2) a simulation cohort study that quantified the potential degree of bias resulting from varying amounts of exposure misclassification to antidiabetic drugs introduced by restrictive drug coverage policies; and 3) a real-world cohort study that measured the effect of exposure misclassification introduced by capturing benefit drug use only on observed associations between exposure and outcome.;We demonstrated that incomplete drug exposure information for drugs with a restrictive coverage policy is more common than previously thought. In fact, we found that on average, drugs with a restrictive coverage had a 40% absolute lower capture rate within one of the most widely used and accepted drug administrative databases, compared to drugs without coverage restrictions. Although our simulation study suggested a large degree of bias might be introduced when drug exposure is differentially misclassified according to a drug policy, results from our cohort study with real-world data demonstrated that a clinically important degree of bias was not apparent, at least for our three study drugs.;In addition to impacting research methods, restrictive drug coverage policies themselves may have unintended clinical consequences at the population-level. Therefore, a second major initiative of the research program was to examine the population-level impact of removing a specific restrictive coverage policy. The fourth study demonstrated that removal of a prior authorization policy for thiazolidinediones significantly influenced drug utilization but did not adversely impact health outcomes.;The results from our program of research highlight the importance of giving serious consideration to the impact of restrictive drug coverage policies when designing, analyzing, and interpreting a pharmacoepidemiologic study. Further, we demonstrate the usefulness of rigorous evaluation for understanding the population-level consequences of the removal of a drug policy.