Date: 2026-08-03. Tests seed hypotheses H4.1–H4.3.
Hypothesis verdicts
H4.1 (methadone's OTP monopoly is a binding legal constraint without strong medical justification) — Supported, with a caveat. Methadone for OUD may only be dispensed through ~2,000–2,100 SAMHSA-certified Opioid Treatment Programs nationally, vs. >60,000 community pharmacies (90% of the population within 5 miles of one). The 2024 42 CFR Part 8 reform (individualized take-home criteria, raised initial dose 30mg→50mg) explicitly left the OTP-only dispensing monopoly untouched — confirmed by both SAMHSA's own materials and Pew Charitable Trusts. Peer countries (Australia since 1985, Canada, UK) dispense methadone at pharmacy scale; Australia's COVID-era expansion of unsupervised dosing showed no mortality or diversion increase. The caveat: comparator-country retention evidence is genuinely mixed — one Australian study found pharmacy dosing improved retention, a Canadian study found the opposite. State the finding as "the safety case for restriction is weak/mixed," not "pharmacy dispensing is unambiguously superior."
H4.2 (X-waiver repeal moved prescribing only modestly; provider willingness/payment, not law, now binds) — Supported. Two independent, methodologically distinct studies converge:
- Chua et al. (NEJM 2024, national IQVIA data, interrupted time series): the Jan. 2023 repeal produced an immediate jump of ~1,938 new prescribers and ~5,245 new patients, but no significant change in the slope of subsequent patient growth — the number of patients receiving buprenorphine in any given month stayed flat at ~810,000–830,000 both before and after.
- A RAND-affiliated Pennsylvania study (difference-in-differences on PDMP data, J. Addict. Med. 2025): found no significant differential effect in counties that had fewer prescribers or higher overdose rates beforehand — the places that most needed new capacity got no extra boost.
Both research teams explicitly attribute the shortfall to non-legal barriers: clinician training/comfort, stigma, and reimbursement friction. This is the cleanest natural-experiment confirmation in the whole filing so far that removing a legal barrier does not by itself move an outcome when the binding constraint lies elsewhere — directly analogous to GBMT-1's workforce-not-money finding and GBMT-2's legal-capacity finding.
H4.3 (strongest-evidence modalities are most legally constrained; weakest-evidence modality is freely supplied) — Supported on the dispensing/payment-ceiling comparison; indeterminate on the causal "because unregulated, therefore freely supplied" claim. See below.
Methadone's regulatory cage (detail)
~2,000–2,100 OTPs nationally is confirmed by two independent sources (SAMHSA administrative statement; Health Affairs 2025 key-informant study). Fewer than 100 OTP "medication units" exist nationally, and "few if any" are located in pharmacies (Pew, Sept. 2025) — the technical legal pathway for pharmacy partnership exists but is essentially unused. Historical UK/Scottish data show supervised dosing reduced methadone-overdose deaths relative to unsupervised dosing, an argument for retaining supervision protocols specifically, not for OTP exclusivity as currently structured.
Buprenorphine X-waiver repeal (detail)
PRIMARY-VERIFIED AND RE-STATED 2026-08-10 by the Phase 2 steelman pass (steelman-log.md, Steelman A-1). Phase 1 could not read Chua et al. — the NEJM correspondence is paywalled and its ledger entries A33 / C67–C68 stayed UNVERIFIABLE. The NIH author manuscript is open in PMC (PMC11103581) and was read in full this pass, so the figures are now primary-verified. Two are re-stated because the summary above transposed them: the 5,245 was the level increase in monthly treatment initiations, not in patients treated; the patients-treated series showed neither a level nor a slope change (level 2,683, 95% CI −6,751 to 12,117; slope −857/month, 95% CI −2,943 to 1,229); and the prescriber series had a significant slope increase (595/month, 95% CI 393 to 786) as well as the 1,938 level increase. Prescribers ran 38,684 → 42,158 across 2022 and reached 53,635 by December 2023. The authors' own verdict, verbatim: the policy "may have reduced barriers to prescribing but was insufficient to meaningfully increase buprenorphine use through the end of 2023." Their stated limitations stand with it: no indication data, and effects may take longer than a year. Cached at method/sources/legal-barrier-removal-natural-experiments.md.
By December 2023, >53,600 clinicians had prescribed buprenorphine, up ~11,500 from a year earlier — real prescriber growth. But total monthly patient volume didn't move. One secondary source (Medscape, citing CMS data) reported buprenorphine prescriptions actually declined post-repeal even as dosage units and Medicaid payments rose — flagged as needing independent verification before citing, since it could reflect a data-composition artifact rather than a genuine patient-level decline.
The rehab industry
Market-size estimates are commercial and inconsistent (16.8B–42B depending on scope; a separate $143.6B figure blends in general mental health and isn't comparable) — no authoritative administrative figure exists. The "Florida shuffle" patient-brokering pattern is well-documented, not fringe: a 2018 House Energy & Commerce hearing, federal prosecutions recovering >$100M, and one Florida facility owner convicted of billing insurers >$58M for unnecessary/unrendered treatment. Congress's response — the Eliminating Kickbacks in Recovery Act (EKRA, 2018 SUPPORT Act) — is a fraud/billing statute, not a clinical-quality one. Accreditation (CARF, Joint Commission) is voluntary: as of 2020 only ~23–30% of facilities held either (single-sourced figure, flagged for re-verification). No federal requirement mandates outcome reporting from licensed rehab facilities — the regulatory floor targets fraud, not whether treatment works.
Contingency management — the payment ceiling, corrected
CORRECTED 2026-08-10 by the Phase 1 verification pass (verification-log.md, headline finding 1). The paragraph below originally described the $75 figure as a federal anti-kickback cap. Primary text says otherwise, and the correction is left visible rather than edited away.
CM (paying for verified abstinence) has the strongest evidence base for stimulant use disorder specifically, since no medication is FDA-approved and robustly effective there the way methadone/buprenorphine are for opioids — a reading OIG itself shares ("research shows that contingency management interventions are the most effective currently available treatment for stimulant use disorders," 85 FR 77684, 77791).
What the $75 figure actually is. It is OIG's nominal value interpretation under the Beneficiary Inducements CMP — "in-kind items and services with a retail value of no more than $15 per item or $75 in the aggregate per beneficiary on an annual basis" — and OIG states plainly that this guidance "applies only with respect to the Beneficiary Inducements CMP and not to the Federal anti-kickback statute." OIG goes further: "There is no OIG-imposed $75 limitation on contingency management program incentives" (85 FR 77684, 77791–92, Dec. 2, 2020). The operative $75 constraint on CM programs was SAMHSA's own grant condition — "Previous SAMHSA Notice of Funding Opportunities limited per patient, per year CM incentive values to $75" (SAMHSA Advisory PEP24-06-001, Jan. 2025), which describes the OIG-cap version as a belief that OIG has since clarified.
What the actual regulatory ceiling is. The patient-engagement-and-support safe harbor, 42 CFR 1001.952(hh), caps the aggregate annual retail value at a $500 base indexed to CPI-U — $591 (2024), $605 (2025), $623 (2026) — and is available only to a participant in a value-based enterprise meeting nine conditions, in-kind only, no cash or cash equivalents. Falling outside it is not illegality: OIG's preamble says an incentive that "does not satisfy an existing safe harbor or exception … does not mean that such incentive automatically violates the statutes and is illegal."
What did change in January 2025: SAMHSA raised its own grant cap to $750/patient/year (a 10x increase) — for State Opioid Response and Tribal Opioid Response grantees only, not for Medicaid fee-for-service or commercial insurance, in-kind only, and expressly "does not authorize or permit new CM activities." SAMHSA's own footnote concedes $750 "would fall outside the safe harbor specified by the current regulation ($605 per patient, per year for 2025)."
Trial-effective doses of $200–300/month for 3–6 months remain the comparator, and the practical gap is real. But the gap is between trial-effective dosing and what a grant condition plus a conditional safe harbor will shelter — not a single federal cap that one office can lift. See ws14-cm-legal-status.md, which reached the same conclusion on 2026-08-06 and whose correction never made it to the public pages.
Contingency management — how it is actually paid for and delivered
ADDED 2026-08-10 by the Phase 2 steelman pass (steelman-log.md, Steelman A-2 to A-4). This section is the delivery-side counterpart to the legal correction above, and it moves the finding further than the legal correction alone does. Sources cached at method/sources/cm-delivery-pathways-and-state-medicaid.md.
CM is already a covered Medicaid benefit in five states, at CMS-approved incentive levels above the safe-harbor cap. Kaufman et al., Subst Abuse Treat Prev Policy 2025;20:47 (PMC12486863, open access; state data verified with the pilot programme managers) records five approvals — California, Washington, Montana, Hawaii, Delaware — two pending (Michigan, Rhode Island) and one denied (West Virginia), with approved maxima of 596–1,092. Washington's $1,092 is 1.80× the 2025 patient-engagement safe-harbor cap; Delaware's rose from $599 to $750 with CMS approval. The route is §1115 demonstration authority, not an OIG safe harbor. Application to approval ran about a year for California and Washington; California went live nine months after approval. This is the executable federal path to CM, it was open and used before this filing was published, and §6's H6.1 ("every binding constraint examined is federal") needs the corresponding narrowing — the operative venue here is state-by-state.
Where the legal barrier never existed, CM is still barely delivered. The VA has run a national CM programme since 2011. Coughlin et al., Am J Psychiatry 2025;182(11):1016–1023 (PMC12872285) identified 138,280 VHA patients diagnosed with stimulant use disorder between July 2018 and December 2020, of whom 1,698 received CM — 1.2%. The paper's own framing: "CM implementation has been slow, especially in comparison to treatments for opioid use disorder." This is the cleanest available test of whether fraud-and-abuse law is CM's binding constraint, and it says no.
The clinical case got stronger, not weaker. The same VHA cohort found CM associated with 41% lower one-year all-cause mortality (aHR 0.59, 95% CI 0.36–0.95; 1,481 recipients vs 1,481 matched controls) — the first real-world mortality evidence for CM, and favourable. It partly answers ws13's caution that the trial base is pre-fentanyl-era and shows no retention effect.
"Trial-effective," in dollars, for the first time. Rash et al., JAMA Psychiatry 2025;82(9):940–945 reviewed 112 published CM protocols and computed inflation-adjusted magnitudes from those with medium-to-large effects: $128/week for voucher protocols and $55/week for prize protocols — $1,536 and $660 over the standard 12 weeks. The $200–300/month comparator above sits between them. Against the voucher benchmark the "held below trial-effective level" framing holds and the gap is large; against the prize benchmark — the design California and Washington actually run — the 2026 safe harbor ($623) is 5.6% short, SAMHSA's $750 clears it, and Washington's $1,092 exceeds it by 65%. The surviving claim is narrower than the filing's: voucher-magnitude CM is not shelterable under current authority.
What is delivered is a staffed service, not a permission. California's protocol (Freese et al., Prev Med 2023;176:107703): "a maximum of $599 over a six-month period, contingent upon 36 stimulant-negative urine test results," CLIA-waived point-of-care tests, a custom incentive- accounting platform, e-gift-card distribution, and — "a significant innovation of the project" — a per-site CM Coordinator, "a designated and highly trained and supervised individual responsible for all aspects of CM operation." Add UCLA-run training, a pre-launch readiness review and ongoing fidelity monitoring (Freese et al., JSAT 2024;167:209513). §10 scores this architecture 3 on fiscal cost — "near-zero net new cost." That score prices the rule change, not the service.
Workforce — the weakest-evidenced section
Median wage for the BLS category bundling addiction and mental-health counselors: $59,190 (2024), corroborated by a secondary $54,650 figure. Beyond pay, the load-bearing numbers (a ~114,000-counselor 2037 shortfall vs. an inconsistent ~77,050 figure from the same search; a ~50% burnout rate) could not be traced to a primary source and are internally inconsistent. Directionally plausible as a GBMT-1-style binding labor constraint, but flagged for a follow-up pass against a primary HRSA or BLS source before citing specific numbers in the whitepaper.
Implications for §10/§11
- The X-waiver natural experiment is the strongest single piece of evidence in the filing so far for the protocol's overarching thesis (H10.1): legal deregulation without addressing the real binding constraint (provider willingness, payment, training) does not move outcomes. Methadone deregulation (§10's seed architecture) should be scored with this caution built in, not assumed to work simply because France's 1995 experience (however contested per Phase 0) suggests deregulation can work under different conditions.
Contingency management's partial 2025 fix is a live case study for §11 sequencing: a cheap, rulemaking-executable fix (no statute required) that materially closed an evidence-practice gap for a subset of programs within about a year — the kind of "rulemaking-executable first" item §11 should prioritize and cite as a working example, not just a hypothetical.Withdrawn 2026-08-10 (Phase 2, steelman-log.md). SAMHSA's January 2025 move was a grant condition, and its own advisory says it "does not authorize or permit new CM activities"; it closed no evidence-practice gap for anyone outside SOR/TOR grantees. The live case study §11 should have cited is the state §1115 route — five CMS approvals, incentive maxima to $1,092, California delivering since March 2023.- The X-waiver lesson generalises further than this file first let it. §4 drew the local conclusion — that the barrier wasn't binding for buprenorphine — and left the filing's thesis ("the binding constraint is law") intact. On the Phase 2 record it is the thesis that needs narrowing: mobile methadone units (2021 rule) hit community resistance, DEA guidance inconsistency and workforce shortages rather than scale; VA contingency management reaches 1.2% of eligible patients with no legal barrier at all; and the one barrier removal that did move access — OTC naloxone — was attached to a product pharmacies already stocked and sold, and still leaves a $56 out-of-pocket price and low stocking. Legal executability (D2) measures what an agency may do. Nothing on the §10 board measures what a clinic can staff, and that is where these cases fail.
Confidence: Moderate
Strong on the two natural-experiment findings (H4.1, H4.2 — cross-corroborated primary/peer-reviewed sourcing) and the CM regulatory-cap history (SAMHSA's own advisory documents). Weaker on rehab-industry market size and workforce shortage/burnout figures, both flagged for follow-up verification before use as citable numbers.