CDC National Center for Health Statistics, provisional overdose mortality data: est. 69,973 U.S. drug overdose deaths in 2025 (−14% from 81,313 in 2024); opioid-involved deaths est. 44,564, down from 55,296. Third consecutive annual decline.
cdc.gov/nchs — provisional data release, May 2026A standing report on treatment continuity in opioid use disorder
Treatment doesn't fail.
Continuity does.
Buprenorphine works. Yet most patients who start it are off it within six months — usually not because the medication stopped working, but because the system connecting patient, physician, and pharmacy broke. BRIDGE makes continuity a shared, auditable obligation of all three parties. This site is the evidentiary record behind that design.
PLATFORM STATUS: PRE-PILOT · BUILD DOCUMENTED · NO OUTCOME CLAIMS MADEI · The Landscape, 2026
The epidemic is bending. The retention problem is not.
Overdose deaths have declined for three consecutive years — a genuine public-health achievement attributed in significant part to naloxone distribution and expanded access to medications for opioid use disorder. But access to a first prescription was never the whole problem. Staying on treatment is.
~69,973
Estimated U.S. drug overdose deaths in 2025 — down almost 14% from 81,313 in 2024, the third consecutive annual decline.
CDC NCHS, provisional, May 2026 · E1
~44,564
Of those deaths involved opioids — still the primary driver of the crisis, down from 55,296 the prior year.
CDC NCHS, provisional, May 2026 · E1
34%
Fewer deaths within one year of diagnosis among OUD patients treated with buprenorphine-naloxone versus untreated, in a propensity-matched cohort of 220,000+.
Cureus / PMC, 2024 · E2
The medication is not the weak link. Buprenorphine reduces mortality, relieves withdrawal, and roughly doubles remission rates relative to no medication. The weak link is what happens after the first fill — and the data on that is unambiguous.
II · The Discontinuity Problem
Discontinuation is not the exception. It is the modal outcome.
Across commercial claims, Medicaid claims, VA data, and health-system cohorts, the same shape appears: most patients who initiate buprenorphine are no longer on it at six months, and the attrition begins in the first month.
51%
Of 58,000+ commercially insured and Medicaid enrollees had discontinued buprenorphine by 180 days; 73% by 365 days. Rates were nearly identical across treatment settings.
Addiction Sci Clin Pract, 2024 · E3
28.4%
Of 17,329 Medicaid enrollees discontinued within the first month; 64.6% before 180 days.
Multi-state Medicaid claims · E3
18–45%
Range of 180-day retention across 12,073 care episodes depending on how "discontinuation" is even defined — the field cannot yet agree on how to measure its own central failure.
Am J Psychiatry, quality-measurement analysis · E3
National quality bodies treat six months as the minimum effective duration. By that standard, the system fails the majority of patients it successfully starts. And the measurement chaos in the third figure is itself part of the indictment: an outcome this consequential is tracked with definitions so inconsistent that reported retention can more than double depending on the permitted gap length. A continuity problem that cannot be measured consistently cannot be managed.
III · Why the Gap Matters
Discontinuation is a medical event with a mortality signature.
Leaving buprenorphine treatment is not a neutral administrative fact. It opens a window of sharply elevated overdose risk — highest in the first weeks after the last dose, when tolerance has fallen and the illicit supply has not gotten safer.
×13.4
Weighted rate ratio for non-fatal overdose requiring hospitalisation during the four weeks following opioid-agonist treatment cessation, versus time in treatment, in a linked primary-care cohort.
Lancet-family cohort, UK · E4
~5%
Of patients retained 6–18 months who then discontinued needed medical treatment for an opioid overdose within just six months of stopping — a rate the study authors called evidence that discontinuation is "a life-threatening event."
Williams et al., Am J Psychiatry / Columbia · E5
>14 days
Gaps in buprenorphine treatment beyond two weeks are associated with increased overdose risk — the threshold BRIDGE's event taxonomy is calibrated around.
VA national cohort literature · E6
A multi-site cohort of 6,550 patients who discontinued found no treatment duration after which stopping becomes safe — post-treatment mortality ran 1.82 per 100 person-years regardless, and patients who discontinued at 91–180 days carried nearly triple the opioid-overdose hazard of those retained past a year. The clinical inference is stark: for many patients the safest available strategy is not to plan an exit, but to prevent an unplanned one.
This reframes the design question. If the gap is the lesion, then the intervention target is not "the patient's motivation." It is every mechanism by which a gap can open.
IV · Where Continuity Breaks
Gaps open at the seams between three parties — not inside any one of them.
The retention literature reflexively locates failure in the patient. The access literature says otherwise: a substantial fraction of interruptions originate at the pharmacy counter or in the physician's workflow, where no relapse has occurred at all.
Patient-side breaks
Life events, not treatment failure
- Lost or stolen medication with no rapid replacement pathway
- Missed appointments during destabilizing events — housing, custody, work
- First-month fragility: over a quarter of Medicaid patients gone before day 30
- Cost exposure and coverage churn interrupting fills mid-episode
Physician-side breaks
Authorization latency
- Refill authorization waiting on an appointment the patient couldn't reach
- No structured channel for the patient to report a discontinuity event in time to act
- Coverage handoffs and prior-authorization cycles with no owner of the interim days
- Risk visible only at the next visit — after the gap has already opened
Pharmacy-side breaks
The counter as a checkpoint
- Roughly a third of telehealth-treated patients missed doses in a year due to a pharmacy-level barrier; a quarter had trouble filling at all
- Studies find 40–50% of U.S. pharmacies without buprenorphine available to dispense
- About half of community pharmacies don't stock it; roughly one in five refuse to order it, citing wholesaler flags and enforcement fear
- "Red tape" verification rituals that convert a valid prescription into days of delay
Sources: Evidence Ledger E7–E9. None of these pharmacy-side failures involve patient relapse, yet each produces the identical clinical object: a gap. A continuity system that monitors only the patient is monitoring one-third of the failure surface.
V · The BRIDGE Answer
Continuity as a shared, auditable contract.
BRIDGE binds the three parties whose seams produce the gaps into one explicit continuity agreement, recorded on an append-only Continuity Ledger. Each party carries obligations. Every continuity-relevant event — a fill, a check-in, a refusal, a lost prescription, a restored pickup — is typed, time-stamped, and visible to the parties who must act on it.
Patient
Commits to the clinical conditions set by their physician
- Structured check-ins and wellness monitoring on the eight-dimension wellness wheel
- Timely reporting of discontinuity events through a typed taxonomy — honesty is rewarded, not punished
- Identity attestation, so the continuity record is theirs alone
Physician
Commits to timely action when the patient's conditions are met
- Refill authorization when obligations are verifiably satisfied
- A closed bridge-script loop: discontinuity event → bridge order → patient acknowledgment → confirmed pickup — with silence escalating back to the physician, never dropped
- A continuity score and drift velocity that surface risk between visits, not after them
Pharmacy
Commits to dispensing when both upstream obligations are met
- A verifiable record that the prescription in hand sits inside an active, physician-supervised continuity contract
- Fill successes and fill failures both enter the ledger — the counter becomes observable
- Documentation that answers the legitimacy question pharmacists are currently forced to improvise
The instrument layer follows from the evidence, not the other way around. The two-week gap threshold in the literature calibrates the event taxonomy. The first-month attrition cliff shapes the check-in cadence. The post-cessation risk spike is why an unplanned interruption triggers an active closed loop rather than a note in a chart. And contingency-management incentives — among the best-evidenced behavioral tools in addiction medicine — reward the behavior that matters here: returning. BRIDGE gamifies return, not perfection.
VI · Design Invariants
What the system is built never to do.
A platform touching controlled-substance treatment must be legible to regulators before it is attractive to users. These invariants are not policies layered on top of the software; they are encoded as machine-checked constraints that gate every release. A build in which any invariant is violated cannot ship.
INV-01
The code never prescribes and never denies.
No pathway in BRIDGE issues, withholds, or terminates medication. Prescribing authority is human, licensed, and singular. There is no automated denial state anywhere in the system.
INV-02
Non-satisfaction routes to human review — always.
When a party's obligations go unmet, the outcome is escalation to the physician's review queue, never an automatic consequence. A lost prescription is a clinical event to be managed, not a violation to be punished.
INV-03
No protected health information in the contract state.
The continuity state machine carries no identifiers and no clinical detail. Verification and clinical content live in their proper, regulated homes; the ledger records typed events, not charts.
INV-04
Safety signals bypass all automation.
Any crisis-shaped signal freezes contract mechanics instantly and routes to a human. No incentive logic, scoring, or contract state touches a safety event.
INV-05
Incentives are firewalled from medication logic.
Contingency-management rewards are capped, monotonic, and structurally incapable of influencing whether medication is authorized or dispensed. Money and medicine do not share a wire.
VII · Alignment
What continuity is worth, in the currencies decision-makers are measured in.
Different institutions fund, regulate, and reimburse this problem in different units — abatement compliance, quality scores, per-member cost, real-world persistence. Retention converts into every one of them.
Settlement Fund Administrators
Retention is a named abatement lever — not an adjacent one
- Exhibit E of the national settlement agreements requires at least 85% of funds go to opioid abatement, with expanded MOUD access listed among Schedule A Core Strategies — the priority tier
- RAND's settlement-allocation modeling found no single intervention cuts overdoses 40%; the combination required is MOUD initiation, increased 6-month retention, and naloxone distribution. Two of those three levers are well funded. Retention is the neglected one
- Exhibit E separately approves evidence-based data collection and research analyzing abatement effectiveness — the Continuity Ledger produces exactly that audit trail, per dollar spent
Medicaid & Managed Care Executives
BRIDGE moves a measure you are already scored on
- The HEDIS Pharmacotherapy for Opioid Use Disorder (POD) measure and MIPS #468 (NQF 3175) both define success as ≥180 days of continuous pharmacotherapy with no gap over 7–8 days — the same gap events, at the same thresholds, the Continuity Ledger records and closes in real time
- Discontinuity is a cost event: 42–50% of patients were seen in the ED within six months of discontinuation, and claims analyses find total health expenditures highest among those with the poorest buprenorphine retention
- Contingency management is now CMS-sanctioned: seven-plus states hold approved or pending §1115 waivers, incentives are excluded from MAGI eligibility determinations, and SAMHSA's 2025 advisory permits up to $750 per client per year. BRIDGE's incentive cap is designed to sit beneath that ceiling, firewalled from medication logic by INV-05
Manufacturers
Persistence is the number your product is judged by
- Real-world persistence — not trial efficacy — is what payers, PBMs, and health systems increasingly contract on. A product discontinued by half its patients at 180 days underperforms its own pharmacology
- Most discontinuation is not clinical failure: a meaningful share originates at the pharmacy counter and in authorization latency (Section IV). A continuity platform addresses the attrition your patient-support programs cannot see
- The ledger produces de-identified, typed interruption data — where gaps open, why, and what closed them — the evidence base for formulary position, REMS-adjacent commitments, and access advocacy
Regulators and licensing boards are addressed throughout — Sections V and VI exist for them. The invariants are not marketing to that audience; they are the compliance architecture: no automated prescribing or denial, mandatory human review, no PHI in contract state, safety bypass, and incentives structurally separated from dispensing.
VIII · Evidence Ledger
The sources behind every number on this page.
In the spirit of the platform it documents, this site keeps its own append-only ledger. Tier A: federal surveillance data and peer-reviewed multi-site cohorts. Tier B: peer-reviewed single-domain studies and claims analyses. Tier C: institutional and expert commentary. Nothing on this page rests on a source below Tier C.
EntrySource
E1 · A
E2 · B
Propensity-matched cohort, 220,000+ patients: buprenorphine-naloxone associated with 34% fewer deaths within one year of OUD diagnosis (2.6% vs 4.0%) and ~1.9× remission rate versus no buprenorphine.
PMC11610725 — Buprenorphine-Naloxone for OUD: Reduction in Mortality and Increased Remission, 2024E3 · A/B
Retention: 51% discontinued by 180 days and 73% by 365 days across 58,000+ multi-state commercial and Medicaid enrollees (Addiction Sci Clin Pract, 2024). Medicaid claims: 28.4% discontinued in month one, 64.6% before 180 days (PMC6354252). Definitional analysis of 12,073 episodes: 180-day retention ranged 18–45% by discontinuation definition (Am J Psychiatry).
ascpjournal.biomedcentral.com 13722-024-00450-0 · PMC6354252 · psychiatryonline.org appi.ajp.20230808E4 · B
Linked primary-care cohort with hospitalisation and mortality records: non-fatal overdose requiring hospitalisation 13.4× more likely during the four weeks after opioid-agonist treatment cessation versus time in treatment.
PMC9399254 — Non-fatal overdose risk during and after opioid agonist treatment, 2022E5 · B
MarketScan Medicaid cohort: among patients retained 6–18 months, ~5% required medical treatment for opioid overdose within six months of discontinuation; ED use, hospitalization, and opioid fills elevated across all discontinuation cohorts; outcomes improved with retention beyond 15 months. Study authors characterized discontinuation as a life-threatening event.
Williams et al., Am J Psychiatry 2020 (PubMed 31786933); Columbia Psychiatry commentaryE6 · B
Multi-site cohort of 6,550 discontinuers across eight U.S. health systems: post-treatment mortality 1.82 per 100 person-years; no duration after which discontinuation was associated with reduced all-cause mortality; discontinuation at 91–180 days carried HR 2.94 for opioid overdose versus >365-day retention. VA-linked literature: treatment gaps >2 weeks associated with increased overdose risk.
Glanz, Campbell et al., Addiction 2023;118(1):97–107 (PMC9722535) · PMC12004863E7 · B
Cross-sectional study of 601 adults in telemedicine OUD treatment: approximately one-third missed buprenorphine doses in the prior 12 months due to a pharmacy-related barrier; one-quarter experienced difficulty filling; the most common barrier was the pharmacy lacking stock. Companion telephone studies found 40–50% of U.S. pharmacies without buprenorphine available.
JAMA-family cross-sectional study, PMC12362223E8 · B
PhARM-OUD expert consensus and systematic reviews: around half of community pharmacies do not stock buprenorphine and roughly one in five refuse to order it; drivers include perceived DEA order caps, wholesaler flags, diversion concern, and stigma. Six-state Medicaid analysis: nearly 1 in 5 pharmacies dispensing other opioids did not dispense buprenorphine.
PMC12301831 · PMC11102015 · J Am Pharm Assoc systematic review, 2025E9 · B/C
Qualitative "red flags and red tape" literature: pharmacist verification rituals — diagnostic-code demands, prescriber phone confirmation, refusal of early refills, informal geographic radius rules with no regulatory basis — documented as bureaucratic barriers converting valid prescriptions into treatment delay.
PMC10962060 · NEJM commentary, Univ. of Kentucky / Emory, 2020E10 · A
National opioid settlement agreements: at least 85% of funds must be used for opioid abatement; Exhibit E Schedule A Core Strategies prioritize expanded MOUD access, with Schedule B approving evidence-based data collection and research analyzing abatement effectiveness. RAND allocation modeling: reducing overdose by 40% in simulated communities required simultaneously scaling MOUD initiation, 6-month treatment retention, and naloxone distribution — no single intervention sufficed.
Distributor Settlement Agreement, Exhibit E (attorneygeneral.gov) · nationalopioidsettlement.com · RAND OPTIC fund-allocation analysisE11 · A
Quality-measure specifications: HEDIS Pharmacotherapy for Opioid Use Disorder (POD) — percentage of OUD pharmacotherapy events lasting ≥180 days with no gap of 8+ consecutive days (NCQA). CMS MIPS Quality Measure #468 / NQF 3175 — ≥180 days of continuous pharmacotherapy with no gap of more than 7 days; telehealth encounters allowable.
NCQA HEDIS POD specification · CMS QPP Measure #468 specification, qpp.cms.govE12 · A/B
Contingency management legitimacy: CMS-approved §1115 demonstration waivers implementing CM for substance use disorder in California, Washington, Montana, Delaware, Hawaii and additional states (seven-plus approved or pending as of 2025); incentives excluded from MAGI-based Medicaid eligibility determinations. SAMHSA January 2025 advisory permits State and Tribal Opioid Response grantees to spend up to $750 per client per grant year on CM incentives, replacing the prior $75 limit. ASAM recognizes CM as standard of care for stimulant use disorder; evidence also supports CM for MOUD adherence. Cost context: annual health expenditures highest among patients with the poorest buprenorphine retention (commercial claims, N=6,444); 42–50% of discontinuing patients seen in the ED within six months (see E5).
KFF §1115 Waiver Watch, updated 2026 · medicaid.gov DE-DSHP CM protocol approval, 2025 · SAMHSA advisory via cminfo.org · Psychiatric Services, appi.ps.201900309Methodological note: retention and discontinuation figures vary with the claims database, population, and gap definition used; ranges are reported where the literature reports ranges. Provisional CDC mortality counts are adjusted with each release. This page states no efficacy or outcome claims for the BRIDGE platform itself, which has not yet completed a clinical pilot. When pilot data exist, they will be appended to this ledger under their own tier — including if they are unfavorable.