Medicaid Unwinding: Association With New and Ending Buprenorphine Treatment Episodes

Abstract

Importance:

Medicaid is an important source of coverage for buprenorphine treatment for opioid use disorder (OUD). The loss of Medicaid coverage through Medicaid unwinding may substantially impact individuals’ ability to initiate and continue buprenorphine, but little is known about how Medicaid unwinding is associated with buprenorphine OUD treatment.

Objective:

Examine the association of Medicaid unwinding overall and by magnitude of unwinding with changes in buprenorphine OUD treatment.

Design, Setting, and Participants:

Retrospective cohort study using 2021–2023 national retail pharmacy data on dispensed buprenorphine prescriptions.

Exposure:

Magnitude of Medicaid disenrollment in the 6 months after unwinding began.

Main Outcome(s) and Measure(s):

Average monthly change in new and ending buprenorphine treatment episodes, defined as starting buprenorphine treatment after 30 days without buprenorphine and ending after 30 days without buprenorphine. Statistical significance was assessed using 95% CI constructed from 2-sample t tests.

Results:

Medicaid unwinding was associated with increases in Medicaid buprenorphine episodes ending (+3.0%, 95% CI: −1.1, 7.0, n = +5350) and decreases in new Medicaid episodes (−2.6%, 95% CI: −5.3, 0.1, n = −5756). The greatest changes were in states with the greatest disenrollment (+5.5%, 95% CI: 0.8, 10.1; n = +2320 for episodes ending and −3.9%, 95% CI: −7.3, −0.6; n = −1843 for episodes starting). Changes in Medicaid-covered episodes were not offset by other payers: among all payers, we observed an increase in episodes ending (+2.7%, 95% CI: 1.9, 3.4, n = +10,300) and a negligible increase in new episodes (+0.1%, 95% CI: −0.7, 1.0, n = +570).

Conclusions and Relevance:

Medicaid unwinding may have resulted in substantial disruptions to buprenorphine treatment.

Medicaid covers ~40% of treatment for Americans with opioid use disorder (OUD).¹ Medication treatment for OUD, including buprenorphine, is the most effective treatment.^2,3

Given state Medicaid programs’ role in covering OUD treatment, loss of Medicaid coverage may substantially affect individuals’ ability to initiate and continue treatment.⁴ States began disenrolling individuals from Medicaid after the end of the continuous Medicaid enrollment provisions of the Families First Coronavirus Act⁵ (hereafter, unwinding). It is critical to understand how unwinding may have affected the treatment of individuals receiving buprenorphine. Studies commonly focus on total prescriptions and find unwinding associated with fewer Medicaid-paid fills.^4,6–8 To provide a more nuanced understanding of unwinding’s effect on individuals, we used national retail pharmacy-dispensed buprenorphine claims to examine associations with individuals starting or terminating buprenorphine treatment.

METHODS

Study Data and Methods

We used IQVIA Real World Data longitudinal prescriptions (LRx) data, capturing 93% of prescriptions dispensed at retail pharmacies in 50 states and the District of Columbia, to identify buprenorphine prescriptions for OUD treatment dispensed January 1, 2021, to December 31, 2023. We excluded buprenorphine formulations approved solely for pain. We constructed buprenorphine treatment episodes (hereafter, episodes) as follows: Episodes started with the first dispensed buprenorphine prescription after a 30-day buprenorphine-free period in which the days’ supply from previously dispensed buprenorphine prescriptions was exhausted. Episodes ended when the days’ supply of buprenorphine was exhausted, followed by at least 30 days with no new dispensed buprenorphine. In sensitivity analyses, we explored an episode definition using 60 days; results were not sensitive to the change. We identified new episodes’ first prescription payer and ending episodes’ last payer. We used Centers for Medicare and Medicaid Services data to determine each state’s Medicaid unwinding effective date⁹ and Medicaid enrollment¹⁰ and Census Bureau data to determine state population.¹¹ The corresponding author’s IRB deemed the study exempt.

Analysis

For each state, we calculated the magnitude of unwinding using Medicaid enrollment in the last month before unwinding began and Medicaid enrollment in December 2023, dividing monthly adult Medicaid enrollment population by state population. We determined the number of new and ending episodes in each month during the study period. We used data for the 6 months before and after unwinding to calculate the monthly relative change in new and ending episodes. We calculated the average monthly new episodes in the 6 months before and 6 months after a state began unwinding for new episodes, and the average monthly episodes ending in the 6 months before and 5 months after a state began unwinding for ending episodes.

We categorized states into terciles based on unwinding magnitude: states with the greatest, moderate, or smallest decline in Medicaid enrollment. We excluded Oregon because we lacked a sufficient observation period after Oregon’s unwinding began. Overall and by state terciles and payer, we calculated the percent change in new/ending episodes post-unwinding by calculating the difference in average monthly new/ending episodes in the 6 months pre-unwinding and average monthly new/ending episodes in the 5 months post-unwinding and dividing by the pre-unwinding 6-month average; we assessed statistical significance using 95% CIs constructed from 2-sample t tests. Analyses were performed using SAS Enterprise Guide 8.3.

RESULTS

We identified 3,616,779 episodes among 2,234,560 individuals from January 1, 2021, through December 31, 2023. Approximately 45% of episodes were paid for by Medicaid, 18% by commercial insurance, 11% by Medicare, 11% by discount cards/vouchers, and 6% were cashpay (Table 1). Approximately 27% of episodes were in high-unwinding states, 41% in moderate-unwinding states, and 31% in low-unwinding states.

TABLE 1.

Payer and State Unwinding Status for New and Ending Buprenorphine Treatment Episodes

	%
	Episodes Starting Between February 1, 2021 and December 31, 2023	Episodes Ending Between January 1, 2021 and November 30, 2023
	N = 2,816,309	N = 2,664,946
Overall
Medicaid	44.2	45.7
Commercial Insurance	18.5	18.1
Medicare	10.9	11.0
Self-pay	6.2	5.4
Discount card/voucher	11.4	10.9
Other	9.0	8.9
State unwinding status
Greatest unwinding	27.7	27.5
Moderate unwinding	41.2	41.7
Least unwinding	31.1	30.8

Medicaid Episodes

On average, there were 3% (95% CI: −1.1, 7.0; n = +5350) more monthly episodes paid for by Medicaid that ended after unwinding began compared with before unwinding. In states with the greatest unwinding, the percentage increase (5.5%, 95% CI: 0.8, 10.1; n = +2320) was more than double that in states with moderate (2.2%, 95% CI: −2.6, 6.9; n = +1806) or low (2.2%, 95% CI: −1.8, 6.3; n = +1224) unwinding (Table 2).

TABLE 2.

Average Monthly and Cumulative Change in Starting and Ending Medicaid Buprenorphine Treatment Episodes Post-Unwinding, by Unwinding Status

	Change in Starting Episodes		Change in Ending Episodes
	Average (%) (95% CI)	Cumulative (n)	Average (%) (95% CI)	Cumulative (n)
State unwinding status
Greatest unwinding	−3.9 (−7.3, −0.6)	−1843	5.5 (0.8, 10.1)	2320
Moderate unwinding	−2.4 (−6.1, 1.2)	−2580	2.2 (−2.6, 6.9)	1806
Least unwinding	−2.0 (−5.6, 1.6)	−1333	2.2 (−1.8, 6.3)	1224
Overall	−2.6 (−5.3, 0.1)	−5756	3.0 (−1.1, 7.0)	5350
Medicaid unwinding status
Greatest unwinding	2.7 (1.6, 3.7)	3515	5.8 (5.0, 6.6)	6172
Moderate unwinding	0.0 (−1.0, 1.1)	86	2.2 (1.3, 3.1)	3485
Least unwinding	−2.0 (−3.2, −0.8)	−3031	0.5 (−0.3, 1.4)	643
Overall	0.1 (−0.7, 1.0)	570	2.7 (1.9, 3.4)	10,300

We found 2.6% (95% CI: −5.3, 0.1; n = −5756) fewer new episodes paid for by Medicaid in the 6 months after unwinding compared with 6 months before unwinding. We observed a significant 3.9% decrease (95% CI: −7.3, −0.6; n = −1843) in episodes beginning in states with the greatest unwinding, compared with a 2.4% (95% CI: −6.1, 1.2; n = −2580) decrease in states with moderate unwinding and a 2.0% (95% CI: −5.6, 1.6; n = −1333) decrease in states with the least unwinding.

Episodes Among All Payers

Across all payers, including self-pay, monthly episodes ending were 2.7% (95% CI: 1.9, 3.4; n = +10,300) higher in the months after unwinding compared with the monthly average before unwinding. States with the greatest unwinding had the greatest increase in episodes ending (5.8%, 95% CI: 6.0, 6.6; n = +6172) compared with a 2.2% (95% CI: 1.3, 3.1; n = +3485) increase in states with moderate unwinding and 0.5% (95% CI: −0.3, 1.4; n = +643) in states with the least unwinding (Table 2).

There was a negligible 0.1% (95% CI: −0.7, 1.0; n = +570) increase in new episodes in the months after unwinding compared with months before unwinding. However, compared with Medicaid, a different pattern emerged across all payers: a 2.7% (95% CI: 1.6, 3.7; n = +3515) increase in new episodes in states with the greatest unwinding, no change in states with moderate unwinding, and a 2.0% (95% CI: −3.2, −0.8; n = −3031) decrease in states with the least unwinding.

DISCUSSION

From April 2023 through September 2024, more than 25 million individuals lost Medicaid coverage as a result of Medicaid unwinding;¹² ~13 million seem not to have re-enrolled.¹³ Our analysis suggests that these changes were associated with a substantial increase in the number of Medicaid enrollees stopping buprenorphine treatment and a substantial decrease in enrollees beginning it. Consistent with research on buprenorphine fills,⁶ effects were greatest in states with the greatest disenrollment.

Changes in Medicaid episodes seem not to be offset by changes in reimbursement from other sources, such as individuals losing Medicaid paying out-of-pocket for buprenorphine or obtaining commercial insurance. Among all payers, we observed an increase in episodes ending after unwinding, with little change in the number of new episodes.

We note several limitations. We observed only retail buprenorphine prescriptions dispensed January 1, 2021, to December 31, 2023; we can neither identify seasonal patterns of buprenorphine episodes before unwinding nor how buprenorphine treatment may have evolved subsequently; we have insufficient data to apply more rigorous causal inference methods. We also have less post-unwinding data for the 11 states that started unwinding after June 2023; we do not know to what extent additional post-unwinding data would influence our results. We do not know if our findings generalize to pharmacies not captured in the IQVIA data, settings dispensing buprenorphine (eg, inpatient hospital or residential treatment), or if individuals were transitioned to methadone.

Despite limitations, our analysis enriches our understanding of how Medicaid unwinding may have affected buprenorphine care for OUD: Medicaid enrollees in states with greater unwinding had reductions in buprenorphine treatment not offset by other payers. This finding is particularly salient at a time when policy changes are likely to cause many individuals to lose Medicaid coverage. To sustain progress against the opioid overdose crisis,¹⁴ it is vital to ensure that individuals who can benefit from this life-saving medication continue to receive it.