Abrupt Discontinuation of Long-term Opioid Therapy Among Medicare Beneficiaries, 2012-2017.

Background: With mounting pressure to reduce opioid use, concerns exist about abrupt withdrawal of treatment for the millions of Americans using long-term opioid therapy (LTOT). However, little is known about how patients are tapered from LTOT nationally.

Objective: Measure national patterns of LTOT discontinuation and adherence to recommended tapering speed.

Design: Observational study of Medicare Part D from 2012 to 2017.

Participants: Using claims for a 20% sample of Medicare beneficiaries, we included patients on LTOT for 1 year or more, defined as those with ≥ 4 consecutive quarters with > 60 days of opioids supplied in each quarter.

Main Measures: Our primary outcome was discontinuation of LTOT, defined as at least 60 consecutive days without opioids supplied. We additionally examined whether discontinuation of LTOT was "tapered" or "abrupt" by comparing LTOT users' daily MME dose in the last month of therapy to their average daily dose in a baseline period of 7 to 12 months before discontinuation. By the last month of therapy, patients with "abrupt" discontinuation had a < 50% reduction in their average daily dose at baseline.

Key Results: From 2012 to 2017, there were 258,988 LTOT users, 17,617 of whom discontinued therapy. Adjusted rates of LTOT discontinuation increased from 5.7% of users in 2012 to 8.5% in 2017, a 49% relative increase (p < 0.001). There was a similar increase in annual discontinuation rate for LTOT users on lower (26-90 MME, 5.8% to 8.7%, p < 0.001) vs. higher doses (> 90 MME, 5.3% to 7.7%, p < 0.001). The majority of LTOT discontinuations were stopped abruptly, and increased over time (70.1% to 81.2%, 2012-2017, p < 0.001).

Conclusions: Medicare beneficiaries on LTOT were increasingly likely to have their therapy discontinued from 2012 to 2017. The vast majority of discontinuing users, even those on high doses, had less than 50% reduction in dose, which is inconsistent with existing guidelines.