One barrier removed, four in five still without buprenorphine: the large count after the X-waiver
- Retrospective observational analysis of de-identified electronic health records from the TriNetX network, which the authors describe as a global network drawing predominantly on United States data, comparing two calendar cohorts of adults aged 18 and over: opioid use disorder recorded in 2021-2022, while the federal waiver was still required, against 2023-2024, after it was removed. 1 041 754 adults carried an opioid use disorder diagnosis (516,508 in the waiver-required and 525,246 in the waiver-elimination period), and 295,675 of them had the diagnosis entered for the first time (146,904 and 148,771). Eligibility required at least two outpatient visits; cancer and skilled nursing or palliative care were excluded. There is no control group and no interrupted time series: the design is one calendar period set against another.
- The result carried in the article title. Among adults with a new diagnosis, buprenorphine was prescribed to 27,665 (18.8%) while the waiver was required and to 29,223 (19.6%) after it was removed, adjusted odds ratio 1.14 (95% CI 1.12 to 1.16). New diagnoses were 2% less likely in the later period, adjusted odds ratio 0.98 (95% CI 0.97 to 0.98). Mean time from a new diagnosis to the first buprenorphine prescription was 56.3 days (standard deviation 176.3) against 46.0 days (standard deviation 115.9), p < 0.001; the standard deviation runs three times the mean in the earlier period and two and a half times the mean in the later one, so those averages sit on a heavily skewed distribution.
- In absolute terms the movement is 0.8 percentage points, and the crude odds ratio implied by 18.8% and 19.6% is 1.0536. The published adjusted estimate of 1.14 carries roughly two and a half times the excess odds above unity that the crude one carries. The two figures also do not describe the same people: the regression was built on 87.1% of the eligible sample, so the gap reflects both the adjustment for baseline characteristics and the narrowing of the sample. Those characteristics did differ between the two cohorts on every variable in Table 1, but with 295,675 people significance is close to automatic and the absolute gaps are minute: females were 45.4% of the earlier cohort and 45.1% of the later one. Nobody examined a patient: the outcome is a prescription order in a record, and "initiation" was defined as at least two prescriptions of a buprenorphine formulation approved for opioid use disorder. Dispensing, ingestion and retention are outside the data.
- What the study does not contain. Retention in treatment, overdose and death are absent, and the authors state that further research with longer follow-up is needed to assess the impact on overdose, mortality and other harms. The buprenorphine model was built on 131,330 people from the waiver-required period and 126,122 from the waiver-elimination period, 87.1% of the 295,675 newly diagnosed taken together, and neither the article text nor the supplementary material states why the remainder was dropped, the supplementary tables having been opened; the loss was larger in the later cohort (15.2% against 10.6%). One number is internally inconsistent: the Results paragraph attaches the denominators to the wrong cohorts, giving 18.8% as 27,665/148,771 and 19.6% as 29,223/146,904, which do not reconcile. The percentages reconcile with the Table 1 denominators, and I followed Table 1.
In January 2023 the United States removed the administrative barrier that American addiction medicine had spent two decades naming as the reason its medicine did not reach people. From 2000 to the end of 2022 a clinician had to complete special training and obtain a federal waiver, the X-waiver, before prescribing buprenorphine for opioid use disorder outside a licensed opioid treatment program. The Consolidated Appropriations Act of 2023 abolished it. A team at Penn State has now asked the arithmetic question that the advocacy did not: how many more people got the medicine.
What the same statute also did
The waiver was repealed by section 1262 of that act. Section 1263, headed in the statute itself as a training requirement for practitioners who prescribe controlled substances, took effect on 27 June 2023 and requires every practitioner applying for or renewing a Drug Enforcement Administration registration, veterinarians aside, to attest to at least eight hours of training on substance use disorders; the short name MATE Act belongs to a separate bill and not to this section. The old eight hours applied to physicians, while nurse practitioners and physician assistants had to complete 24, and that physician-facing requirement for those who wanted to prescribe buprenorphine was replaced by an eight-hour requirement aimed at everyone who prescribes controlled substances, landing halfway through the study's later period. The paper does not discuss this, and the design cannot separate it from anything else that moved between 2022 and 2023: telehealth flexibilities, the return of routine clinical operations, the unwinding of pandemic Medicaid coverage.
The headline number, read at both scales
The headline is real and should not be waved away. More people with a fresh diagnosis were started on buprenorphine, they were started sooner, and the authors describe the change as beneficial. They also, to their credit, call the increase modest and list the barriers they think remain: clinician discomfort, thin training, insurance limits, transport. Now the second scale. Nineteen point six per cent against eighteen point eight per cent is eight tenths of a percentage point. Four newly diagnosed adults in five received no buprenorphine before the repeal, and four in five received none after it. The adjusted odds ratio of 1.14 carries roughly two and a half times the excess odds above unity that the crude ratio of 1.0536 carries, the crude ratio being what the two raw proportions give on their own. The two estimates are not computed on the same people: the regression dropped 12.9% of the eligible sample, and it dropped more of the later cohort than of the earlier one, so the gap between 1.0536 and 1.14 reflects both the adjustment for baseline characteristics and the narrowing of the sample, and cannot be charged to the adjustment alone. The authors make a neighbouring point themselves in the discussion: several of the risk factors they identified have odds ratios close to 1, and they write that such associations, while statistically significant, may have limited clinical relevance and likely reflect small effect sizes with modest practical impact.
The other half of the authors' own headline is the 2% decline in new diagnoses, and they read it as exhaustion rather than as a downturn: most of the previously undiagnosed, on their account, were identified during and shortly after the pandemic, when screening effort was at its peak, while the later period caught clinics returning to routine operations with screening dialled back. The larger total number of adults carrying a recorded diagnosis after the repeal is consistent with that reading.
What a prescription record is not
The denominator here is people already inside a health system, with at least two outpatient visits and a coded diagnosis. Everyone who never reached care is invisible by construction. The numerator is two prescription orders. It is not a pharmacy fill, not a swallowed dose, and above all not a person still in treatment at ninety or a hundred and eighty days. The authors also flag that the F11.2 code covers opioid physical dependence, which is expected in long-term opioid therapy for pain without opioid use disorder, and that buprenorphine is increasingly prescribed for pain; both push in the direction of counting the wrong people. And the record is not evenly distributed: on the authors' own adjusted estimates the odds of buprenorphine initiation against White patients were 0.68 for Black patients, 0.76 for the Other category and 0.92 for Hispanic patients, while their odds of receiving the diagnosis in the first place were higher, which the authors read as persistent systemic and socioeconomic barriers rather than as clinical need.
Funding, interests and registration
Funding and competing interests are declared separately and both are worth reading. The work was funded by the Penn State Inter-Campus Health and Medicine Research Program, with access to TriNetX supported by a Clinical and Translational Science Award from the National Center for Advancing Translational Sciences (UL1 TR002014). Under competing interests the senior author declares having been a member of the Board of Directors and Vice President of the American Society of Addiction Medicine, the professional body that campaigned for the repeal this paper evaluates. The declaration is there and the reader can weigh it. The study was determined exempt from institutional review board oversight; no protocol registration is named, and the published article carries no author contributions section, so the division of labour is not stated on the pages available.
What changes in the room
The practical reading is short. A federal waiver is no longer required for a colleague with a Schedule III registration to prescribe buprenorphine, though the section 1263 attestation and state-level requirements remain, and the paper shows that when the paperwork went, prescribing moved a little and started a little sooner. It also shows how little of the problem the paperwork was. If four in five people who receive the diagnosis in an American clinic still leave without buprenorphine, the remaining obstacle is inside the consultation and inside the service: who feels competent to start it, who will follow the patient, who pays. That is the part no statute repeals, and it is the part a clinician can act on this week.
The waiver disappeared and the share of newly diagnosed adults given buprenorphine moved from 18.8% to 19.6%, while retention, overdose and death were not measured at all.
This is a before-and-after comparison of two calendar periods with no control group, so anything else that changed between 2022 and 2023 is inside the estimate, including the universal eight-hour training requirement that took effect in June 2023. The outcome is a prescription order in the TriNetX network, not a dispensed or taken medication, and prescriptions written outside participating health systems are not captured; "initiation" required only two prescriptions, and retention was never an outcome. The F11.2 code may capture opioid physical dependence in patients treated for chronic pain who do not have opioid use disorder. The adjusted odds ratio of 1.14 carries roughly two and a half times the excess odds above unity that the crude ratio of 1.0536 carries, and the two are not computed on the same people: the regression used 87.1% of the eligible sample, dropping more of the later cohort than of the earlier one, with no stated reason for the exclusions, so the gap reflects both the adjustment and the narrowing of the sample. Socioeconomic variables were unavailable in the database. The article reports that people from racial minority groups were more likely to receive the diagnosis and less likely to receive buprenorphine, and supplementary Table S2 carries the underlying estimates: against White patients the adjusted odds of buprenorphine initiation were 0.68 (0.66 to 0.70) for Black patients, 0.76 (0.74 to 0.78) for the Other category and 0.92 (0.89 to 0.95) for Hispanic patients, while the adjusted odds of receiving the diagnosis were 1.02 (1.01 to 1.03), 1.33 (1.31 to 1.35) and 1.20 (1.18 to 1.22) respectively.