Differential Responses of Minorities to the Medicare Part D Coverage Gap

Poor compliance with medications is endemic, particularly among the elderly.  Nearly half of all patients prescribed pharmaceutical therapies do not take sufficient doses to experience a therapeutic effect, and a meta-analysis on medication adherence among the elderly indicated that 29% to 59% of patients did not properly adhere to their outpatient medications.  Patients fail to adhere to their medications for numerous reasons, but economic incentives matter.  Higher patient cost-sharing is associated with lower rates of compliance, particularly among low-income, minority populations.   This research investigates whether minority groups respond differentially to the gap in prescription drug coverage in Medicare Part D, the so-called “doughnut hole.”

Using a differences-in-differences-in-differences approach, we compare medication use of blacks and Hispanics, versus whites, in widely-used drug classes before and after reaching the coverage gap.  The recipients of the full Medicare Part D Low Income Subsidy (LIS) serve as a control group.  Their deductibles and monthly premiums are paid by the federal government, as well as almost all of their drug costs throughout the year – and thus they are unaffected by the coverage gap.  By contrast, the treatment group is comprised of non-LIS beneficiaries who face the full cost of their medications once they reach a threshold level of spending in the year. 

We examine whether the response to the coverage gap differs by race/ethnicity after controlling for variation in socioeconomic status, using zip-code level measures from the American Community Survey.  Preliminary analyses on a subsample of persons with diabetes show that less wealthy Hispanics react more strongly to the coverage gap compared with whites and blacks.  Looking over nine common diabetes drug classes, we observe Hispanics decreasing their use of medications most, followed by blacks, then whites.  Blacks reduce medication use by 4 percentage points (pp) and Hispanics by 6pp, versus 3pp for whites. Among nine of the most widely utilized non-diabetes drug classes, adherence of blacks falls by 8pp, Hispanics by 9pp, while whites decrease utilization by 6pp.  The largest effects for both diabetes and non-diabetes drug classes are associated with the most costly drugs.

We find that about 40% of the differential decline in medication use can be attributed to stopping therapy altogether. More specifically, 7% of Hispanics stop therapy completely in the coverage gap in 2007, compared with 4% of blacks and 2% of whites.  Only 13% of Hispanics that stop resume therapy once benefits reset in the new calendar year (versus 6% of blacks and 7% of whites).  

We see non-LIS beneficiaries of all races switching to generic drugs at modestly higher rates than their LIS counterparts after reaching the coverage gap.  However, we do not observe differential take-up of generics by race.  On average, all three racial groups enter the doughnut hole in late August, leaving them less than five months until their coverage clock restarts.  It is possible that beneficiaries are unwilling to expose themselves to the risk of switching to a new medication, instead deciding to remain on their more costly branded drug.