The Effects of Medicare Advantage on Opioid Use

Background: Health insurance plans may be able to influence opioid use by beneficiaries, but research investigating the role of plans is limited. We study whether the form of insurance coverage affects opioid use in Medicare Part D. Beneficiaries obtain Part D coverage through either a stand-alone prescription drug plan (PDP) that supplements traditional fee-for-service Medicare, or a Medicare Advantage prescription drug (MA-PD) plan that is integrated with a Medicare Advantage (MA) plan to cover all Medicare benefits including drugs. While PDP administrators have essentially no incentive or ability to manage the physicians who prescribe covered drugs, MA-PD plan administrators have an incentive to account for the spillover effects of prescription drug use on the cost of care overall and can choose network physicians and how the plan manages care. We investigate whether beneficiaries in MA-PD plans and PDPs differ in their receipt of opioid prescriptions, and whether they differ in the use of physicians who are high opioid prescribers.

Data and Methods: We use Medicare claims data to identify beneficiaries in either an MA-PD plan or a PDP for all of 2014, and identify those who had any opioid prescription and those who received a >7-day supply in 2014. We also identify the top 1% of physician prescribers in terms of the number of opioid prescriptions, and determine whether beneficiaries received any opioid or a >7 day supply from a top prescriber. We estimate individual-level regression models that relate measures of opioid use to enrollment in an MA-PD plan vs a PDP. Enrollment in an MA-PD plan may be endogenous. Accordingly, we use IV, instrumenting for MA-PD plan enrollment using an indicator for whether a beneficiary resided in a county subject to the “urban floor,” which increases payments to MA plans in counties in MSAs over 250,000 population, but not for counties in MSAs with populations under 250,000, creating a discontinuity. These higher payments increase the likelihood of enrollment in MA and thus in an MA-PD plan. To improve the performance of the instrument, we focus the analysis on beneficiaries who reside in counties in MSAs with populations between 100,000 and 400,000.

Results: We identify 536,481 beneficiaries meeting study criteria, 37% in an MA-PD plan and 63% in a PDP. 31% of the sample had any opioid prescription in 2014. IV estimates show that MA-PD plan enrollees were 11 percentage points, more than 30%, less likely to receive any opioid prescription than a PDP enrollee. There was no difference in the likelihood of receiving >7 days supply conditional on receiving any prescription. MA-PD enrollees are more than 6 percentage points less likely to receive an opioid prescription from a top prescriber, accounting for more than half of the overall (11 percentage point) reduction.

Conclusions: MA-PD plan enrollment is associated with significantly reduced opioid use for Medicare beneficiaries, suggesting that health plan activities can have an impact. An important mechanism appears to be management of the use of high prescribing physicians by MA-PD plans.