Gabapentin Use in Medicaid After Descheduling

Abstract

This cross-sectional study investigates the association of descheduling of gabapentin with use among Medicaid enrollees in Michigan.

Introduction

Gabapentin has consistently ranked as one of the top prescribed medications in the US.¹ This high level of prescribing, driven by increasing off-label use² and pushes to reduce opioid prescribing for pain treatment, has sparked concern of misuse and exposure-related harms.³ Given the documented risks, including respiratory depression and death with concomitant psychotropic medications such as opioids,⁴ scheduling gabapentin as a controlled substance has been proposed as a possible supply-side policy solution.

Gabapentin is not scheduled as a controlled substance at the federal level, despite the closely related gabapentinoid medication pregabalin being scheduled under the Controlled Substance Act since its approval. To regulate gabapentin prescribing and reduce misuse-related harms, 8 states enacted legislation to classify gabapentin as a schedule V controlled substance between 2016 and 2024.⁵ Michigan became the fourth state to schedule gabapentin in January 2019 but subsequently removed it from the state-level controlled substance schedule in May 2024. According to the regulatory impact statement filed by the Michigan Bureau of Professional Licensing, this decision was driven partly by a desire to align Michigan’s classification with those of neighboring states and the federal government.⁶

Evidence suggests that state gabapentin-scheduling policies decrease use among multiple populations, including Medicaid enrollees.⁷ We assessed whether descheduling gabapentin was associated with increased use among Medicaid enrollees in Michigan. To our knowledge, no studies to date have evaluated the descheduling policy approach and associated changes in prescription gabapentin use.

Methods

This cross-sectional study is reported following the STROBE reporting guideline. The study used publicly available, aggregate, deidentified data; thus, institutional review board review and informed consent were not required per 45 CFR §46.104(d)(4)(i).

We conducted a repeated cross-sectional study using a controlled interrupted time-series design to assess the association between Michigan’s descheduling of gabapentin and immediate-level and trend changes in gabapentin dispensing among Michigan Medicaid enrollees. Pregabalin dispensing was selected as the internal negative control given its similar indications to gabapentin. Drug dispensing data were ascertained from the Medicaid State Drug Utilization Data,⁸ a publicly available dataset providing quarterly dispensing data for eligible outpatient prescriptions.

Medicaid and Children’s Health Insurance Program enrollment data were extracted from Centers for Medicare & Medicaid monthly enrollment reports.⁹ Dispensing rates were measured quarterly from quarter 1 (Q1) 2023 to Q3 2025 per 1000 Medicaid enrollees. Michigan descheduled gabapentin on May 28, 2024⁵; therefore, we coded the closest quarter (Q3 2024) as the interruption.

We modeled a segmented ordinary least-square linear regression to calculate immediate and trend changes in gabapentin dispensing relative to the control group, with Newey-West standard errors (lag = 2) evaluating group by postpolicy level and group by postpolicy trend interactions as the primary parameters of interest. The statistical test was 2-sided, and statistical significance was defined as a 2-sided P value < .05. Analyses were performed in R version 4.5.1 (R Project for Statistical Computing).

Results

Gabapentin dispensing in the Michigan Medicaid program ranged from 33 to 43 prescriptions per 1000 enrollees across the study period, with pregabalin dispensing consistently lower than that of gabapentin (Figure). Descheduling gabapentin was associated with an immediate increase in gabapentin dispensing of 4.00 prescriptions per 1000 enrollees (95% CI, 2.33 to 5.67 prescriptions) relative to pregabalin dispensing (Table). No evidence of a change in trend was observed (0.49 gabapentin prescriptions per 1000 enrollees per quarter; 95% CI, −0.15 to 1.13 prescriptions). The descheduling policy was associated with a small uptake in pregabalin dispensing, but absolute dispensing of gabapentin remained more than 3-fold higher at the end of the study period.

Figure. Line Graph of Quarterly Dispensing Rate of Gabapentin and Pregabalin.

Orange dashed line indicates the counterfactual trend from the controlled interrupted time-series model; vertical dashed line, the descheduling policy change.

Table. Association Between Descheduling Policy and Quarterly Gabapentin Prescription Rates.

Estimatea	Level (95% CI)b	Trend (95% CI)c
Prepolicy period in control group (pregabalin)	9.81 (9.33 to 10.29)	0.34 (0.21 to 0.46)
Prepolicy difference, gabapentin vs pregabalin	21.98 (20.41 to 23.54)	0.09 (−0.30 to 0.47)
Postpolicy change in control group (pregabalin)	0.68 (−0.01 to 1.36)	−0.29 (−0.55 to −0.04)
Postpolicy change, gabapentin vs control	4.00 (2.33 to 5.67)	0.49 (−0.15 to 1.13)

^a Model specification: Yᵢ = β₀ + β₁(time) + β₂(group) + β₃(group × time) + β₄(level) + β₅(trend) + β₆(group × level) + β₇(group × trend) + εᵢ. Group = 1 for gabapentin and 0 for pregabalin. Time = continuous quarter (1-12). Level = 0 prepolicy and 1 postpolicy period. Trend = 0 prepolicy and 1 to 6 postpolicy period. β₆ and β₇ are the primary parameters of interest, representing the policy-associated change in level and slope, respectively.

^b Quarterly gabapentin or pregabalin prescriptions per 1000 Medicaid enrollees.

^c Quarterly gabapentin or pregabalin prescriptions per 1000 Medicaid enrollees per quarter.

Discussion

This cross-sectional study found that Michigan’s descheduling policy of gabapentin products was associated with an increase in gabapentin dispensing among Medicaid enrollees relative to the pregabalin control. Along with prior evidence, this outcome supports a reciprocal association between gabapentin scheduling and use, with prescribing shown to decrease after scheduling⁷ and increase after descheduling. These findings suggest that prescribers may be more hesitant to prescribe gabapentin under stricter regulations but that this hesitancy loosens once restrictions are lifted. Given documented harms associated with gabapentin misuse, including involvement in drug overdose,¹⁰ the observed increase in gabapentin prescribing after descheduling may carry adverse implications.

We recognize several limitations. First, we used dispensing data, which cannot capture actual medication taken by patients or gabapentin obtained outside formal dispensing channels. Second, the analysis is limited to relatively short pre- and post-descheduling periods; while we found a significant level change, the model may lack enough statistical power to detect a meaningful change in the trend. The use of Newey-West standard errors improves robustness given this limited number of observations. Third, other interventions may have occurred during our study period that may have been associated with gabapentin use in this population, precluding causal inference.

Our findings suggest that drug scheduling may serve as a supply-side policy to prevent overprescribing. Future research should evaluate how prescriber behaviors are influenced by a medication’s controlled substance status and assess the association of such policies with access among patients with legitimate need.

Supplement.

Data Sharing Statement