At the Median Opioid Treatment Program, One in Twenty New Patients Is Back in Acute Care Within 90 Days

Opioid treatment programs have been measured on whether patients start medication and whether they stay on it. They have not been measured on what happens to patients afterward in any standard way — a gap that has left roughly 1,900 federally regulated clinics with no common yardstick for the outcome families actually care about.

A team at RTI International has now built one, and published it in the journal Addiction on Oct. 25. Using Medicaid claims from 1,226 opioid treatment programs during the 2024 measurement year, the researchers calculated the share of each program's newly initiated patients who had a substance use disorder-related emergency department visit or hospitalization within 90 days. The dataset covered 412,411 new treatment episodes among 257,986 patients.

At the median program, that share was 5.0 percent. The middle half of programs ran from 2.8 percent to 8.0 percent.

Across more than 103,300 qualifying acute care events in the measurement year, 80.2 percent were emergency department visits and 19.8 percent were inpatient stays. Opioid-related disorders were the most frequent principal diagnosis, at 47.4 percent of events.

"Quality improvement starts with measurement," said Tami L. Mark, a distinguished fellow in behavioral health financing at RTI. "Opioid treatment programs often do not know when their patients are hospitalized or visit an emergency department after beginning care." Co-author Barrett Montgomery, a drug use epidemiologist, framed the measure as a complement to existing metrics rather than a replacement.

Why Claims Data Changes the Politics of the Measure

The design choice is the interesting part. Because the measure is built entirely from claims, it requires no new reporting from clinics — a state Medicaid agency or a health plan can compute it from data it already holds, and a program cannot decline to participate.

That reduces one objection and creates another. Case-mix adjustment reshuffled the rankings more than a casual reader might expect: 46 of the 1,226 programs moved from the 25th percentile to the 50th once patient characteristics were accounted for, and 20 moved from the 50th to the 75th. Raw rankings, in other words, are unstable for a meaningful minority of clinics, and any state that publishes a league table without adjustment will mislabel them.

The Blind Spot Is the Worst Outcome

Two limits deserve to travel with the number. The cohort is Medicaid enrollees only, excluding people dually eligible for Medicare and anyone uninsured or paying privately — which strips out a substantial share of older patients with opioid use disorder.

The second is harder. An acute care event has to generate a claim to be counted, which means the measure captures a patient who relapsed and was treated, and misses a patient who relapsed and died. A program whose staff know their patients and push them toward emergency care when things go wrong may look worse than a program that loses contact. High rates can mean better access, not worse care — which is why the authors present the measure as a starting point for inquiry rather than a verdict.

For anyone weighing medication-assisted treatment options, that distinction is the useful one. A clinic's 90-day number is a question to ask about, not an answer.