Systematic review of untreated remission from alcohol problems: estimation lies in the eye of the beholder

Background: Treatment planners require a detailed understanding of the potential population in need of alcohol treatment. One common and data-friendly approach is to estimate treatment need from the number of people who meet diagnostic criteria. However, many people meeting diagnostic criteria will not require treatment because they will remit without treatment. This is where estimates of untreated remission become fundamental for accurate treatment planning. Remission from alcohol problems in the absence of treatment is common, yet there are wide ranging estimates of the extent to which this occurs, depending on method. A systematic review of the literature on untreated remission from alcohol problems was conducted to analyse the ways different definitions and study designs impact on estimates of untreated remission from alcohol problems, and to explore the implications for treatment planning.

Methods: 2,103 texts were identified through systematic searches of databases (Medline, PsycINFO, EMBASE) and other searches. Peer-reviewed journal articles published since 1975 which provided numeric estimates of untreated remission from alcohol problems were included (n=27). A narrative synthesis was undertaken.

Results: 124 estimates of untreated remission from alcohol problems were extracted from 27 studies. Three different methods were identified which differed based on the starting sample: method 1 (‘alcohol problems sample’) takes a sample of people with alcohol problems and estimates what proportion remit without treatment; method 2 (‘untreated sample’) takes a sample of untreated people with alcohol problems and estimates what proportion remit; and method 3 (‘remitted sample’) takes a sample of people already in remission from alcohol problems and estimates what proportion have not received treatment. In addition to this sampling diversity, the definitions of an alcohol problem, definitions of remission, and definitions of treatment varied between studies. The combination of the methods and definitional issues impacted on estimates of untreated remission from alcohol problems, which can range between a low of 2.7%; and a high of 98.3% depending on method and definition.

Conclusion: As treatment planners are interested in the number of people in the population who are in need of alcohol treatment, estimates derived from ‘remitted samples’ (method 3) are not useful for treatment planners. Estimates derived from an ‘alcohol problems sample’ (method 1) are best incorporated into treatment planning which aims to estimate the total number of people who will need treatment in a given year. Estimates derived from an ‘untreated sample’ (method 2) start with a sample of untreated people with alcohol problems, and as such are best incorporated into treatment planning which aims to estimate how many additional treatment places are required for people not receiving treatment. Over and above these sampling considerations, how a treatment planner defines what treatment is, what remission is, and how an alcohol problem is defined all matter for estimates of untreated remission from alcohol problems.