Does drinking spread through social networks, and do treatments that change a person's network help?

Does drinking cluster in social networks?

The Framingham Heart Study followed 12,067 people in the United States from 1971 to 2003. At every point in time, drinkers and non-drinkers formed clusters, reaching up to three degrees of separation (Ann Intern Med 2010).

In the same group of people, a change in drinking among a person's network was significantly linked to that person's later drinking. This held for relatives and friends, but not for neighbours or co-workers. The authors say the pattern seems to reflect interpersonal influence.

This is an observational study. It shows clustering, not cause. Drinking was not measured with a clinical assessment, and not all of each person's ties were observed. Shared surroundings, rather than influence, may explain part of the pattern.

Is that proof that drinking spreads from person to person?

No. A 2008 study in the BMJ used the same kind of method on 4,300–5,400 US teenagers. It found that acne, headaches and height also "spread" through friendships. After adjusting for shared surroundings, these effects became uniformly smaller and not significant.

This means the method can give a false signal. It does not prove that drinking does not pass between people, and that study did not look at alcohol. What the network studies show is that drinkers cluster together; why they cluster is not settled.

Do treatments that change a person's network help?

In a US randomized trial, 210 people were assigned to one of two network-support treatments (one adding rewards for not drinking, called contingency management) or to case management; 186 were followed up at 15 months. At 15 months, drinking outcomes were better in both network-support groups than with case management. At the end of treatment, support for not drinking and AA involvement had risen, while support for drinking had not fallen (J Consult Clin Psychol 2007). The network change that was measured was a rise in support for not drinking; support for drinking did not fall. Whether this change caused the better outcomes is not shown — the links are correlations. This is one site, the comparison was case management, and the links between network change and drinking are correlations. We read only the abstract.

At two years, the network-support group without contingency management had up to 20% more days without drinking than both other groups (Litt et al. 2009). 172 of the 210 people were interviewed at the last follow-up, 27 months after intake. This is the same study, not an independent repeat.

In the UK Alcohol Treatment Trial, 742 clients received either social behaviour and network therapy or motivational enhancement therapy; 617 (83.2%) were followed up at 12 months. The two did not differ significantly. In both groups, drinking, dependence and alcohol problems fell over 12 months (BMJ 2005). There was no untreated group, and "no difference" does not mean the two are equal. A correction to the paper was published later (BMJ 2019;367:l6608).

In the same trial, among 608 clients with full cost data, spending on health, social care and criminal justice fell by about five times the cost of therapy; with no untreated group, the fall cannot be credited to the therapy itself. The network therapy averaged £221 and the motivational therapy £129. The network therapy cost significantly more, but neither net savings nor cost-effectiveness differed significantly. The figures cover 12 months.

What about mutual-help groups like AA?

A 2020 Cochrane review (27 studies, 10,565 participants) compared Alcoholics Anonymous and twelve-step facilitation with other treatments. Manualized AA and twelve-step facilitation led to more continuous abstinence at 12 months than other treatments such as cognitive behavioural therapy (CBT) (relative risk, RR — how many times more likely: 1.21, 95% CI 1.03–1.42; 2 studies, 1,936 participants; high certainty).

The high certainty applies only to continuous abstinence. For other outcomes, such as how much people drank, the review says AA may perform as well as other treatments. The review covers adults who were not forced into treatment. Two of the authors have received US government (NIH and Veterans Health Administration) grants for research that includes AA.

Risk of bias from how people were selected was high in 11 of the 27 studies, and risk of bias from people dropping out was high in 9. This describes the quality of the evidence, not the size of the effect.

What this evidence does not show

These studies do not show that drinking passes from person to person: the network data are observational, shared surroundings are not ruled out, and the same kind of method finds that height "spreads".

They do not show that network treatment beats other good treatment. The US trial is one site compared with case management, and its two-year result is the same study. The UK trial found no significant difference from motivational therapy, had no untreated group, and its cost data cover only 12 months.

The AA evidence is high certainty only for continuous abstinence at 12 months, covers adults who were not forced into treatment, and many studies had a high risk of bias.

This page does not review medication for alcohol problems.

How many people with alcohol use disorder get help at all?

That is a separate question, covered in our briefing "Alcohol use disorder: in Australia, an estimated 1 in 4 will ever seek help, the fewest of any disorder in the survey" (in the sources below). In Australia's 2020–22 national survey, an estimated 25% of people with alcohol use disorder will ever make contact with help for it, against 94% with major depressive disorder (Birrell 2025). This is a projection from people's recall, and "help" means any contact, not adequate care.