Effectiveness of brief alcohol interventions in primary care populations

Eileen FS Kaner; Fiona R Beyer; Colin Muirhead; Fiona Campbell; Elizabeth D Pienaar; Nicolas Bertholet; Jean B Daeppen; John B Saunders; Bernard Burnand

doi:10.1002/14651858.CD004148.pub4

. 2018 Feb 24;2018(2):CD004148. doi: 10.1002/14651858.CD004148.pub4

Effectiveness of brief alcohol interventions in primary care populations

Eileen FS Kaner ^1,^✉, Fiona R Beyer ¹, Colin Muirhead ¹, Fiona Campbell ², Elizabeth D Pienaar ³, Nicolas Bertholet ⁴, Jean B Daeppen ⁴, John B Saunders ⁵, Bernard Burnand ⁶

Editor: Cochrane Drugs and Alcohol Group

PMCID: PMC6491186 PMID: 29476653

Abstract

Background

Excessive drinking is a significant cause of mortality, morbidity and social problems in many countries. Brief interventions aim to reduce alcohol consumption and related harm in hazardous and harmful drinkers who are not actively seeking help for alcohol problems. Interventions usually take the form of a conversation with a primary care provider and may include feedback on the person's alcohol use, information about potential harms and benefits of reducing intake, and advice on how to reduce consumption. Discussion informs the development of a personal plan to help reduce consumption. Brief interventions can also include behaviour change or motivationally‐focused counselling.

This is an update of a Cochrane Review published in 2007.

Objectives

To assess the effectiveness of screening and brief alcohol intervention to reduce excessive alcohol consumption in hazardous or harmful drinkers in general practice or emergency care settings.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, and 12 other bibliographic databases to September 2017. We searched Alcohol and Alcohol Problems Science Database (to December 2003, after which the database was discontinued), trials registries, and websites. We carried out handsearching and checked reference lists of included studies and relevant reviews.

Selection criteria

We included randomised controlled trials (RCTs) of brief interventions to reduce hazardous or harmful alcohol consumption in people attending general practice, emergency care or other primary care settings for reasons other than alcohol treatment. The comparison group was no or minimal intervention, where a measure of alcohol consumption was reported. 'Brief intervention' was defined as a conversation comprising five or fewer sessions of brief advice or brief lifestyle counselling and a total duration of less than 60 minutes. Any more was considered an extended intervention. Digital interventions were not included in this review.

Data collection and analysis

We used standard methodological procedures expected by Cochrane. We carried out subgroup analyses where possible to investigate the impact of factors such as gender, age, setting (general practice versus emergency care), treatment exposure and baseline consumption.

Main results

We included 69 studies that randomised a total of 33,642 participants. Of these, 42 studies were added for this update (24,057 participants). Most interventions were delivered in general practice (38 studies, 55%) or emergency care (27 studies, 39%) settings. Most studies (61 studies, 88%) compared brief intervention to minimal or no intervention. Extended interventions were compared with brief (4 studies, 6%), minimal or no intervention (7 studies, 10%). Few studies targeted particular age groups: adolescents or young adults (6 studies, 9%) and older adults (4 studies, 6%). Mean baseline alcohol consumption was 244 g/week (30.5 standard UK units) among the studies that reported these data. Main sources of bias were attrition and lack of provider or participant blinding. The primary meta‐analysis included 34 studies (15,197 participants) and provided moderate‐quality evidence that participants who received brief intervention consumed less alcohol than minimal or no intervention participants after one year (mean difference (MD) ‐20 g/week, 95% confidence interval (CI) ‐28 to ‐12). There was substantial heterogeneity among studies (I² = 73%). A subgroup analysis by gender demonstrated that both men and women reduced alcohol consumption after receiving a brief intervention.

We found moderate‐quality evidence that brief alcohol interventions have little impact on frequency of binges per week (MD ‐0.08, 95% CI ‐0.14 to ‐0.02; 15 studies, 6946 participants); drinking days per week (MD ‐0.13, 95% CI ‐0.23 to ‐0.04; 11 studies, 5469 participants); or drinking intensity (‐0.2 g/drinking day, 95% CI ‐3.1 to 2.7; 10 studies, 3128 participants).

We found moderate‐quality evidence of little difference in quantity of alcohol consumed when extended and no or minimal interventions were compared (‐20 g/week, 95% CI ‐40 to 1; 6 studies, 1296 participants). There was little difference in binges per week (‐0.08, 95% CI ‐0.28 to 0.12; 2 studies, 456 participants; moderate‐quality evidence) or difference in days drinking per week (‐0.45, 95% CI ‐0.81 to ‐0.09; 2 studies, 319 participants; moderate‐quality evidence). Extended versus no or minimal intervention provided little impact on drinking intensity (9 g/drinking day, 95% CI ‐26 to 9; 1 study, 158 participants; low‐quality evidence).

Extended intervention had no greater impact than brief intervention on alcohol consumption, although findings were imprecise (MD 2 g/week, 95% CI ‐42 to 45; 3 studies, 552 participants; low‐quality evidence). Numbers of binges were not reported for this comparison, but one trial suggested a possible drop in days drinking per week (‐0.5, 95% CI ‐1.2 to 0.2; 147 participants; low‐quality evidence). Results from this trial also suggested very little impact on drinking intensity (‐1.7 g/drinking day, 95% CI ‐18.9 to 15.5; 147 participants; very low‐quality evidence).

Only five studies reported adverse effects (very low‐quality evidence). No participants experienced any adverse effects in two studies; one study reported that the intervention increased binge drinking for women and two studies reported adverse events related to driving outcomes but concluded they were equivalent in both study arms.

Sources of funding were reported by 67 studies (87%). With two exceptions, studies were funded by government institutes, research bodies or charitable foundations. One study was partly funded by a pharmaceutical company and a brewers association, another by a company developing diagnostic testing equipment.

Authors' conclusions

We found moderate‐quality evidence that brief interventions can reduce alcohol consumption in hazardous and harmful drinkers compared to minimal or no intervention. Longer counselling duration probably has little additional effect. Future studies should focus on identifying the components of interventions which are most closely associated with effectiveness.

Plain language summary

Effectiveness of brief alcohol interventions in primary care populations

What is the aim of this review?

We aimed to find out whether brief interventions with doctors and nurses in general practices or emergency care can reduce heavy drinking. We assessed the findings from 69 trials that involved a total of 33,642 participants; of these 34 studies (15,197 participants) provided data for the main analysis.

Key messages

Brief interventions in primary care settings aim to reduce heavy drinking compared to people who received usual care or brief written information. Longer interventions probably make little or no difference to heavy drinking compared to brief intervention.

What was studied in the review?

One way to reduce heavy drinking may be for doctors and nurses to provide brief advice or brief counselling to targeted people who consult general practitioners or other primary health care providers. People seeking primary healthcare are routinely asked about their drinking behaviour because alcohol use can affect many health conditions.

Brief interventions typically include feedback on alcohol use and health‐related harms, identification of high risk situations for heavy drinking, simple advice about how to cut down drinking, strategies that can increase motivation to change drinking behaviour, and the development of a personal plan to reduce drinking. Brief interventions are designed to be delivered in regular consultations, which are often 5 to 15 minutes with doctors and around 20 to 30 minutes with nurses. Although short in duration, brief interventions can be delivered in one to five sessions. We did not include digital interventions in this review.

Search date

The evidence is current to September 2017.

Study funding

Funding sources were reported by 60 (87%) studies. Of these, 58 studies were funded by government institutes, research bodies or charitable foundations. One study was partly funded by a pharmaceutical company and a brewers association, another by a company developing diagnostic testing equipment. Nine studies did not report study funding sources.

What are the main results of the review?

We included 69 controlled trials conducted in many countries. Most studies were conducted in general practice and emergency care. Study participants received brief intervention or usual care or written information about alcohol (control group).

The amount of alcohol people drank each week was reported by 34 trials (15,197 participants) at one‐year follow‐up and showed that people who received the brief intervention drank less than control group participants (moderate‐quality evidence). The reduction was around a pint of beer (475 mL) or a third of a bottle of wine (250 mL) less each week.

Longer counselling probably provided little additional benefit compared to brief intervention or no intervention.

One trial reported that the intervention adversely affected binge drinking for women, and two reported that no adverse effects resulted from receiving brief interventions. Most studies did not mention adverse effects.

Quality of the evidence

Findings may have been affected because participants and practitioners were often aware that brief interventions focused on alcohol. Furthermore, some participants could not be contacted at one‐year follow‐up to report drinking levels. Overall, evidence was assessed as mostly moderate‐quality. This means the reported effect size and direction is likely to be close to the true effect of these interventions.

Summary of findings

Background

Description of the condition

Excessive drinking is a significant cause of mortality, morbidity and social problems in many countries, with a greater global cost to health than for tobacco (WHO 2014). The true impact of alcohol on the health of individuals and the wider community is difficult to estimate because of the many effects resulting from alcohol use, including increased levels of violence, accidents and suicide (GBD 2017). The heavy burden that alcohol use imposes on health, and its significant economic consequences, has led to national and international programmes and policies that seek to reduce consumption levels and so reduce a primary cause of avoidable ill health (Bailey 2011; UK Government 2012).

The impetus for a preventive approach to alcohol problems has been reinforced by epidemiological research. At a population level, most alcohol‐related harm is not due to drinkers with severe alcohol dependence but attributable to a much larger group of excessive (hazardous and harmful drinkers) whose consumption exceeds recommended drinking levels (Anderson 1991). Therefore, at a population level, the greatest impact on alcohol‐related problems can be made by addressing interventions aimed for excessive rather than dependent drinkers (McGovern 2013).

Description of the intervention

Early identification and secondary prevention of alcohol problems, using screening and brief interventions in primary care, has long been advocated as a strategy to reduce excessive drinking and is the focus of a great deal of research (O'Donnell 2014).

Brief intervention is grounded in social‐cognitive theory and typically incorporates some or all of the following elements:

feedback on the person's alcohol use and any alcohol‐related harm clarification as to what constitutes low risk alcohol consumption;
information on the harms associated with risky alcohol use;
benefits of reducing intake;
advice on how to reduce intake;
motivational enhancement; analysis of high risk situations for drinking and coping strategies; and
the development of a personal plan to reduce consumption.

Brief intervention is typically structured according to the FRAMES approach which involves practitioners: giving Feedback on the person’s intake, impressing the Responsibility for change onto them, offering Advice, listing a Menu of options for behavioural change, having an Empathic approach and building Self‐efficacy in the person receiving the brief intervention (Miller 1994).

Some brief intervention trials have included motivational interviewing (Rollnick 1995) or lifestyle counselling approaches. Although forms of brief intervention vary among studies (Heather 1995), core features in primary care are delivery by generalist healthcare workers, targets excessive (hazardous and harmful) drinkers who tend not to be seeking help for alcohol problems, and aims for reduced consumption and alcohol‐related harms. Brief interventions in primary care have focused less frequently on dependent drinkers because these people often need more intensive treatment than is available routinely and are likely to require a goal of total abstinence.

How the intervention might work

Excessive drinking can be identified routinely in general practice and emergency care. People are often asked about alcohol consumption during new patient registrations, general health checks, specific disease clinics (e.g. hypertension, diabetes) and other health screening procedures. This identification process is often referred to as screening and typically involves asking a relatively small number of standardised questions about alcohol consumption (e.g. quantity, frequency and intensity of use) and any associated effects using a validated questionnaire or screening tool. Screening and brief alcohol intervention in routine primary care typically occurs opportunistically ‐ when the main purpose of the appointment is something other than help with drinking.

The brief intervention must be delivered within the limited time frame of a standard consultation (typically 5 to 15 minutes for general practitioners (GPs), or up to 30 minutes for nurses). It also needs to fit in with routine practice (e.g. initial screening plus either referral to a practice colleague or later return for intervention). However, brief intervention trials have evaluated a wide range of activity. The shortest of these is a single 5 to 10 minute session of structured advice delivered by GPs or nurses. More intense interventions can provide multiple sessions of motivational interviewing or some other form of counselling, accompanied by repeated follow‐up and delivered by primary healthcare workers. Other variations relate to the type of population being treated, the amount of training and support received by therapists, the theoretical basis underlying the intervention, and the use of accompanying written material.

Why it is important to do this review

Although previous reports and reviews have indicated beneficial outcomes of screening and brief intervention for excessive drinkers, crucial questions remain concerning its impact in routine practice and applicability to the broader population (Agosti 1995; Bien 1993; Moyer 2002; NHS CRD 1993; Poikolainen 1999; UK Government 2012; Wilk 1997). Whilst there appears to be little doubt that screening and brief intervention with excessive drinkers can work successfully in research settings where intervention delivery and follow‐up is carefully managed (Flay 1986), there has been uncertainty about extrapolation to real world routine primary care (Heather 2014; Holder 1999; Kaner 2001). However, if health professionals are to be encouraged to adopt and administer brief interventions in routine practice, it is necessary to establish a realistic effect size for brief intervention delivered in clinically‐relevant contexts.

The term 'hazardous and harmful drinkers' contains a number of subgroups (e.g. young people, older people, and ethnic minorities). Little is known about how these subgroups respond to brief intervention in primary care. Differential loss of participants from early brief intervention trials led to a call for caution in generalising results to routine practice (Edwards 1997) but types of participants who were lost remains unclear. Therefore, there is need to characterise the types of drinkers for whom brief interventions have a positive impact and any subgroups who have not been represented in the trials to date.

This is an update of our 2007 review (Kaner 2007). An update was necessary because many trials have been conducted since initial publication and relevant developments have occurred in the wider literature; consequently we have added 42 new trials and conducted new subgroup analyses. Firstly, specific tools to measure efficacy and effectiveness have been published (Gartlehner 2006; Koppenaal 2011), which supercede the scale used in the 2007 review. Secondly, the concepts of screening reactivity and assessment reactivity have been defined (McCambridge 2011). These definitions describe how the very process of screening for alcohol consumption or having the effects of drinking assessed may influence reported drinking behaviour, independent of any further brief intervention input. Furthermore, recent literature contains a bigger subgroup of trials that focus on emergency care rather than general practice‐based primary care, whereas previously the vast majority of the trials took place in primary care. Finally, interventions in more recent trials are often based on counselling techniques such as motivational interviewing rather than simpler advice‐based input.

An important development has been in the digitisation of interventions to reduce hazardous and harmful alcohol consumption, using technologies such as websites and smart phone apps. These interventions (and their comparison with face‐to‐face brief interventions) are the focus of another Cochrane Review (Kaner 2017), and were excluded from this review.

Objectives

Main objective

To assess the effectiveness of screening and brief alcohol intervention to reduce excessive alcohol consumption in hazardous or harmful drinkers in general practice or emergency care settings.

Secondary objectives

Specific questions addressed by this review were:

Are brief interventions superior to minimal or no intervention?
Are extended brief interventions containing more or longer sessions superior to no intervention or to standard brief interventions?

Methods

Criteria for considering studies for this review

Types of studies

Randomised controlled trials and cluster‐randomised controlled trials were eligible for inclusion.

Types of participants

People who routinely presented to general practice, emergency care or other primary care settings for a range of health problems, whose alcohol consumption was identified by a screening tool as being excessive, or who had experienced harm as a result of their drinking behaviour. Studies that recruited participants who were seeking treatment specifically for an alcohol problem or who were mainly dependent on alcohol were excluded.

We defined primary care as all immediately accessible, general healthcare facilities. People needed to be able to access services on demand rather than through a specialist referral, and services needed to cover a broad range of problems. Participants recruited in emergency departments and trauma centres were included if this was the first contact following the emergency event.

Types of interventions

Experimental condition: Brief intervention comprised a single session and up to a maximum of five sessions of verbally‐delivered information, advice or counselling that was designed to achieve a reduction in risky alcohol consumption, alcohol‐related problems, or both (Babor 1994).

Control conditions: screening or assessment only, usual care for the presenting condition or written information such as a health or alcohol education leaflet (described as minimal intervention).

Psychology‐based counselling aimed at reducing alcohol consumption or alcohol‐related problems that was unlikely to occur in routine practice, for reasons of length or intensity, were referred to as extended intervention. We defined extended interventions as those that consisted of more than five sessions or total combined session durations was more than 60 minutes.

Interventions specifically aimed at people who were dependent on alcohol were excluded. Digital interventions (e.g. websites, smart phone apps or computer programmes) were excluded because these were investigated in another Cochrane Review (Kaner 2017).

Types of outcome measures

Primary outcomes

The primary outcome was consumption of alcohol. This was most often reported as:

Self‐reported or other reports of drinking quantity (e.g. drinks per week).
Self‐reported or other reports of binge drinking frequency (e.g. number of binges per week).
Self‐reported or other reports of drinking frequency (e.g. drinking days per week).
Self‐reported or other reports of drinking intensity (e.g. number of drinks per drinking day).
Self‐reported or other reports of drinking within recommended limits (e.g. government recommended limits). Although limits vary among countries, practitioners tend to use the national government recommended limits as a guide.

Although not specified in the protocol, we also noted the following consumption outcomes if these were reported.

Proportion of heavy drinkers (in Kaner 2007 a common example was > 35 units/week, but these definitions have reduced recently).
Proportion of binge drinkers.

Secondary outcomes

Levels of laboratory markers of reduced alcohol consumption (e.g. serum gamma‐glutamyltransferase (GGT), mean corpuscular volume (MCV)).
Alcohol‐related harm to drinkers or others affected (e.g. via questionnaires such as the drinking problems index).
Patient satisfaction measures.
Health‐related quality of life.
Economic measures including use of health services.

Search methods for identification of studies

Electronic searches

Current update searches

We searched the following databases from 2005 to September 2017:

Cochrane Central Register of Controlled Trials (CENTRAL) (Issue 8, 2017; searched 25 September 2017);
Cochrane Database of Systematic Reviews (CDSR) (Issue 9, 2017; searched 25 September 2017);
Database of Abstracts of Reviews of Effects (DARE) (Issue 2, 2015; searched 25 September 2017);
MEDLINE (Ovid, 1966 to September week 2, 2017; searched 21 September 2017);
Embase (Ovid 1980 to 2017 week 38; searched 21 September 2017);
PsycINFO (Ovid 1840 to September week 3 2017; searched 21 September 2017);
CINAHL (EBSCO, 1982 to 25 September 2017);
Science Citation Index Expanded (SCI‐EXPANDED) (Web of Science, 2005 to 25 September 2017);
Conference Proceedings Citation Index ‐ Science (CPCI‐S) (2005 to 25 September 2017);
Social Sciences Citation Index (SSCI) (Web of Science, 2005 to 25 September 2017);
Emerging Sources Citation Index (ESCI) (Web of Science, 2015 to 25 September 2017);
NHS‐EED (Wiley, issue 2 of 4, 2015; searched 25 September 2017).

Search strategies are reported in Appendix 1; Appendix 2; Appendix 3; Appendix 4; Appendix 5; Appendix 6; Appendix 7; Appendix 8; and Appendix 9.

We applied no language or publication restrictions.

Previous searches

We searched the following sources from the earliest available date to 2006:

Cochrane Drug and Alcohol Group specialised register (February 2006);
Cochrane Central Register of Controlled Trials (CENTRAL);
MEDLINE (Ovid, 1966 to 2005);
Embase (Ovid, 1980 to 2005);
PsycINFO (Ovid, 1840 to 2005);
CINAHL (EBSCO, 1982 to 2005);
Social Sciences Citation Index (SSCI) (Web of Knowledge, 1970 to 2005);
Science Citation Index (SCI) (Web of Knowledge, 1970 to 2005);
Cochrane Effective Practice and Organisation of Care Group specialised register (2005);
Alcohol Education and Research Council (AERC) alcohol library, searched 2005;
HEED (searched 3 December 2014 ‐ no access for the September 2017 search); and
Alcohol and Alcohol Problems Science Database, ETOH (1972 to 2003, after which the database was discontinued).

Searching other resources

We handsearched the reference lists of included studies and relevant systematic reviews. We contacted key informants and experts to enquire about unpublished work and ongoing research, particularly through links with the International Network on Brief Interventions for Alcohol and other drugs (INEBRIA).

We also searched clinicaltrials.gov (25 September 2017).

We searched organisational websites for reports of eligible trials on 26 September 2017:

USA Department of Health and Human Services, Substance Abuse and Mental Health Services Administration (SAMHSA), National Registry of Evidence Based Programs and Practices (NREPP);
SAMHSA Screening Brief Intervention and Referral to Treatment (SBIRT);
Information on Drugs and Alcohol (IDA);
Alcohol Concern;
Drug and Alcohol Findings;
International Network on Brief Interventions for Alcohol and Other Drugs (INEBRIA); and
National Treatment Agency for Substance Misuse.

Data collection and analysis

Selection of studies

Two review authors independently assessed titles and abstracts in EndNote (EndNote 2015). If the title, abstract and keywords did not yield enough information to ascertain potential for inclusion then the full paper was retrieved.

We initially piloted the inclusion criteria on six retrieved papers for the original review (Kaner 2007). Two review authors independently assessed the study eligibility for both the original review and this update. Any disagreement was resolved by discussion and consensus, or adjudication by a third review author.

Data extraction and management

Two review authors independently extracted data using a piloted data extraction form. We extracted citation information, participants' characteristics (e.g. age, gender, baseline alcohol consumption), intervention descriptions (e.g. number, content and frequency of brief intervention sessions), setting and outcome data to RevMan 5 (Review Manager 2014). We also extracted methodological information to enable critical appraisal. Where data were missing or unclear we emailed study authors to request clarification or further data.

In the protocol, we specified the use of an intention‐to‐treat analysis as a criterion of quality. However, this was revised in light of current Cochrane Handbook guidance (Higgins 2011b). Intention‐to‐treat analysis is usually understood to mean that participants were analysed in the groups to which they were randomised, regardless of the treatment they actually received. However, it is also sometimes understood to imply that all participants were included regardless of whether their outcomes were actually collected, which requires imputation of missing outcomes. Rather than using intention‐to‐treat analysis as a quality criterion, we attempted to extract data for participants in the groups to which they were randomised, regardless of the treatment they actually received, i.e. corresponding to the more widely agreed definition of intention‐to‐treat analysis.

Outcome data on quantity of alcohol consumed in a specific time period were converted to grams per week for each study. Drinks and units were converted to grams using either a conversion factor reported in the relevant paper or, if none was reported, using the conversion factor appropriate for the country where the study was conducted (Furtwaengler 2013; Gual 1999; Heather 2006; Miller 1991). Months were converted to weeks by multiplying by 52/12. Drinking intensity, drinking days, drinking sessions and occasions were all assumed to be equivalent to drinking days. For laboratory markers of gamma‐glutamyltransferase (GGT) (Israel 1996; Romelsjö 1989; Scott 1990; Wallace 1988), microkatals/litre were converted to international units/litre (IU/L) by multiplying by 60. Where relevant, values from analyses that involved adjustment for missing data (e.g. through the imputation of baseline values for participants lacking follow‐up data) were used in preference to unadjusted values.

Assessment of risk of bias in included studies

Two review authors independently assessed potential risk of bias resulting from the trial design according to Cochrane's 'Risk of bias' tool as described in the Cochrane Handbook (Higgins 2011). This is a two‐part tool, addressing the following domains.

Sequence generation and allocation concealment (selection bias).
Blinding of participants and providers (performance bias).
Blinding of outcome assessor (detection bias).
Incomplete outcome data (attrition bias).
Selective outcome reporting (reporting bias).
Other source of bias.

The first part of the tool involves describing what was reported to have happened in the study. The second part of the tool involves assigning a judgement relating to the risk of bias in terms of low, high or unclear risk. We used the criteria indicated by the Cochrane Handbook adapted to the addiction field to make these judgments (see Appendix 10).

Any discrepancies between review authors were resolved by discussion to achieve consensus.

Measures of treatment effect

We calculated the mean difference (MD) and standard deviation (SD) between the value of the outcome measure at 12 months following the brief intervention and the corresponding value following the control intervention for each continuous outcome. If standard deviations of final values were not available, the change score (i.e. the difference between the final and initial value of the outcome measure) was used if its standard deviation was available. If no standard deviations were available, these trials were omitted from the primary analysis but included in a sensitivity analysis using imputed standard deviations.

The risk difference (RD) with 95% confidence intervals were calculated for dichotomous outcomes because 95% CIs are intuitively clearer and this was consistent with the approach applied previously (Kaner 2007).

Unit of analysis issues

We extracted a direct estimate of the desired treatment effect and its standard error where analyses accounted for clustering in cluster‐randomised trials. We assigned imputed standard deviations to the treatment and control groups, such that the standard error of the treatment effect estimated by the weighted mean difference method in RevMan 5 (Review Manager 2014) was the same as that reported in the analysis accounting for clustering. If the analysis did not account for the cluster design, we extracted the number of clusters randomised to each intervention, the average cluster size in each intervention group, and the outcome data for all participants in each intervention group.

A design effect was estimated using an external estimate of the intra‐cluster coefficient (ICC). In this way we inflated the variance of the effect estimate. In the case of dichotomous outcomes, this involved reducing the total number of participants and the number of participants with events, whilst keeping the proportion with events fixed. It was then possible to enter data to RevMan 5 (Review Manager 2014), and combine cluster‐randomised trials with individually randomised trials in the same meta‐analysis.

Dealing with missing data

We contacted study authors to obtain missing data and seek clarification where appropriate. Studies with missing standard deviations or for which the number of participants in each arm was not reported were excluded from the main analysis for the associated continuous measure. These studies were included in a sensitivity analysis, using imputed values for the standard deviations or the number of participants in each arm.

Assessment of heterogeneity

The magnitude of heterogeneity among trials was assessed using the I² statistic (Higgins 2002; Higgins 2003). The statistical significance of heterogeneity was assessed using P values derived from Chi² tests (Deeks 2001).

Assessment of reporting biases

We assessed whether studies appeared to have incomplete reporting bias by noting in the risk of bias assessments whether the reported outcomes matched methods sections or any published protocols. We made every effort to minimise publication bias by searching a wide range of databases and sources of grey literature and not restricting by language or publication status. We constructed funnel plots (plots of the effect estimate from each study against the effect standard error) to assess potential for bias related to the size of the trials, which could indicate possible publication bias.

Data synthesis

The weighted mean difference method was used to estimate pooled effect sizes and 95% CI, if sufficient studies reporting the outcome were available. Most trials reported weekly or monthly alcohol consumption and few reported drinking frequency or intensity. Hence, the meta‐analysis of alcohol quantity consumed per week provided most information and constituted the primary meta‐analysis. Statistical analyses were performed using RevMan 5 (Review Manager 2014). Because the populations and interventions evaluated by the trials were so heterogeneous, it was deemed more appropriate to use a random‐effects model for all analyses (DerSimonian 1986). Random‐effects meta‐regression modelling was conducted using the metareg command in Stata version 14.1 (Stata 2015). Meta‐regression was used to assess any differences in calculated effect associated with the publication date of studies, baseline consumption of participants, duration of treatment, and efficacy/effectiveness score.

For dichotomous outcomes (participant classified as a heavy or binge drinker), RDs and 95% CIs were calculated and pooled in a random‐effects meta‐analysis using Mantel‐Haenszel test weighting.

We addressed variable risk of bias using sensitivity analysis (see Sensitivity analysis).

If trials had more than one control arm and the various control arms were very similar (e.g. Sommers 2013), the results for these arms were combined by calculating weighted means of continuous outcomes and summing dichotomous outcomes; likewise for very similar treatment arms (D'Onofrio 2012).

'Summary of findings' tables

We used the GRADE approach to assess the quality of the evidence. GRADE takes into account issues related to both internal and external validity, such as directness of results (Atkins 2004; Guyatt 2008; Guyatt 2011). The 'Summary of findings' tables present main review findings in a transparent and simple tabular format. In particular, 'Summary of findings' tables provide key information concerning the quality of evidence, the magnitude of effect of the interventions examined and the sum of available data on the main outcomes.

The GRADE system uses the following criteria for assigning grades of evidence.

High: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

We used GRADEpro GDT 2015 to import data from Review Manager 2014 for the main outcomes of quantity of drinking (g/week), frequency of drinking (days per week and binges per week) and intensity of drinking (drinks/drinking day) for each of the comparisons (brief intervention versus minimal or no intervention, extended intervention versus minimal or no intervention, extended intervention versus brief intervention) (Table 1; Table 2; Table 3).

Summary of findings for the main comparison. Brief intervention compared to no or minimal intervention for people with hazardous or harmful alcohol consumption.

Brief intervention compared to no or minimal intervention for people with hazardous or harmful alcohol consumption
Patient or population: people with hazardous or harmful alcohol consumption as identified by a screening tool  Setting: primary care (directly accessible to participant, no referral required), mostly high income countries  Intervention: brief intervention  Comparison: no or minimal intervention
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Quality of the evidence  (GRADE)	Comments
Outcomes	Risk with control	Risk with Brief intervention	Relative effect  (95% CI)	№ of participants  (studies)	Quality of the evidence  (GRADE)	Comments
Quantity of drinking (g/week) at 12 months	The mean quantity of drinking (g/week) at 12 months was 238 g/week	MD 20.08 g/week lower  (28.36 lower to 11.81 lower)	‐	15197  (34 RCTs)	⊕⊕⊕⊝  MODERATE ¹	Heterogeneity was substantial (73%) but not unexplained; interventions differed in content and delivery. The direction of effect favoured the intervention in 82% of the studies
Frequency of drinking (no. binges/wk) at 12 months	The mean frequency of drinking (no. binges/wk) at 12 months was 0.98 binges/week	MD 0.08 binges/week lower  (0.14 lower to 0.02 lower)	‐	6946  (15 RCTs)	⊕⊕⊕⊝  MODERATE ¹
Frequency of drinking (no. days drinking/wk) at 12 months	The mean frequency of drinking (no. days drinking/wk) at 12 months was 2.73 drinking days/week	MD 0.13 drinking days/week lower  (0.23 lower to 0.04 lower)	‐	5469  (11 RCTs)	⊕⊕⊕⊝  MODERATE ¹
Intensity of drinking (g/drinking day) at 12 months	The mean intensity of drinking (g/drinking day) at 12 months was 55 g/drinking day	MD 0.18 g/drinking day lower  (3.09 lower to 2.73 higher)	‐	3128  (10 RCTs)	⊕⊕⊕⊝  MODERATE ¹
Adverse effects	Only five trials reported adverse effects. No participants experienced any adverse effects in two trials; one trial reported that the intervention increased binge drinking for women; and two trials reported adverse events related to driving outcomes but concluded they were equivalent in both study arms.		‐	(5 RCTs)	⊕⊝⊝⊝ VERY LOW 12
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).    CI: Confidence interval; RR: Risk ratio; OR: Odds ratio;
GRADE Working Group grades of evidence  High quality: We are very confident that the true effect lies close to that of the estimate of the effect  Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different  Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect  Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

Extended intervention compared to no or minimal intervention for people with hazardous or harmful alcohol consumption
Patient or population: people with hazardous or harmful alcohol consumption  Setting: primary care (directly accessible to participant, no referral required)  Intervention: extended intervention  Comparison: no or minimal intervention
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Quality of the evidence  (GRADE)	Comments
Outcomes	Risk with control	Risk with Extended intervention	Relative effect  (95% CI)	№ of participants  (studies)	Quality of the evidence  (GRADE)	Comments
Quantity of drinking (g/week) at 12 months	The mean quantity of drinking (g/week) at 12 months was 236 g/week	MD 19.50 g/week lower  (40.47 lower to 1.48 higher)	‐	1296  (6 RCTs)	⊕⊕⊕⊝  MODERATE ¹
Frequency of drinking (no. binges/wk) at 12 months	The mean frequency of drinking (no. binges/wk) at 12 months was 1.3 binges/week	MD 0.08 binges/week lower  (0.28 lower to 0.12 higher)	‐	456  (2 RCTs)	⊕⊕⊕⊝  MODERATE ¹
Frequency of drinking (no. days drinking/week) at 12 months	The mean frequency of drinking (no. days drinking/week) at 12 months was 2.1 drinking days/week	MD 0.45 drinking days/week lower  (0.81 lower to 0.09 lower)	‐	319  (2 RCTs)	⊕⊕⊕⊝  MODERATE ¹
Intensity of drinking (g/drinking day) at 12 months	The mean intensity of drinking (g/drinking day) at 12 months was 76.6 g/day	MD 8.51 g/day lower  (25.69 lower to 8.67 higher)	‐	158  (1 RCT)	⊕⊕⊝⊝  LOW ^{1 2}
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).    CI: Confidence interval; RR: Risk ratio; OR: Odds ratio;
GRADE Working Group grades of evidence  High quality: We are very confident that the true effect lies close to that of the estimate of the effect  Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different  Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect  Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

Extended compared to brief intervention for people with hazardous or harmful alcohol consumption
Patient or population: people with hazardous or harmful alcohol consumption  Setting: primary care (directly accessible to participant, no referral required)  Intervention: extended intervention  Comparison: brief intervention
Outcomes	*Anticipated absolute effects^ (95% CI)**		Relative effect  (95% CI)	№ of participants  (studies)	Quality of the evidence  (GRADE)	Comments
Outcomes	Risk with brief intervention	Risk with Extended	Relative effect  (95% CI)	№ of participants  (studies)	Quality of the evidence  (GRADE)	Comments
Quantity of drinking (g/week) at 12 months	The mean quantity of drinking (g/week) at 12 months was 251 g/week	MD 1.54 g/week higher  (42.01 lower to 45.10 higher)	‐	552  (3 RCTs)	⊕⊕⊝⊝  LOW ^{1 2}
Frequency of binge drinking (no. binges/wk) at 12 months ‐ not measured	‐	‐	‐	‐	‐
Frequency of drinking (no. days drinking/week) at 12 months	The mean frequency of drinking (no. days drinking/week) at 12 months was 2.82 days drinking/week	MD 0.51 drinking days/week lower  (1.21 lower to 0.19 higher)	‐	147  (1 RCT)	⊕⊕⊝⊝  LOW ^{2 3}
Intensity of drinking (g/drinking day) at 12 months	The mean intensity of drinking (g/drinking day) at 12 months was 70 g/drinking day	MD 1.7 g/drinking day lower  (18.86 lower to 15.46 higher)	‐	147  (1 RCT)	⊕⊕⊝⊝  VERY LOW ^{1 2 3}
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).    CI: Confidence interval; RR: Risk ratio; OR: Odds ratio;
GRADE Working Group grades of evidence  High quality: We are very confident that the true effect lies close to that of the estimate of the effect  Moderate quality: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different  Low quality: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect  Very low quality: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect

Scale item	Score	Meaning
Patients and problems	2	Clinically representative people initially present with a typically wide range of problems via self‐referral or invitation for a health check
	1	Mixed: e.g. routine patients but paid for participation in study, or patients prescreened then invited research representative subjects may be paid patients
	0	Researcher‐solicited volunteers (e.g. via advertisement) or referrals from specialist services
Practice context	2	Clinically representative is a community‐based setting in which a range of clinical services are usually provided to patients
	1	Mixed
	0	Research representative is a setting in which the research function clearly dominates any clinical one (e.g. clinic at a university or hospital)
Practitioners and therapists	2	Clinically representative practitioners are practising doctors, nurses and qualified therapists who earn their main living by providing health services in primary care
	1	Qualified clinician but specifically recruited for the study
	0	Research representative practitioners are non‐clinicians, or clinicians in training, who are contracted to deliver interventions for the purposes of the study
Intervention content	2	Clinically representative intervention fits with current practice in terms of timing, content or style (e.g. for primary care 5 to 15 minutes for a GP; 20 to 30 minutes for a nurse or initial screening accompanied by a return visit for brief intervention; for emergency settings motivational interviewing‐style intervention fits in here, e.g. 45 minutes
Intervention content	0	Research representative treatment would not normally occur in routine practice e.g. unusually long consultations
Therapeutic flexibility	1	Clinically representative: allows professional judgement in how an intervention is delivered e.g. freedom to focus on particular issues according to patient need
	½	Flexible protocol (tailored to participants)
	0	Research representative: strict adherence to a prescribed protocol or script that does not allow for variability in practice
Pre‐therapy training	1	Clinically representative training in intervention procedures occurs according to typical CPD/CME procedures (e.g. outreach visits, seminars, one‐off training days). Full day off‐site (for emergency care staff)
	½	Full day off‐site (for primary care staff)
	0	Research representative training is unusually intensive or requiring of atypical levels of interest or motivation, e.g. prolonged or intensive courses, formal qualification
Intervention support	1	Clinically representative support occurs within standard practice resources (e.g. colleague assistance with screening, IT flagging, provision of general (uncustomised) manual. Note for emergency care settings some procedures are clinically representative (e.g. taking bloods and doing various tests)
Intervention support	0	Research representative support would not typically be available (e.g. researcher help to flag notes, extra staff for period of the trial)
Intervention monitoring	1	Clinically representative monitoring of intervention delivery does not interfere with practitioners' behaviour or their relationships with patients
Intervention monitoring	0	Research representative monitoring would be direct observation of therapist behaviour or ongoing/immediate feedback to practitioners after each session

Trial	Reported units	Conversion factor	Source of conversion	Efficacy score	Treatment exposure¹
Aalto 2000	g/week	1	NA	10.5	45, 105
Altisent 1997	Units/week	8	Altisent 1997	8.5	20
Babor 2006	Drinks/week	14	Babor 2006	11	4
Bazargan‐Hejazi 2005	Alcohol consumption	NR	NA	9	17.5
Beckham 2007	Drinks/day	11.671 x 7	Miller 1991	4.5	52.5
Beich 2007	Drinks/week	12	Beich 2007	11	10
Bernstein 2010	Drinks/week	11.671	Miller 1991	6.5	32.5
Bischof 2008	g/day	7	NA	6.5	60
Blow 2006	Drinks/week	13	Blow 2009*	7.5	7.5
Cherpitel 2009	Drinks/week	11.671	Miller 1991	9	17.5
Córdoba 1998	Units/week	8	Córdoba 1998	11	15
Crawford 2004	Units/week	8	Miller 1991	10.5	30
Crawford 2010	Alcohol consumption	NR	NA	8.5	30
Curry 2003	Drinks/week	11.671	Miller 1991	8.5	22.5
D'Amico 2008	Basic statistics	NR	NA	7.5	25
D'Onofrio 2008	Drinks/week	11.671	Miller 1991	10.5	6.7
D'Onofrio 2012	Drinks/week	11.671	Miller 1991	10.5	7;17
Daeppen 2007	Drinks/week	10	Daeppen 2007	7	15
Dent 2008	Alcohol consumption	NR	NA	11.5	5; 45
Désy 2010	Drinks/week	11.671	Miller 1991	10.5	7.5
Díez 2002	Units/week	8	Díez 2002	10.5	10
Drummond 2009	Drinks previous 180 days	8 x (7/180)	Drummond 2009	10.5	40; 200
Drummond 2014	Drinks/day	8 x 7	Drummond 2009	8	26
Ettner 2014	Drinks/week	11.671	Miller 1991	11	108
Fernández 1997	Units/week	10	Miller 1991	7.5	10
Field 2010	Drinks/week	11.671	Miller 1991	7.5	17.5
Fleming 1997	Drinks/week	12	Fleming 1997	10.5	40
Fleming 1999	Drinks/week	12	Fleming 1999	9	40
Fleming 2004	Drinks/month	11.671 x (12/52)	Miller 1991	4.5	40
Fleming 2010	Drinks in previous 28 days	11.671 x (7/28)	Miller 1991	8.5	42.5
Heather 1987	Units/month	8 x (12/52)	Heather 1987	8.5	7.5
Helstrom 2014	Drinks/day	11.671 x 7	Miller 1991	7	37.5
Huas 2002	Units/week	10	Heather 2006	10	7.5
Israel 1996	Drinks/month	13.456 x (7/28)	Miller 1991	7.5	150
Kaner 2013	Drinks/day	8 x 7	Kaner 2013	11	26
Kunz 2004	Drinks/week	11.671	Miller 1991	6	37.5
L'Engle 2014	Alcohol consumption	NR	NA	8	180
Lane 2008	Alcohol consumption	NR	NA	10.5	7.5
Lock 2006	Drinks/week	8	Miller 1991	12	7.5
Longabaugh 2001	Alcohol consumption	NR	NA	6	50
Maisto 2001	Drinks/month	17.01 x (7/30)	Miller 1991	5	13.5; 65
McDevitt‐Murphy 2014	Drinks/week	11.671	Miller 1991	5.5	60
McIntosh 1997	Drinks/month	13.456 x (12/52)	Miller 1991	10.5	60
Mello 2008	Alcohol consumption	NR	NA	7	45
Mertens 2014	Alcohol consumption	NR	NA	7.5	10
Monti 2007	Drinks/week	11.671	Miller 1991	7.5	85
Moore 2011	Standard drinks/week	11.671	Miller 1991	9.5	83
Noknoy 2010	Drinks/week	10	Furtwaengler 2013	11	45
Ockene 1999	Drinks/week	12.8	Ockene 1999	10	7.5
Richmond 1995	Drinks/week	10	Richmond 1995	9.5	5, 57.5
Rodríguez 2003	Alcohol consumption	NR	NA	10.5	17.5
Romelsjö 1989	g/day	1 x 7	NA	4.5	7.5
Rubio 2010	Drinks/week	12.8	Rubio 2010	9.5	32.5
Schaus 2009	Drinks/week	11.671	Miller 1991	8	40
Scott 1990	Units/week	8	Miller 1991	11	10
Segatto 2011	Alcohol consumption	NR	NA	6	45
Senft 1997	Drinks/3 months	11.671 x (4/52)	Miller 1991	9	16
Seppa 1992	Alcohol consumption	NR	NA	8.5	15
Soderstrom 2007	Drinks/last 90 days	11.671 x (7/90)	Miller 1991	6.5	32.5
Sommers 2006	Alcohol consumption	NR	NA	7	40
Sommers 2013	Drinks/week	11.671	Miller 1991	7.5	40
Spirito 2004	Standard drinks/month	11.671 x (12/52)	Miller 1991	7	40
Spirito 2011	Alcohol consumption	NR	NA	7.5	52.5
Tait 2004	alcohol consumption	NR	NA	8	37.5
Tomson 1998	g/week	1	NA	8.5	37.5
Wallace 1988	Units/week	8	Miller 1991	9.5	7.5
Walton 2010	Alcohol consumption	NR	NA	7	37
Watson 2013	Drinks/week	8	Watson 2013	9.5	20
Woolard 2013	Alcohol consumption	NR	NA	5.5	72.5

Date	Event	Description
25 September 2017	New citation required and conclusions have changed	1. The addition of 42 new studies changed conclusions with respect to the effect for women, and enabled further subgroup analyses. 2. The review authorship changed.
25 September 2017	New search has been performed	1. The search was updated in September 2017 and 42 new trials (24,057 participants) were included, bringing the total to 69 included studies (33,642 participants). Two trials in the original review were excluded from this update. 2. The review has been substantially revised in accordance with Cochrane guidance, including the implementation of GRADE assessment, creation of 'Summary of findings' tables for the main comparisons, and further subgroup analyses.
28 July 2009	Feedback has been incorporated	We have replied to feedback from Dr Anders Beich
19 June 2008	Amended	corrected figures, estimates and add minor changes
29 April 2008	Amended	little changes
29 April 2008	New search has been performed	Add one study but conclusions are not changed
15 February 2007	New citation required and conclusions have changed	Substantive amendment

Item	Judgment	Description
1. Random sequence generation (selection bias)	Low risk	The investigators describe a random component in the sequence generation process such as: random number table, computer random number generator, coin tossing, shuffling cards or envelopes, throwing dice, drawing of lots, minimisation
	High risk	The investigators describe a non‐random component in the sequence generation process such as: odd or even date of birth; date (or day) of admission; hospital or clinic record number; alternation; judgement of the clinician; results of a laboratory test or a series of tests; availability of the intervention.
	Unclear risk	Insufficient information to permit judgement of low or high risk.
2. Allocation concealment (selection bias)	Low risk	Investigators enrolling participants could not foresee assignment because one of the following, or an equivalent method, was used to conceal allocation: central allocation (including telephone, web‐based, and pharmacy‐controlled, randomisation); sequentially numbered, opaque, sealed envelopes. Very few studies which used envelopes to conceal allocation described them as serially numbered, opaque and sealed, but some studies nevertheless reported enough to suggest adequate allocation concealment. We used all the text describing allocation concealment to inform our judgement of this item.
	High risk	Investigators enrolling participants could possibly foresee assignments because one of the following method was used: open random allocation schedule (e.g. a list of random numbers); assignment envelopes without appropriate safeguards (e.g. if envelopes were unsealed or nonopaque or not sequentially numbered); alternation or rotation; date of birth; case record number; any other explicitly unconcealed procedure.
	Unclear risk	Insufficient information to permit judgement of low or high risk.
3. Blinding of providers (performance bias)	Low risk	No blinding or incomplete blinding, but the review authors judge that the outcome is not likely to be influenced by lack of blinding. Blinding of providers and unlikely that the blinding could have been broken. If the study was a cluster randomised study where the arms were physically separated and the intervention providers only had contact with a single arm. If the pre‐intervention steps (e.g. screening, randomisation, baseline assessment) were carried out by an independent person (e.g. researcher, research or practice nurse), and the intervention provider only had contact with the intervention arm, so that there was no chance of 'contamination' in the control arm.
	High risk	No blinding or incomplete blinding, and the outcome is likely to be influenced by lack of blinding; Blinding of key study participants and personnel attempted, but likely that the blinding could have been broken, and the outcome is likely to be influenced by lack of blinding.
	Unclear risk	Insufficient information to permit judgement of low or high risk.
4. Blinding of participants (performance bias)	Low risk	Blinding of participants and unlikely that the blinding could have been broken. At screening or assessment, participants were unaware that the trial specifically focused on alcohol consumption, e.g. tools encompassed wider health or lifestyle questions.
	High risk	No blinding or incomplete blinding, and the outcome is likely to be influenced by lack of blinding; Blinding of key study participants and personnel attempted, but likely that the blinding could have been broken, and the outcome is likely to be influenced by lack of blinding. Studies received "high risk" as default unless there was an explicit attempt to blind participants.
	Unclear risk	Insufficient information to permit judgement of low or high risk.
5. Blinding of outcome assessors (detection bias)	Low risk	No blinding of outcome assessment, but the review authors judge that the outcome measurement is not likely to be influenced by lack of blinding. Blinding of outcome assessors and unlikely that the blinding could have been broken.
	High risk	No blinding or incomplete blinding
	Unclear risk	Outcome collection was not automated and insufficient information is provided to assess blinding.
6. Incomplete outcome data (attrition bias) For all outcomes except retention in treatment or drop out	Low risk	No missing outcome data; Reasons for missing outcome data unlikely to be related to true outcome (for survival data, censoring unlikely to be introducing bias); Missing outcome data balanced in numbers across intervention groups, with similar reasons for missing data across groups; For dichotomous outcome data, the proportion of missing outcomes compared with observed event risk not enough to have a clinically relevant impact on the intervention effect estimate; For continuous outcome data, plausible effect size (difference in means or standardised difference in means) among missing outcomes not enough to have a clinically relevant impact on observed effect size; Missing data have been imputed using appropriate methods All randomised patients are reported/analysed in the group they were allocated to by randomisation irrespective of non‐compliance and co‐interventions (intention to treat).
	High risk	Reason for missing outcome data likely to be related to true outcome, with either imbalance in numbers or reasons for missing data across intervention groups; For dichotomous outcome data, the proportion of missing outcomes compared with observed event risk enough to induce clinically relevant bias in intervention effect estimate; For continuous outcome data, plausible effect size (difference in means or standardised difference in means) among missing outcomes enough to induce clinically relevant bias in observed effect size; ‘As‐treated’ analysis done with substantial departure of the intervention received from that assigned at randomisation.
	Unclear risk	Insufficient information to permit judgement of low or high risk (e.g. number randomised not stated, no reasons for missing data provided; number of drop out not reported for each group).
7. Selective reporting (reporting bias)	Low risk	The study protocol is available and all of the study’s pre‐specified (primary and secondary) outcomes that are of interest in the review have been reported in the pre‐specified way; The study protocol is not available but it is clear that the published reports include all expected outcomes, including those that were pre‐specified (convincing text of this nature may be uncommon).
	High risk	Not all of the study’s pre‐specified primary outcomes have been reported; One or more primary outcomes is reported using measurements, analysis methods or subsets of the data (e.g. subscales) that were not pre‐specified; One or more reported primary outcomes were not pre‐specified (unless clear justification for their reporting is provided, such as an unexpected adverse effect); One or more outcomes of interest in the review are reported incompletely so that they cannot be entered in a meta‐analysis; The study report fails to include results for a key outcome that would be expected to have been reported for such a study.
	Unclear risk	Insufficient information to permit judgement of low or high risk.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Quantity of drinking (g/week) at 12 months	34	15197	Mean Difference (IV, Random, 95% CI)	‐20.08 [‐28.36, ‐11.81]
2 Quantity of drinking (g/week) at 12 months, restricted to trials at low risk of bias due to allocation concealment	19	8856	Mean Difference (IV, Random, 95% CI)	‐19.98 [‐30.45, ‐9.51]
3 Quantity of drinking (g/week) at 12 months, restricted to trials at low risk of bias due to attrition	19	8496	Mean Difference (IV, Random, 95% CI)	‐20.69 [‐31.78, ‐9.60]
4 Quantity of drinking (g/week) at 12 months, with imputation of unknown standard deviations	39	17514	Mean Difference (IV, Random, 95% CI)	‐19.35 [‐26.90, ‐11.80]
5 Quantity of drinking (g/week) at 12 months, subgrouped by cluster/individual randomisation	34	15197	Mean Difference (IV, Random, 95% CI)	‐20.08 [‐28.36, ‐11.81]
5.1 Cluster‐randomised	6	2588	Mean Difference (IV, Random, 95% CI)	‐25.30 [‐42.69, ‐7.91]
5.2 Individually randomised	28	12609	Mean Difference (IV, Random, 95% CI)	‐18.77 [‐28.01, ‐9.53]
6 Quantity of drinking (g/week) at 12 months, subgrouped by cluster/individual randomisation, varying imputed ICC	34	15197	Mean Difference (IV, Random, 95% CI)	‐19.41 [‐27.59, ‐11.22]
6.1 Cluster‐randomised	6	2588	Mean Difference (IV, Random, 95% CI)	‐20.32 [‐33.50, ‐7.14]
6.2 Individually randomised	28	12609	Mean Difference (IV, Random, 95% CI)	‐18.77 [‐28.01, ‐9.53]
7 Quantity of drinking (g/week) at 12 months, subgrouped by gender	11	4836	Mean Difference (IV, Random, 95% CI)	‐37.42 [‐54.39, ‐20.45]
7.1 Men	11	3486	Mean Difference (IV, Random, 95% CI)	‐42.21 [‐64.53, ‐19.89]
7.2 Women	7	1350	Mean Difference (IV, Random, 95% CI)	‐30.27 [‐58.99, ‐1.55]
8 Quantity of drinking (g/week) at 12 months, subgrouped by gender, excluding trials of men only	7	4036	Mean Difference (IV, Random, 95% CI)	‐33.12 [‐51.47, ‐14.76]
8.1 Men	7	2686	Mean Difference (IV, Random, 95% CI)	‐35.16 [‐60.51, ‐9.80]
8.2 Women	7	1350	Mean Difference (IV, Random, 95% CI)	‐30.27 [‐58.99, ‐1.55]
9 Quantity of drinking (g/week) at 12 months, subgrouped by adolescents/young adults only versus other	34	15197	Mean Difference (IV, Random, 95% CI)	‐20.08 [‐28.36, ‐11.81]
9.1 Adolescents/young adults only	3	1638	Mean Difference (IV, Random, 95% CI)	‐7.09 [‐17.18, 2.99]
9.2 Other trials	31	13559	Mean Difference (IV, Random, 95% CI)	‐22.79 [‐32.42, ‐13.16]
10 Quantity of drinking (g/week) at 12 months, subgrouped by general practice/emergency setting	34	15197	Mean Difference (IV, Random, 95% CI)	‐20.08 [‐28.36, ‐11.81]
10.1 General practice setting	24	8811	Mean Difference (IV, Random, 95% CI)	‐25.69 [‐37.30, ‐14.08]
10.2 Emergency care setting	10	6386	Mean Difference (IV, Random, 95% CI)	‐9.73 [‐17.52, ‐1.93]
11 Quantity of drinking (g/wk) at 12 months, subgrouped by intervention modality (advice versus counselling)	32	13780	Mean Difference (IV, Random, 95% CI)	‐21.01 [‐30.06, ‐11.96]
11.1 Advice‐based	20	8243	Mean Difference (IV, Random, 95% CI)	‐32.68 [‐45.75, ‐19.60]
11.2 Counselling‐based	12	5537	Mean Difference (IV, Random, 95% CI)	‐0.17 [‐2.96, 2.61]
12 Quantity of drinking (g/week) at 12 months, subgrouped by type of control	34	15197	Mean Difference (IV, Random, 95% CI)	‐20.08 [‐28.36, ‐11.81]
12.1 No alcohol content	18	8606	Mean Difference (IV, Random, 95% CI)	‐24.41 [‐36.47, ‐12.35]
12.2 Some alcohol content	16	6591	Mean Difference (IV, Random, 95% CI)	‐12.80 [‐22.93, ‐2.66]
13 Quantity of drinking (g/week), subgrouped by length of follow‐up	38		Mean Difference (IV, Random, 95% CI)	Subtotals only
13.1 At 6 months	21	10313	Mean Difference (IV, Random, 95% CI)	‐21.56 [‐31.56, ‐11.55]
13.2 At 8 to 9 months	2	352	Mean Difference (IV, Random, 95% CI)	‐15.70 [‐34.19, 2.79]
13.3 At 12 months	34	15197	Mean Difference (IV, Random, 95% CI)	‐20.08 [‐28.36, ‐11.81]
13.4 At 36 months	1	277	Mean Difference (IV, Random, 95% CI)	‐6.11 [‐79.55, 67.33]
14 Quantity of drinking (g/week), subgrouped by length of follow‐up (6 and 12 months) and restricted to trials with information at both times	18		Mean Difference (IV, Random, 95% CI)	Subtotals only
14.1 At 6 months	18	9739	Mean Difference (IV, Random, 95% CI)	‐21.28 [‐31.82, ‐10.75]
14.2 At 12 months	18	9645	Mean Difference (IV, Random, 95% CI)	‐22.25 [‐34.04, ‐10.47]
15 Quantity of drinking (g/week) at 12 months, subgrouped by effectiveness/efficacy	34	15197	Mean Difference (IV, Random, 95% CI)	‐20.08 [‐28.36, ‐11.81]
15.1 Effectiveness trials	16	7091	Mean Difference (IV, Random, 95% CI)	‐27.06 [‐41.49, ‐12.63]
15.2 Efficacy trials	18	8106	Mean Difference (IV, Random, 95% CI)	‐14.37 [‐23.97, ‐4.76]
16 Frequency of binge drinking (number binges/week) at 12 months	15	6946	Mean Difference (IV, Random, 95% CI)	‐0.08 [‐0.14, ‐0.02]
17 Frequency of drinking (number binges/week), subgrouped by length of follow‐up (6 & 12 months) and restricted to trials with information at both times	9		Mean Difference (IV, Random, 95% CI)	Subtotals only
17.1 At 6 months	9	4354	Mean Difference (IV, Random, 95% CI)	‐0.15 [‐0.22, ‐0.08]
17.2 At 12 months	9	4333	Mean Difference (IV, Random, 95% CI)	‐0.12 [‐0.23, ‐0.02]
18 Frequency of drinking (number days drinking/week) at 12 months	11	5469	Mean Difference (IV, Random, 95% CI)	‐0.13 [‐0.23, ‐0.04]
19 Frequency of drinking (number days drinking/week), subgrouped by length of follow‐up (6 & 12 months) and restricted to trials with information at both times	6		Mean Difference (IV, Random, 95% CI)	Subtotals only
19.1 At 6 months	6	3637	Mean Difference (IV, Random, 95% CI)	‐0.12 [‐0.24, 0.00]
19.2 At 12 months	6	3658	Mean Difference (IV, Random, 95% CI)	‐0.16 [‐0.28, ‐0.04]
20 Intensity of drinking (g/drinking day) at 12 months	10	3128	Mean Difference (IV, Random, 95% CI)	‐0.18 [‐3.09, 2.73]
21 Heavy drinkers at 12 months	18	7623	Risk Difference (M‐H, Random, 95% CI)	‐0.09 [‐0.13, ‐0.04]
22 Binge drinkers at 12 months	10	4456	Risk Difference (M‐H, Random, 95% CI)	‐0.07 [‐0.12, ‐0.02]
23 Laboratory markers ‐ GGT (IU/L) at 12 months	3	1166	Mean Difference (IV, Random, 95% CI)	‐0.89 [‐3.86, 2.08]
24 Laboratory markers ‐ GGT (IU/L) at 12 months, subgrouped by gender	2	1094	Mean Difference (IV, Random, 95% CI)	‐0.03 [‐2.52, 2.46]
24.1 Men	2	764	Mean Difference (IV, Random, 95% CI)	‐2.08 [‐6.10, 1.95]
24.2 Women	2	330	Mean Difference (IV, Random, 95% CI)	1.23 [‐1.94, 4.40]

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Quantity of drinking (g/week) at 12 months	6	1296	Mean Difference (IV, Random, 95% CI)	‐19.50 [‐40.47, 1.48]
2 Frequency of binge drinking (number binges/week) at 12 months	2	456	Mean Difference (IV, Random, 95% CI)	‐0.08 [‐0.28, 0.12]
3 Frequency of drinking (number days drinking/week) at 12 months	2	319	Mean Difference (IV, Random, 95% CI)	‐0.45 [‐0.81, ‐0.09]
4 Intensity of drinking (g/drinking day) at 12 months	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
5 Heavy drinkers at 12 months	1		Risk Difference (M‐H, Random, 95% CI)	Subtotals only
6 Binge drinkers at 12 months	2	777	Risk Difference (M‐H, Random, 95% CI)	‐0.02 [‐0.07, 0.03]
7 Laboratory markers ‐ GGT (IU/L) at 12 months	1		Mean Difference (IV, Random, 95% CI)	Subtotals only

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Quantity of drinking (g/week) at 12 months	3	552	Mean Difference (IV, Random, 95% CI)	1.54 [‐42.01, 45.10]
2 Quantity of drinking (g/week) at 12 months, with imputation of unknown standard deviations	4	635	Mean Difference (IV, Random, 95% CI)	1.13 [‐37.00, 41.26]
3 Frequency of drinking (number days drinking/week) at 12 months	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
4 Intensity of drinking (g/drinking day)	1		Mean Difference (IV, Random, 95% CI)	Subtotals only
5 Binge drinkers at 12 months	2	339	Risk Difference (M‐H, Random, 95% CI)	0.02 [‐0.08, 0.12]
6 Laboratory markers ‐ MCV (fl) at 12 months	1	95	Mean Difference (IV, Random, 95% CI)	0.62 [‐1.58, 2.81]
6.1 Men	1	81	Mean Difference (IV, Random, 95% CI)	0.80 [‐1.70, 3.30]
6.2 Women	1	14	Mean Difference (IV, Random, 95% CI)	0.0 [‐4.57, 4.57]

Methods	Parallel group RCT: extended brief intervention versus brief intervention versus active control.  ITT: outcome data were based on imputation of baseline values to participants lost to follow‐up.
Participants	Setting: Finland; primary care clinic. Participants 20 to 60 years, consuming ≥ 280 g absolute ethanol/week or CAGE ≥ 3 for men, ≥ 190 g absolute ethanol/week or CAGE ≥ 2 for women; excluded if severe psychiatric disease, or at least one detox treatment, or alcohol dependence, or alcohol‐related disease; screened by self‐administered health questionnaire including CAGE and quantity‐frequency consumption questions.  Number randomised = 414; 71% male; mean age = 41.6 years; 18% comprehensive school, 7.3% vocational school, 12.7% college or university; 18.7% working/studying, 13.6% unemployed, 6% retired.  At baseline: mean drinking amount per week = 286 g for men, 165.5 g for women, 259.8 g overall; mean drinking times per week = 2.2 for men, 2.1 for women; mean usual drinking amount per occasion = 139.2 g for men, 85.8 g for women; mean CAGE = 3.2 for men, 2.8 for women.
Interventions	Group A (N = 149) received brief intervention from GP or nurse at baseline, 2, 6, 12, 18, 24 and 30 months. Intervention was 10 to 20 minutes based on FRAMES according to the needs of individual participants.  Group B (N= 137) received the same intervention less frequently: at baseline, 12 and 24 months.  Group C (N = 128) received advice to reduce drinking and contact their GP in the event of health problems; were not told about 36 month follow‐up.
Outcomes	Mean drinking amount per week; drinking times per week; usual drinking amount per occasion; CDT, AST, ALT, GGT, MCV. Assessed at 36 months (blood tests additionally assessed at each brief intervention).
Funding source	Not reported.
Declaration of interests	Not reported.
Notes	Lahti project. Loss to follow‐up: Group A: 61/149 (41%). Group B: 55/137 (40%). Group C: 55/128 (43%).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomisation was performed by each participating general practitioner: before a participant arrived for a feedback session, the general practitioner drew a card from a mixed pack that included equal numbers of A, B, and C signs to refer the participant to the group which the participant was to be allocated" (p. 1682).
Allocation concealment (selection bias)	High risk	GP could see allocation and it was possible to subvert.
Blinding of treatment providers	High risk	GP provided intervention to all arms therefore risk of contamination between arms was high.
Blinding of participants	High risk	No reported attempt to blind participants.
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Unclear whether outcome assessors were blinded.
Incomplete outcome data (attrition bias)   All outcomes	High risk	Loss to follow‐up > 30%.
Selective reporting (reporting bias)	Unclear risk	Did not specify exact alcohol outcome measures in the methods.

Methods	Parallel group RCT: brief intervention versus active control. ITT: no.
Participants	Setting: Spain; primary care clinic. Participants aged 15 to 75 years with a weekly consumption of > 280 g for men and > 168 g for women; excluded if current treatment for alcohol problems or hepatologic problems or concomitant diseases requiring alcohol abstinence or MALT > 11; screened by MALT scale for alcohol dependence.  Number randomised = 139; 100% male; mean age = 45 years; 47% comprehensive school, 19% vocational school, 33% college/university; 49% working/studying, 36% unemployed, 15% retired.  At baseline: mean weekly alcohol consumption = 57 units (1 unit = 8 g alcohol).
Interventions	Intervention group (N = 75) received five minutes general advice from GP with support material plus a five‐visit program over the year. NB, 21 participants were subsequently excluded; 54 received intervention treatment.  Control group (N = 64) received a single session of brief advice from GP. NB, 19 were subsequently excluded; 45 received the control treatment.
Outcomes	Percentage reduction in alcohol consumption; MALT test; Goldberg score; per cent drinking < 35 units/week. Assessed at 12 months.
Funding source	Este trabajo se realizo con la ayuda proporcionada por la Beca FS 93/0882. [This work was conducted with support provided by scholarship/grant number FS 93/0882].
Declaration of interests	Not reported.
Notes	Loss to follow‐up: Intervention group: 20/54 (37%). Control group: 15/45 (33%).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	States only that assignment was made by simple randomisation ("aleatorizacion simple", p. 122).
Allocation concealment (selection bias)	Low risk	Consecutively numbered, sealed envelopes ("los sobres cerrados y numerados", p. 122).
Blinding of treatment providers	High risk	Not blinded.
Blinding of participants	High risk	No reported attempt to blind participants.
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Not reported.
Incomplete outcome data (attrition bias)   All outcomes	High risk	Loss to follow‐up > 30%.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods were reported in results.

Methods	Cluster RCT: brief intervention (two arms delivered by different professionals) versus usual care; 15 practices randomised. Used a pre‐post repeated measures group design. ITT: yes.
Participants	Setting: USA; family or internal medicine clinics in a managed care organisation (MCO)  Participants ≥ 18 years were screened with Health Assessment Survey including AUDIT and selected if they scored 8+ (men) or 7+ (women). Number of clinics randomised = 15; number of participants randomised = 50,411, 60% male; mean age followed up at 3 months (n = 1379) = 45.9 years. At baseline: mean drinks per week (defined by AUDIT questions 1 to 2) = 13.8 (men); 8.7 (women).
Interventions	Group P (N = 17,257 randomised, 1151 sampled for follow‐up) received brief intervention (according to their level of drinking ‐ see below for detail) from physician or physician assistant. Group S (N = 15,938 randomised, 1124 sampled for follow‐up) received brief intervention from specialist, i.e. mostly nurses. Group C (N = 17,216 randomised, 1955 sampled for follow‐up) was the control group and received "patient care as usual" (not defined). Within each group treatment allocated by drinking level: Zone I (modified AUDIT 7 to 15 for women and men aged > 65 years; 8 to 15 for men < 65 years) participants received simple advice (3 to 5 minutes) and an information brochure; Zone II (AUDIT 16 to 19) received advice and a more extensive self‐help manual; Zone III (AUDIT > 19) specified advice and referral to specialty care for alcohol assessment and treatment. Advice comprised: feedback of screening scores and any problems or symptoms identified; discussion of the participant’s risk level; connection of drinking risk to actual or potential problems and information about safer drinking limits; commitment to a goal of cutting back or stopping; and presentation of the brochure and encouragement to reach the established goal.
Outcomes	Number of drinks per week; per cent at risk drinkers who had changed drinking at three months; health care utilisation (all days, outpatients visits, inpatient days, emergency department visits, ADM‐related days (mental health). Assessed at 3 and 12 months.
Funding source	This research was supported by grant no. 029620 from the Robert Wood Johnson Foundation.
Declaration of interests	Not reported.
Notes	Cutting Back programme. Payment: the MCOs were reimbursed; participants received no payment. Loss to follow‐up: 81/575 (14%) overall (not reported by arm).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random numbers from table generated by independent statistician (p. 625).
Allocation concealment (selection bias)	Low risk	Allocation by independent statistician – not a member of the team (p. 625).
Blinding of treatment providers	Low risk	Cluster‐randomised trial and all participants from a particular clinic received the same intervention, so therefore risk of contamination between arms was low.
Blinding of participants	High risk	No reported attempt to blind participants.
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Computer‐assisted telephone interview by independent survey organisation – interviewers not aware of the study condition (p. 625).
Incomplete outcome data (attrition bias)   All outcomes	High risk	Only reported data from those who completed, not those randomised.
Selective reporting (reporting bias)	High risk	Only primary outcome fully reported.

Methods	Parallel group RCT: 'full care' telephone‐based brief intervention versus stepped care telephone‐based brief intervention versus inactive control. ITT: yes.
Participants	Setting: Germany; 84 primary care practices.  Participants: Aged 18 to 64 years with average consumption > 20/30 g alcohol per day for women/men within the last four weeks, or regular heavy drinking episodes (“binge drinking”) defined as > 60/80 g alcohol for women/men on at least two occasions within the last four weeks; included if assessed as alcohol dependent; excluded if they had an acute or terminal illness, or severe drug dependence, or no telephone, or did not understand German, or unable to read, or were in alcohol treatment, or reported no alcohol consumption in the last four weeks; screened with AUDIT and LAST (Luebeck Alcohol dependence and abuse Screening Test). Number randomised = 408; 68% male; mean age not reported for total sample. Baseline characteristics not reported for total sample.
Interventions	All three groups received a health questionnaire incorporating AUDIT and LAST for screening; this served as baseline assessment. Full care group (N = 131) received computerised feedback immediately after assessment; followed by one 30 minute session of brief intervention counselling by telephone by trained psychologists based on motivational interviewing and structured elements of behaviour change counselling straight after assessment; followed by three similar booster sessions at 1, 3 and 6 months. Stepped care group (N = 138) received the same computerised feedback as full care immediately after assessment but no intervention at this point; they received a maximum of three 30 to 40 minute brief intervention sessions at 1, 3 and 6 months. If stepped care participants reported a reduction of alcohol consumption below the inclusion criteria for the trial and indicated high self‐efficacy to maintain the change, no further contact was made until the 12 month follow‐up. Control participants (N = 139) received a booklet on health behaviour.
Outcomes	Grams per day of alcohol, percentage meeting binge criteria. Assessed at 12 months.
Funding source	"Funding for this study was provided by German Federal Ministry of Research and Education grant no. 01 EB 0121. The German Federal Ministry of Research and Education had no further role in study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the paper for publication", p. 250.
Declaration of interests	Conflict of interest: None.
Notes	SIP (Stepped Interventions for Problem drinkers) Loss to follow‐up: Full care group: 15/131 (11%). Stepped care group: 6/138 (4%). Control group: 13/139 (9%). We contacted the authors, who supplied missing data.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomisation of subjects to one of the three groups was carried out prior to the telephone interview using sealed cards drawn by study staff from one container" [analogous to card shuffle described in Cochrane Handbook as low risk of selection bias] (p. 245).
Allocation concealment (selection bias)	Low risk	Final study participants were randomly allocated, using coloured cards in sealed, non‐transparent envelopes (drawn envelopes were not put back), to either the control group or to one of two intervention groups (p. 245).
Blinding of treatment providers	High risk	"Due study design, blinding could not be conducted for the individual study participant and the staff member providing the counselling session" (p. 245).
Blinding of participants	High risk	"Due to our study design, blinding could not be conducted for the individual study participant and the staff member providing the counselling session" (p. 245).
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	"A blinded personal interview was conducted by study staff who had no contact with the participant prior to the outcome assessment which mainly was assessed by telephone" (p. 247).
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Loss to follow‐up reported with reasons.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods were reported in results.

Methods	Parallel group RCT: brief intervention and tailored booklet versus brief intervention and generic booklet versus tailored booklet only (active control) versus generic booklet only (active control). ITT: yes.
Participants	Setting: USA; emergency department level 1 trauma centre.  Participants: Aged ≥ 19 years and injured, displaying at risk or heavy episodic drinking. At risk drinking was defined as: ≥ 15 drinks per week for men aged < 65 years; ≥ 12 drinks per week for women aged < 65 years and men aged ≥ 65 years; and ≥ 9 drinks per week for women aged ≥ 65 years in the previous 3 months. Heavy episodic drinking was defined as ≥ 5 drinks/occasion on > 4 occasions in the last month (i.e. weekly episodes) for men < 65 years; ≥ 4 drinks per occasion on > 4 occasions in the last month for men ≥ 65 years and all women. Excluded if: severely injured (e.g. unconscious); in need of immediate life‐saving procedures; blood alcohol level > 200 mg/dL; self‐inflicted injury, sexual assault, overdose, poisoning, near‐drowning, chronic injury without specific event associated with re‐injury; pregnant; prisoner; or non‐English speaking. Screened with computerised health survey incorporating AUDIT‐C and DrInC; participants also received an alcohol salivary test before intervention, which proceeded once blood alcohol concentration reached 100 mg/dL or less. Number randomised = 575 (but following data are for 494 participants who were followed‐up); 71% male; mean age 27.8 years; 86% White, 6% Black, 8% other race; 80% some college education or higher. At baseline: mean number of drinks per week = 21.
Interventions	Participants were screened with a health survey incorporating AUDIT‐C and DrInC which served as baseline assessment (duration was not reported). Tailored manual/brief advice group (N = 129) received brief intervention before leaving the emergency department from a research social worker, during which the tailored booklet was reviewed with the participant. The tailored booklet was generated according to responses to the computerised screening tool. Tailored manual/no brief advice group (N = 121) received a tailored booklet from the research social worker who told them that, based on their responses to the health screen, they scored as at‐risk for hazardous drinking and should review the booklet. Generic manual/brief advice group (N = 124) received brief intervention before they left the emergency department from a research social worker, during which the generic booklet was reviewed with the participant. The generic booklet was identical to the tailored booklet in length, content and graphics but included standard rather that tailored text/graphics. Generic manual/no brief advice group (N = 120) received the generic booklet from the research social worker who told them that, based on their responses to the health screen, they scored as at‐risk for hazardous drinking and should review the booklet. Duration of intervention was not reported.
Outcomes	Mean number of drinks per week, number of binge drinking episodes (> 5 drinks for men, > 4 drinks for women) in previous month, DrInC score (alcohol‐related consequences). Assessed at 3 and 12 months.
Funding source	This study was supported by NIAAA grant AA111629.
Declaration of interests	Not reported.
Notes	Participants willing to complete the survey were entered into a monthly draw for USD 100. Participants were paid USD 20 for three month telephone follow‐up interview and USD 30 for 12 month telephone follow‐up. Loss to follow‐up: 81/575 (14%) (not reported according to study arms). Data (including baseline data) were reported only for participants who were followed up.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"A computer program automatically determined eligibility for the RCT and randomly assigned participants to one of four intervention conditions" (Blow 2006, p. 570). "To prevent the possibility that staff could unwittingly manipulate assignment to intervention condition, participants were randomly assigned to conditions via the computer using an urn randomisation technique" (Blow 2009, p. 487).
Allocation concealment (selection bias)	Unclear risk	Not reported.
Blinding of treatment providers	Low risk	Cluster randomised ‐ treatment providers only had contact with one arm and only delivered one intervention, reducing the chance of cross contamination.
Blinding of participants	High risk	No reported attempt to blind participants.
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Not reported.
Incomplete outcome data (attrition bias)   All outcomes	High risk	Reported data only from those who completed not those randomised.
Selective reporting (reporting bias)	Low risk	Outcomes stated in methods were reported.

Methods	Parallel group RCT: brief negotiated interview versus assessment only active control versus screening only active control. ITT: yes.
Participants	Setting: Poland; emergency department. Participants: Aged ≥ 18 years not presently being treated for an alcohol‐related problem, willing to give informed consent to be randomised into one of three groups, willing to provide contact information for at least two individuals who would always know participant’s whereabouts, screened as consuming ≥ 11 drinks per week for men (≥ 6 for women); or ≥ 4 drinks on one occasion for men (≥ 3 for women); or ≥ 1 on the RAPS (Remorse, Amnesia, Perform, Starter) tool. Participants who reported drinking an average of more than six drinks per day at baseline on both quantity/frequency screening and 30 day timeline follow back were excluded from the analysis. Number randomised = 446; 85% male; mean age not reported for total sample. Baseline characteristics not reported for total sample.
Interventions	Participants were screened with the RAPS tool and quantity/frequency questions. The assessed group and brief negotiated interview groups underwent assessment (duration not reported) as follows: blood alcohol concentration measure with a breathalyser, self‐reported drinking in the previous six hours using the Timeline Followback, consequences of drinking using the Short Inventory of Problems SIPs +6, stage of change using the Readiness to Change Ruler, risk‐taking assessed with questions from Eysenk and Jackson, and sensation seeking assessed with questions adapted from Zuckerman. Brief negotiated interview group (N = 147) received a 15 to 20 minute brief negotiated interview from a trained emergency department nurse, which included engagement and permission, feedback, information and norms, decisional balance and pros and cons, readiness to change, menus of options, and prescription for change. A list of AA groups and specialised services for alcohol treatment and counselling was provided to the participant. Assessed group (N = 152) received the same list of AA groups and specialised alcohol services. Screened group (N = 147) received no assessment or intervention.
Outcomes	Per cent exhibiting at‐risk drinking, per cent with RAPS > 1, drinking days per week, drinks per drinking day, maximum drinks on one occasion, number of negative consequences. Assessed at 3 and 12 months.
Funding source	This research was supported by NIAAA grant R21 AA 016081‐01.
Declaration of interests	Not reported.
Notes	Loss to follow‐up at three months: Brief negotiated interview group: 26/147 (18%). Assessed group: 23/152 (15%). Screened group: 12/147 (8%). Loss to follow‐up at 12 months: Brief negotiated interview group: 60/147 (41%). Assessed group: 53/152 (35%). Screened group 55/147 (37%).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomisation process not described.
Allocation concealment (selection bias)	Unclear risk	"Participants were first randomised to the screen‐only or assessment condition by the study interviewer, who drew an envelope [does not specify sealed or opaque] with the condition assignment. The envelope of those receiving an assessment contained a second envelope, which was opened by the interviewer following assessment to determine whether the participant was assigned to the intervention condition" (p. 984).
Blinding of treatment providers	Low risk	Nurse who delivered the intervention only saw intervention participants.
Blinding of participants	High risk	No reported attempt to blind participants.
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Assessment interviewer was blind to group status (p. 986).
Incomplete outcome data (attrition bias)   All outcomes	High risk	Loss to follow‐up > 30%.
Selective reporting (reporting bias)	Low risk	RAPS+1 not reported at three months, but because we focused on 6 and 12 month outcomes, there was little impact.

Methods	Parallel group RCT: brief motivational message plus booster telephone calls versus usual care  ITT: paper reported multiple imputation to impute outcome data for non‐respondents; unpublished data supplied to the reviewers was based on 222/333 (67%) of those randomised.
Participants	Setting: USA, primary care clinic. Participants: (no age restrictions), presenting for pre‐booked appointment; screened by telephone interviews to select those with AUDIT score <= 15 and consuming ≥ 2 alcoholic drinks per day in past month (chronic drinking), or ≥ 2 episodes of binge drinking (≥ 5 drinks) in past week (binge drinkers), or ≥ 1 episode of driving after ≥ 3 drinks; excluded if alcoholic, pregnant, terminally ill, or cognitively impaired.  Number randomised = 333; 65% male; mean age 46.9 years; 16% unemployed; 91% post‐high school education; 68% income > $35,000 per year; 80% Caucasian.  Number assessed = 222 (67%).  At baseline: mean drinking amount = 166 g/week; 42% chronic drinkers; 33% binge drinkers.
Interventions	Intervention group (N = 166) received: a) a brief motivational message of 1 to 5 minutes from the primary care physician during the planned routine visit; b) self‐help manual; c) written personalised feedback; d) up to three telephone counselling calls over 10 weeks by a psychology graduate. Control group (N = 167) received usual care (no intervention or any information about their participation in the study in their notes).
Outcomes	Drinks per week, drinking days per week, binges per week, grams of alcohol per drinking day, percentage of binge drinkers, percentage of heavy drinkers (average of > 1 drink per day for women or > 2 drinks per day for men. Assessed at 3 and 12 months.
Funding source	This study was supported by National Institute on Alcohol Abuse and Alcoholism Grant RO1 AA09175 (to Susan J. Curry, Principal Investigator).
Declaration of interests	Not reported.
Notes	Loss to follow‐up:  Intervention group: 66/166 (40%).  Control group: 45/167 (27%).  Analyses of frequency and intensity of drinking are based on unpublished data on 222 cases  Analysis of quality of alcohol consumed/week are based on published means and unpublished SDs.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of sequence generation not reported.
Allocation concealment (selection bias)	Unclear risk	Method of allocation concealment not described.
Blinding of treatment providers	Low risk	Treatment providers only interacted with intervention participants.
Blinding of participants	Low risk	During recruitment participants were blinded to the focus of the study on alcohol.
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Unclear if outcome assessor was blinded (p. 157).
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Loss to follow‐up reported per arm but not reasons. Uneven follow‐up between arms; used multiple imputation which uses a regression‐type approach that can remove bias that is due to differential non‐response if the imputation model contains variables that are good predictors of the outcome and of non‐response.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods are reported.

Methods	Cluster RCT: brief intervention versus active control; 74 primary care practices were randomised, average cluster size = 6.9.  ITT: unclear.
Participants	Setting: Spain, primary care practices. Participants: Men aged 14 to 50 years with either weekly alcohol consumption over 35 units, or over 10 units on any given day at least once a month (1 unit = 8 g); excluded if alcohol dependency or previous advice to reduce drinking or chronic pathology/treatment requiring > 3 months abstinence or CAGE score > 1; recruited from general practice; screened by lifestyle questionnaire with embedded CAGE  Number randomised = 546 (all data reported for 229 followed up participants); 100% male; mean age 36.5 years; 100% Hispanic; 70.1% married; 78.7% middle‐lower or lower social status; 95.6% employed; 64.1% further education  At baseline: mean weekly alcohol consumption = 54.0 units; CAGE = 1 for 63.2% of participants (data given for 229 heavy throughout‐week drinkers only)
Interventions	Both groups were screened using a lifestyle questionnaire incorporating CAGE questions, this served as baseline assessment Intervention group (N = 104 participants) received from the GP 15 minutes cognitive‐behavioural therapy consisting of a self‐informative booklet including day diary for registration of consumption, individualised agreement of consumption targets, and offer of follow‐up and support  Control group (N = 125) received from the GP five minute 'simple advice' which reproduced usual care
Outcomes	Per cent of participants cutting down to under 35 units per week Assessed at 12 months
Funding source	Not reported
Declaration of interests	Not reported
Notes	EBIAL trial  Loss to follow‐up not reported by arm
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of sequence generation is unclear
Allocation concealment (selection bias)	Low risk	All research workers in one centre conducted the same intervention, and they were not aware of the details of the other intervention; allocation unit was the primary care practice (p.563)
Blinding of treatment providers	Low risk	Cluster randomised: treatment providers only had contact with one arm and only delivered one intervention, reducing the chance of cross contamination
Blinding of participants	High risk	No reported attempt to blind participants
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Not reported
Incomplete outcome data (attrition bias)   All outcomes	High risk	Overall loss to follow‐up 49%
Selective reporting (reporting bias)	Low risk	Outcomes described in the methods section are reported

Methods	Parallel RCT: motivational interview versus control. ITT: no.
Participants	Setting: USA; primary care clinic providing free care to underserved populations, described as uninsured (working poor, homeless, runaway and high‐risk youth). Participants: 12 to 18 years who reported alcohol consumption or drug use and consequences due to use, English‐speaking, able to return for 15 to 20 minute appointment. Number randomised = 64 (of which 42 completed and are in the analysis, following percentages are of 42 completers); 48% male; mean age of total sample not reported; 85.7% Hispanic/Latino, 9.5% African American, 4.8% White; 42.9% completed high school, 42.9% didn't complete high school, 14.3% don't know; 31.0% attended at least some college, 52.4% didn't attend at least some college, 16.7% don't know (some are below college age). At baseline: 78.6% reported alcohol use in the last 30 days; 100% reported alcohol use in their lifetime.
Interventions	Participants were screened with CRAFFT plus six filler questions on health, and both groups completed a baseline survey (duration not reported). MI group (N = 38 randomised, 22 completers in analysis) received a 15 to 20 minute motivational interviewing intervention plus a 5 to 10 minute booster telephone call one month after intervention from case managers who worked in the mental health division of the clinic. The motivational interviewing session focused on assessing motivation to change (five to seven minutes), enhancing motivation for change (5 to 7 minutes), and making a plan (5 to 7 minutes). The booster call briefly reiterated what the teen had discussed in the Project CHAT session, reviewed their goals and whether they had been able to implement any of the strategies they had discussed during the session, and revised goals as necessary. Control group (N = 26 randomised, 20 completers in analysis) were screened and assessed; no detail is given about their care.
Outcomes	Number of days consumed alcohol in the last month, how many drinks consumed, number of days consumed more than three drinks. Assessed at 3 months.
Funding source	Work on this article was supported by a grant from the National Institute on Drug Abuse (R21DA018854) to Elizabeth J D'Amico.
Declaration of interests	Not reported.
Notes	Project CHAT. Participants received USD 15 for baseline survey; USD 25 for 3‐month survey; USD 15 for completion. During the study incentive was increased to USD 30, and 22 teens received this payment. Loss to follow‐up: MI group: 22/38 (42%). Control group: 20/26 (23%).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	"Initially the youth were randomised on a one‐to‐one basis... as the trial progressed, we recognised that dropout rates were unequal between groups... thus, to maximise power, we altered the allocation schedule such that the probability of being allocated to the intervention was higher." i.e. the researchers changed the randomisation procedure (p. 56).
Allocation concealment (selection bias)	High risk	As above, researchers changed allocation procedure whilst the trial was in progress to alter the allocation.
Blinding of treatment providers	Unclear risk	Unclear whether treatment provider has contact with both intervention and control participants.
Blinding of participants	Low risk	Screening questionnaire had filler questions to mask recruitment criteria (p. 57).
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Not described.
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Differential in missing data across groups. This is one of the few studies that did not exclude hard to reach populations such as homeless people.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods are reported.

Methods	Parallel RCT: brief negotiated interview versus scripted discharge instructions. ITT: yes.
Participants	Setting: USA, emergency department. Participants: ≥ 18 years who reported consuming in excess of NIAAA limits, or whose emergency department visit related to an injury associated with alcohol use; excluded if non–English speaking, AUDIT > 19, or using drugs daily, or currently enrolled in a substance abuse treatment program, or seeking treatment for an acute psychiatric complaint or hospitalised for a psychiatric problem in the past year, or critically ill or injured, or cognitively impaired. Number randomised = 500 (of which 494 in analysis, following percentages are of 494); 68% male; mean age of total group not reported; 67% White, 20% Black, 11% Hispanic, 2% Other; 47% high school or less, 36% some college, 17% college degree or more. Baseline characteristics for total sample not reported.
Interventions	Participants were screened with an 18‐item health screening questionnaire, including multiple health factors and the NIAAA quantity/frequency questions embedded. Both groups received baseline assessment (duration not reported), including alcohol consumption for the past 30 days as measured by Time Line Follow‐Back method, and questions concerning drinking‐related consequences such as drinking after driving, injury sustained while drinking, and arrests. Readiness to change was assessed by the Contemplation Ladder, and the Short Form Health Survey was used to assess health status in mental and physical domains. BNI group (N = 250 randomised, 247 analysed) received brief negotiated interview from one of a group of trained emergency practitioners ‐ a manual‐guided intervention using techniques based on motivational interviewing, brief advice, and behavioral contracting and designed to be delivered in less than ten minutes. The four primary steps were: (1) raise the subject of alcohol; (2) provide feedback by reviewing the participant’s screening data, make a connection between alcohol and the visit/illness or injury if possible, review National Institute on Alcohol Abuse and Alcoholism guidelines for low‐risk drinking; (3) enhance motivation with motivational interviewing techniques; and (4) negotiate and advise by summarising the participant’s reasons for change and negotiating a drinking goal. Participants were then asked to complete and sign a drinking agreement. The control group (N = 250 randomised, 247 analysed) received scripted discharge instructions read by the emergency practitioner, designed to be less than one minute in length (although actual mean length was 1.4 minutes). This included a statement recommending that the participant decrease alcohol intake and, if appropriate, use seatbelts, exercise regularly, and stop smoking. A handout was provided with more information related to all identified health risks.
Outcomes	Average drinks per week, binge episodes in last month, proportion over NIAA guidelines, number who have driven after drinking, number arrested or pulled over for driving under the influence, number who had motor vehicle crash whilst intoxicated, number who were injured while drinking one or more times, contact with legal system, missed 1 workday in past 30 days. Assessed at 12 months.
Funding source	Funded by National Institute on Alcohol Abuse and Alcoholism grant R01 AA12417‐01A1 (Dr D’Onofrio), National Institute on Drug Abuse grant K23 DA15144 (Dr Pantalon), and Robert Wood Johnson Generalist Physician Faculty Scholar Award (Dr Fiellin).
Declaration of interests	None (says all conflicts must be reported but none were specified).
Notes	Project ED Health. USD 20 at intake visit (from research associate); USD 40 at six months by post; USD 50 at 12 months by post. 95% follow‐up at six months; 92% at 12 months.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Participants were randomly assigned with 100 block randomisation but does not say how the sequence was generated.
Allocation concealment (selection bias)	Low risk	"Research associates were provided with sealed, opaque envelopes for each of the 500 randomised study identification numbers. The identification number appeared on the outside, and the assigned treatment condition was specified inside the envelope" (p. 744).
Blinding of treatment providers	Unclear risk	Treatment provider provides both intervention and control, but control statement is scripted and reported duration of control arm suggests contamination is unlikely.
Blinding of participants	Low risk	"The screen included questions related to smoking, exercise, and seatbelt use to mask alcohol as the central focus" (p. 744).
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	"Interviews were conducted by research associates blinded to subject treatment assignment" (p. 744).
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Follow‐up rates were high and comparable, but reasons for loss to follow‐up not reported.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods are reported.

Methods	Parallel RCT: brief alcohol intervention versus control with assessment versus control without assessment. ITT: unclear.
Participants	Setting: Switzerland, emergency department. Participants: ≥ 18 years who were admitted to emergency department for an injury; excluded if history of alcohol‐related treatment over the last 12 months, or clinically intoxicated, or medical condition that precluded a face‐to‐face interview, or if not qualifying as hazardous drinker over the last 30 days. Number randomised = 987; 78% male; mean age = 36.7 years. Baseline data not reported for total sample.
Interventions	Participants were screened with a 10‐item lifestyle questionnaire including three alcohol‐related questions. BAI and CA groups received a 30 minute assessment. Brief Alcohol Intervention (BAI) group (N = 310) received a single motivation style brief intervention session lasted approximately 15 minutes and covered six steps: (1) to thank the participant for participation and provide reassurance about confidentiality; (2) to provide feedback about participant’s alcohol use compared to similar measures for men and women in the Swiss community and ask the participant’s opinion of the feedback; (3) to ask the participant to explore the pros and cons of their alcohol use; (4) to use a 1 to 10 scale to explore participant’s importance and readiness to change their drinking patterns; (5) to ask if the participant feels ready to set an objective and provide positive reinforcement about the ability to achieve this objective; and (6) to give each participant written material, including their AUDIT score, drinking pattern percentiles compared to the Swiss community and their drinking pattern objectives. Control with assessment (CA) group (N = 342) received assessment and "usual care" (not defined). Control no assessment (C) group (N = 335) received "usual care" only.
Outcomes	Per cent hazardous drinkers, mean number of days drinking per week in previous 12 months, mean number of drinks per occasion in last 12 months, mean number of binge drinking occasions per month in last 12 months, mean number of drinks in last seven days, mean AUDIT score, mean SF‐12 score (mental and physical). Assessed at 12 months.
Funding source	Funded by grant no. 3200–067949 of the Swiss National Science Foundation. The Swiss National Science Foundation is the government agency funding scientific research in Switzerland.
Declaration of interests	Not reported.
Notes	Loss to follow‐up: BAI: 74/310 (24%). CA: 65/342 (19%). C: 78/335 (23%).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Each participant was randomised and assigned to experimental or control group based by block of ten participants on a computer‐generated allocation method" (p. 1226).
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding of treatment providers	High risk	"A group of five female and two male research assistants were recruited and selected to do the screening, assessment and BAI at baseline, depending on their ability to apply research procedures and to conduct BAI... Research assistants conducted the motivational style BAI right after the assessment" (p. 1228).
Blinding of participants	Low risk	"One of the goals of the trial was to blind subjects assigned to the control groups to minimize possible intervention effects. The subjects in both control groups were told that the trial focused on general health behaviours, including alcohol use" (p. 1228).
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	"Research assistants at follow up were blinded to the participant assignment, and because this interview conducted no evaluation of the BAI they were unaware from which group the participants came" (p. 1228).
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Loss to follow‐up reported but reasons not reported by arm.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods are reported.

PERMALINK

Effectiveness of brief alcohol interventions in primary care populations

Eileen FS Kaner

Fiona R Beyer

Colin Muirhead

Fiona Campbell

Elizabeth D Pienaar

Nicolas Bertholet

Jean B Daeppen

John B Saunders

Bernard Burnand

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Background

Description of the condition

Description of the intervention

How the intervention might work

Why it is important to do this review

Objectives

Main objective

Secondary objectives

Methods

Criteria for considering studies for this review

Types of studies

Types of participants

Types of interventions

Types of outcome measures

Primary outcomes

Secondary outcomes

Search methods for identification of studies

Electronic searches

Current update searches

Previous searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of risk of bias in included studies

Measures of treatment effect

Unit of analysis issues

Dealing with missing data

Assessment of heterogeneity

Assessment of reporting biases

Data synthesis

'Summary of findings' tables

Summary of findings for the main comparison. Brief intervention compared to no or minimal intervention for people with hazardous or harmful alcohol consumption.

Summary of findings 2. Extended intervention compared to no or minimal intervention for people with hazardous or harmful alcohol consumption.

Summary of findings 3. Extended intervention compared to brief intervention for people with hazardous or harmful alcohol consumption.

Subgroup analysis and investigation of heterogeneity

Applicability issues ‐ population characteristics, setting and mode of intervention

Variability in treatment exposure, control condition, year of publication, baseline consumption and follow up time scales

Effectiveness and efficacy

1. Effectiveness ‐ efficacy scale.

Sensitivity analysis

Results

Description of studies

Results of the search

1.

Included studies

Population

Setting

Interventions

Brief intervention

Extended interventions

Total treatment exposure

Control group content

Efficacy/effectiveness scores

2. Conversion factors for alcohol consumption and efficacy scores of trials.

Reporting of outcomes

Quantity of alcohol consumed in a specified time period

Frequency of drinking (number of drinking sessions in a specified time period)

Intensity of drinking (amount of alcohol consumed in a drinking session)

Methods	Parallel RCT: motivational interview (extended) versus brief intervention versus standard care; only brief intervention and standard care groups used because the motivational interview was not comparable to other trials, and very few participants received the intervention. ITT: yes.
Participants	Setting: Australia, emergency department. Participants: Aged ≥ 18 years presenting to emergency department for any reason; excluded if inability or refusal to give informed consent (including lack of communication in English, intoxication preventing coherent answers to questions), or previous enrolment and inability to be followed up (e.g. travellers, no phone contact). Number randomised = 468; 78% male; mean age of total sample not reported. Baseline data not reported for total sample.
Interventions	Participants were screened with PAT (Paddington Alcohol Test), and this served as baseline data. BI group (N = 159) received a semi scripted informative discussion from a trained nurse or doctor, median five minutes (range 1 to 30 minutes), including the risks of high‐risk drinking, a definition of a standard drink, advice on safe drinking, tips on cutting down and incorporated empathy and encouragement. A purpose‐designed pamphlet reinforcing the information discussed was also provided. SC group (N = 161) received standard care, i.e. no counselling unless the clinician would have accessed addiction services as part of clinical duty or the participant requested counselling. MI group (N = 148) received an appointment at a specialist drug and alcohol centre. A separate control group (not randomised) was comprised of non‐consenting participants who screened positive.
Outcomes	Self‐reported most drinks in a day, number of participants exceeding daily alcohol limit once per week, emergency department attendances. Assessed at 1 and 3 months.
Funding source	An Alcohol Education and Rehabilitation Foundation Grant supported the present study.
Declaration of interests	Not reported.
Notes	Loss to follow‐up: BI group: 83/159 (52%). SC group: 95/161 (59%).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer generated block‐randomisation process (p. 122).
Allocation concealment (selection bias)	Low risk	Sequentially‐numbered, opaque, sealed envelope (p. 122).
Blinding of treatment providers	High risk	"Group allocation was not blinded to the participant, clinicians performing interventions or the researchers conducting follow‐up interviews" (p. 123).
Blinding of participants	High risk	"Group allocation was not blinded to the participant, clinicians performing interventions or the researchers conducting follow‐up interviews" (p. 123).
Blinding of outcome assessment (detection bias)   All outcomes	High risk	"Group allocation was not blinded to the participant, clinicians performing interventions or the researchers conducting follow‐up interviews" (p. 123).
Incomplete outcome data (attrition bias)   All outcomes	High risk	Loss to follow‐up > 30%.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods are reported.

Methods	Parallel RCT: stepped care brief intervention versus brief advice. ITT: yes.
Participants	Setting: Wales, primary care practices. Participants: Aged ≥ 18 years presenting to primary care clinic and screening AUDIT ≥ 8, may also have diagnosis of alcohol use disorder, or be drinking above Department of Health guidelines (> 21 units per week or > 8 units per day), must live within commuting distance of PC practice, and not have received treatment for alcohol use disorder in previous 180 days; excluded if primary drug dependence other than alcohol, or severe mental or physical illness, or severe cognitive impairment or legal issues that might interfere with follow‐up. Number randomised = 112; 100% male; mean age = 42.7 years. At baseline: mean AUDIT score = 13.7.
Interventions	Stepped Care group (N = 54) received: Step 1. 40‐minute session of behaviour change counselling from a trained practice nurse. Each participant was invited for follow‐up with the same nurse 28 days after the initial session. Participants who had consumed > 21 units of alcohol in any one week or > 10 units in any one day during the 28‐day period were referred to step 2. Step 2. All participants referred to step 2 were offered a maximum of four 50‐minute sessions of motivational enhancement therapy conducted by a trained alcohol counsellor. Sessions were held at the primary care practice and scheduled for one per week over a four week period. Each participant was invited for a follow‐up with the practice nurse 28 days after the final session. Participants who had consumed > 21 units in any one week or > 10 units in any one day in the 28‐day period were referred to step 3. Step 3. Participants were referred to the local community alcohol team for specialist intervention. There was no limit on duration or intensity of treatment, which could encompass, where necessary, detoxification, inpatient treatment, outpatient counselling, relapse prevention and drug therapy. The clinical protocol included a caveat that any participant in the intervention group who needed urgent referral to step 3 at any stage could be referred without going through intermediate care steps. Minimal intervention group (N = 58) received a five minute directive advice session from a practice nurse addressing the need for them to reduce their alcohol consumption. Participants also received a short self‐help booklet outlining the consequences of excessive alcohol consumption and including details on where to seek help locally for alcohol problems.
Outcomes	Total drinks in previous 180 days, mean drinks per drinking day, per cent days abstinent, alcohol problems questionnaire (APQ), quality of life (SF‐12), health utility (EQ‐5D), mean costs. Assessed at 6 months.
Funding source	The study was funded by the Wales Office for Research and Development. All of the authors are independent of the funding body.
Declaration of interests	Declaration of interest: None.
Notes	Loss to follow‐up: Stepped Care: 15/54 (28%) (52 received step 1, 17 received step 2, 1 received step 3). Minimal intervention: 6/58 (10%) (all received intervention).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomisation was conducted remote from the research centres by the trial randomisation service at the University of York" (p. 449).
Allocation concealment (selection bias)	Unclear risk	"Block randomisation with seeded random elements was used to minimise the possibility of subversion" (p. 449).
Blinding of treatment providers	High risk	Treatment providers not blinded.
Blinding of participants	Low risk	Participants completed "an AUDIT questionnaire embedded within a general lifestyle questionnaire" (p. 449).
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	"Follow‐up was conducted by a researcher masked to the allocated group" (p. 449).
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Loss to follow‐up is reported with reasons. Loss to follow‐up is 10% versus 28% but stepped care design means that not all participants received all steps.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods are reported.

Methods	Cluster RCT: patient information leaflet versus brief advice versus brief lifestyle counselling, average cluster size = 135. ITT: yes.
Participants	Setting: England, emergency departments. Participants: Aged ≥ 18, who were screened positive on an alcohol screening test, sufficiently alert and orientated to provide informed consent, living within the catchment area of the ED, and being able to speak, read or write English sufficiently well to complete study questionnaires; excluded if already seeking alcohol treatment, or participating in another study of alcohol interventions, or having sustained a severe injury, or suffering from a serious mental health problem, or grossly intoxicated, or being of no fixed abode; screened by one of three alcohol screening tools: the modified Single Alcohol Screening Question (M‐SASQ), or the FAST Alcohol Screening Test, or a modified version of Paddington Alcohol Test (SIPS‐PAT) (EDs were randomly assigned to one of the three screening approaches). Number randomised: 1204; 65% male; mean age = 34.6 years, 88% white, 65% continued with education after the age of 16 years, over 45% current smokers. At baseline: mean AUDIT score = 12.4 (SD 6.9).
Interventions	Brief Lifestyle Counseling (BLC) group (N = 395) received the patient information leaflet (PIL) and brief advice (BA) from emergency department staff. The following day (or as soon as possible thereafter), they received 20 minutes of lifestyle counselling alcohol intervention based on the 'How much is too much?' intervention pack used for the PIL group, delivered by Alcohol Health Workers (AHW) with specialist training and experience in alcohol motivational interventions. B) group (N = 403) received five minutes of brief advice about drinking using the SIPS brief advice tool ('Brief Advice About Alcohol Risk') developed for the trial and based on the 'How much is too much?' intervention pack. Following brief advice, the PIL was delivered in the same manner as in the minimal intervention group. PIL group (N = 406) received simple clinical feedback using a standard script that their test result indicated they were drinking above the government's 'safe' drinking levels, and were given a PIL ‐ the Department of Health's 'Drinking and You: How Much is Too Much?' leaflet, including information on local alcohol services where further help could be sought by the patient themselves.
Outcomes	AUDIT status (score of < 8 versus ≥ 8) on the extended item AUDIT questionnaire; average number of drinks per day using the quantity‐frequency questions of the extended AUDIT, alcohol related problems using the Alcohol Problems.  Alcohol Screening and Brief Intervention in Emergency Care Questionnaire (APQ), readiness to change using a modified Readiness Ruler, patient satisfaction using a modified version of the Patient Satisfaction Questionnaire measured (at 12 months only). Assessed at 6 and 12 months.
Funding source	The study was funded by the UK Department of Health. The Department of Health, proposed the general study design, but the details of the methodology were determined by the study team. The sponsor had no role in data collection, data analysis, interpretation of the results or writing the report. The corresponding author had full access to the data and had final responsibility for the decision to submit for publication.
Declaration of interests	Competing interests: all authors had financial support from the Department of Health in England (Alcohol Policy Unit) for the submitted work; no financial relationships with any organisations that might have an interest in the submitted work in the previous three years; no other relationships or activities that could appear to have influenced the submitted work.
Notes	Loss to follow‐up at 6 months: Patient information leaflet: 114/406 (28%). Brief advice: 99/403 (25%). Brief lifestyle counselling: 128/395 (32%). Loss to follow‐up at 12 months: Patient information leaflet: 142/406 (35%). Brief advice: 108/403 (27%). Brief lifestyle counselling: 144/395 (36%). Payment: GBP 10 voucher for baseline assessment.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomisation was conducted using a remote secure randomisation service (p. 3).
Allocation concealment (selection bias)	Low risk	Cluster randomised trial.
Blinding of treatment providers	Low risk	Treatment providers not blinded but cluster randomisation means little chance of contamination.
Blinding of participants	High risk	Participants were not blinded.
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	"Researchers conducting 6 and 12 month follow up were blinded to the participants allocated treatment condition and efforts were made to prevent participants from inadvertently revealing the intervention they received" (p. 3).
Incomplete outcome data (attrition bias)   All outcomes	High risk	Loss to follow‐up > 30%.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods were reported.

Methods	Parallel RCT: SBIRT versus usual care. ITT: no.
Participants	Setting: USA; emergency department. Participants: ≥ 18 years presenting to emergency department with non life‐threatening injuries; excluded due to age, language, level of consciousness or acuity; screened with NIAA quantity/frequency questions and CAGE.  Number randomised = 94; 60% male; mean age 39 years, age range 19 to 69 years; 77% White, 13% African American, 9% Hispanic, 1% Other ethnicity; education and employment status not reported.  Baseline data not reported for total sample.
Interventions	SBIRT group (N = 49) received one session of 5 to 10 minutes of motivational counselling from an emergency department staff nurse, plus educational brochures and a list of community resources where further intervention and treatment could be sought, plus special discharge instructions that provided resources for follow‐up care and community services appropriate to their alcohol use risk level.  Control group (N = 42) received usual care.
Outcomes	Mean number of drinks per week, mean drinking days per week, per cent with recurring emergency department visits. Assessed at 3 months.
Funding source	This work was supported by a grant from the National Highway Traffic Administration.
Declaration of interests	Not reported.
Notes	Loss to follow‐up: SBIRT group: 23/49 (47%).  Control group: 22/42 (52%).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Before the start of the study, subject numbers were computer generated in blocks of 100" (p. 541).
Allocation concealment (selection bias)	Unclear risk	"Before the start of the study, subject numbers were... placed in sequentially numbered sealed enrolment packets" (p. 541).
Blinding of treatment providers	Low risk	"Nurses... were blinded to the group assignment for all study subjects" (p. 541).
Blinding of participants	High risk	Participants were not blinded, screening questions concern alcohol consumption (National Institute on Alcohol Abuse and Alcoholism quantity and frequency questions and the CAGE questionnaire).
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Not reported.
Incomplete outcome data (attrition bias)   All outcomes	High risk	Loss to follow‐up was high compared to other studies, although participants commonly considered 'hard‐to‐reach' (such as homeless or non‐local people) were not excluded.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods were reported.

Methods	Parallel RCT: brief intervention versus control.  ITT: unclear.
Participants	Setting: Spain, recruited from university hospital or urban general practice or rural general practice or industrial occupational health clinic. Participants: men between 18 to 65 years with a weekly alcohol consumption of 21 to 95 units (1 unit = 8 g); screened by an evaluation survey with drinking questions embedded with general health questions; excluded if alcohol dependence (defined as alcohol consumption > 95 units per week) or psychiatric disorders.  Number randomised = 1022; 100% male; mean age 42.4 years (of those evaluated not randomised).  At baseline: mean weekly consumption = 47.1 units; % risk drinkers (> 35 units/wk) = 62%.
Interventions	Intervention group (N = 592) received the evaluation survey (10 minutes) plus a self‐help manual containing methods to evaluate their drinking and its effects on their lifestyle, and guidelines for consumption, with an extra 10 minutes of advice and explanation of the manual  Control group (N = 430) received the evaluation survey only (10 minutes), with no comment or advice.
Outcomes	Weekly alcohol consumption; per cent at‐risk drinkers (> 35 units per week); all reported by setting at six months for the university hospital, urban general practice, and rural general practice, and 12 months for the industrial occupational health clinic.
Funding source	Not reported.
Declaration of interests	Not reported.
Notes	Loss to follow‐up:  Intervention group: 111/592 (19%).  Control group: 84/430 (20%).  Extracted data for C Urbano and C Rural groups only, others were not based in primary health care.  Loss to follow‐up for these two groups:  Intervention group: 49/255 (19%).  Control group: 43/229 (19%).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Random allocation was performed by order of entering the study, controlling by age (p. 17).
Allocation concealment (selection bias)	Unclear risk	Not reported.
Blinding of treatment providers	High risk	Not blinded; no reported attempt to reduce bias.
Blinding of participants	High risk	Not blinded; no reported attempt to reduce bias.
Blinding of outcome assessment (detection bias)   All outcomes	Unclear risk	Not reported.
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Loss to follow‐up is reported with reasons.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods are reported.

Methods	Cluster RCT: brief intervention plus telephone boosters versus usual care. ITT: yes.
Participants	Setting: USA, primary care clinic. Participants: aged > 60 years invited by letter, who had consumed at least one alcoholic beverage in the past three months and was physically and cognitively well enough to participate; excluded if severe cognitive impairment, or terminally ill or deceased, or residing in a skilled nursing facility, or moving out of the area in the next year, or did not speak English, or no longer a patient of the physician, or did not drink alcohol; patients could also be excluded at the discretion of the physician; screened with CARET (an updated and revised version of the Alcohol‐Related Problems Survey). Number randomised = 1186; 66% male; 96% White, < 1% Black, < 1% Asian/Pacific, 1.4% American Indian; 3% less than high school, 10% high school graduate, 27% some college, 25% college graduate, 35% graduate degree. At baseline: mean drinks per week = 13.6 (SD 8.0).
Interventions	Intervention group (N = 546) received a mailed personalised patient report, an educational booklet on alcohol and aging, a drinking diary to track alcohol consumption, and, depending on the individual patient's reported alcohol‐associated risks (as identified on the CARET screening tool), up to 13 'tip sheets' (e.g. on drinking sensibly, sleep, preventing falls and fractures, gout, etc.). Their physician received a copy of the patient report attached to the front of the patient's notes, and used it in the patient's appointment to discuss the patient's drinking and associated risks, and to advice them to reduce consumption if appropriate. Patients then received telephone calls from a health educator who answered questions about the written materials and engaged in the following five steps: (a) assessment and direct feedback; (b) negotiation and goal setting; (c) behavioural modification techniques; (d) self‐help‐directed bibliotherapy, and; (e) follow‐up and reinforcement. Control group (N = 640) received usual care, which could include alcohol‐related advice.
Outcomes	At‐risk drinking, drinks per week, recollection of alcohol‐related discussions, health care utilisation, costs.
Funding source	This project was funded by National Institute on Alcohol Abuse and Alcoholism (NIAAA) Grant 1RO1AA013990 (Principal investigator: Susan L. Ettner). Alison A. Moore’s time was additionally supported by NIAAA Grants R01 AA15957 and K24 AA15957 (Principal investigator: Alison A. Moore). O. Kenrik Duru’s time was supported by National Institute on Aging Grants 5P30AG021684‐12 and 5K08AG033630‐05.
Declaration of interests	Conflict of interest: all authors declared no potential conflicts of interest.
Notes	Project SHARE. Loss to follow‐up at 12 months: Intervention group: 107/546 (20%). Control group: 30/640 (5%).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"We aimed to match each participating physician to another with the same specialty and clinic site. Physicians from each of the matched pairs were randomly assigned by a statistician who drew random numbers from a uniform [0,1] distribution for the pair; the physician having the lower number was assigned to the intervention group. The patient’s treatment assignment was then based on the random assignment of the patient’s primary care physician" (p. 448).
Allocation concealment (selection bias)	Low risk	Cluster randomised.
Blinding of treatment providers	Low risk	"Randomisation at the level of participating physicians was used so that physicians would not see patients in both the intervention and control arms of the study, thereby avoiding contamination effects" (p. 448).
Blinding of participants	High risk	Not blinded; no reported attempt to reduce bias.
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	"Research assistants, blinded to treatment allocation, entered all data collected" (p. 448).
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Loss to follow‐up is reported with reasons, but big differential between arms.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods were reported.

Methods	Parallel RCT: brief motivational intervention versus treatment as usual + assessment. ITT: unclear.
Participants	Setting: USA, trauma unit. Participants: Aged ≥ 18 years who presented with an injury, demonstrated orientation to person, place and time, and identified themselves as white, Hispanic or black, screened with stepwise procedure: (i) clinical indication of acute intoxication or alcohol use or positive blood alcohol concentration; (ii) self‐reported drinking six hours prior to injury; (iii) at‐risk drinking as per NIAAA guidelines (7 drinks per week for women, 14 drinks per week for men; > 4 drinks per day in men, > 3 drinks per day in women); (iv) positive on CAGE questionnaire, 40 minute assessment procedure; excluded if they spoke neither English nor Spanish, or no identifiable residence, or under arrest or in police custody at admission, or actively suicidal or psychotic, or victims of sexual assault, or their medical condition precluded face‐to‐face interview, or Glasgow Coma Score ≥ 14 that did not resolve prior to discharge. Number randomised = 1493; 82% male; mean age of total sample not reported; 45% White, 19% Black, 36% Hispanic.
Interventions	BMI group (N = 737) received brief motivational intervention lasting 15 minutes from a trained health educator; the primary components consist of acknowledging the patient's responsibility for changing drinking, encouraging the patient to explore the pros and cons of drinking, assessing importance, confidence and readiness to change drinking behaviour, reinforcing the patient’s sense of self‐efficacy and providing support for any efforts or intention to quit drinking or reduce harm associated with drinking, including injury. TAU + group (N = 756) received treatment as usual, which included referral to a drug and alcohol counsellor or other appropriate hospital services, and a patient handout regarding the effect of alcohol.
Outcomes	Volume consumed per week, maximum amount consumed per occasion, percentage days abstinent, percentage days heavy drinking. Assessed at 6 and 12 months.
Funding source	This work was supported by a grant (R01 013824; principle investigator: R. Caetano) from the National Institute on Alcohol Abuse and Alcoholism to the University of Texas School of Public Health. The lead author would like to acknowledge the support of the NIH Health Disparities Loan Repayment Program funded by the National Center of Minority Health and Health Disparities.
Declaration of interests	Declarations of interest: none.
Notes	MARIA study USD 25 for baseline assessment and USD 50 for 6 month and 12 month follow‐up assessments Loss to follow‐up: BMI: 317/737 (43.0%). TAU+: 309/756 (40.9%).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	"Randomised… using a permuted block design (block size 6) to ensure approximately equal distribution of patients according to their race/ethnicity" (Field 2010, p. 65).
Allocation concealment (selection bias)	Low risk	"Study clinicians were blinded to patient randomisation prior to completion of baseline assessment. Treatment assignment was generated offsite and was provided to study clinicians in sealed opaque envelopes" (Roudsari 2009, p. 287).
Blinding of treatment providers	High risk	Although study clinicians were blinded to patient randomisation prior to completion of the baseline assessment, they then had contact with both arms (Field 2010, p. 65).
Blinding of participants	High risk	No reported attempt to blind participants.
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	"Research staff blind to treatment assignment conducted follow‐up assessments" (Field 2010, p. 66).
Incomplete outcome data (attrition bias)   All outcomes	High risk	Loss to follow‐up > 30%.
Selective reporting (reporting bias)	High risk	Authors specified typical quantity consumed as a primary outcome but did not report these data.

Methods	Parallel RCT: brief physician advice versus health booklet control. ITT: only in 1 sub analysis paper.
Participants	Setting: USA. Participants: patients with regularly scheduled physician appointments, aged 18 to 65 years, consuming > 14 drinks per week for men and > 11 drinks per week (1 drink = 8 g) for women; screened by health screening survey provided by reception personnel containing four sets of parallel questions on exercise, smoking, weight and alcohol use; excluded if pregnant, under 18, over 65, had attended an alcohol treatment program in previous year, reported symptoms of alcohol withdrawal in previous year, advice in previous three months from GP to change alcohol use, consumption > 50 drinks per week, reported symptoms of suicide.  Number randomised = 774; (following numbers are for those assessed): 62% male; 92% White, 1% Hispanic, 4% African American, 3% other; 41% high school or less, 40% some college, 20% college degree or more.  At baseline: mean consumption in previous seven days = 19.0 drinks; number of binge drinking episodes (defined as > 5 drinks for men or > 4 drinks for women on one occasion) in previous 30 days = 5.5; binge drinkers in previous 30 days = 78%; excessive drinkers (defined as > 20 drinks per week for men or > 13 drinks per week for women) in previous 7 days = 43%.
Interventions	Intervention group (N = 392) received booklet on general health issues, and scheduled to see their personal physician for two 15 minute advice sessions (one brief intervention and one reinforcement session) one month apart, and a workbook containing feedback regarding current health behaviours, review of prevalence of problem drinking, adverse effects of alcohol, worksheet on drinking cues, drinking agreement and diary cards (based on MRC trial). Participants received a follow‐up telephone call from the clinic nurse two weeks after each meeting with GP.  Control group (N = 382) received a health booklet on general health issues, instructed to address health concerns in their usual manner.
Outcomes	Mean drinks in previous seven days; binge drinking (defined as > 5 drinks for men or > 4 drinks for women on one occasion); excessive drinking (defined as > 20 drinks per week for men or > 13 drinks per week for women); assessed at 6 and 12 months (further paper gives 48 month data); health care utilisation (hospital days and emergency department visits), changes in health status (smoking status, depression, motor vehicle crashes, unintentional injuries).
Funding source	This work was supported by grant AA 08512‐01A from the National Institutes of Health, National Institute on Alcohol Abuse and Alcoholism, and was assisted by the Wisconsin Research Network, Madison, and the Wisconsin Institute of Family Medicine, Milwaukee.
Declaration of interests	Not reported.
Notes	TrEAT trial Participants paid USD 50, practices paid USD 300.  Loss to follow‐up:  Intervention group = 39/392 (10%).  Control group = 12/382 (3%).  Separate papers on 48 month data; cost‐benefit analysis; subgroup analysis of women of childbearing age; subgroup analysis of young adults.
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Computer generated random sequence (p. 1040).
Allocation concealment (selection bias)	Low risk	Physicians were not told which of their patients were in control group.
Blinding of treatment providers	Low risk	Researchers carried out baseline assessment; physicians (treatment providers) only saw the intervention participants and were not told which patients were in the control group, therefore little chance of contamination.
Blinding of participants	Low risk	"The survey was designed as a general lifestyle questionnaire to... minimise the intervention effect of the alcohol questions" (p. 1040).
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Follow‐up included telephone interview by researcher not assigned to subject's clinic (p. 1040).
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Loss to follow‐up 10% in intervention group and 3% in control group.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods were reported.

Methods	Parallel RCT: physician delivered counselling versus health booklet. ITT: yes.
Participants	Setting: USA, primary care clinic. Participants: patients > 65 years with regularly scheduled appointments at community based primary health care clinics; consuming > 11 drinks for men or > 8 drinks (1 drink = 8 g) per week for women, or ≥ 2 positive responses to CAGE, or binge drinking (defined as ≥ 4 drinks per occasion for men two or more times in the last three months, or ≥ 3 drinks per occasion for women); screened using modified Health Screening Survey; baseline assessment 30 minute face‐to‐face interview in primary care clinic conducted by trained researcher; excluded if attendance at an alcohol treatment programme or reported symptoms of alcohol withdrawal in the last year, or physician advice received in previous three months to change alcohol use, or consumption > 50 drinks per week, or reported thoughts of suicide.  Number randomised = 158; 66% male; age range = 65 to 75 years; 26% college educated  At baseline: mean weekly alcohol consumption = 16.0 drinks; mean binge drinking episodes (defined as ≥ 4 drinks per occasion for men or ≥ 3 for women) in previous 30 days = 3.7; binge drinkers in previous 30 days = 44.9%; excessive drinkers (defined as > 20 drinks per week for men and > 13 for women) in previous seven days = 29.7%.
Interventions	Intervention group (N = 87) received general health booklet, scheduled to see their personal physicians for two 10 to 15 minute sessions one month apart (1 brief intervention, 1 reinforcement session) received follow‐up phone call from nurse two weeks after each session and a workbook containing feedback on patients health behaviours, review of problem drinking prevalence, reasons for drinking, adverse consequences of alcohol, drinking agreement, drink diary cards. Used same protocols as Medical Research Council trial and Project TrEAT  Control group (N = 71) received a general health booklet only.
Outcomes	Mean drinks in previous seven days; binge drinking episodes (defined as > 4 drinks per occasion for men or > 3 for women) in previous 30 days; percentage of participants binge drinking in previous 30 days; percentage of participants drinking excessively (defined as > 20 drinks per week for men and > 13 for women) in previous seven days. Assessed at 3, 6, 12 months.
Funding source	This work was supported by NIAA grant 3R01 AA08512‐0351, the American Academy of Family Physicians, and the Dean Foundation for Health Research and Education.
Declaration of interests	Not reported.
Notes	GOAL trial Physicians paid USD 250.  Loss to follow‐up:  Intervention group: 9/87 (9%).  Control group: 4/71 (6%).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Not reported.
Allocation concealment (selection bias)	Low risk	Physicians were not told which of their patients were in control group (p. 379).
Blinding of treatment providers	Low risk	Researchers carried out baseline assessment; physicians (treatment providers) only saw the intervention participants and were not told which patients were in the control group, therefore little chance of contamination.
Blinding of participants	Low risk	Patients completed a modified Health Screening Survey with alcohol questions embedded.
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Follow‐up telephone interviews done by physician not linked to patient's clinic (p. 379).
Incomplete outcome data (attrition bias)   All outcomes	Low risk	Loss to follow‐up is reported with reasons.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods are reported.

Methods	Parallel RCT: brief primary care clinician advice versus usual care. ITT: yes.
Participants	Setting: USA, primary care clinics. Participants: Aged 30 to 60 years who were on medication for diabetes or hypertension with CDT level > 2.5%, or ≥ 50 drinks in the previous 30 days for men or ≥ 30 for women; identified by patient database search, screened by medical records review and telephone interview; baseline interview conducted at clinic participating in the study excluded if current symptoms of alcohol withdrawal, or participation in alcohol treatment programme in previous 12 months; recruited from existing study of %CDT test in 8 primary care clinics.  Number randomised = 151; 45% male; mean age = 48.7 years; 88% white, 8% black, 4% other; high school or less = 41%, college degree = 21%, technical degree = 18%, advanced degree = 19%.  At baseline: current alcohol abuse = 7.3%, current alcohol dependence = 9.9%, mean drinks in previous 30 days = 33.2, percentage heavy drinkers (defined as ≥ 30 drinks in previous 30 days for men or ≥ 25 for women) = 39.1%, mean frequency of binge drinking (defined as ≥ 5 drinks in one occasion for men or ≥ 4 for women) in previous 30 days = 2.6.
Interventions	Intervention group (N = 81) received two 15 minute sessions (one month apart) from nurse practitioners or physician assistants and two five minute follow‐up phone calls from the office nurse. Sessions (based on TrEAT) followed a scripted workbook reviewing prevalence of problem drinking, adverse effects of alcohol, %CDT test result, drinking diary cards and a drinking agreement in the form of a prescription.  Control group (N = 70) received a general health booklet and were told by the researcher to contact the physician with any health concerns.
Outcomes	Mean percentage of heavy drinkers, mean drinks in previous 30 days, mean frequency of binge drinking, proportion of subjects who reduced %CDT. Assessed at two months, four months and 12 months with change scores reported.
Funding source	This study was funded by Axis Shield (CT‐C0003), by the National Institute on Alcohol Abuse and Alcoholism (R01 AA8512‐03) and by grants from the Department of Family Medicine, University of Wisconsin. [Axis Shield: "A pioneering organisation focused on the development and manufacturing of important and innovative in vitro diagnostic tests for use in clinical laboratories"].
Declaration of interests	Not reported.
Notes	Loss to follow‐up at 12 months:  Intervention group: 11/92 (12%).  Control group: 5/75 (7%).
*Risk of bias*
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Method of sequence generation not reported.
Allocation concealment (selection bias)	Unclear risk	Allocation concealment not reported.
Blinding of treatment providers	Low risk	Nurse practitioners or physician assistants carried out the brief intervention but had no contact with control patients, so little chance of contamination.
Blinding of participants	High risk	Participants not blinded (p. 632).
Blinding of outcome assessment (detection bias)   All outcomes	Low risk	Data were collected by a researcher blind to group assignment (p. 632).
Incomplete outcome data (attrition bias)   All outcomes	Unclear risk	Loss to follow‐up is reported but not reasons.
Selective reporting (reporting bias)	Low risk	Outcomes specified in methods are reported.