Twelve-step programs kept more people sober in alcohol trials, and staying, sponsors, faith and helping others went with the gains

What the trials found

Every study in this article is about alcohol. We found no controlled trial of a twelve-step program for pornography. We think the alcohol work is still worth reading closely, because it is the best evidence there is on how a twelve-step program works over years, and P-Minus runs one.

In 2020 a Cochrane review by Kelly, Humphreys and Ferri gathered 27 studies and 10,565 adults with alcohol use disorder¹. Two of those studies were randomised trials that set a twelve-step program, run from a manual, against another established treatment such as CBT for a full year. Together they had 1,936 adults¹.

In Cochrane's estimate, 41.8% of people in the twelve-step program stayed sober for the whole year, against 34.5% on the other treatment¹. The 95% interval for the twelve-step group ran from 35.6% to 49%¹. The reviewers rated the certainty of this result high¹.

The detail we find most telling is when the gap appeared. At the end of treatment, the one trial that measured it, with 1,726 adults, found no evidence of a difference between the groups¹. The gap opened over the following year¹. That suggests something people kept doing after treatment ended made the difference. Two trials in the review found the twelve-step treatment raised AA participation, which then went with better drinking outcomes¹. Studies that followed people for years point at four things. Staying in, keeping a sponsor, growing in faith and helping someone else.

Staying in

Moos and Moos recruited 628 adults with an alcohol use disorder who had never been treated, through referral centres and detox programmes, and followed them for 16 years². 507 were alive at the end, and the analysis covers the 461 of them who completed at least two follow-ups or the 16-year one². Abstinent meant no alcohol in each of the past 6 months².

The more weeks of AA people went to in their first year, the more of them were abstinent 16 years later². Of the 192 who did not go in the first year, 33.9% were². Of the 66 who went for 1 to 8 weeks, 37.9% were, not a significant difference². The line climbed after that, to 53.4% for 9 to 26 weeks and 67% for 27 weeks or more². Only 9 weeks or more differed significantly from none². Half a year or more went with about twice the rate of those who did not go².

Staying past the first year went with more abstinence too, on top of the first year². Of those who went for 27 weeks or more in years 2 and 3, 72% were abstinent at 16 years, against 38% of those who did not go². The study counted attendance and nothing else, so it cannot say what people did at the meetings, whether they worked the steps or had a sponsor². People also chose whether to go and for how long, and those who stayed may have been more motivated from the start².

A sponsor, and keeping one

The next study picked up one of the things attendance leaves out. Witbrodt, Kaskutas, Bond and Delucchi followed 495 adults with alcohol dependence in Northern California for 7 years, recruited from treatment programs and the general population³. At years 1, 3, 5 and 7 they asked whether people had a sponsor and how many days they had drunk in the past 30³. They then grouped people by the pattern of their answers³.

Three sponsor patterns came out. About 75 people kept a sponsor through year 7, and 75% of them were in the high abstinence group³. About 70 had a sponsor early and then dropped it³. 56% of them were in that group³. About 350 rarely had a sponsor at all, and 30% of them were³. In the high abstinence group, the chance of no drinking in the past 30 days rose from 75% at year 1 to nearly 100% at year 7³.

The authors held meeting attendance and other factors equal, which the 16-year study could not do³. Those who kept a sponsor still had 7 times the odds of being in the high rather than the low abstinence group, compared with those who rarely had one³. Those who dropped theirs had 2 times the odds in the paper's table, though its text says 3.3³. A sponsor added something on top of showing up.

A second study by Tonigan and Rice followed 182 adults with no more than 16 weeks of past AA⁴. Having a sponsor at 3 months went with 3.67 times the odds of no drinking in months 4 to 6, after accounting for meetings, step work and motivation⁴. Having one at 9 months did not predict drinking at 12 months⁴. The sponsor seemed to count most early on. Both studies asked a yes or no question and never measured how often people talked to their sponsor or how close they were³⁴.

Faith that grows

Kelly, Stout, Magill, Tonigan and Pagano tested what they call AA's principal theory of how it changes behaviour, spiritual growth⁵. They used data on 1,726 adults from Project MATCH, a large US treatment trial⁵. Spiritual practice was measured with six questions on how often people thought about God, prayed, meditated, attended worship, read scripture and had direct experiences of God, plus one on belief in God⁵.

More AA in the first 3 months went with more spiritual practice at 9 months, most of all for people who started with little⁵. More practice at 9 months went with less drinking at 15 months⁵. Among 774 people in aftercare, that rise in practice explained 28% of AA's link with more sober days and 24% of its link with fewer drinks a day⁵. Among 952 outpatients the shares were 14% and 13%⁵.

That is a real share of the benefit, and it leaves the rest to routes the study did not measure⁵. The authors add that their measure may partly stand in for intensive AA involvement⁵. A smaller study points the same way. Robinson and colleagues followed 364 adults with alcohol dependence in the US Midwest, and 8 of 12 measures of spirituality and religiousness changed over 6 months, by amounts the authors call small⁶. Each point gained in forgiving oneself went with 14% lower odds of any heavy drinking at 9 months, after accounting for drinking and AA involvement at the start⁶. The authors add that less drinking may also raise spirituality, so the link may run both ways⁶. For us that result sits close to the centre of recovery. A person who can turn back to God and accept forgiveness has somewhere to go after a fall.

Helping someone else

Pagano, Friend, Tonigan and Stout looked at service to others in the same Project MATCH trial, among 1,501 people⁷. Only 8%, 120 people, were helping other alcoholics when treatment ended⁷. 99 of them said they had completed the twelfth step and 27 were sponsors⁷.

In the 12 months after treatment, 40% of the helpers did not take a drink, against 22% of those who were not helping⁷. About 75% of the whole sample drank again within the year, so staying dry was the exception in both groups⁷. Allowing for meetings attended and study arm, outpatient or aftercare, helpers had a 0.73 risk ratio of relapse against non-helpers⁷. In the outpatient arm alone the result was not significant⁷.

The authors say a finer measure would probably have found more people helping⁷. They also note that the twelfth step covers a spiritual awakening and practising the principles, so the effects of those are in the result too⁷. A separate study by Zemore, Kaskutas and Ammon followed 279 adults through four day-treatment programs in the San Francisco Bay Area, and asked how much time each had spent helping others in recovery the day before⁸. Those who helped more were more involved in twelve-step groups 6 months later, and that involvement went strongly with staying sober⁸. Helping did not predict abstinence directly⁸. Among the 110 still drinking at 6 months, helping went with less binge drinking but not with more involvement⁸. Helping may work by keeping people involved, which loops back to staying in.

What this does and does not tell us

The four findings fit each other. People who stayed in longer, kept a sponsor, grew in spiritual practice and helped others did better, and the two studies that held attendance equal still found an extra benefit³⁷. But in all four, people chose for themselves. Nobody was assigned a sponsor or told to help others. The 16-year study says part of its result reflects who chose to go and how motivated they were², and the sponsor study calls its findings more correlational than causal³. Only the trials in the Cochrane review were randomised, and they compared whole programs, not the parts¹.

The samples were narrow too. Project MATCH was mostly white men, 76% male and 83% non-Hispanic white⁷. Drinking was self-reported throughout²³⁵. And we found no trial of an Islamic version of the twelve steps. P-Minus is built on that version, with faith in one God at its foundation. The alcohol studies give us reason to take each part seriously. They do not prove it works for porn, and we do not claim they do.

How they found it

• The Cochrane review covered 27 studies and 10,565 adults with alcohol use disorder. The headline result pools the two randomised trials that set a manual-based twelve-step program against another established treatment for a full year. The reviewers rated its certainty high.
• 628 adults with an alcohol use disorder who had never had treatment were followed for 16 years. The attendance chart counts 461 of the 507 still alive at the end, those who completed at least two follow-ups or the 16-year one. For 56 of them the 16-year outcome was imputed. Abstinent meant no alcohol in each of the past 6 months. Group sizes come from the paper's Table 3. Its Table 1 prints two of them as 60 and 188, which do not add up to 461.
• Interviews at 1, 3, 5 and 7 years asked whether people had a sponsor now and how many days they had drunk in the past 30. The researchers grouped people by the pattern of their answers and held attendance and other factors equal. The paper's text and its Table 1 give slightly different sponsor group sizes, 77, 72 and 352 against 75, 71 and 349, so we give them as about 75, 70 and 350.
• Spiritual practice was measured with six questions on how often people thought about God, prayed, meditated, attended worship, read scripture and had direct experiences of God, plus one on belief in God, at the start and at 9 months, with drinking at 15 months.
• Helping was read from a questionnaire on AA involvement, 3 months in, when treatment ended. Of the 120 helpers, 99 said they had completed the twelfth step and 27 were sponsors.
• The sponsor, spirituality and helping findings each have a second, smaller study behind them, of 182, 364 and 279 adults. The study of 279 measured helping from a single day of recall, and people stayed in treatment a median of 13 days.

What it cannot tell you

• Alcohol only. We found no controlled trial of a twelve-step program for pornography, and no trial of an Islamic version of the twelve steps.
• At the end of treatment, the one trial that measured it found no evidence of a difference between the groups. The gap opened over the following year.
• Attendance, sponsorship, spiritual practice and helping were not assigned at random. People who did more of them may have been more motivated or committed from the start.
• The sponsor studies asked a yes or no question and did not measure how often people talked to their sponsor or how close they were. The attendance study measured attendance only.
• Spiritual practice carried part of AA's benefit. The rest went through routes the study did not measure.
• Drinking and practice were self-reported. Project MATCH was mostly white men, 76% male and 83% non-Hispanic white.

In P-Minus

P-Minus runs its own twelve steps, rewritten around faith in one God. Nobody has run a trial of that version yet, so we do not call it proven. Worship sits on the same daily list as the habit work, and asking God's forgiveness and counted praise before sleep are both daily tasks. The help button shows one person you chose and the words you wrote to say to them.

Sources

1. Kelly JF, Humphreys K, Ferri M. Alcoholics Anonymous and other 12-step programs for alcohol use disorder. Cochrane Database Syst Rev. 2020;3:CD012880.
2. Moos RH, Moos BS. Participation in treatment and Alcoholics Anonymous: a 16-year follow-up of initially untreated individuals. J Clin Psychol. 2006;62(6):735-50.
3. Witbrodt J, Kaskutas L, Bond J, Delucchi K. Does sponsorship improve outcomes above Alcoholics Anonymous attendance? A latent class growth curve analysis. Addiction. 2012;107(2):301-11.
4. Tonigan JS, Rice SL. Is it beneficial to have an Alcoholics Anonymous sponsor? Psychol Addict Behav. 2010;24(3):397-403.
5. Kelly JF, Stout RL, Magill M, Tonigan JS, Pagano ME. Spirituality in recovery: a lagged mediational analysis of Alcoholics Anonymous' principal theoretical mechanism of behavior change. Alcohol Clin Exp Res. 2011;35(3):454-63.
6. Robinson EAR, Krentzman AR, Webb JR, Brower KJ. Six-month changes in spirituality and religiousness in alcoholics predict drinking outcomes at nine months. J Stud Alcohol Drugs. 2011;72(4):660-8.
7. Pagano ME, Friend KB, Tonigan JS, Stout RL. Helping other alcoholics in Alcoholics Anonymous and drinking outcomes: findings from Project MATCH. J Stud Alcohol. 2004;65(6):766-73.
8. Zemore SE, Kaskutas LA, Ammon LN. In 12-step groups, helping helps the helper. Addiction. 2004;99(8):1015-23.

Sources checked October 2026.