
Weekly AA Attendance in Inpatient Alcohol Treatment Fell by Half. The Gaps by Race Are Wider Than the Decline.
In 2015, 53.9 percent of people discharged from inpatient alcohol treatment in the United States had attended four or more mutual-help meetings in the previous 30 days. By 2022, that figure was 39.2 percent. Outpatient treatment moved too, from 19.7 percent to 16.0 percent, and it stopped moving in 2021.
Those are the raw shares from a study published this week in the Journal of Studies on Alcohol and Drugs, led by Sarah Zemore of the Alcohol Research Group, a program of the Public Health Institute. The analysis covers 2.8 million discharges from publicly funded alcohol use disorder treatment in every state except Oregon, from 2015 through 2022, and includes adults treated for alcohol alone or alongside other drugs.
The decline is the headline. The distribution is the story.
What counts as attendance, and what the data can see
The measure is a threshold, not a gradient. A patient counts as a weekly attendee if they report four or more Alcoholics Anonymous or similar meetings in the 30 days before discharge. Someone who went to three meetings in a month is recorded the same as someone who went to none. That matters when a study is used to argue that programs have stopped encouraging meetings: the data captures a fairly committed level of participation, not casual exposure.
The source is the federal Treatment Episode Data Set, which covers publicly funded facilities — roughly half of all U.S. substance use treatment providers — and excludes private programs and most federal ones. It counts treatment episodes rather than individual people, though the authors report that the main findings held when they limited the analysis to first-time patients. Oregon does not appear in the release's description of the sample, and no explanation for its absence is given.
Taken together, these limits describe a particular system: the publicly funded one, serving people with the least ability to buy an alternative. Whatever the data shows about mutual-help attendance is therefore a statement about the safety-net treatment system, not about American drinking.
The drop came in one year and then stayed
The steepest single-year change was 2020. Inpatient attendance went from 49.8 percent in 2019 to 41.7 percent in 2020, and then it did not recover through 2022, even as in-person life resumed.
That detail undercuts the most obvious explanation. If the decline were a supply problem — meetings unavailable, rooms closed, schedules disrupted — 2021 and 2022 should have shown a rebound, because online meeting directories expanded enormously during the pandemic and many groups kept hybrid schedules afterward. A patient in 2022 could attend a meeting from a phone in a treatment facility. They largely did not.
The authors point to two other forces, and both are cultural rather than logistical. Treatment has shifted away from abstinence-only goals as the field's organizing principle, which weakens the traditional handoff from discharge to a 12-step room. And medications for opioid and alcohol use disorder have become far more common, while some 12-step culture still treats medication as a substitute for surrender.
Katherine Karriker-Jaffe, a co-author and senior director at RTI International, put the two in the same sentence: "Medications and mutual-help groups aren't competing options. Many people do best with both, so programs should encourage support group participation whether or not a person is taking medication."
The adjusted numbers are worse than the raw ones
Raw percentages can mislead when the population changes. If inpatient treatment in 2022 served younger, more employed, more medically complicated patients than it did in 2015, attendance could fall without anyone's referral behavior changing at all.
The authors adjusted for exactly that, and the result is a larger decline than the unadjusted shares suggest. Compared with 2015, the odds of weekly attendance in 2022 were about one-third lower for outpatients and roughly one-half lower for inpatients, after accounting for other factors.
A steeper adjusted estimate implies the patients themselves shifted in ways that should have pushed attendance up — so the drop is not explained by a different case mix. The referral stopped happening.
Four gaps, and one group going the other way
Compared with White patients, and after adjustment, the odds of weekly attendance were 36 to 39 percent lower for Latinx and Hispanic patients, 28 percent lower for Black patients, 22 to 28 percent lower for American Indian and Alaska Native patients, and 10 to 11 percent lower for patients of other races and ethnicities.
Then the direction reverses. Asian and Pacific Islander patients had 60 to 66 percent higher odds of weekly attendance than White patients. That is the largest effect in the study, in the opposite direction from the rest, and it appeared in almost none of the coverage.
Why one group bucks the trend
The study cannot answer this, and the authors say so. But the hypothesis they raise fits the pattern better than a story about group cultures. Earlier research suggests some patients avoid meetings they perceive as built for White men, or where they have met mistrust. Yet other studies have found higher participation among Black clients in some urban areas — which points away from the groups themselves and toward a simpler mechanism: what matters is whether you walk into a room where you are the only person who looks like you.
That reading explains the reversal without requiring assumptions about any group's attitude toward AA. Where a recovery community of a given background is dense enough to support its own meetings, attendance can run above the White baseline. Where it isn't, the patient is a demographic minority at every meeting available, in a room where anonymity is supposed to make that irrelevant and often does not.
The medication problem underneath the meeting problem
The study's own framing ties the decline to the spread of medication treatment, and this is where the numbers get uncomfortable for the field.
Medication for alcohol use disorder — naltrexone, acamprosate — and for opioid use disorder carries a stigma in some 12-step rooms that dates to an era when taking a drug meant you were not really sober. That stigma has softened in many places and has not disappeared. Meanwhile the evidence base for combining medication with psychosocial support has strengthened, and the blog has covered that literature repeatedly, including the finding that medication and meetings are not in competition.
The practical result is a referral gap with two directions. Programs that embraced medication sometimes stopped pressing patients toward meetings, on the reasonable view that a patient stabilized on medication does not need a room full of strangers. Programs that kept the 12-step handoff sometimes sent medicated patients into rooms where they were advised to stop. Neither is a strategy. The study's data cannot separate these two paths, but the drop began in the same years medication became routine, and the decline concentrated in inpatient care, where the discharge referral is the entire mechanism.
The correlates that should stop you scoring programs on this
Buried under the racial findings is a set of associations that reads backwards from intuition.
Women had slightly higher odds of weekly attendance than men. Compared with adults 45 to 54, young adults 18 to 24 and adults 65 and older had lower odds. So did married patients, employed patients, and patients without a college degree.
The employment and marriage findings are the ones worth sitting with. In most health research, employment and marriage are advantages. Here they predict less attendance, which is unsurprising once you think about a nightly meeting in a church basement: the people with jobs and families have less unstructured time and more alternative sources of support, including a spouse who notices when they are struggling.
Which means attendance is a measure of what someone did with their evenings, not of how well their treatment worked. Any payer or accreditation body that adopts weekly mutual-help attendance as a quality metric would be, on this data, rewarding programs that discharge patients with the most free time. That is a real risk as state Medicaid programs look for measurable recovery outcomes.
What the researchers say would actually close the gap
Zemore's conclusion is a warning against the obvious fix. Expanding standard programs that push 12-step participation may not narrow these gaps, the authors write, because those programs have not proven especially effective for racial and ethnic minority participants.
What they call for instead is specific: more Spanish-language meetings, referrals to culturally adapted groups, and connections to peer support specialists and recovery community centers rather than only to meeting lists. The distinction matters. A referral to 12-step programs in general is an address. A referral to a meeting where a patient's language is spoken and their neighbors are present is a plan.
There is an uncomfortable implication for the field in the shape of this decline. Mutual-help groups are free, available at night, and one of the few recovery resources that scale without a workforce. For seven years, the public treatment system has been quietly using them less, while the patients with the fewest alternatives have been using them least of all. The finding is not that AA stopped working. It is that the handoff from treatment to a room full of people got weaker at exactly the moment treatment got more clinical — and nobody built a replacement.
Sources
- Zemore SE, et al. Race/ethnicity and Other Sociodemographic Correlates of Weekly Mutual-Help Group Attendance During Alcohol Treatment in 2015-2022 National Data. Journal of Studies on Alcohol and Drugs (2026). DOI 10.15288/jsad.26-00104
- Public Health Institute / Alcohol Research Group, via News-Medical, "Support group attendance drops among minority alcohol treatment patients" (Nov. 16, 2026)
- Medical Xpress, "Mutual-help group attendance in alcohol treatment drops sharply, with wide racial and ethnic gaps" (Nov. 16, 2026)
- Substance Abuse and Mental Health Services Administration, Treatment Episode Data Set - Discharges (TEDS-D)
Editorial Board
LADC, LCPC, CASAC
The Rainier Rehab editorial team consists of licensed addiction counselors, healthcare journalists, and recovery advocates dedicated to providing accurate, evidence-based information about substance abuse treatment and rehabilitation.
Related Articles

Seventy Percent of Patients Get Asked About Drinking. Four Percent of Heavy Drinkers Get Told to Cut Back.
A national NSDUH analysis finds 69% of US adults are screened for alcohol use, but only 4% of heavy drinkers get a brief intervention, and women fare worse.

The States Where the Most Women Drink Are the Ones With the Least Treatment Built for Them
A Boston University study finds women's alcohol use disorder prevalence varies from 6.2 to 12.7 percent by state, while only 58.5 percent of treatment facilities offer women-specific care.

Cocaine Now Appears in 61% of Cuyahoga County Overdose Deaths. Fentanyl Fell Too Fast for Anyone to Notice the Swap.
County medical examiner records show cocaine in 221 overdose deaths in 2025 and 124 more through Oct. 19, 2026 — and in most of them, no fentanyl at all. The stimulant side of the epidemic is now the leading one.