PNS69 THE MEDICARE PART D COVERAGE GAP CLOSURE, PRICE SENSITIVITY, AND UTILIZATION: WHICH DRUG CLASSES SAW THE LARGEST EFFECT?

Abstract

The Medicare Part D coverage gap began to close in 2011, starting with a 50% discount for brandname drugs the initial year, with discounts gradually increasing thereafter. Policies to lower the cost of drugs should theoretically increase utilization, but it is unclear whether there are heterogeneous effects across different types of drugs based on their characteristics (e.g. brand dominance, tolerability, use for mental health, etc.). We examined differences in response to the coverage gap closure by drug class. We examined 100% of Medicare Claims from 2006-2016. For each drug class defined by USC-2 codes, we ran a separate difference-in-difference regression model comparing utilization before and after the coverage gap closure, across beneficiaries who did and did not have a coverage gap. Notably, beneficiaries who receive low-income subsidies (LIS) do not face a coverage gap. We used the regression results from each of the separate models to run a meta-regression analysis to examine drug characteristics associated with a larger response to the coverage gap closure. The coverage gap closure was associated with an increase in utilization of 2.5% across all classes examined in the non-LIS group relative to the LIS group (P<0.01). The estimates by drug class ranged from a 1.4% decline in utilization (diabetes drugs) to a 7.9% increase in utilization (genitourinary drugs) of the non-LIS group relative to the LIS group. There was no statistically significant relationship between any of the factors examined (potential for abuse, mental health drug, or brand dominance) and the effect of the gap closure on changes in utilization. Prescription drugs are a large contributor to healthcare costs, and access critically depends on out-of-pocket costs. The fact that policies designed to lower out-of-pocket costs have wide ranging effects across drug classes may mean that more targeted policies are necessary to affect particular outcomes of interest.