Effectiveness of brief alcohol interventions in primary care populations

In the paper

“Incomplete outcome data (attrition bias) All outcomes High risk”

What the sources say

• Disputed

  The Limitations section acknowledges missing data at 6 and 12 months but says the mixed-model analyses were designed to handle them with little bias. The Methods section likewise describes including participants with missing data; the paper does not judge attrition bias to be high risk for all outcomes. No page numbers were available in our copy. (LLM, conf=0.84)

  Checked by a reasoning model — may be wrong.

  cites D'Onofrio 2012 sourceExperimental — accuracy not yet measured

In the paper

“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).”

What the sources say

• Supported

  The Introduction describes primary healthcare as an ideal setting for early detection and secondary prevention of alcohol-related problems, and screening and brief intervention as a preventative approach aimed at changing drinking behaviour. It also describes years of efforts to promote brief interventions and an extensive primary-healthcare research literature. No page number is available in our copy. (LLM, conf=0.92)

  cites O'Donnell 2014 source

In the paper

“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.”

What the sources say

• Supported

  The Conclusions/Significance state that answering questions about drinking in brief-intervention trials appears to alter subsequent self-reported behaviour, including in non-intervention control groups. The Discussion describes small, somewhat equivocal assessment effects, consistent with the claim’s qualified 'may influence' wording. (LLM, conf=0.92)

  cites McCambridge 2011 source

In the paper

“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).”

What the sources say

• Partly supported

  The Alcohol section (page not available) reports a substantial health burden from alcohol use, and the Limitations section says estimates of alcohol consumption rely on limited data and carry significant uncertainty. These passages support the broad point that alcohol’s impact is difficult to estimate, but they do not establish the claim’s stated reason—that difficulty arises from its many effects—or clearly link alcohol use to the listed increases in violence, accidents, and suicide. (LLM, conf=0.82)

  cites GBD 2017 sourceExperimental — accuracy not yet measured

In the paper

“the concepts of screening reactivity and assessment reactivity have been defined (McCambridge 2011).”

What the sources say

• Partly supported

  The Discussion describes assessment reactivity as a possible effect of answering research-assessment questions on drinking behaviour. The Results distinguish minimal screening from longer assessments and from no screening, but the paper does not define screening reactivity as a separate concept. No page numbers are available in our copy. (LLM, conf=0.78)

  cites McCambridge 2011 sourceExperimental — accuracy not yet measured

In the paper

“We included 14 trials that reported baseline measures of frequency of drinking which could be converted to days drinking per week; the mean value was 2.07 days/week (Aalto 2000;Bernstein 2010;Cherpitel 2009;Daeppen 2007;Fleming 1997;Fleming 2010;Helstrom 2014;Monti 2007;Noknoy 2010;Rubio 2010;Schaus 2009;Senft 1997;Soderstrom 2007;Spirito 2004).”

What the sources say

• ?Not checked — no full text of the sourcecites Aalto 2000 source
• Supported

  Bernstein 2010 is cited as one of the trials contributing baseline drinking-frequency data, not as the source of the 14-trial mean. Table 2 reports baseline drinking days per month for its study groups, a measure convertible to days per week. The 2.07 days/week figure is the citing authors’ aggregate across trials. (LLM, conf=0.91)

  cites Bernstein 2010 source
• ?Not checked — no full text of the sourcecites Cherpitel 2009 source

In the paper

“One trial reported mean corpuscular volume (MCV) for this comparison (Seppa 1992).”

What the sources say

• Supported

  The Outcome section’s Table II reports mean MCV values for the intervention and control groups at the start and at 12 months. The Results section also compares MCV values between the groups, so this trial reported MCV for the comparison. (LLM, conf=0.94)

  cites Seppa 1992 source

In the paper

“A further six studies reported the change between baseline and the end of follow-up (change score) in the quantity of alcohol consumed in a specified time period and the corresponding standard deviation (Bischof 2008;Fernández 1997;Field 2010;Huas 2002;Ockene 1999;Romelsjö 1989).”

What the sources say

• ?Not checked — no full text of the sourcecites Bischof 2008 source
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Fernández 1997
• Supported

  The Results (Volume per Week) report changes from baseline to the 12-month follow-up in standard drinks consumed per week, with corresponding standard deviations. This confirms the claim for Field 2010, one of the six studies cited; it does not assess the other five. (LLM, conf=0.93)

  cites Field 2010 source

In the paper

“Frequency of drinking in terms of number of days drinking each week or each month was reported by 11 trials at 6 or 12 months (Bernstein 2010;Cherpitel 2009;Crawford 2004;Curry 2003;Daeppen 2007;Field 2010;Fleming 2010;Helstrom 2014;Maisto 2001;Senft 1997;Spirito 2004).”

What the sources say

• Supported

  Table 6 (section labeled CONCLUSIONS; page not provided) reports drinking days per month at the 12-month follow-up. This supports citing Bernstein 2010 as one of the trials that reported the stated frequency measure. (LLM, conf=0.92)

  cites Bernstein 2010 source
• ?Not checked — no full text of the sourcecites Cherpitel 2009 source
• Partly supported

  The Sample size and data analysis section says the trial measured percentage of days abstinent over a 13-week period at both 6 and 12 months; the outcomes table reports that measure. This captures drinking frequency indirectly, but the paper does not report it as the number of days drinking each week or month. As one of the cited trials, it also cannot establish the citing sentence’s total of 11 trials. (LLM, conf=0.82)

  cites Crawford 2004 sourceExperimental — accuracy not yet measured

In the paper

“Of these, 27 studies reported quantity and corresponding standard deviation at 12 months (Altisent 1997;Beich 2007;Bernstein 2010;Blow 2006;Córdoba 1998;Crawford 2004;Daeppen 2007;D'Onofrio 2008;D'Onofrio 2012;Drummond 2014;Fleming 1997;Fleming 1999;Fleming 2004;Fleming 2010;Helstrom 2014;Kaner 2013;Lock 2006;Maisto 2001;Richmond 1995;Rubio 2010;Schaus 2009;Scott 1990;Senft 1997;Soderstrom 2007;”

What the sources say

• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Altisent 1997
• Supported

  The paper reports 12-month alcohol-consumption quantities with corresponding standard deviations in a table on p. 6, so it qualifies as a study reporting those data. The total of 27 studies is the citing paper’s count, not a claim that Beich et al. counted 27 studies. (LLM, conf=0.90)

  cites Beich 2007 source
• Supported

  Table 7 (indexed under CONCLUSIONS) reports 12-month alcohol-consumption quantities with corresponding standard deviations, including drinking days and mean drinks per week. This supports citing Bernstein 2010 as one of the studies providing those data; the count of 27 studies is the citing paper’s own tally. (LLM, conf=0.92)

  cites Bernstein 2010 source

In the paper

“A measure of alcohol-related consequences or harm (e.g.Drinker Inventory of Consequences -DrInC; Alcohol Problems Quesionnaire -APQ) was reported by 21 studies (Blow 2006;Cherpitel 2009;D'Onofrio 2012;Drummond 2009;Drummond 2014;Fleming 2010;Heather 1987;Helstrom 2014;Kaner 2013;Lane 2008;Longabaugh 2001;McDevitt-Murphy 2014;McIntosh 1997;Mello 2008;Monti 2007;Romelsjö 1989;Schaus 2009;Spirito 200”

What the sources say

• ?Not checked — no full text of the sourcecites Blow 2006 source
• ?Not checked — no full text of the sourcecites Cherpitel 2009 source
• Supported

  The Methods (Primary and Secondary Outcomes; Measurements) identify the Short Inventory of Problems (SIP) as a collected secondary outcome. The Results (Negative Behaviors and Consequences) report SIP scores, confirming that this paper is one of the studies reporting a measure of alcohol-related consequences. The paper does not establish the citing review’s total of 21 studies, but that total is not attributed to this individual study. (LLM, conf=0.92)

  cites D'Onofrio 2012 source

In the paper

“Two trials were not included in this subgroup analysis because they contained arms that included both intervention types (Drummond 2014;Kaner 2013).”

What the sources say

• Supported

  The Methods > Procedures passage describes intervention arms that combined components: brief advice was followed by an information leaflet, while the lifestyle-counseling arm first received both brief advice and the leaflet. This supports citing Drummond 2014 as a trial with combined intervention arms; the decision to exclude it from the subgroup analysis belongs to the citing authors, not to Drummond et al. (LLM, conf=0.85)

  cites Drummond 2014 source
• ?Not matched

  Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  cites Kaner 2013

In the paper

“Therefore, it is possible that screening alone might represent an impetus for change (McCambridge 2011).”

What the sources say

• Supported

  The Introduction (page unavailable) says that reporting on one’s own behaviour may induce reflection and actual change. The Discussion (page unavailable) describes a possible effect of answering alcohol-related questions without a behaviour-change intervention, supporting the hedged suggestion that screening alone might prompt change. (LLM, conf=0.87)

  cites McCambridge 2011 source

In the paper

“Allocation concealment (selection bias) Low risk Screening questionnaires went into a sealed envelope which had a symbol on the front which had to be uncovered scratch card style (p.594)”

What the sources say

• Supported

  The Protocol (p. 2) says participants sealed their completed screening questionnaires in envelopes before staff uncovered a symbol on each envelope to assign them to an arm. This supports the cited description of the allocation-concealment procedure; “low risk” is the citing paper’s assessment of that procedure. (LLM, conf=0.90)

  cites Beich 2007 source

In the paper

“Random sequence generation (selection bias) Unclear risk Sequence generation not described.”

What the sources say

• Supported

  The Protocol (p. 2) says staff assigned participants by uncovering a symbol on a sealed envelope, but does not describe how the assignment sequence was generated. That supports the citing paper’s assessment of unclear risk for random sequence generation; the risk rating is the citing authors’ assessment, not a label used by Beich et al. (LLM, conf=0.88)

  cites Beich 2007 source

In the paper

“Blinding of treatment providers Unclear risk Appears that peer educators who delivered the intervention only saw intervention participants but this is not clear”

What the sources say

• Partly supported

  The Procedures section (page not available) says peer educators delivered the intervention to the intervention group, and the Study Protocol section says research assistants who performed assessments were blinded. The paper does not say whether peer educators also saw control participants or explicitly rate the risk of bias for provider blinding; the cited assessment is therefore only partly grounded in its account. (LLM, conf=0.78)

  cites Bernstein 2010 sourceExperimental — accuracy not yet measured

In the paper

“Blinding of treatment providers High risk”

What the sources say

• Supported

  The Protocol (p. 2) states that blinding was not feasible for GPs, who delivered the intervention. This supports the citing document’s high-risk assessment for blinding of treatment providers. (LLM, conf=0.94)

  cites Beich 2007 source

In the paper

“ITT: no.”

What the sources say

• Partly supported

  The Outcomes Analysis section describes the primary comparisons but does not identify them as intention-to-treat analyses; the Results section reports only 72% follow-up at 12 months. However, the Sensitivity Analysis includes participants lost to follow-up by counting them as having made no quit attempt. Thus, 'ITT: no' may describe the primary analysis, but it overlooks an analysis that accounts for all enrollees. (LLM, conf=0.78)

  cites Bernstein 2010 sourceExperimental — accuracy not yet measured

In the paper

“Blinding of participants High risk No reported attempt to blind participants.”

What the sources say

• Supported

  The Study Protocol describes blinding research assistants to randomization status and instructing participants not to reveal their status to follow-up assessors; it does not report an attempt to blind participants. The Procedures section describes distinct experiences for the intervention and control groups, consistent with the citing document’s high-risk assessment. Page numbers are unavailable in our copy. (LLM, conf=0.86)

  cites Bernstein 2010 source

In the paper

“ITT: yes.”

What the sources say

• Supported

  The Table 2 caption (p. 5) explicitly says the consumption changes were analysed both by intention-to-treat (ITT) and by protocol. The paper also states that missing follow-up values were substituted with baseline values. (LLM, conf=0.94)

  cites Beich 2007 source

In the paper

“Random sequence generation (selection bias) Low risk”

What the sources say

• Supported

  The Study Protocol’s Randomization section states that assignment used computer-generated lists, blocked after every nine subjects and stratified by age group. This supports a low-risk assessment for random sequence generation. (LLM, conf=0.92)

  cites Bernstein 2010 source

In the paper

“Incomplete outcome data (attrition bias) All outcomes Unclear risk Loss to follow-up reported, but reasons not given.”

What the sources say

• Partly supported

  The Results section (page not available) reports 70% follow-up at 3 months and 72% at 12 months, and the Sensitivity Analysis describes attrition as considerable. These passages support the reported loss to follow-up, but the paper does not itself assign an ‘unclear risk’ rating for incomplete outcome data; that is the citing authors’ assessment. (LLM, conf=0.82)

  cites Bernstein 2010 sourceExperimental — accuracy not yet measured

In the paper

“Blinding of participants High risk”

What the sources say

• Supported

  The Protocol (p. 2) explicitly says blinding was not feasible for patients. This supports the citing document’s high-risk assessment for blinding of participants. (LLM, conf=0.94)

  cites Beich 2007 source

In the paper

“Allocation concealment (selection bias) Unclear risk”

What the sources say

• Partly supported

  The Procedures section (page unavailable) says allocation used computer-generated lists and opaque envelopes bearing patient identification numbers. It does not say whether the envelopes were sealed or opened in sequence, and it does not itself judge selection-bias risk. The reported method provides a basis for uncertainty, but not the stated risk judgment. (LLM, conf=0.76)

  cites Crawford 2004 sourceExperimental — accuracy not yet measured

In the paper

“Selective reporting (reporting bias) Low risk”

What the sources say

• ?Could not verify

  The Patient assessment and Sample size and data analysis sections describe the trial’s outcome measures and analyses, but do not judge the risk of selective reporting. The Discussion describes limitations without making a reporting-bias assessment. A low-risk judgment cannot be verified from our copy of this paper. (LLM, conf=0.73)

  cites Crawford 2004 sourceExperimental — accuracy not yet measured

In the paper

“Blinding of treatment providers Low risk”

What the sources say

• ?Could not verify

  The Methods describe the trial as single-blind, and the Results report a test of researchers’ blinding. Neither passage says treatment providers were blinded or judges their blinding to be at low risk of bias. A search for provider blinding and risk-of-bias language found no such assessment in our copy. (LLM, conf=0.75)

  cites Crawford 2004 sourceExperimental — accuracy not yet measured

In the paper

“Random sequence generation (selection bias) Low risk”

What the sources say

• Supported

  The Procedures section states that patients were randomised using lists derived from a computer program. That directly supports a low-risk judgment for random sequence generation, although the paper does not use the risk-of-bias label itself. (LLM, conf=0.90)

  cites Crawford 2004 source

In the paper

“Blinding of participants High risk”

What the sources say

• Supported

  The Methods describe the trial as single-blind, and Patient assessment specifies that the follow-up researcher—not the participants—was blind to allocation. The Procedures describe visibly different cards for the two groups: an appointment card or a blank card. These passages support the citing paper’s high-risk judgment for participant blinding, although Crawford et al. do not use that risk-of-bias label. (LLM, conf=0.87)

  cites Crawford 2004 source

In the paper

“Blinding of outcome assessment (detection bias) All outcomes Low risk”

What the sources say

• Supported

  The Patient assessment section states that follow-up interviews were conducted by a researcher blind to allocation status. The Results section reports that researchers correctly guessed allocation in only 41.6% of a test sample, supporting the citing paper’s low-risk judgment for outcome assessment. (LLM, conf=0.88)

  cites Crawford 2004 source

In the paper

“ITT: yes.”

What the sources say

• Supported

  The Sample size and data analysis section states that differences in alcohol consumption between the experimental and control groups were examined on an intention-to-treat basis. (LLM, conf=0.94)

  cites Crawford 2004 source

In the paper

“Incomplete outcome data (attrition bias) All outcomes High risk”

What the sources say

• Partly supported

  The Results report that only 384 of 599 randomised patients (64.1%) completed the 12-month interview, and the Discussion identifies difficult follow-up as a limitation. This supports concern about attrition, but the paper does not judge incomplete data to be at high risk of bias for all outcomes; it also notes that follow-up rates were similar between trial arms. (LLM, conf=0.82)

  cites Crawford 2004 sourceExperimental — accuracy not yet measured

In the paper

“Random sequence generation (selection bias) Unclear risk”

What the sources say

• ?Could not verify

  The Methods (Setting and Participants; page unavailable) report a 100-block randomization procedure and sealed envelopes. Our copy does not judge random sequence generation to be at unclear risk of selection bias; searches for that assessment and for how the sequence was generated did not find it. (LLM, conf=0.75)

  cites D'Onofrio 2012 sourceExperimental — accuracy not yet measured

In the paper

“Allocation concealment (selection bias) Unclear risk”

What the sources say

• ?Could not verify

  The Methods (Setting and Participants; page unavailable) say research associates were provided sealed envelopes for a block-randomization procedure. The paper does not judge allocation concealment to be at unclear risk of selection bias, so that risk-of-bias rating cannot be verified from our copy. (LLM, conf=0.76)

  cites D'Onofrio 2012 sourceExperimental — accuracy not yet measured

In the paper

“ITT: yes.”

What the sources say

• Supported

  The Methods section (Primary Data Analyses) explicitly says the primary analyses were based on the intention-to-treat principle and included all 740 patients in the BNI, BNI with Booster, and SC groups. (LLM, conf=0.95)

  cites D'Onofrio 2012 source

In the paper

“Blinding of treatment providers Low risk”

What the sources say

• ?Could not verify

  The Methods (Measurements; page unavailable) say that independent raters of intervention recordings were blinded, but those raters were not the treatment providers. The Methods (Study Interventions; page unavailable) describe the practitioners who delivered the intervention without stating whether they were blinded or assessing their risk of bias. Our copy therefore does not verify the low-risk judgment for provider blinding. (LLM, conf=0.72)

  cites D'Onofrio 2012 sourceExperimental — accuracy not yet measured

In the paper

“Selective reporting (reporting bias) Low risk”

What the sources say

• ?Could not verify

  The Methods section, “Primary and Secondary Outcomes” (page unavailable), lists the study outcomes, but does not judge the risk of selective reporting. Searches for “selective reporting,” “reporting bias,” and prespecified outcomes found no low-risk assessment in our copy. (LLM, conf=0.76)

  cites D'Onofrio 2012 sourceExperimental — accuracy not yet measured

In the paper

“Blinding of outcome assessment (detection bias) All outcomes Low risk”

What the sources say

• Supported

  The Methods > Measurements section (page unavailable) says follow-up alcohol and SIP outcomes were collected through an Interactive Voice Response system, while research assistants collected the other secondary outcomes and were blinded to study-group assignment. These procedures support the citing document’s low-risk judgment for outcome-assessment blinding. (LLM, conf=0.88)

  cites D'Onofrio 2012 source

In the paper

“Blinding of participants Low risk”

What the sources say

• ?Could not verify

  The Methods (Measurements; page unavailable) say that research associates and independent adherence raters were blinded, but do not say whether participants were blinded. The Methods (Study Interventions; page unavailable) describe different care across groups, without assessing the risk of bias from participant blinding. Our copy therefore does not establish the claimed low-risk judgment. (LLM, conf=0.77)

  cites D'Onofrio 2012 sourceExperimental — accuracy not yet measured

In the paper

“Not blinded.”

What the sources say

• Supported

  The Method > Procedure section (page not provided) describes clinicians delivering a motivational interviewing session in one condition, while the feedback-only condition had no counseling session. Treatment providers therefore knew which intervention they were delivering. The high-risk-of-bias rating is the citing authors’ assessment, not a rating reported by the trial. (LLM, conf=0.87)

  cites McDevitt-Murphy 2014 source

In the paper

“Research interviewers conducting followup were blinded to randomisation status”

What the sources say

• Supported

  The Methods section, under “Trial design,” explicitly states that research interviewers conducting follow-up were blinded to randomization status. This supports the blinding basis given for the citing document’s low-risk-of-bias assessment. (LLM, conf=0.95)

  cites Mertens 2014 source

In the paper

“Two studies took place in college health clinics (Fleming 2010; Schaus 2009)”

What the sources say

• Supported

  The Discussion identifies this trial as conducted in a college health setting and describes recruitment of students attending college health clinics for routine care. This supports citing Fleming 2010 as one of the two studies conducted in such clinics; the other citation is not assessed here. (LLM, conf=0.95)

  cites Fleming 2010 source
• Supported

  Schaus 2009 is one of the two studies cited for this sentence. Its Study participants section says the trial was conducted at university health services, and the abstract (p. 131) describes interventions delivered in a college student health center. This confirms the clinic setting for the Schaus study; the other cited study is outside this verdict. (LLM, conf=0.95)

  cites Schaus 2009 source

In the paper

“the trial is included in a Cochrane Review of brief interventions in patients admitted to hospital.”

What the sources say

• Partly supported

  The Methods describe a randomized trial of a brief intervention for patients at a trauma center, conducted on or near hospital discharge. Our copy does not say whether the trial was included in a Cochrane Review; a search for that review found no confirming passage. (LLM, conf=0.82)

  cites Gentilello 1999 sourceExperimental — accuracy not yet measured

In the paper

“one in a public sexual health clinic (Lane 2008)”

What the sources say

• Supported

  The METHOD section states that Sydney South West Area Health Service provides public sexual health clinics and that the study recruited patients during clinic sessions. This supports the cited description of the study setting. (LLM, conf=0.94)

  cites Lane 2008 source

In the paper

“one in a veterans' affairs medical cen- tre (McDevitt-Murphy 2014)”

What the sources say

• Supported

  The Method—Participants section says the study recruited veterans seeking care at an urban Veterans’ Affairs Medical Center, primarily through its primary care clinic. This supports identifying it as a study conducted in a veterans’ affairs medical centre. (LLM, conf=0.94)

  cites McDevitt-Murphy 2014 source

In the paper

“two trials reported adverse events related to driving outcom es but concluded they were equivalent in both study arm s”

In the paper

“Heterogeneity was substantial (73%) but not unexplained; interventions dif f ered in content and delivery.”

In the paper

“Half the included trials (51%) reported participants' ethnicity; most participants were Caucasian (n = 28 trials, 78%).”

In the paper

“Extended interventions were based on motivational interviewing (five studies), MET (two studies), multiple FRAMES sessions (two studies), or CBT (one study) approaches.”

In the paper

“General practitioners, nurse practitioners, health psychologists or trainee health psychologists administered the interventions.”

In the paper

“In the remaining studies, there were between two and five sessions, where individual sessions varied from one to a maximum of 60 minutes.”

In the paper

“All interventions provided feedback on the screening outcome plus structured advice about potential risks of heavy drinking and ways to reduce consumption.”

In the paper

“Extended interventions were based on motivational interviewing (five studies), MET (two studies), multiple FRAMES sessions (two studies), or CBT (one study) approaches.”

In the paper

“The median duration was 25 minutes and IQR 7.5 to 30.0 minutes.”

In the paper

“The median was 8.5 and IQR 7 to 10.5 (Table 2).”

In the paper

“We could not include 22 studies in metaanalyses (Aalto 2000;Bazargan-Hejazi 2005;Beckham 2007;Crawford 2010;D'Amico 2008;Dent 2008;Désy 2010;Díez 2002;Drummond 2009;Heather 1987;Kunz 2004;L'Engle 2014;Lane 2008;McDevitt-Murphy 2014;Mello 2008;Mertens 2014;Noknoy 2010;Rodríguez 2003;Segatto 2011;Sommers 2006;Tait 2004;Woolard 2013), either because outcomes were not reported at 12 months, or outcome”

What the sources say

• ?Not checked — no full text of the sourcecites Aalto 2000 source
• ?Not checked — no full text of the sourcecites Bazargan-Hejazi 2005 source
• ?Not checked — no full text of the sourcecites Beckham 2007 source

In the paper

“Previously, the overall mean baseline consumption in 21 trials reporting these data was 313 g/week”

In the paper

“Eight trials focused on adolescents, young adults, or both (D'Amico 2008;Mertens 2014;Monti 2007;Segatto 2011;Spirito 2004;Spirito 2011;Tait 2004;Walton 2010).”

In the paper

“The mean percentage of male participants was 70%.”

In the paper

“described as brief Intervention (and assumed to be based on FRAMES where not reported) (n = 27)”

In the paper

“Among the 45 trials that reported participants' age, the mean age was 40 years (SD 11.18).”

In the paper

“Five categories of control condition were reported.”

In the paper

“Extended interventions were based on motivational interviewing (five studies), MET (two studies), multiple FRAMES sessions (two studies), or CBT (one study) approaches.”

In the paper

“or based on Cognitive Behavioural Therapy (CBT) techniques (n = 2)”

In the paper

“This review included 69 studies (112 reports) (Figure 1; Characteristics of included studies)”

In the paper

“based on or informed by motivational interviewing, Motivational Enhancement Therapy (MET), or Brief Negotiated Interview (BNI) (n = 32)”

In the paper

“Searches identified 4004 potentially relevant records which were screened by title and abstract for eligibility.”

In the paper

“The total duration of extended intervention sessions ranged from 60 minutes to 180 minutes.”

In the paper

“Extended interventions were based on motivational interviewing (five studies), MET (two studies), multiple FRAMES sessions (two studies), or CBT (one study) approaches.”

In the paper

“We retrieved 165 records for full text assessment.”

In the paper

“Only 11 trials added for this update reported overall mean baseline consumption, which was 181 g/week (about 23 UK units).”

In the paper

“Four trials focused on older adults (aged > 55 years, aged > 60 years, or aged > 65 years) (Ettner 2014;Fleming 1999;Moore 2011;Watson 2013).”

In the paper

“Baseline intensity of drinking was reported in 11 trials, in which the mean baseline value was 69 g/drinking day.”

In the paper

“Previously, the overall mean baseline consumption in 21 trials reporting these data was 313 g/week (about 39 UK units).”

In the paper

“We added 42 studies (73 reports) for this update.”

In the paper

“There was substantial heterogeneity among trials in terms of the mechanisms of screening participants for inclusion and in the content of both control and intervention arms.”

In the paper

“Extended interventions were evaluated in 10 trials (Aalto 2000;Bischof 2008;Israel 1996;L'Engle 2014;Longabaugh 2001;Maisto 2001;Monti 2007;Moore 2011;Richmond 1995;Spirito 2011), in which the total duration was greater than 60 minutes and the number of sessions delivered to participants ranged from two to seven.”

What the sources say

• ?Not checked — no full text of the sourcecites Aalto 2000 source
• ?Not checked — no full text of the sourcecites Bischof 2008 source
• ?Not checked — no full text of the sourcecites Israel 1996 source

In the paper

“Extended interventions also provided feedback and structured advice but comprised either more than five sessions or more than 60 minutes in total.”

In the paper

“In the extended intervention conditions, the treatment exposures ranged from 65 minutes to 175 minutes .”

In the paper

“The previous version of this review included 29 studies (39 reports) (Kaner 2007).”

In the paper

“Most studies (n = 61) compared brief intervention with minimal or no intervention.”

In the paper

“Three of these studies also compared an extended intervention to a brief intervention (Bischof 2008;Maisto 2001;Richmond 1995).”

What the sources say

• ?Not checked — no full text of the sourcecites Bischof 2008 source
• ?Not checked — no full text of the sourcecites Maisto 2001 source
• ?Not checked — no full text of the sourcecites Richmond 1995 source

In the paper

“Two trials reported this outcome for extended versus no or minimal intervention (Maisto 2001;Monti 2007), and one for extended versus brief intervention (Maisto 2001).”

What the sources say

• ?Not checked — no full text of the sourcecites Maisto 2001 source
• ?Not checked — no full text of the sourcecites Monti 2007 source

In the paper

“One trial reported this outcome for extended intervention versus no or minimal intervention and versus brief intervention (Maisto 2001).”

What the sources say

• ?Not checked — no full text of the sourcecites Maisto 2001 source

In the paper

“Romelsjö 1989 reported change scores whereas the other studies reported final values; only three studies reported at 12 months.”

What the sources say

• ?Not checked — no full text of the sourcecites Romelsjö 1989 source

In the paper

“One study reported this outcome for extended versus no or minimal intervention (Israel 1996).”

What the sources say

• ?Not checked — no full text of the sourcecites Israel 1996 source

In the paper

“Longabaugh 2001 reported the number of days drinking but the number of participants assessed was not reported.”

What the sources say

• ?Not checked — no full text of the sourcecites Longabaugh 2001 source

In the paper

“Two trials reported this outcome for extended versus no or minimal intervention (Longabaugh 2001;Monti 2007).”

What the sources say

• ?Not checked — no full text of the sourcecites Longabaugh 2001 source
• ?Not checked — no full text of the sourcecites Monti 2007 source

In the paper

“Many studies reported at several time points, although not all outcomes were reported at all time points.”

In the paper

“With two exceptions, funding was provided by government institutes, research bodies or charitable foundations.”

In the paper

“the intervention or comparator did not fit our prespecified criteria (n = 5)”

In the paper

“We excluded two trials from the original review for this update.”

In the paper

“One study was partly funded by a pharmaceutical company and a brewers association (Wallace 1988)”

In the paper

“Nine studies reported measures of healthcare utilisation (Babor 2006;Crawford 2004;Dent 2008;Désy 2010;Ettner 2014;Fleming 1997;Fleming 2010;Kaner 2013;Tait 2004).”

In the paper

“Richmond 1995 reported outcomes at six months and one year.”

In the paper

“We excluded 46 studies.”

In the paper

“Curry 2003 reported outcomes at three months and one year.”

In the paper

“two studies reported economic evaluations (Lock 2006;Watson 2013).”

In the paper

“Babor 2006 recruited five managed care organisations, which provided 15 practices.”

In the paper

“The HEALTH study reported at six-months and one-year follow-up (Ockene 1999).”

In the paper

“Older studies tended to define higher limits (e.g.> 35 units/week by Wallace 1988) than more recent studies (e.g.> 18 drinks/week by Rubio 2010).”

In the paper

“Kaner 2013 recruited 34 general practices with a target of 31 participants per practice.”

In the paper

“Most studies reported sources of funding (n = 60, 87%).”

In the paper

“Two studies reported on male and female participants in separate papers (Aalto 2000;Scott 1990).”

In the paper

“The percentage of heavy or binge drinkers at 12 months was reported by 19 studies (Altisent 1997;Beich 2007;Cherpitel 2009;Córdoba 1998;Curry 2003;Daeppen 2007;D'Onofrio 2008;Drummond 2014;Ettner 2014;Fernández 1997;Fleming 1997;Fleming 1999;Fleming 2004;Kaner 2013;Rubio 2010;Scott 1990;Sommers 2013;Wallace 1988;Walton 2010).”