Content uploaded by Nick Heather
Author content
All content in this area was uploaded by Nick Heather on Oct 01, 2019
Content may be subject to copyright.
REVIEW
The effectiveness of brief alcohol interventions in primary care
settings: A systematic review
EILEEN F. S. KANER1, HEATHER O. DICKINSON1, FIONA BEYER1, ELIZABETH PIENAAR2,
CARLA SCHLESINGER3, FIONA CAMPBELL4, JOHN B. SAUNDERS5, BERNARD BURNAND6
& NICK HEATHER7
1Institute of Health and Society, Newcastle University, Newcastle upon Tyne, UK, 2South African Cochrane Centre,
Tygerberg, South Africa, 3Alcohol and Drug Ser vice, Brisbane, Australia, 4School of Health and Related Research,
University of Sheffield, Sheffield, UK, 5Faculty of Medicine, University of Sydney, Sydney, Australia, 6Centre
d’épidémiologie clinique, University of Lausanne, Lausanne, Switzerland, and 7Division of Psychology, Northumbria
University, Newcastle upon Tyne, UK
Abstract
Issues.Numerous studies have reported that brief interventions delivered in primary care are effective in reducing
excessive drinking. However, much of this work has been criticised for being clinically unrepresentative.This review
aimed to assess the effectiveness of brief interventions in primary care and determine if outcomes differ between
efficacy and effectiveness trials. Approach.A pre-specified search strategy was used to search all relevant electronic
databases up to 2006.We also hand-searched the reference lists of key articles and reviews. We included randomised
controlled trials (RCT) involving patients in primary care who were not seeking alcohol treatment and who received
brief intervention. Two authors independently abstracted data and assessed trial quality. Random effects meta-
analyses, subgroup and sensitivity analyses and meta-regression were conducted. Key Findings.The primary
meta-analysis included 22 RCT and evaluated outcomes in over 5800 patients. At 1 year follow up, patients receiving
brief intervention had a significant reduction in alcohol consumption compared with controls [mean difference:
-38 g week-1, 95%CI (confidence interval): -54 to -23], although there was substantial heterogeneity between trials
(I2=57%). Subgroup analysis confirmed the benefit of brief intervention in men but not in women. Extended
intervention was associated with a non-significantly increased reduction in alcohol consumption compared with brief
intervention.There was no significant difference in effect sizes for efficacy and effectiveness trials.Conclusions.Brief
interventions can reduce alcohol consumption in men, with benefit at a year after intervention, but they are unproven
in women for whom there is insufficient research data. Longer counselling has little additional effect over brief
intervention. The lack of differences in outcomes between efficacy and effectiveness trials suggests that the current
literature is relevant to routine primary care. [Kaner EFS, Dickinson HO, Beyer F, Pienaar E, Schlesinger C,
Campbell F, Saunders JB, Burnand B, Heather N. The effectiveness of brief alcohol interventions in primary care
settings: A systematic review. Dr ug Alcohol Rev 2009;28:301–323]
Key words: alcohol, brief intervention, systematic review.
Eileen F. S. Kaner PhD, Professor of Public Health Research, Heather O. Dickinson PhD, Reader in Epidemiology, Fiona Beyer PGDip
Information & Library Studies, Information Specialist, Elizabeth Pienaar MPH, Senior Scientist, Carla Schlesinger PhD (Clinical Psych.),
Programme Manager, Fiona Campbell BSc. (Hons.), Systematic Reviewer, John B. Saunders MD, Professor of Internal Medicine and Addiction,
Bernard Burnand MD, Associate Professor, Nick Heather PhD, Emeritus Professor of Alcohol and Drug Studies. Correspondence to Professor
Eileen Kaner, Institute of Health and Society, the Medical School, Newcastle University, Framlington Place, Newcastle on Tyne, NE2 4HH, UK.
Tel: +41-21-314-7255; Fax: +41-21-314-4954; E-mail: e.f.s.kaner@newcastle.ac.uk
Received 16 May 2008; accepted for publication 15 September 2008.
Drug and Alcohol Review (May 2009), 28, 301–323
DOI: 10.1111/j.1465-3362.2009.00071.x
© 2009 Australasian Professional Society on Alcohol and other Drugs
Background
Alcohol contributes 4% to the total disease burden
worldwide, as measured by disability-adjusted life years
[1].This burden is more evident in developed countries
(9% disability-adjusted life years), where alcohol ranks
third after smoking and hypertension as the lead cause
of morbidity and premature death [2]. However, the
full impact of alcohol on the health of individuals and
the wider community is difficult to estimate because of
many hidden effects resulting from its use, including
increased levels of violence, accidents and suicide [3].
The heavy burden that alcohol use places on the health
of populations, and its significant economic conse-
quences, has led to national and international pro-
grammes and policies that seek to reduce consumption
levels and thus reduce a primary cause of avoidable ill
health [4,5]. The impetus for a preventive approach to
alcohol problems has been reinforced by epidemiologi-
cal research which shows that, on a population level,
most alcohol-related harm is not due to drinkers with
severe alcohol dependence but attributable to a much
larger group of hazardous or harmful drinkers whose
consumption exceeds recommended drinking levels
and who experience a wide range of physical, psycho-
logical or social problems [6].
Secondary prevention of alcohol problems, using
screening to identify risk or harm at an early stage and
followed by brief intervention to help reduce such prob-
lems, has been the focus of a great deal of research
[7,8]. Most of this research on screening and brief
intervention has taken place in primary care. Brief
alcohol 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; infor-
mation on the harms associated with risky alcohol use;
benefits of reducing intake; motivational enhancement;
analysis of high risk situations for drinking and coping
strategies; and the development of a personal plan to
reduce consumption. Although the form that brief
intervention takes might vary between studies [9], core
features of brief interventions in primary care are that
they: are delivered by generalist health-care workers;
target a population of hazardous and harmful drinkers
who tend not to be seeking help for alcohol problems
and aim for reductions in consumption and related
harm.
In this review, we define primary care as all immedi-
ately accessible, general health care which covers a
broad range of presenting problems, and which can be
accessed by a wide range of patients on demand, and
not as the result of a referral for specialist care. Thus
primary care includes both general practice settings
where 20% patients are likely to be hazardous or
harmful drinkers [10] and emergency care where
30–70% patients can present for reasons related to
alcohol [11,12]. In both contexts, brief alcohol inter-
vention typically occurs opportunistically because
drinking problems are generally not the primary reason
for the presentation and because patients are usually
not seeking alcohol treatment. In addition, brief inter-
vention needs to be time-limited because primary care
clinicians have just a short amount of time to spend
with individual patients. Despite this, brief alcohol
interventions have ranged from single, 5 min sessions of
structured advice delivered by generalist clinicians
through to multiple sessions of prolonged motivational
counselling delivered by specialists working in primary
care.Thus there is a need to establish more precisely the
necessary treatment duration for brief interventions in
primary care practice.
Although a number of reviews have indicated benefi-
cial outcomes of brief intervention for hazardous and
harmful drinkers [13–20], crucial questions remain
concerning the impact of brief interventions in routine
primary care. It has been suggested that much of the
published literature on brief alcohol interventions con-
sists of ‘ideal world’, efficacy trials [21], that is studies
carried out in tightly controlled research conditions
designed to optimise internal validity [22]. Efficacy
studies are an important component of research evalu-
ation, particularly in ‘proof of concept’ contexts where
new or early stage treatments are being considered.
However, if clinicians are to deliver an intervention in
routine practice, it is necessary to establish if it is effec-
tive when delivered in a clinically relevant context; that
is in less tightly controlled effectiveness trials (some-
times called pragmatic trials) which more closely rep-
resent routine practice.
A further challenge for brief alcohol intervention
work relates to the of range patients that present to
clinicians in primary care. A number of population
subgroups (e.g. young people, the elderly and ethnic
minorities) exist within the categories of hazardous or
harmful drinkers and little is known about how these
subgroups respond to brief intervention in primary
care. Differential loss of subjects from brief intervention
trials has led to a call for caution in generalising these
results to routine practice [23], but little emphasis has
been placed on characterising patients who were not
included in the studies at the outset. Thus there is a
clear need to clarify not only the types of drinkers for
whom brief interventions have a positive impact but
also any subgroups that have not been represented in
the trials to date.
This paper reports findings from an updated version
of a Cochrane Collaboration systematic review [24]
which aimed to determine:
302 E. F. S. Kaner et al.
© 2009 Australasian Professional Society on Alcohol and other Drugs
•The effect of brief intervention compared with
control interventions at reducing excessive alcohol
consumption in routine primary care.
•The type of drinkers for whom these interventions
are effective.
•Whether brief intervention outcomes differ
between efficacy and effectiveness trials.
•The impact of extended intervention compared
with brief intervention in primary care settings.
Our specific focus on primary care as a clinical
setting meant that this review did not consider seminal
brief intervention work which recruited subjects from
the entire population, such as the Malmo trial [25]. In
addition, given our interest in the relevance of brief
interventions for primary care clinicians, we focused
on trials that reported specific primary care data,
even though this meant excluding important work in
primary care where data were aggregated with findings
from social care, occupational and educational settings,
for example, the large World Health Organisation trial
based across 10 countries worldwide [26].
Methods
The systematic review was conducted using Cochrane
Collaboration method and the details have been
reported elsewhere [24]. In brief, the review focused on
identifying randomised controlled trials, including
cluster randomised controlled trials involving patients
attending primary care and who were identified as
heavy, problematic or excessive drinkers. Dependent
users of alcohol were not the main focus of this review.
We specified that brief intervention could consist of up
to five sessions [14] of time-limited engagement with a
patient in primary care which involved the provision
of information, advice and/or counselling that was
designed to achieve a reduction in alcohol consumption
or alcohol-related problems. Control conditions were
typically assessment only, treatment as usual and/or the
delivery of written information.
Search strategy
An optimal search strategy for identifying randomised
controlled trials was used [27] and combined with rel-
evant keywords and medical subject headings (MeSH)
terms (see Appendix 1). This strategy was applied to a
wide range of electronic and specialist databases up to
2006 (see Appendix 2).There were no language restric-
tions. Additional searching included: hand and archive
searches of relevant journals, the reference lists of
included papers and relevant systematic reviews. Key
informants and experts were also contacted to enquire
about unpublished work and ongoing research.
Identification of included studies
Two researchers independently selected potentially rel-
evant trials from the search outputs by reading the
study titles and abstracts. Full copies of all selected
papers were obtained and checked against the inclusion
criteria. All studies meeting the inclusion criteria
underwent data extraction to capture the characteristics
of the study population, interventions and follow up, to
assess methodological quality and to record outcomes
relevant to the review. Data extraction was indepen-
dently carried out by two researchers and any disagree-
ment about inclusion criteria or study details was
resolved by a third reviewer.
Quality assessment
The following aspects of method were assessed and
tabulated:
•Selection bias—adequacy of randomisation and
allocation concealment.
•Performance bias—blinding of patients or
masking of clinicians.
•Attrition bias—differential loss of subjects from
comparison groups.
•Detection bias—blinding of the individuals assess-
ing outcomes.
Efficacy/effectiveness
In order to establish if the studies were efficacy (tightly
controlled) or effectiveness (real world) trials, an eight
item coding scale was developed (see Appendix 3)
which was informed by the work of Shadish and col-
leagues [28]. Four items considered issues that were
crucial for routine primary care including whether: trial
subjects were genuine health service patients with a full
range of possible presenting conditions; practices deliv-
ered the full range of medical services to patients, prac-
titioners routinely worked in the health service rather
than being funded by the trial; the intervention could
be delivered in standard consultation times. The pos-
sible scores on these items were 0, 1, 2, where the
higher score indicated a higher effectiveness rating.
Four further items considered issues that were desirable
in routine primary care including: an allowance for
therapeutic flexibility in the delivery of the intervention;
pre-intervention training that could fit within clinicians’
regular continuing professional development schemes;
intervention monitoring that did not overly intrude on
the consultation and support systems that could gener-
ally be available in practices/departments. The possible
scores on these items were 0, 1/2, 1 where a higher score
indicated a higher effectiveness rating. If a paper did not
Brief intervention effectiveness 303
© 2009 Australasian Professional Society on Alcohol and other Drugs
give full information on an item then the midpoint
score was allocated.The possible score range was 0–12
and a binary variable was created by dividing the scores
into those above and below the median. Each trial was
independently rated by two researchers and any dis-
agreement over an item rating was resolved by a third
researcher.
Treatment exposure
A measure of treatment exposure was calculated as the
sum of the duration of the initial brief intervention plus
the total duration of all booster sessions, in minutes.
Statistical methods
The primary meta-analysis focused on quantity of
alcohol consumed per week as this outcome was most
consistently reported. For each trial, outcome data on
quantity of alcohol consumed in a specific time period
were converted to g per week if necessary. 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 [29,30].
Outcomes considered in secondary meta-analyses
included frequency and intensity of drinking; here
‘drinking days’, ‘drinking sessions’ and ‘occasions’
were all assumed to be equivalent to drinking days. In
addition, we considered laboratory indicators of heavy
alcohol consumption, particularly gammaglutamyl
transferase (GGT) and mean corpuscular volume
(MCV).
If trials had more than one control arm, they were
combined by calculating weighted means of continuous
outcomes and summing dichotomous outcomes; like-
wise for multiple treatment arms within our definition
of brief intervention.
For cluster randomised controlled trials, if the analysis
accounted for the cluster design a direct estimate of the
treatment effect and its standard error was extracted.
If the analysis did not account for the cluster design,
the variance of the effect of treatment was inflated using
an external estimate of the intra-cluster correlation
coefficient.
For continuous outcome measures, the findings of
included trials were aggregated in meta-analyses using
the weighted mean difference method.
For dichotomous outcomes, relative risks and 95%CI
(confidence interval) were calculated and pooled in a
meta-analysis using Mantel–Haenzel weighting.
The magnitude of heterogeneity between trials was
assessed using the I2statistic [31]; the statistical signifi-
cance of heterogeneity was assessed using P-values
derived from c2-tests [32]. The populations and inter-
ventions evaluated by the trials were so heterogeneous
that we used a random effects model for all analyses
[33].
Forest plots are presented with trials ordered by their
efficacy/effectiveness score, that is, effectiveness trials is
at the top and the more tightly controlled efficacy trials
at the bottom.
Funnel plots (plots of the effect estimate from each
study against the sample size or effect standard error)
were used to assess the potential for bias related to the
size of the trials, which could indicate possible publica-
tion bias.
Sensitivity analyses were carried out based on meth-
odological quality: trials which did not report adequate
concealment of allocation were excluded and the analy-
sis repeated to check the robustness of findings; analysis
was repeated imputing data for participants who were
lost to follow up for the two trials which reported these
data; trials which did not report standard deviations
were included with imputed standard deviations; and
analysis was repeated varying the assumptions about
the intra-cluster correlation coefficient of cluster ran-
domised trials which did not allow for the cluster ran-
domisation in the analysis.
Subgroup analyses were undertaken, grouping trials
as (i) either efficacy or effectiveness trials, dichotomis-
ing the efficacy/effectiveness score at the median; and
(ii) trials of either low or high exposure to treatment,
dichotomising the treatment exposure at the median.
Meta-regression was carried out, relating the effect of
brief intervention in each trial to (i) its efficacy/
effectiveness score; and (ii) the treatment exposure.
Analyses were carried out using Review Manager
4.2.8 (Clicktime.com Inc., San Francisco, CA, USA)
[34] and Stata 9.2 (StataCorp, College Station, TX,
USA) [35].
Results
The search strategy identified 1060 potentially relevant
references (see Figure 1) which we electronically
screened and 280 were retrieved for detailed evaluation.
Of these, 243 were excluded because the study was
either not a randomised controlled trial, did not take
place in primary care or did not involve brief alcohol
intervention.
Twenty-nine unique trials (reported in 39 papers)
met the inclusion criteria for the review [36–64]. Where
there were multiple reports of a single trial, such as
project TrEAT [44,65–68], project ELM [54,69],
project HEALTH [56,70], the Lahti trial [36,71], the
Cut Down on Drinking trial [60,72] and the Barcelona
trial [58,73,74], we used one key reference for the trial.
Among the included trials, three [53,55,58] did not
report the number of participants assessed by treatment
arm so they could not be included in any meta-analysis.
304 E. F. S. Kaner et al.
© 2009 Australasian Professional Society on Alcohol and other Drugs
Twenty-eight trials compared a brief intervention with
a control intervention: four of these also included an
extended intervention arm [36,53,54,57]. One of the
trials included two control arms [48] and one included
two intervention arms identical in substance but deliv-
ered by different health professionals [55]. One further
trial compared an extended intervention with a brief
intervention [50].
A description of the trials in the review (Table 1)
Country
Eleven trials took place in the USA [38,41,44–
47,51,53,54,56,61], five in the UK [40,48,52,60,64],
five in Spain [37,39,42,43,58], two each in Canada
[50,55], Finland [36,62] and Sweden [59,63], one in
France [49] and one in Australia [57].
Clinical setting
Most interventions (n=24 trials) were administered in
general practice-based primary care [36–39,41–46,48–
50,52–57,59–64]. Five trials were carried out in acci-
dent and emergency departments [40,47,51,53,58].
One trial [42] reported findings for two primary care
settings and two other settings; only the former data
were included in the meta-analyses.
Potentially relevant references
identified from health databases
n =708
Potentially relevant references
identified from Health and ETOH
databases, excluding duplicates
n =1060
References data abstracted
n= 63
References excluded after
data abstraction
n =24
References included in review
n = 39
References excluded on basis
of title and abstract
n =780
Duplicates excluded
n =716
Potentially relevant references
identified from ETOH
n = 1068
Potentially relevant references
identified from Health and ETOH
databases
n =1776
References excluded after
reading
n =217
References retrieved for more
detailed evaluation
n=280
RCTs included in review
n =29
RCTs excluded from meta -
analysis
n =4
RCTs included in meta-analysis
n =25
r
Figure 1. Flow chart showing number of potentially relevant references identified by searches and number meeting inclusion criteria and
included in meta-analyses. ETOH refers to the Alcohol and Alcohol problems science database run by the National Institute of Alcohol Abuse
and Alcoholism.
Brief intervention effectiveness 305
© 2009 Australasian Professional Society on Alcohol and other Drugs
Ta b l e 1 . Summar y of the brief alcohol intervention trials included in the meta-analysis
Authors (year)
Trial name
Setting (country,
PHC context)
Methodology: randomisation,
blinding, allocation
concealment
Efficacy/
effectiveness
score
Participants: number
randomised (% male);
mean age; baseline
alcohol consumption Intervention details Outcomes
Aalto et al. (2000,
2001) [36,71]
Lahti trial
Finland, General
practice.
Randomisation adequate
Blinding unclear
Allocation concealment
inadequate
10.5 n=414 (71%); 41.6 years;
Mean consumption
week-1=286 g (men)/165 g
(women)
Extended intervention: 10–20 min
FRAMES-based BI from GP or nurse, & 6
booster sessions over 30 months;
Brief intervention: same BI with 2 follow-up
sessions over 24 months;
Control: advice to reduce drinking and contact
their GP if any health problems, were not
told approximately 36 month follow-up
Mean drinks week-1
Drinking occasions week-1
Drinking amount/occasion
CDT
ASAT
ALAT
GGT
MCV
Altisent et al. (1997)
[37]
Spain, General
practice.
Randomisation adequate
Blinding unclear
Allocation concealment
adequate
8.5 n=139 (100%); 45 years;
Mean consumption
week-1=57 units
Brief intervention: 5 min BI from GP with
support material, 5 follow-up sessions over
1 year;
Control: single brief advice session from GP
0% reduction in alcohol
consumption
MALT test
Goldberg score
% drinking <35 units week-1
Chang et al. (1997)
[38]
USA, General
practice.
Randomisation adequate
Blinding unclear
Allocation concealment unclear
9n=24 (0%); 39.3 years;
Mean consumption
week-1=10.3 drinks
Brief intervention: BI with study psychiatrist
(duration not reported);
Control (alcohol treatment group): referral
from the research assistant
Mean drinking days week-1
Mean drinks per drinking day
Mean drinks week-1
Mean binges
Cordoba et al. (1998)
[39]
EBIAL trial
Spain, General
practice.
Randomisation unclear
Practices cluster randomised
Blinding unclear
Allocation concealment
adequate
11 n=229 (100%); 36.5 years;
Mean consumption
week-1=54.0 units
Brief intervention: 15 min cognitive-behavioural
therapy BI from GP;
Control: 5 min ‘simple advice’ (reproducing
standard care) from GP
% reducing to <35 units week-1
Crawford et al. (2004)
[40]
UK, Emergency
care.
Randomisation adequate
Blinding adequate
Allocation concealment
inadequate
10.5 n=599 (78.1%); 44 years;
Mean consumption/
occasion =21.2 units
Brief intervention: 30 min BI from alcohol
worker plus health information leaflet;
Control: health information leaflet only
Mean units week-1
Mean units per drinking day
Mean proportion abstinent days
Curry et al. (2003)
[41]
USA, General
practice.
Randomisation unclear
Blinding unclear
Allocation concealment unclear
9n=333 (65%); 46.9 years;
Mean consumption
week-1=166 g
Brief intervention: BI of 1–5 min from GP, plus
up to 3 follow-up phone calls over 10 weeks
from researcher (psychology graduate), also
self-help manual and written personalised
feedback;
Control: received standard care
N drinks week-1
N drinking days week-1
N binges week-1
Grams per drinking day
% binge drinking
% heavy drinking (>1 drink day-1
for women or >2 drinks day-1
for men
Diez et al. (2002) [42] Spain, General
practice.
Randomisation unclear
Blinding unclear
Allocation concealment unclear
10.5 n=1022 (100%); 42.4 years;
Mean consumption
week-1=47.1 units
Brief intervention: received an evaluation
survey, a self-help manual with guidelines
for consumption plus 10 min BI
(interventionist not reported);
Control: evaluation survey only with no
comment or advice
Drinks week-1
% risk drinkers (>35 units week-1)
Fernandez et al.
(1997) [43]
Spain, General
practice.
Randomisation adequate
Blinding adequate
Allocation concealment
inadequate
7.5 n=152 (100%); 40.3 years;
ⱖ35 units per week
Brief intervention: 10 min BI (interventionist
not reported);
Control: no intervention
N participants with weekly
intake>35 IU and >21 IU
306 E. F. S. Kaner et al.
© 2009 Australasian Professional Society on Alcohol and other Drugs
Fleming et al. (1997,
2000, 2002)
[44,65,66]
Grossberg et al. (2004)
[67];
Manwell et al. (2000)
[68];
TrEAT trial
USA, General
practice.
Randomisation adequate
Blinding adequate
Allocation concealment
adequate
10.5 n=774 (62%);
Mean consumption in last
wk =19.0 drinks; subgroup
analysis of n=226 young
adults, aged 18–30 years;
subgroup analysis of n=205
women, aged 18–40 years
Brief intervention:two15minBIfromGP
1 month apart plus workbook, then
follow-up phone call from nurse 2 weeks
after each GP meeting;
Control: general health booklet only
Mean drinks in last 7 days
Binge drinking (>5 drinks/occasion
for men or >4 for women)
Excessive drinking (>20 drinks
week-1for men or >13 for
women)
Fleming et al. (1999)
[45]
GOAL trial
USA, General
practice.
Randomisation unclear
Blinding adequate
Allocation concealment
adequate
9n=158 (66%); age range
65–75 years;
Mean consumption
week-1=16.0 drinks
Brief intervention: two 10–15 min BI plus
workbook from GP, then follow-up phone
call from nurse 2 weeks after each visit;
Control: general health booklet only
N drinks in last 7 days
N binges (>4 drinks/occasion for
men or >3 for women) in last
30 days
% drinking excessively (>20 drinks
week-1for men or >13 for
women) in last 7 days
Fleming et al. (2004)
[46]
USA, General
practice.
Randomisation unclear
Blinding adequate
Allocation concealment unclear
4.5 n=151 (45%); 48.7 years;
Mean consumption in last
30 days =33.2 drinks
Brief intervention: two 15 min BI from nurse
or GP assistant;
Control: self-help manual, told by researcher to
contact GP with health concerns
Mean % heavy drinkers
Mean drinks last 30 days
Mean frequency of binge drinking
Proportion of subjects who
reduced %CDT
Gentilello et al. (1999)
[47]
USA, Emergency
care.
Randomisation adequate
Blinding adequate
Allocation concealment
adequate
6n=762 (82%); 36.1 years;
Mean BAC =152 mg dL-1
Brief intervention: 30 min BI from research
psychologist plus a handwritten summary
letter 1 month later;
Control: helped to find assistance with
drinking problem if requested
Trauma recurrence after hospital
discharge
Mean drinks week-1
Heather et al. (1987)
[48]
DRAMS trial
UK, General
practice.
Randomisation unclear
Blinding adequate
Allocation concealment unclear
8.5 n=104 (75%); 36.4 years;
Mean consumption in last
month =194.4 units
Brief intervention (DRAMS): BI from GP
(duration not reported) plus self-help book
& manual for controlled drinking, 1 or
more follow-up sessions;
Control 1 (Advice group): BI (‘strong advice’)
from GP (duration not reported), no
follow-up sessions;
Control 2: no intervention
Units consumed in last month
Units consumed in heaviest month
of the last 6
Huas et al. (2002) [49] France, General
practice.
Randomisation unclear
Blinding unclear
Allocation concealment unclear
10 n=541 (100%); 51.8 years;
consumption level not
reported
Brief intervention: 10 min BI (interventionist
not reported), patients with physical or
biological symptoms received 4 follow-up
sessions;
Control: received standard care
Mean drinks week-1
Israel et al. (1996) [50] Canada, General
practice.
Randomisation adequate
Blinding unclear
Allocation concealment
inadequate
7.5 n=105 (sex unclear)
age range 30–60 years;
Mean consumption in last
4 weeks =145.2 drinks
Extended intervention (Brief counselling
group): one 30 min BI from nurse
practitioner plus guideline pamphlet,
6¥20 min follow-up sessions with nurse in
1 year;
Brief intervention (Advice group):
recommended to reduce consumption plus
guideline pamphlet
Mean alcohol consumption in last
4 weeks
Serum GGT
Kunz et al. (2004) [51] USA, Emergency
care.
Randomisation unclear
Blinding adequate
Allocation concealment unclear
6n=294 (81%); 41.7 years;
Mean weekly consumption in
last 3 months =34.1 drinks
Brief intervention: BI from researchers plus
action plan & health information pack, 1
follow-up session (duration not reported);
Control: health information pack only
Mean weekly alcohol consumption
% binge drinkers in last month
AUDIT score
(continued)
Brief intervention effectiveness 307
© 2009 Australasian Professional Society on Alcohol and other Drugs
Ta b l e 1 . (Continued)
Authors (year)
Trial name
Setting (country,
PHC context)
Methodology: randomisation,
blinding, allocation
concealment
Efficacy/
effectiveness
score
Participants: number
randomised (% male);
mean age; baseline
alcohol consumption Intervention details Outcomes
Lock et al. (2006) [52] UK, General
practice.
Randomisation adequate
Practices cluster randomised
Blinding adequate
Allocation concealment
adequate
12 n=127 (50%); 44.1 years;
Mean consumption
week-1=24.6 units
Brief intervention: 5–10 min BI from nurse
plus self-help booklet;
Control: standard care (advice to cut down)
AUDIT score
Units week-1
SF-12 physical health
SF-12 mental health
Longabaugh et al.
(2001) [53]
USA, Emergency
care
Randomisation unclear
Blinding adequate
Allocation concealment
inadequate
6n=539 (78%); 27 years;
consumption level not
reported
Extended intervention (BI): 40–60 min BI from
trained clinician plus change plan
worksheet;
Brief intervention (BIB): same BI plus 40-min
booster session 7–10 days later from trained
clinician
Control: standard care
N heavy drinking days per week
Alcohol related injuries
Negative consequences from
drinking
Maisto et al. (2001)
[54]; Gordon et al.
(2003) [69]
ELM trial
USA, General
practice.
Randomisation adequate
Blinding unclear
Allocation concealment
adequate
5n=301 (70%); 45.6 years;
Mean consumption in last
30 days =75.3 drinks
Extended intervention (motivational
enhancement): 30–45 min BI plus booklet,
plus 2 ¥15–20 min follow-up sessions from
trained interventionist;
Brief intervention (BI): 10–15 min BI from
trained interventionist plus booklet;
Control: standard care, but selected data from
baseline assessments were forwarded to GP
for action if desired
Mean number of days abstinent
Mean no drinks
Mean no days consumed 1–6
drinks
Mean drinks per drinking day
McIntosh et al. (1997)
[55]
Canada, General
practice.
Randomisation unclear
Blinding inadequate
Allocation concealment unclear
10.5 n=159 (52%);
31.1 years
consumption level not
reported
Brief intervention (GP): 2 ¥30 min BI from
GP (not their own) 2 weeks apart plus
booklet;
Brief intervention (nurse): same BI but from
nurse practitioner;
Control: 5 min advice from their own GP plus
handout
Mean monthly Q/F
Ockene et al. (1999)
[56]; Reiff-Hekking
et al. 2005 [70]
HEALTH trial
USA, General
practice.
Randomisation adequate
Practices cluster randomised
Blinding adequate
Allocation concealment
adequate
11 n=530 (64.7%); 43.9 years;
Mean consumption
week-1=17.8 drinks
Brief intervention: 5–10 min BI from trained
interventionists plus health booklet;
Control: health booklet only
Mean drinks week-1
Mean binge drinking episodes (>5
drinks/occasion for men or >4
for women)
Richmond et al. (1995)
[57]
Alcoholscreen trial
Australia, General
practice.
Randomisation unclear
Blinding adequate
Allocation concealment unclear
9.5 n=378 (57%); 37.7 years;
Mean consumption
week-1=36.8 units
Extended intervention (Alcoholscreen) 5 min
session where patients given self-help
manual, 1 week later 15–20 min BI,
1 month later 5–25 min consultation, 2
further 5 min follow-up sessions;
Brief intervention (Minimal intervention):
5 min BI from GP;
Control 1 (No intervention): assessed at
baseline;
Control 2 (No assessment): no assessment or
intervention
Mean drinks week-1
Mean Q/F
GGT
308 E. F. S. Kaner et al.
© 2009 Australasian Professional Society on Alcohol and other Drugs
Rodriguez et al. (2003)
[58]; Rodriguez
et al. 2005 [73];
Rodriguez et al.
(2006) [74]
Barcelona trial
Spain, Emergency
care.
Randomisation unclear
Blinding adequate
Allocation concealment
inadequate
10.5 n=85 (88%); median age
26 years; consumption level
not reported
Brief intervention (BI): 15–20 min BI
(interventionist not reported) plus
information leaflet;
Control (minimal intervention): 5 min
empathic advice plus information leaflet
AUDIT-C positive or negative
% participants who reduced
consumption
% reduction in hazardous drinkers
Romelsjö et al. (1989)
[59]
Sweden, General
practice.
Randomisation unclear
Blinding adequate
Allocation concealment
adequate
4.5 n=83 (84%); 46.3 years;
Consumption per day =29.1 g
(100% ethanol)
Brief intervention: BI from GP (duration not
reported), GP decided frequency of
follow-up sessions (mean was 3);
Control: told by GP to cut down on alcohol
and that a follow-up examination would
occur after 1 year
Change in GGT
Change in self-reported alcohol
consumption
Change in a combined measure of
alcohol problems ‘problem
index’
Scott & Anderson
(1991) [60],
Anderson & Scott
(1992) [72]
Cut Down on
Drinking trial
UK, General
practice.
Randomisation adequate
Blinding adequate
Allocation concealment
adequate
10 n=226 (68%); 44.7 years;
Mean consumption last wk
(interview) =526 g (men)/
293 g (women); mean
quantity frequency drinking
last wk (HSQ) =439 g/
247 g
Brief intervention: 10 min BI from GP plus
booklet;
Control: no advice except at their own request
Change in weekly alcohol
consumption
Senft et al. (1997) [61] USA, General
practice.
Randomisation unclear
Blinding adequate
Allocation concealment unclear
9n=516 (71%); 42.4 years;
Mean weekly drinking
days =3.4; mean drinks per
drinking day =4.9
Brief intervention: 30 second message from
primary care clinician (GP, GP assistant or
nurse practitioner), plus 15 min BI from
health counsellor plus printed pack;
Control: standard care
Units in last 3 months
Drinking days week-1in last
6 months
Drinks per drinking day in last
6 months
Seppä (1992) [62] Finland, General
practice.
Randomisation unclear
Blinding unclear
Allocation concealment unclear
8.5 n=178 (79%); 53.2 years;
consumption level not
reported
Brief intervention: 5 BI sessions from GP
(duration not reported);
Control: no intervention
Self-report reduction in
consumption
MCV values
Tomson et al. (1998)
[63]
Sweden, General
practice.
Randomisation inadequate
Blinding unclear
Allocation concealment unclear
8.5 n=222 (81%); 45.2 years;
Mean consumption week-1at
baseline only for
intervention group =337 g
Brief intervention:2¥BI from nurse (duration
not reported);
Control: 1 appointment with GP to discuss
lifestyle in general
GGT
Wallace et al. (1988)
[64]
UK, General
practice.
Randomisation unclear
Blinding adequate
Allocation concealment unclear
8.5 n=909 (71%); 42.4 years;
Consumption week-1
(interview) =55.0 units;
(Q/F) =44.1 units
Brief intervention: BI from GP (duration not
reported) plus information booklet, 1–4
follow-up sessions;
Control: no intervention
Drinks week-1
ALAT/ALT, alanine aminotransferase (blood test); ASAT/AST, asparate aminotransferase (blood test); AUDIT, Alcohol Use Disorders Identification Test, self-repor t measure of alcohol intake, symptoms of
dependence, tolerance and alcohol-related negative consequences; AUDIT-C, consumption questions from AUDIT; BAC, blood alcohol content; CAGE, self-report measure of alcohol use, with CAGE being an
acronym formed by taking the first letter of keywords in four questions: cutting down, being annoyed by criticisms of one’s drinking, feeling guilty about one’s drinking and having an ‘eye-opener’ drink in the morning;
CDT, carbohydrate-deficient transferrin (blood test); ED, emergency department; FRAMES, framework for counselling encompassing Feedback, Responsibility, Advice, Menu of alternatives, Empathy, Self-efficacy;
GGT, gammaglutamyl transpeptidase (blood test); IU, international units; MALT, Munich Alcoholism Test; MAST, Michigan Alcoholism Screening Test, self-report measure of alcohol problems; MCV, mean
corpuscular volume; min, minute(s); N, number; PHC, primary health care; SF-12, health survey;T-ACE, self-report measure of alcohol problems, with T-ACE being an acronym formed by taking the first letter of
keywords in four questions: tolerance, annoyance, cut down and ‘eye-opener’; Q/F, quantity multiplied by frequency of drinking.
Brief intervention effectiveness 309
© 2009 Australasian Professional Society on Alcohol and other Drugs
Inclusion/exclusion criteria
Inclusion criteria in terms of alcohol consumption
varied and included the number of standard drink units
per week, a screening tool score, or evidence of binge
drinking. Patients were usually excluded if they were
severely alcohol dependent or already on an alcohol
treatment programme.
Screening
Screening questionnaires included: general health
questionnaires, such as the Health and Habits Survey
with embedded alcohol questions; established alcohol
screening tools, such as AUDIT, MAST, CAGE or
variations and/or combinations of these. Most trials
administered the screening tool in the clinic as soon as
the patient had registered for their appointment, but
one [46] administered the questionnaires by telephone
afterwards.
Control conditions
Categories of control treatment included: assessment
only [41–43,47,48,54,57,60,62]; ‘usual care’ which
included general practitioner or nurse advice to cut
down drinking [36–39,49,53,59,61–64]; the delivery of
a leaflet either on general health issues or specifically
about alcohol [40,44–46,51,56]; usual care plus a
leaflet [36,52,58] or treatment referral [38]. One trial
did not have a control condition [50], but compared
extended intervention with advice from a nurse who
gave the participants feedback about their GGT levels.
Treatment duration in the control groups ranged from
0 [38] to 10 min [42,58].
Brief interventions
Brief intervention took the form of: simple structured
advice, motivational counselling; cognitive behavioural
therapy; self-completed action plans; leaflets, either on
general health issues or specifically about alcohol;
requests to keep drinking diaries and exercises to com-
plete at home. Most trials (n=14) evaluated a single
brief intervention session [38,40,42,43,47,50–54,56–
58,60].The number of sessions ranged from one to five
sessions and lasted between 1 and 50 min, while total
intervention exposure time ranged from 5 [52] to
60 min [55]. The median brief intervention treatment
exposure was 25 min (interquartile range 7.5–30 min).
Extended interventions
Five trials evaluated extended interventions
[36,50,53,54,57] in which the number of sessions
delivered to patients ranged from two [53] to seven
[36,50]. The total duration of extended interventions
ranged from 52 [57] to 175 min [50]. The median
extended intervention treatment exposure was 100 min
(interquartile range 65–105 min).
Efficacy/effectiveness
Scores ranged from 4.5 [46,59] to 12 [52]; the median
was 9 and the interquartile range 7.5–10.5.
Baseline drinking
Baseline drinking was not reported in 8 trials
[40,43,47,53,58,61–63]. In the remaining 21 trials, the
mean baseline consumption ranged from 89 to 456 g
per week, with an overall mean across trials of 313 g per
week. Six trials reported baseline consumption for men
only and also reported the number of men randomised;
in these the mean baseline consumption was 377 g of
alcohol per week. Five trials reported baseline con-
sumption for women only and also reported the
number of women randomised; in these the mean base-
line consumption was 219 g of alcohol per week.
Three trials reported baseline measures of frequency
of drinking in terms of days drinking per week or month
[36,41,61] and the mean value was 2.9 days week-1.
Three trials reported baseline measures of frequency of
drinking in terms of number of binges per week
[44–46] and the mean value was 0.9 binges week-1.
Baseline intensity of drinking was measured in five
trials [36,40,41,54,61], in which the mean baseline
value was 110 g per drinking day.
Primary outcome
Twenty-four trials reported the quantity of alcohol
consumed in a specified time period, usually a week
or a month. Sixteen trials [36,37,39,40,42,44–
46,48,51,52,54,57,60,61,64] reported a final quantity
value with a corresponding standard deviation. One
trial [41] reported the final value of the quantity of
alcohol consumed per week and supplied us with
unpublished data on the corresponding standard devia-
tion. A further five trials [43,47,49,56,59] 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. These 22 trials were included in the primary
meta-analysis comparing the effects of brief interven-
tion against controls on the quantity of alcohol con-
sumed per week. Two of these trials [60,64] reported
the quantity of alcohol consumed per week both as
assessed by structured interview and a self-reported
questionnaire; we used the interview data.
310 E. F. S. Kaner et al.
© 2009 Australasian Professional Society on Alcohol and other Drugs
Two further trials [38,63] that reported the final
values of the quantity of alcohol consumed in a speci-
fied time period, but without the corresponding stan-
dard deviations, were included in the sensitivity analysis
with imputation of the standard deviations. Another
trial [55] reported final quantity values in a specified
time period but neither the corresponding standard
deviations nor the number of participants assessed and
so it could not be included in any meta-analysis.
Secondary outcomes
Nine trials reported outcomes on heavy drinking
[37,39,41–46,64]. However, the definition of heavy
drinking varied considerable between these trials and
included: the percentage of participants (generally
men) who drank over 35 units week-1[37,39,43,64];
the percentage of men and women who drank over 20
and 13 drinks week-1, respectively [44,45]; the percent-
age of men and women who drank over 30 and 25
drinks week-1, respectively [46]; and the percentage of
men and women who drank over 2 and 1 drinks day-1,
respectively [41].
Four trials [41,44,45,51] reported the final percent-
age of participants who were binge drinkers and three
[44,45,46] the frequency of binge drinking episodes.
Another common outcome was the frequency of
drinking in terms of number of days drinking each
week or month [36,41,61], or in terms of number of
binge drinking occasions each week or each month
[44,45,46]. One trial [38] reported the number of days
drinking each week but not the corresponding standard
deviations.
Five trials reported final values and standard devia-
tion of intensity of alcohol consumption in terms of
number of drinks per occasion [36,40,41,54,61]. One
trial [38] reported the number of drinks per occasion
but not the corresponding standard deviations and so it
could not be included in meta-analysis of this outcome.
Two trials [40,47] reported new injuries necessitat-
ing further visits to the Emergency Department or hos-
pital readmission, by treatment arm.
Three trials [52,53,59] reported other measures of
alcohol-related harm including: the Drinking Problems
Index [52]; the Drinker Inventory of Consequences
and a revised Injury Behaviour checklist [53] and a
problem index consisting of a summary measure con-
structed from the answers to six frequency questions
[59].
Five trials reported the final values and stand-
ard deviation of laboratory markers: either GGT
[36,50,59,64] or MCV [62].
Two trials reported health and quality of life out-
comes including: General Health Questionnaire scores
at 6 months and Euroqol EQ5D (standardised instru-
ment to measure health outcomes) at 1 year [40]; and
health-related quality of life via SF-12 plus an eco-
nomic evaluation based on health service use at 1 year
[52]. One of the reports of the TrEAT trial also
reported a benefit–cost analysis [65].
Outcomes by sex, age and time
Two trials reported on male and female participants in
separate papers [36,71,72,60] while five trials [57,60–
62,64] reported male and female outcomes separately
in the same paper. For the primary meta-analysis, data
on men and women in the same trial were combined.
Two trials [44,56] reported the final values of the quan-
tity of alcohol consumed in a specified time period by
sex but did not report the number of men and women
assessed and so could not be included in the meta-
analysis of quantity of alcohol consumed in a specified
time period, subgrouped by sex. Project TrEAT [44]
reported a 4 year follow up of a female subgroup from
the trial [68].
There was little analysis by age of subjects. One trial,
project GOAL focused specifically on older adults [45],
while the ELM trial [54] reported data on an elderly
subgroup in the trial [69]. Grossberg et al. reported on
outcomes for young adults in the TrEAT trial [67],
although these ‘young’ patients were aged between 18
and 30 years old.
Several trials reported outcomes at different time-
points: TrEAT reported after 1 and 4 years [44,75];
HEALTH reported after 6 months and 1 year [56,70]
and Alcoholscreen reported after 6 months and a year
[57]. Curry et al. [41] reported outcomes at 3 months
and 1 year. If outcomes were reported at several time-
points, data for 1 year follow up were used in the meta-
analyses if available. Aalto et al. [36] reported outcomes
at 1 year for the intervention groups but not for the
control group, so the outcomes at 3 years (reported for
all arms) were used in the meta-analyses.
Methodological quality
The overall methodological quality of the trials was
variable, perhaps not surprising since they took place
over a 20 years period. Randomisation was confirmed
as adequate in 12 trials [36–38,40,43,44,47,50,52,
54,56,60]. Allocation concealment was confirmed as
adequate in 10 trials [37,39,44,45,47,54,56,52,59,60].
In 18 trials the outcome assessors were blinded
[40,41,43–48,51–53,56–61,64]. Eleven trials reported
loss to follow up by individual arms [36,37,41,47,51,
52,54,57,58,60,62] and the overall loss to follow up
was 27%.
Four trials [39,49,52,56] were cluster randomised
although two [39,49] did not allow for the cluster ran-
domisation in their analysis.
Brief intervention effectiveness 311
© 2009 Australasian Professional Society on Alcohol and other Drugs
Eleven of the included trials reported an intention-
to-treat analysis [36,40,46,47,52,53,56,57,58,60,61].
Three of these [36,57,60] imputed final outcomes to be
the same as baseline values for participants who were
lost to follow up; these trials also reported data exclud-
ing participants lost to follow up, which we used in the
primary meta-analysis. Fleming et al. [46] imputed
values from the interview with the longest follow up to
the 11% of participants who were lost to follow up.The
remaining trials that reported an intention-to-treat
analysis excluded participants who were lost to follow
up. In the TrEAT trial [44], intention-to-treat was used
in one sub-analysis paper but not in the primary refer-
ence. In the remaining trials, it was not possible to
determine whether or not intention-to-treat analysis
was carried out [37–39,42,43,45,48–51,54,55,59,62–
64,76].
Meta-analytic findings
Primary outcome—Impact of brief intervention on drinks
per week
The primary meta-analysis included 22 trials [36,37,
39,40–49,51,52,54,56,57,59,60,61,64] which enrolled
7619 participants (median 247, range 83–909), with a
mean age of 43 years. All trials except one [42] reported
on sex and 67% of the participants were male. Only
eight trials [39,41,44,46,51,54,56,61] reported on eth-
nicity and 71% of the participants were white. Eighteen
trials reported outcomes after follow up of 1 year, while
four trials reported follow up at 3 years [36], 18 months
[43], 6 months [48] and 3 months [51].
Meta-analysis of follow-up data at 1 year included
5856 patients and showed that participants receiving
brief intervention drank less alcohol per week than those
receiving a control intervention (mean difference
=-38 g, 95%CI: -54 to -23 g week-1). There was no
significant difference between the pooled findings of the
effectiveness trials and the pooled findings of the efficacy
trials (see Figure 2). The 10 effectiveness trials showed
significant benefits of brief intervention (mean
difference =-33 g, 95%CI: -51 to -16 g week-1), and
the 12 efficacy trials showed a similar benefit of brief
intervention (mean difference =-45 g, 95%CI: -70 to
-19 g week-1). There was substantial heterogeneity
(I2=58%) between the findings of the trials.While all but
two trials [36,57] reported a benefit of brief intervention
compared with controls, the estimated benefit varied
substantially between the trials.
The forest plot showed no obvious relationship
between the effect of brief intervention and the efficacy/
effectiveness score and this was confirmed by meta-
regression, which showed a non-significant increase in
the effect of treatment by increasing score. A funnel plot
showed no evidence of publication bias (see Figure 3).
Meta-analysis restricted to the 10 trials that con-
firmed concealment of allocation [37,39,44,45,47,
52,54,56,59,60] to treatment group showed similar
results both for the effectiveness trials (mean
difference =-48 g, 95%CI: -65 to -31 g week-1), the
efficacy trials (mean difference =-71 g, 95%CI: -115
to -26 g week-1) and for all trials (mean difference
=-56 g, 95%CI: -75 to -36 g week-1), with moderate
heterogeneity (I2=33%) between trials. Seven of these
trials reported blinding of the outcome assessor and
seven reported adequate randomisation. Other sensitiv-
ity analyses showed similar outcomes after accounting
for subjects lost to follow up (assuming final values
were the same as baseline values) and missing standard
deviations (imputing the median SD values). The four
cluster randomised trials [39,49,52,56] showed similar
results (mean difference =-41 g, 95%CI: -73 to
-9gweek
-1) to the 18 individually randomised trials
(see above).
Treatment exposure
Although high treatment exposure resulted in a greater
net reduction in alcohol consumption than low treat-
ment exposure (the high exposure mean differences was
-51 g week-1, 95%CI: -75 to -27 g week-1compared
with the low exposure mean difference of -23 g week-1,
95%CI: -38 to -8gweek
-1), this difference was not
statistically significant. Meta-regression (see Figure 4)
showed a non-significant increase in the effect of treat-
ment with increasing treatment exposure (an increase
in the reduction in alcohol consumption of 1.0 g,
95%CI: -0.1 to 2.2 g week-1,P=0.09, for each
increase of 1 min in the treatment exposure). Although
there was no heterogeneity between the results of trials
that had low exposure to treatment, substantial hetero-
geneity (I2=72%) remained among trials with high
exposure to treatment.
Sex differences
In the trials that reported sufficient information (mean,
standard deviation and number of participants assessed
by treatment arm) to assess drinking outcomes by sex,
men experienced significant benefits of brief inter-
vention (mean difference =-57 g, 95%CI: -89 to
-25 g week-1) [36,37,39,49,57,60,61,64], but women
did not (mean difference =-10 g, 95%CI: -48 to
29 g week-1) [36,57,60,61,64]; nevertheless, the differ-
ence between men and women was not statistically
significant (see Figure 5). These results were based on a
sample of only 499 women.
Secondary outcomes
All nine trials that reported heavy drinking outcomes
[37,39,41–46,64] showed a reduction in the percentage
312 E. F. S. Kaner et al.
© 2009 Australasian Professional Society on Alcohol and other Drugs
of heavy drinkers in participants receiving the brief
intervention, although this reduction was statistically
significant for only six trials.
Four trials reporting on binge drinking outcomes
[41,44,45,51] showed a significant reduction in the
percentage of binge drinkers in the brief intervention
group compared with the control group (risk differ-
ence =-11%, 95%CI: -19% to -3%). However, three
trials reporting on the frequency of binge drinking
episodes [44–46] showed no significant reduction as a
result of brief intervention (mean difference =-0.3
binges week-1, 95%CI: -0.6 to 0.0 binges week-1);
with little difference between the findings of the two
effectiveness trials [44,45] and the one efficacy trial
[46].
Two effectiveness trials [36,41] and one efficacy trial
[61] reported the number of drinking days per week
and, overall, these showed no significant effect of
brief intervention compared with control (mean
difference =0.04 days, 95%CI: -0.5 to 0.4 drinking
days week-1).
Five trials reporting the amount of alcohol consumed
per drinking day [36,40,41,54,61] found no significant
reduction in intensity of drinking as a result of brief
intervention (mean difference =-3.1 g, 95%CI: -8.8 to
2.6 g per drinking day); and no statistically significant
difference between effectiveness [36,40,41] and effi-
cacy [54,61] trials.
Three trials reported GGT [36,59,64] and found no
significant difference between brief intervention and
control [mean difference =-1.1 IU (international
unit), 95%CI: -3.9 to 1.7 IU L-1]. One trial [62]
reported MCV; this showed no significant difference
between brief intervention and control, both overall
(mean difference =0.6 fL, 95%CI: -1.6 to 2.8 fL) and
for each sex separately.
One trial reported 0.5 fewer visits to Emergency
Department by the intervention group during the year
after randomisation [40]. A further trial reported a 47%
reduction in new injuries requiring either treatment in
the emergency department or readmission to trauma
services 1 year after brief intervention (hazard ratio
194 -254 (601) 215 -78 (993)
-38.4 (-54.2, -22.7)
-28.2 (-158.7, 102.4)
-65.2 (-148.9, 18.5)
-32.8 (-60.0, 5.5)
-17.0 (-53.3, 19.3)
-44.6 (-70.2, -19.1)
-108.8 (-235.4, 17.8)
-176.2 (-333.6, -18.9)
-42.8 (-183.6, 98.0)
-81.8 (-111.7, -51.9)
-2.4 (-27.8, 23.3)
-47.8 (-69.0, -26.5)
-77.5 (-172.5 17.5)
WMD (95% CI)
-19.7 (-54.1, 14.6)
-33.5 (-51.2, -15.9)
-65.3 (-137.7, 7.25)
15.5 (-76.2, 107.2)
-13.3 (-59.3, 32.7)
-92.8 (-144.4, -41.2)
36.0 (-35.9, 107.9)
-9.0 (-43.6, 25.6)
-76.2 (-115.8, -36.6)
-8.3 (-37.3, 20.6)
-33.7 (-121.3, 53.9)
-112.0 (-196.0, -28.0)
100.00
1.3
2.6
7.9
6.6
1.3
0.9
1.1
7.5
8.1
%
8.7
2.2
Weight
6.9
3.2
2.3
5.5
4.9
3.3
6.9
6.2
7.6
2.5
2.6
0
-300 -200 -100 0100
Alcohol intake in g week–1
Favours brief intervention Favours control
Study
Overall (I-squared = 57.8%, P = 0.000)
Lock 2006
Heather 1987
Ockene 1999
Huas 2002
Effectiveness trials
Subtotal (I-squared = 67.2%, P = 0.000)
Efficacy trials
Crawford 2004
Gentillelo 1999
Fernandez 1997
Wallace 1988
Curry 2003
Fleming 1997
Romelsjo 1989
Senft 1997
Subtotal (I-squared = 44.4%, P = 0.063)
Scott 1991
Aalto 2000
Maisto 2001
Cordoba 1998
Richmond 1995
Diez 2002
Fleming 1999
Fleming 2004
Kunz 2004
Altisent 1997
NMean(SD) NMean(SD)
36 129 (293) 42 157 (293)
104 202 (183) 125 295 (215)
235 -73 (146) 210 -41 (146)
206 294 (186) 186 303 (163)
189 458 (547) 195 566 (710)
353 138 (136) 370 186 (155)
82 278 (283) 73 263 (301)
80 245 (192) 70 311 (253)
270 -109 (165) 149 -92 (190)
70 326 (211) 61 290 (208)
78 119 (84) 67 195 (146)
100 109 (99) 122 111 (93)
196 141 (177) 215 161 (177)
363 304 (201) 385 386 (233)
29 253 (156) 62 318 (247)
34 168 (167) 30 280 (174)
38 -107 (370) 50 -65 (278)
90 201 (309) 104 234 (312)
74 134 (148) 85 147 (148)
36 -35 (209) 36 43 (202)
81 58 (106) 70 66 (74)
46.3
53.7
WMD (95% CI)
Brief intervention Control
Figure 2. Forest plots corresponding to primary meta-analysis: estimated mean difference in alcohol intake between brief intervention and
control groups in each trial and overall effect, subgrouped by effectiveness/efficacy trials. N, number of participants assessed in each group;
Mean (SD), final value or change score (and its standard deviation) for alcohol intake in g week-1in each group;WMD, weighted mean
difference in alcohol intake (or change in alcohol intake) between brief intervention and control groups; %Weight, weight given to this trial
in random effects meta-analysis.
Brief intervention effectiveness 313
© 2009 Australasian Professional Society on Alcohol and other Drugs
0.53, 95%CI: 0.26–1.07, P=0.07) and a 48% reduc-
tion in inpatient hospital readmissions for injury treat-
ment at 3 years follow up (hazard ratio 0.52, 95%CI:
0.21–1.29) but no significant differences in the death
rate between the intervention and control groups (2.7%
and 2.3%, respectively) [47].
Two trials that reported health-related quality of life
measures found no significant differences between
-200
-150
-100
-50
0
50
4681012
Effectiveness/efficacy score of trial
Treatment effect
Efficacy trials Effectiveness trials
-200
-150
-100
-50
0
50
4681012
Effectiveness/efficacy score of trial
Treatment effect
Efficacy trials Effectiveness trials
Figure 3. Estimated treatment effect versus effectiveness/efficacy score for trials comparing brief intervention with control: the predicted
meta-regression line and its 95%CI.The treatment effect is the net reduction in alcohol intake in g week-1;the effectiveness/efficacy score was
estimated as described in Appendix 3.
-200
-150
-100
-50
0
50
010 20 30 40 50
-200
-150
-100
-50
0
50
010 20 30 40 50
Treatment exposure in trial
Treatment effect
Figure 4. Estimated treatment effect versus treatment exposure for trials comparing brief intervention with control: the predicted
meta-regression line and its 95%CI. The treatment effect is the net reduction in alcohol intake in g week-1;the treatment exposure is the
estimated duration of the brief intervention in minutes.
314 E. F. S. Kaner et al.
© 2009 Australasian Professional Society on Alcohol and other Drugs
treatment and control groups at 12 months [40,52].
However, Fleming et al. [66] reported cost–benefits of
brief intervention, particularly in the reduced use of
public services, at a 4 year follow up in the USA.
Extended intervention
Finally, a meta-analysis of four trials [36,50,54,57]
enrolling 684 participants showed that participants
receiving an extended intervention drank less alcohol
per week than those receiving a brief intervention
(mean difference =-28 g, 95%CI: -62 to 6 g week-1),
with little heterogeneity (I2=0%), but this finding was
not statistically significant.
Discussion
Do brief interventions work?
The primary meta-analysis of 22 trials assessed out-
comes in 5856 patients and showed that, compared
with control conditions, brief intervention reduced the
quantity of alcohol drunk by 38 g per week, which
equates to approximately 4-5 UK standard drink units.
These results were robust: several sensitivity analyses
were carried out and all yielded similar results. All of
these sensitivity analyses showed a statistically signifi-
cant benefit of brief intervention and a funnel plot
showed no evidence of publication bias. Thus we con-
clude that brief alcohol interventions are effective at
reducing excessive drinking in primary care settings.
Weekly alcohol consumption was the most com-
monly reported outcome in this field of work. However,
a wide range of other outcomes was also reported. Of
these secondary outcomes, we found benefits of brief
intervention in terms of reduced percentages of patients
who were classified as binge drinkers or as heavy
drinkers. Although we did not meta-analyse the latter
outcome because of different definitions of heavy drink-
ing across trials, all nine trials that considered this
outcome reported a significant decrease in heavy drink-
ing in the brief intervention group. However, there were
non-significant decreases in the frequency and intensity
WMD (95% CI)WMD (95% CI)
-38.7 (-64.9, -12.4)
-42.0 (-155.0, 71.0)
31.0 (-79.5, 141.5)
WMD (95% CI)
-54.4 (-101.5, -7.3)
27.0 (-47.9, 101.9)
-23.2 (-94.8, 48.4)
-9.5 (-48.3, 29.2)
-77.0 (-172.8, 18.8)
-92.8 (-130.6, -55.0)
-112.0 (-196.0, -28.0)
-30.5 (-73.0, 12.0)
-17.0 (-53.3, 19.3)
-3.6 (-43.4, 36.2)
-57.1 (-88.7, -25.4)
133.0 (-34.2, 300.2)
-92.8 (-144.4, -41.2)
100.00
4.0
4.1
Weight
10.2
6.8
7.1
37.4
5.0
11.5
5.9
10.8
11.7
11.2
62.6
%
2.1
9.5
0-200 -100 0100
Alcohol intake in g week–1
Favours brief intervention Favours control
Overall (I-squared = 58.9%, P = 0.004)
Aalto 2000
Richmond 1995
Wallace 1988
Richmond 1995
Scott 1991
Subtotal (I-squared = 45.0%, P = 0.122)
Men
Scott 1991
Women
Wallace 1988
Altisent 1997
Senft 1997
Huas 2002
Senft 1997
Subtotal (I-squared = 55.5%, P = 0.028)
Study
Aalto 2000
Cordoba 1998
NMean(SD) NMean(SD)
104 202 (183) 125 295 (215)
57 278 (217) 49 320 (350)
55 363 (224) 45 440 (258)
270 -109 (165) 149 -92 (190)
39 393 (220) 31 362 (245)
143 158 (184) 147 189 (184)
257 352 (205) 273 445 (238)
34 168 (167) 30 280 (174)
25 279 (395) 24 146 (158)
25 190 (118) 25 213 (139)
31 242 (169) 30 215 (127)
53 96 (111) 68 100 (111)
106 189 (124) 112 243 (220)
WMD (95% CI)
Brief intervention Control
Figure 5. Forest plots showing estimated mean difference in alcohol intake between brief intervention and control groups in each trial and
overall effect, subgrouped by trials enrolling men only and women only. N, Number of participants assessed in each group; Mean (SD), final
value or change score (and its standard deviation) for alcohol intake in g week-1in each group;WMD, weighted mean difference in alcohol
intake (or change in alcohol intake) between brief intervention and control groups; % Weight, weight given to this trial in random effects
meta-analysis.
Brief intervention effectiveness 315
© 2009 Australasian Professional Society on Alcohol and other Drugs
of drinking and in laboratory markers of alcohol con-
sumption; this lack of statistically significant findings
might be due to a lack of statistical power as only five
trials reported frequency or intensity outcomes and
only four trials reported laboratory markers of alcohol
consumption. Given the wide range of outcome mea-
sures reported in trials of brief alcohol intervention and
the fact that some are reported in just a small number
of trials, it would be helpful if future research focused
on fewer, key indicators of alcohol-related risk and
harm and on standardising the definition of these key
indicators. In particular, we found a confusing variabil-
ity in definitions of heavy and binge drinking.
Who do brief interventions work for?
Overall, approximately 70% of the participants in the
brief intervention trials were men. Only eight trials
reported sufficient information to analyse outcomes
separately by sex, of which three reported data for men
only and five included separable data on women. In
these trials, brief intervention significantly reduced the
quantity of alcohol consumed per week by men, but not
by women. This result of no significant benefit of brief
intervention among women is in contrast to the previ-
ous meta-analytic findings of Ballasteros and colleagues
[77], partly because of the inclusion of different trials
and partly because of different analytical approaches.
We included the Lahti trial [36,71] whereas the earlier
review did not, while the latter included two trials
[44,56] which we excluded from the sex analysis
because we could not ascertain the numbers of men
and women followed up (we did not assume that these
were the same as the numbers enrolled in the study). In
addition, Ballesteros et al. [77] used a fixed effects
model in their meta-analysis whereas we used a more
conservative random effects model, primarily because
of the heterogeneity that we found between trials. Nev-
ertheless, it is clear that the current published data do
not provide definitive evidence on the impact of brief
interventions in women. Although approximately 1700
subjects followed up in brief intervention trials were
women, we could only include just over a quarter of this
group in the sex-specific meta-analysis. As other com-
mentators have reported greater non-specific reactivity
in women (indicated by consistent reductions in drink-
ing in both intervention and control groups) [78], it is
essential to have an adequately sized population of
women in which to evaluate potentially small effects of
brief alcohol intervention. Thus there is a real need
either for previous trials to report their sex-specific
outcomes or for more prospective brief alcohol inter-
vention research specifically focused on women.
Of the 29 trials identified by this review, 19 (65%)
were based in English-speaking countries (USA, UK,
Canada, Australia), six (21%) in continental Europe,
and four (14%) in Scandinavia. No studies were based
in transitional or developing countries. In addition, eth-
nicity was poorly reported; in those trials which did
report it, approximately 70% of participants were
white. Thus there is a need for more brief intervention
research in a wider range of countries and in different
ethnic groups within populations.
We also found a paucity of data relating to older and
younger people who are hazardous or harmful drinkers.
Just one trial specifically focused on older people [45]
and another reported an older subgroup analysis from a
larger trial [69]. However, there is a need for more
research on older hazardous and harmful drinkers. In
addition, there was only one subgroup analysis of young
adults which came from the TrEAT trial which was
conducted in the USA [67]. However, the definition of
young adults in this study included 18–30 year-old
people. Thus this report has limited applicability to
underage drinkers.
How much is too much alcohol?
On entering the trials, participants consumed, on
average, 310 g of alcohol per week, but this varied
between trials from 90 to 460 g per week. Looking
across this body of work, there appears to have been a
reduction over time in the levels of alcohol consump-
tion that are regarded as being excessive.This might be
due to broader notions of what constitutes alcohol-
related risk and particularly an increasing focus on
hazardous (asymptomatic) as well as harmful (symp-
tomatic) drinking. However, in some trials, participants
receiving brief alcohol intervention were clearly below
the ‘at risk’ threshold used in other trials. For instance
in Curry et al. [41], participants consumed an average
of 14.2 drinks per week and just 5% of the study popu-
lation averaged four or more drinks per day. Indeed,
Curry et al. [41] reported selecting drinking patterns to
encompass primary and secondary prevention goals.
However, traditionally, brief intervention has been sec-
ondary preventive work [59], addressing early-stage
disease in a subset of patients who are hazardous or
harmful drinkers. Primary prevention, by contrast,
focuses on whole population approaches and includes
people who might not currently be ‘at risk’ because of
their drinking. This changing focus for brief interven-
tion work might explain why some recent trials of brief
intervention report more modest changes in drinking
behaviour than earlier trials.
How long do brief interventions need to be?
Trials involving longer brief interventions (high treat-
ment exposure) did not report significantly greater
316 E. F. S. Kaner et al.
© 2009 Australasian Professional Society on Alcohol and other Drugs
reductions in alcohol consumption than trials evaluat-
ing shorter brief interventions (low treatment expo-
sure). Subsequent comparison of extended and brief
interventions allowed a direct comparison of low and
high treatment exposure, free of confounding from
other factors. Again there was no significant benefit of
extended brief intervention in reducing alcohol con-
sumption; although this result was based on just four
trials. One trial [36] did report a benefit of extended
intervention in terms of a significant reduction in the
frequency of drinking. Thus there is some weak evi-
dence that longer counselling might produce greater
reductions in alcohol consumption. However, this was
of the order of a possible reduction of one standard
drink or less per week for 10 min extra counselling.
Thus, given this weak relationship between duration of
counselling and outcome, it seems likely that the struc-
ture and content brief interventions might have more
influence on patients’ drinking than the total duration
of intervention delivery.
Are brief interventions real or ideal world interventions?
Most of the trials in this field appear to have been
designed and conducted in a clinically meaningful way.
In our efficacy or effectiveness classification, the major-
ity of trials scored over the midpoint value (which was
6) and so had specific design features which increased
the external generalisability of the findings. In addition,
there was no significant difference between trials clas-
sified as efficacy and effectiveness trials in the effect of
brief intervention on the quantity of alcohol consumed.
Our meta-regression showed no significant relationship
between the estimated treatment effect and the efficacy/
effectiveness score of the trial. Thus not only do brief
interventions appear to be effective at reducing alcohol
consumption in primary care patients, but this body of
published work also seems to be relevant to the real
world of clinical practice.
The lack of difference in outcomes from efficacy and
effectiveness trials could indicate insensitivity in our
classification scheme. In some papers, authors did not
report information relating to certain items. In these
cases, we ascribed a mid-value score for that item so as
not to tip the trial towards either the efficacy or effec-
tiveness domain. This might have reduced variation in
the final scores (there were no extreme scores in the
efficacy domain) and led to a clustering of trials
towards the middle of this scale. However, the range of
scores was 4.5–12 (of a possible 0–12) and there was no
obvious clustering of scores on a visual plot of the trials.
In retrospect, it is unlikely that there would be any pure
efficacy studies (score 0), as a trial protocol would need
to be acceptable and relevant to clinicians (and ethics
committees) before it could be enacted in a health
service context. Nevertheless, future work will explore
the properties of our classification scheme including the
validity of this method of assessing efficacy and effec-
tiveness trials. In addition, it is possible that the brief
intervention ‘treatment effect’ might be related to some
of the individual factors which were combined in the
overall (summative) score. Thus investigating the rela-
tionship between individual items in the classification
scheme and treatment effect will be the subject of
future work focused on trying to identify the ‘active
ingredients’ of brief intervention effects.
Are there screening and/or assessment effects?
It was clear in this review that many studies reported
reductions in alcohol consumption in the control
groups. As other studies have reported similar findings,
it has been suggested that screening alone can reduce
alcohol consumption. McIntosh et al. [55] reported
that a significant proportion of patients reduced their
drinking between screening and assessment, and thus
that brief intervention was delivered to some patients
who were no longer eligible for it. However, apparent
‘screening effects’ are most likely to be explained by
regression to the mean [79,80]. If a group of people are
selected on the basis of having a measure of a particular
characteristic (e.g. alcohol consumption) above a
certain value, then a second measure of this character-
istic will, on average, be lower than the first value and
closer to the mean of the entire population. Genuine
reductions in alcohol consumption because of an inter-
vention can only be separated from regression to the
mean by comparing participants receiving the interven-
tion with a randomised control group. Thus inferences
made on the basis of trends in both groups (or trends in
the control group alone) are likely to be flawed.
However, recent trial-based work [81] has shown that
assessment processes in brief intervention trials can
positively affect alcohol consumption. In some respects,
this might not be surprising as assessment processes
can last much longer than the brief interventions them-
selves and they generally focus on alcohol and its
effects. This evidence of an ‘assessment effect’ and our
finding that even very short brief interventions can
reduce excessive drinking clearly indicate a benefit to
patients of clinician enquiry and advice about alcohol-
related risk and harm.
Strengths and weaknesses of this review
There was substantial heterogeneity between trials in
the settings (primary care or accident and emergency),
populations enrolled, screening instruments used, base-
line consumption of alcohol and the active and control
interventions delivered. Hence, the statistical heteroge-
Brief intervention effectiveness 317
© 2009 Australasian Professional Society on Alcohol and other Drugs
neity in our meta-analyses is not surprising. Subgroup
analyses showed that heterogeneity of findings was
greatest in individually randomised trials and trials with
high treatment exposure.
Empirical research has shown that failure to conceal
from participants and treatment providers the alloca-
tion of participants to treatment groups is often related
to over-estimation of the treatment effect [82,83], and
that trials where the participant and treatment provider
are not blinded might be more likely to report signifi-
cant effects of the intervention [82]. Although the 22
trials in the primary meta-analysis were of variable
quality, 10 reported adequate allocation concealment
and seven of these reported adequate blinding.
However, sensitivity analysis restricted to trials which
reported adequate concealment of allocation showed a
significant benefit of brief intervention similar to that
found in the primary meta-analysis, but with less
heterogeneity between trials. Thus poor quality trials
are unlikely to have introduced much bias in our
meta-analysis.
The most likely source of bias in the trials was the
loss to follow up of subjects, which was approximately
27% overall and significantly higher in brief interven-
tion than control groups (difference in rates of 3%,
95%CI: 1–6%). If participants who dropped out of the
brief intervention groups had higher alcohol consump-
tion than those who did not, the estimated reduction in
alcohol consumption because of brief intervention
would be an overestimate of the real effect. Neverthe-
less, the estimated reduction in the quantity of alcohol
consumed per week was sufficiently marked that the
real effect is likely to be a reduction in alcohol con-
sumption.
Furthermore, the random effects model which was
used assumed that the effect of treatment is different in
different populations and that the estimated reduction
in alcohol consumption of 38 g per week is the mean
treatment effect, averaged over all populations. There-
fore, the findings provide strong evidence that brief
interventions are effective in many populations.
Comparison with other meta-analyses
Since the early 1990s, there has been a steadily growing
number of reviews of brief alcohol interventions
[13–20,84–90]. Each review is bound by the time that
it was conducted, and by specific research or clinical
questions. These parameters affect both the number of
trials identified by the review and the precise set of
studies included in pooled analyses. Nevertheless, it is
important to set our findings into the context of this
large body of work and consider any discrepancies.
Moyer et al. has been the most comprehensive
review, identifying 56 controlled trials of brief alcohol
intervention [16]. This review [16] incorporated both
subjects who were seeking treatment for alcohol prob-
lems which is typical in secondary care and specialist
alcohol work (n=22 trials) and non-treatment seekers
(n=34 trials) who would be more typical of patients in
primary care. However, the non-treatment seeking indi-
viduals in these 34 trials [16] came from a range of
social care, occupational health settings, hospitals and
educational contexts; just 20 trials were based in
primary care [16]. Moreover, in some of the primary
care trials, brief interventions were provided by a range
of doctors, nurses, medical assistants, psychiatrists,
social workers and psychologists (e.g. the WHO multi-
national collaborative study 1996 which was reported
as seven separate trials). As we could not clearly identify
specific primary care data (either the setting or the
delivery agent), we excluded these trials from our
review.
Eight reviews have specifically focused on brief inter-
ventions in primary care [17,18,84–89]; six in general-
practice-based care [17,18,84,85–88] and one in
emergency care [89]. Two early reviews included a
limited number of trials [17,18] and two more did not
conduct a formal meta-analysis of alcohol-specific out-
comes [84,87]. A recent review based in emergency
care included both controlled and uncontrolled evalu-
ations and intervention by a range of delivery agents
including computers.Thus it is difficult to compare our
findings with these reports. A further systematic review
and meta-analysis focused on the question of whether
screening was an efficient means of identifying patients
for subsequent brief intervention [86]. The authors
concluded that screening was not practical in general
practice as they estimated that just two or three patients
per thousand might ultimately benefit from screening
and brief intervention. However, this analysis was ques-
tionable as it used data from trials designed to evaluate
the impact of brief intervention per se (with various
inclusion and exclusion criteria governing patients’ par-
ticipation in the trials) and not designed to address the
question of screening in routine practice.
Two recent reviews [85,88] focused specifically on
general practice-based primary care and reported meta-
analysis of brief intervention outcomes from 19 trials
and 13 trials, respectively. Of these, Bertholet et al. [88]
conducted the most recent and directly comparable
meta-analysis to our work. In this review, 12 entries
were included in their pooled analysis as shown in a
forest plot although three trials contributed two sepa-
rate entries each; single and multiple-session interven-
tions from a single trial were considered separately.
There were several reasons for the different trials
included in this earlier meta-analysis and ours. The
search strategy in the earlier work [88] went up to 2003
whereas ours went up to 2006. Hence, we found four
318 E. F. S. Kaner et al.
© 2009 Australasian Professional Society on Alcohol and other Drugs
more recent trials. Bertholet et al. [88] focused only on
general practice-based studies whereas our definition of
primary care included four Accident and Emergency
trials. Betholet et al. [88] also included two of the
studies from the WHO collaborative trial (see above)
which were not in our review. Finally, Bertholet et al.
[88] excluded three trials [59,60,64] which we
included; their exclusion was based on the fact that
screening involved a postal questionnaire survey of lif-
estyle issues which they did not consider to be repre-
sentative of routine primary care. However, the patients
in these trials were genuine practice patients who were
not seeking alcohol treatment and so were likely to be
unaware of alcohol-related risk or harm. Moreover, we
felt that our efficacy/effectiveness classification would
address the additional work carried out by practices in
the postal screening stage; hence, we included these
trials as did other meta-analyses [16,18,85].
A further contrast between our review and earlier
work is that previous meta-analyses have reported male
and female outcomes from a single trial (judged by the
trial protocol) as separate studies [16,18,77,85,88]
while we combined them in our primary meta-analysis.
Treating findings from male and female participants in
the same trial as the results of independent trials gives
slightly more weight to the trial than it should be
accorded, but is unlikely to lead to any major difference
in the overall pooled findings of the meta-analysis.Also,
earlier reviews [85,88] reported intention-to-treat
analyses which imputed zero change to all subjects lost
to follow up. In addition, Moyer et al. [16] noted that if
study results were described as, or inferred to be non-
significant, the an effect size of zero was assigned to the
trial; this related to 13% of studies.These assumptions
seem to be very conservative. Our analysis attempted to
analyse all participants in the groups to which they were
randomised but included imputed outcomes only in
sensitivity analyses.
Nevertheless, our meta-analysis yielded similar
results to previous findings. Bertholet et al. [88] used a
random effects model to analyse alcohol consumption
and reported an adjusted intention-to-treat analysis
showing a mean pooled difference of -38 g alcohol
per week (95%CI -51 to -24 g week-1), equating to
approximately four fewer drinks per week. Six years
earlier, Poikolainen [18] reported that multi-session
brief interventions produced a pooled effect estimate of
change in alcohol consumption of -51 g (95%CI -74
to -29 g week-1). At least some of the change from 1999
to 2005 might reflect the fact that earlier trials of brief
alcohol intervention tended to focus on heavier (or
harmful) drinkers whereas recent work has included
less heavy or hazardous drinkers with a reduced range
for consumption to fall within recommended sensible
drinking limits [85].
Implications for the future
In the field of brief alcohol intervention, it has been
reported that most of the trials to date have been tightly
controlled efficacy studies and not particularly repre-
sentative of routine clinical practice [91]. One could
argue that any clinical trial can never be a true analogue
for clinical practice. However, randomised controlled
trials remain the gold standard for evaluating the out-
comes of psychosocial or pharmacological interven-
tions in health care. Our analysis suggests that most of
the randomised controlled trials of brief alcohol inter-
vention tend to be effectiveness studies, which include
patients, clinicians and practices that are representative
of primary care. Thus this body of brief alcohol inter-
vention research seems to be applicable to routine clini-
cal practice.
There is a need for more research on brief interven-
tions with women, older and younger drinkers and
those from minority ethnic groups. In addition, there
should be more work carried in transitional and devel-
oping countries where drinking cultures might differ
from the developed world. However, given the large
number of trials of brief alcohol intervention showing
consistently positive outcomes in men, there is no need
for more of the same before such interventions are
delivered in primary care. Indeed, as extended treat-
ment had little additional benefit over brief interven-
tion, it seems that primary care intervention for
alcohol-risk reduction can be short and effective. Thus
we recommend that brief interventions should be deliv-
ered to hazardous and harmful drinkers in both general
practice-based primary care and emergency depart-
ments to promote public health improvement. Future
research should focus on promoting the uptake and use
of brief interventions in primary care and on identifying
the element of brief interventions that promote success-
ful behaviour change in patients.
Acknowledgements
Professor Kaner would like to thank all her co-authors
who carried out this extensive piece of research without
specific funding for much of the work and with a tre-
mendous amount of good will. Professor Kaner was
supported by a personal NHS Career Scientist fellow-
ship and small grants from the Cochrane Collaboration
and Newcastle University. The research team would
also like to thank Dr Marta Roque of the Iberian
Cochrane Centre for assistance with language transla-
tion and data abstraction of studies published in
Spanish. We thank Dr Jean-Bernard Daeppen of the
Centre de Traitement en Alcoologie, Lausanne, and Dr
Nicolas Bertholet of the Service de Psychiatrie de
Liaison, Lausanne, for comments on the review.
Brief intervention effectiveness 319
© 2009 Australasian Professional Society on Alcohol and other Drugs
This paper is based on a Cochrane review published
in the Cochrane Library 2007, Issue 2 and updated in
2008, Issue 1 (see http://www.thecochranelibrary.com
for information). Cochrane reviews are regularly
updated as new evidence emerges and in response to
feedback, and the Cochrane Library should be con-
sulted for the most recent version of the review.
References
[1] Rehm J, Room R, Monteiro M, et al. Alcohol as a risk factor
for global burden of disease. Eur Addict Res 2003;9:157–
64.
[2] Ezzati M, Rodgers A, Vander Hoorn S, Murray C. Selected
major risk factors and global and regional burden of disease.
Lancet 2002;360:1347–60.
[3] Anderson P. Alcohol as a key area. BMJ 1991;303:766–8.
[4] Prime Minister’s Strategy Unit. Alcohol harm reduction
strategy for England. London: Cabinet Office, 2004.
[5] World Health Organisation. Global Status Report: alcohol
policy. Geneva, Switzerland 2004.
[6] Kreitman N. Alcohol consumption and the preventive
paradox. Br J Addict 1986;81:353–63.
[7] Anderson P. Alcohol and primary health care. World Health
Organisation Regional Publications European Series
1996;64:1–90.
[8] Wutzke SE, Conigrave KM, Saunders JB, Hall WD. The
long-term effectiveness of brief interventions for unsafe
alcohol consumption: a 10-year follow-up. Addiction
2002;97:665–75.
[9] Heather N. Interpreting the evidence on brief interventions
for excessive drinkers: the need for caution. Alcohol Alcohol
1995;30:287–96.
[10] Kaner EFS, Lock CA, McAvoy BR, Heather N, Gilvarr y
EA. RCT of three training and support strategies to en-
courage implementation of screening and brief alcohol
intervention by general practitioners. Br J Gen Pract
1999;49:699–703.
[11] Thom B, Herring R, Judd A. Identifying alcohol-related
harm in young drinkers: the role of accident and emergency
departments. Alcohol Alcohol 1999;34:910–5.
[12] Drummond C, Phillips T, Coulton S, et al. National preva-
lence survey of alcohol-related attendances at accident and
emergency departments in England. Alcohol Clin Exp Res
2005;29:6A.
[13] Agosti V. The efficacy of treatments in reducing alcohol
consumption: a meta-analysis. Int J Addict 1995;30:1067–
77.
[14] Bien TH, Miller WR, Tonigan JS. Brief interventions for
alcohol problems: a review. Addiction 1993;88:315–35.
[15] Department of Health. The health of the nation: a summary
of the strategy for health in England. London:HMSO, 1992.
[16] Moyer A, Finney JW, Swearingen CE, Vergun P. Brief
interventions for alcohol problems: a meta-analytic review
of controlled investigations in treatment-seeking and non-
treatment-seeking populations. Addiction 2002;97:279–92.
[17] NHS Centre for Reviews and Dissemination. Brief inter-
vention and alcohol use. Eff Healthc 1993;1:13.
[18] Poikolainen K. Effectiveness of brief interventions to reduce
alcohol intake in primary health care populations: a meta-
analysis. Prev Med 1999;28:503–9.
[19] Wilk AI, Jensen NM, Havighurst TC. Meta-analysis of ran-
domized control trials addressing brief interventions in
heavy alcohol drinkers. J Gen Intern Med 1997;12:274–83.
[20] Whitlock EP, Polen MR, Green CA, Orleans T, Klein J.
Clinical guidelines. Behavioral counseling interventions in
primary care to reduce risky/harmful alcohol use by adults:
a summary of the evidence for the U.S. Preventive Services
Task Force. Ann Intern Med 2004;140:557–68.
[21] Babor TF, Higgins-Biddle JC, Dauser D, Burleson JA,
Zarkin GA, Bray J. Brief interventions for at-risk drinking:
patient outcomes and cost-effectiveness in managed care
organizations. Alcohol Alcohol 2006;41:624–31.
[22] Flay BR. Efficacy and effectiveness trials (and other phases
of research) in the development of health promotion pro-
grams. Prev Med 1986;15:451–74.
[23] Edwards AG, Rollnick S. Outcome studies of brief alcohol
intervention in general practice: the problem of lost sub-
jects. Addiction 1997;92:1699–704.
[24] Kaner E, Beyer F, Dickinson H, et al. Brief interventions
for excessive drinkers in primary health care settings.
Cochrane Database Syst Rev 2007; issue 2; doi:10.1002/
14651858.CD004148.pub3.
[25] Kristenson H, Ohlin H, Hulten-Nosslin MB, Trell E, Hood
B. Identification and intervention of heavy drinking in
middle-aged men: results and follow-up of 24–60 months of
long-term study with randomized controls. Alcohol Clin
Exp Res 1983;203–9.
[26] Babor TF, Grant M, Acuda W, et al. Randomized clinical
trial of brief interventions in primar y health care: summary
of a WHO project. Addiction 1994;89:657–60.
[27] Alderson P, Green S, Higgins JPT, eds. Cochrane Review-
ers’ Handbook 4.2.2 [updated March 2004]. In: The
Cochrane Library, Issue 1. Chichester, UK: John Wiley &
Sons Ltd, 2004.
[28] Shadish WR, Matt GE, Navarro AM, Phillips G.The effects
of psychological therapies under clinically representative
conditions: a meta-analysis. Psychol Bull 2000;126:512–
29.
[29] Miller WR, Heather N, Hall W. Calculating standard drink
units: international comparisons. Br J Addict 1991;86:43–7.
[30] Gual A, Rodriguez MA, Lligona A, Llopis J. Does the
concept of a standard drink apply to viticultural societies?
Alcohol Alcohol Int J Med Counc Alcohol 1999;34:153–60.
[31] Higgins JPT, Thompson SG. Quantifying heterogeneity in
a meta-analysis. Stat Med 2002;21:1539–58.
[32] Deeks JJ, Altman DG, Bradburn MJ. Statistical methods for
examining heterogeneity and combining results from
several studies in meta-analysis. Systematic reviews in
health care, 2nd edn. London: BMJ books, 2001.
[33] DerSimonian R, Laird N. Meta-analysis in clinical trials.
Control Clin Trials 1986;7:177–88.
[34] Review Manager (RevMan). Version 4.2 for Windows
Copenhagen: The Nordic Cochrane Centre. The Cochrane
Collaboration; 2003.
[35] StataCorp. Stata Statistical Software: Release 9.2; 2006.
[36] Aalto M, Saksanen R, Laine P, et al. Brief intervention for
female heavy drinkers in routine general practice: a 3-year
randomized, controlled study. Alcohol Clini Exp Res
2000;24:1680–6.
[37] Altisent R, Cordoba R, Delgado MT, Pico MV, Melus E,
Aranguren F. Multicenter study on the efficacy of advice for
the prevention of alcoholism in primary health care. Med
Clin (Barc) 1997;109:121–4.
[38] Chang G, Behr H, Goetz MA, Hiley A, Bigby J. Women and
alcohol abuse in primary care: identification and interven-
tion. Am J Addict 1997;6:183–92.
[39] Cordoba R, Delgado MT, Pico V, Altisent R, Fores D,
Monreal A. Effectiveness of brief intervention on non-
320 E. F. S. Kaner et al.
© 2009 Australasian Professional Society on Alcohol and other Drugs
dependent alcohol drinkers (EBIAL): a Spanish multi-
centre study. Fam Pract 1998;15:562–8.
[40] Crawford MJ, Patton R, Touquet R, et al. Screening and
referral for brief intervention of alcohol-misusing patients in
an emergency department: a pragmatic randomised con-
trolled trial. Lancet 2004;364:1334–9.
[41] Curr y SJ, Ludman EJ, Grothaus LC, Donovan D, Kim E. A
randomized trial of a brief primary-care-based intervention
for reducing at-risk drinking practices. Health Psychol
2003;22:156–65.
[42] Diez JF, Pena C, Garcia E, Gaite L. Brief intervention in
Cantabria (Spain) in alcohol related problems. [Interven-
cion breve en Cantabria en problemas relacionados con el
alcohol]. Adicciones 2002;14:13–24.
[43] San Martin F Bermejo Caja MI, Alonso CJ, et al. Effective-
ness of brief medical counseling to reduce drinkers’ alcohol
consumption. Aten Primaria 1997;19:127–32.
[44] Fleming MF, Barry KL, Manwell LB, Johnson K, London
R. Brief physician advice for problem alcohol drinkers: a
randomized controlled trial in community-based primary
care practices. JAMA 1997;277:1039–45.
[45] Fleming MF, Manwell LB, Barry KL, Adams W, Stauf-
facher EA. Brief physician advice for alcohol problems in
older adults: a randomized community-based trial. J Fam
Pract 1999;48:378–84.
[46] Fleming M, Brown R, Brown D. The efficacy of a brief
alcohol intervention combined with %CDT feedback in
patients being treated for type 2 diabetes and/or hyperten-
sion. J Stud Alcohol 2004;65:631–7.
[47] Gentillelo LM, Rivara FP, Donovan DM, et al. Alcohol
interventions in a trauma center as a means of reducing the
risk of injury recurrence. Ann Surg 1999;230:473–83.
[48] Heather N, Campion PD, Neville RG, Maccabe D. Evalu-
ation of a controlled drinking minimal intervention for
problem drinkers in general practice (the DRAMS scheme).
J R Coll Gen Pract 1987;37:358–63.
[49] Huas D, Pessione F, Bouix JC, Demeaux JL, Alemand H,
Rueff B. Efficacite a un an d’une intervention breve aupres
des consommateurs d’alcool a problemes. Rev Pract Med
Gen 2002;16:1343–8.
[50] Israel Y, Hollander O, Sanchez-Craig M, Booker S. Screen-
ing for problem drinking and counseling by the primary
care physician-nurse team. Alcohol Clin Exp Res 1996;20:
1443–50.
[51] Kunz MF, French MT, Bazargan-Hejazi S. Cost-
effectiveness analysis of a brief intervention delivered to
problem drinkers presenting at an inner-city hospital emer-
gency department. J Stud Alcohol 2004;65:363–70.
[52] Lock CA, Kaner E, Heather N, et al. Effectiveness of nurse-
led brief alcohol intervention: a cluster randomized con-
trolled trial. J Adv Nurs 2006;54:426–39.
[53] Longabaugh R, Woolard RF, Nirenberg TD, Minugh AP,
Becker B, Clifford PR, et al. Evaluating the effects of a brief
motivational intervention for injured drinkers in the emer-
gency department. J Stud Alcohol 2001;62:806–16.
[54] Maisto SA, Conigliaro J, McNeil M, Kraemer K, Conigliaro
RL, Kelley ME. Effects of two types of brief intervention
and readiness to change on alcohol use in hazadous drink-
ers. J Stud Alcohol 2001;62:605–14.
[55] McIntosh MC, Leigh G, Baldwin NJ, Marmulak J. Reduc-
ing alcohol consumption. Comparing three brief methods
in family practice. Can Fam Physician 1997;43:1959–
62.
[56] Ockene JK, Adams A, Hurley TG, Wheeler EV, Hebert JR.
Brief physician- and nurse practitioner-delivered counseling
for high-risk drinkers: does it work? Arch Intern Med
1999;159:2198–205.
[57] Richmond R, Heather N, Wodak A, Kehoe L, Webster I.
Controlled evaluation of a general practice-based brief inter-
vention for excessive drinking. Addiction 1995;90:119–32.
[58] Rodriguez A, Santa-marina E, Martinez X, et al.
Identificaci¢n precoz e intervenci¢n breve en lesionados de
trafico con presencia d alcohol: primeros resultados. Adic-
ciones 2003;15:191–202.
[59] Romelsjo A, Andersson L, Barrner H, et al. A randomized
study of secondary prevention of early stage problem drink-
ers in primary health care. Br J Addict 1989;84:1319–27.
[60] Scott E, Anderson P. Randomized controlled trial of
general practitioner intervention in women with excessive
alcohol consumption. Drug Alcohol Rev 1991;10:313–21.
[61] Senft RA, Polen MR, Freeborn DK, Hollis JF. Brief inter-
vention in a primary care setting for hazardous drinkers.
Am J Prev Med 1997;13:464–70.
[62] Seppa K. Intervention in alcohol abuse among macrocytic
patients in general practice. Scand J Prim Health Care
1992;10:217–22.
[63] Tomson Y,Romelsjo A, Aberg H. Excessive drinking—Brief
intervention by a primary health care nurse: a randomized
controlled trial. Scand J Prim Health Care 1998;16:188–92.
[64] Wallace P, Cutler S, Haines A. Randomised controlled trial
of general practitioner intervention in patients with exces-
sive alcohol consumption. BMJ 1988;297:663–8.
[65] Fleming MF, Mundt MP, French MT, Manwell LB, Stauf-
facher EA, Kl B. Benefit-cost analysis of brief physician
advice with problem drinkers in primary care settings. Med
Care 2000;38:7–18.
[66] Fleming MF, Mundt MP, French MT, Manwell LB, Stauf-
facher EA, Barry KL. Brief physician advice for problem
drinkers: long-term efficacy and benefit-cost analysis.
Alcohol Clin Exp Res 2002;26:36–43.
[67] Grossberg PM, Brown DD, Fleming MF. Brief physician
advice for high-risk drinking among young adults. Ann Fam
Med 2004;2:474–80.
[68] Manwell LB, Fleming MF, Mundt MP, Stauffacher EA,
Barry KW. Treatment of problem alcohol use in women of
childbearing age: results of a brief intervention trial. Alcohol
Clin Exp Res 2000;24:1517–24.
[69] Gordon AJ, Conigliaro J, Maisto SA, McNeil M, Kraemer
KL, Kelley ME. Comparison of consumption effects of
brief interventions for hazardous drinking elderly. Subst
Use Misuse 2003;38:1017–35.
[70] Reiff-Hekking S, Ockene JK, Hurley TG, Reed GW. Brief
physician and nurse practitioner-delivered counseling for
high-risk drinking. Results at 12 months follow-up. J Gen
Intern Med 2005;20:7–13.
[71] Aalto M, Mattila P, Mustonen H, et al. Brief intervention
for male heavy drinkers in routine general practice: A three-
year randomized controlled study. Alcohol Alcohol
2001;36:224–30.
[72] Anderson P, Scott E. The effect of general practitioners’
advice to heavy drinking men. Br J Addict 1992;87:891–900.
[73] Rodriguez-Martos A. [Short-term effectiveness of brief
interventions in alcohol-positive traffic casualties].
[Spanish]. Gac Sanit 2005;19:45–9.
[74] Rodriguez-Martos Dauer A. Brief intervention in alcohol-
positive traffic casualties: is it worth the effort? Alcohol
Alcohol 2006;41:76–83.
[75] Fleming MF, Mundt MP, French MT. A brief intervention
reduced alcohol drinking for up to 48 months in problem
drinkers. Evid Based Med 2002;7:151.
Brief intervention effectiveness 321
© 2009 Australasian Professional Society on Alcohol and other Drugs
[76] Ockene JK, Wheeler EV, Adams A, Hurley TG, Hebert J.
Provider training for patient-centered alcohol counseling in
a primary care setting 128. Arch Intern Med 1997;157:
2334–41.
[77] Ballesteros J, Gonzalez-Pinto A, Querejeta I, Arino J. Brief
interventions for hazardous drinkers delivered in primary
care are equally effective in men and women. Addiction
2004;99:103–8.
[78] Babor TF. Avoiding the horrid and beastly sin of drunken-
ness: does dissuasion make a difference? J Consult Clin
Psychol 1994;62:1127–40.
[79] Bland J, Altman D. Regression towards the mean. BMJ
1994;308:1499.
[80] Bland J, Altman D. Some examples of regression towards
the mean. BMJ 1994;309:780.
[81] Kypri K, Langley JD, Saunders J, Cashell-Smith ML.
Assessment may conceal therapeutic benefit: Findings from
a randomized controlled trial for hazardous drinking.
Addiction 2006;102:62–70.
[82] Schulz KF, Chalmers I, Hayes RJ, Altman DG. Empirical
evidence of bias. Dimensions of methodological quality
associated with estimates of treatment effects in controlled
trials. JAMA 1995;273:408–12.
[83] Moher D, Pham B, Jones A, et al. Does quality of reports of
randomised trials affect estimates of intervention efficacy
reported in meta-analyses? Lancet 1998;352:609–13.
[84] Ashenden R, Silagy C, Weller D. A systematic review of the
effectiveness of promoting lifestyle change in general prac-
tice. Fam Pract 1997;14:160–76.
[85] Ballesteros J, Duffy JC, Querejeta I, Arino J, Gonzalez-Pinto
A. Efficacy of brief interventions for hazardous drinkers in
primary care: systematic review and meta-analyses. Alcohol
Clin Exp Res 2004;28:608–18.
[86] Beich A, Thorsen T, Rollnick S. Screening in brief
intervention trials targeting excessive drinkers in general
practice: systematic review and meta-analysis. Br Med J
2003;327:06.
[87] Berglund M, Thelander T, Salaspuro M, Franck J,Andreas-
son S, Ojehagen A. Treatment of alcohol abuse: an evidence
review. Alcohol Clin Exp Res 2003;27:1645–56.
[88] Bertholet N, Daeppen J-B, Wietlisbach V, Fleming M,
Burnand B. Reduction of alcohol consumption by brief
alcohol intervention in primary care: systematic review and
meta-analysis. Arch Intern Med 2005;165:986–95.
[89] Havard A, Shakeshaft A, Sanson-Fisher R. Systematic
review and meta-analyses of strategies targeting alcohol
problems in emergency departments: interventions
reduce alcohol-related injuries. Addiction 2008;103:368–
76.
[90] Kahan M, Wilson L, Becker L. Effectiveness of physician-
based interventions with problem drinkers: a review. CMAJ
1995;152:851–9.
[91] Babor T, Higgins-Biddle JC, Dauser D, Burleson JA,
Zarkin GA, Bray J. Brief intervention for at-risk drinking
patient outcomes and cost effectiveness in managed care
organization. Alcohol Alcohol 2006;41:624–31.
Appendix 1
Search strategy
1. family practice/
2. family pract$.tw.
3. general practice/
4. general pract$.tw.
5. primary health care/
6. primary care/
7. community health services/
8. community care
9. shared care
10. patient care team
11. family medicine/
12. family physician/
13. family phys$.tw
14. exp alcohol/
15. alcohol$.tw
16. 1–13
17. 14–15
18. 16 and 17
Interventions
1. alcohol reduction
2. brief intervention
3. early intervention
4. minimal intervention
5. alcohol therapy
6. alcohol treatments
7. harm reduction
8. screening
9. counseling
10. controlled drinking
11. brief counseling
12. physician-based intervention
13. general practice intervention
14. secondary prevention
15. general practitioner’s advice
16. brief physician-delivered counseling
17. brief nurse-delivered counseling
18. identification
19. intervention
20. #1 or #2 or #3 or #4 or #5 or #6 or #7 or #8 or
#9 or #10 or # 11 or #12 or #13 or #14 or #15 or
#16 or #17 or #18 or #19
21. alcohol or alcohol consumption
22. #20 and #21
Appendix 2
Databases searched
MEDLINE (1966 to 2006)
EMBASE (1980 to 2006)
PsycINFO (1840 to 2006)
CINAHL (1982 to 2006)
Social Sciences Citation Index (SSCI) (1970 to 2006)
Science Citation Index (SCI) (1970 to 2006)
Cochrane Drug and Alcohol Group specialised register
(2006)
322 E. F. S. Kaner et al.
© 2009 Australasian Professional Society on Alcohol and other Drugs
Cochrane Effective Practice and Organisation of Care
Group specialised register (2006)
Cochrane Central Register of Controlled Trials
(CENTRAL—Cochrane Library 2006, issue 2)
Alcohol and Alcohol Problems Science Database,
ETOH (http://etoh.niaaa.nih.gov/) (1972 to 2003)
Appendix 3
Efficacy-effectiveness classification scale
Trials were classified as more clinically representative
(Effectiveness trial) or less clinically relevant (Efficacy
trial) on 8 criteria using the following scoring system:
1. Patients and problems
More clinically relevant (score 2)
Subjects self-present with a wide range of problems to
routine primary care.
Less clinically relevant (score 0)
Subjects might be paid, researcher-solicited volunteers
or referrals from specialists.
2. Practice context
More clinically relevant (score 2)
Setting is community-based setting, a range of clinical
services are provided.
Less clinically relevant (score 0)
Setting has a research or specialist focus (e.g. university
clinic, hospital).
3. Practitioner attributes
More clinically relevant (score 2)
Practitioners are regularly practising doctors or nurses
providing general health care.
Less clinically relevant (score 0)
Practitioners are non-clinicians, trainees and/or con-
tracted for the purpose of the trial.
4. Intervention content
More clinically relevant (score 2)
Intervention fits within routine consultations in terms
of timing, content or style*.
Less clinically relevant (score 0)
Intervention could not occur in routine practice e.g.
unusually long consultations.
5. Therapeutic flexibility
More clinically relevant (score 1)
Professional judgement is allowed in how an interven-
tion is delivered.
Less clinically relevant (score 0)
Strict adherence to a protocol, no allowance for vari-
ability in practice.
6. Pre-therapy training
More clinically relevant (score 1)
Training fits with typical CPD/CME procedures, e.g.
outreach visit, time-limited.
Less clinically relevant (score 0)
Training is intensive, requiring atypical interest, moti-
vation or formal qualification.
7. Intervention support
More clinically relevant (score 1)
Support occurs within standard practice resources i.e.
no new staff provided.
Less clinically relevant (score 0)
Support atypical e.g. researcher help to flag notes, extra
staff for period of the trial.
8. Intervention monitoring
More clinically relevant (score 1)
Intervention monitoring does not interfere with
clinician–patient interaction.
Less clinically relevant (score 0)
Intervention monitoring via direct observation or
ongoing/immediate feedback.
If a paper scored 12 then it was likely to be highly
clinically relevant and so considered to be an effective-
ness trial with high external validity. Conversely, if a
trial scored 0 then it was less clinically relevant and
considered to be an efficacy trial with high internal
validity. If an item appeared to be partially clinically
representative on any item, then a midpoint score was
given (either 1 or 0.5 as applicable). If the authors did
not report data relating to a particular item, then an
intermediate score was allocated so as not to bias the
trial towards the effectiveness or efficacy domain. For
the purpose of subgroup analysis a binary variable was
created with a cut-off point at the median.
*(5–15 min for GPs; 20–30 min for nurses; initial
screening and return for brief intervention delivery).
Brief intervention effectiveness 323
© 2009 Australasian Professional Society on Alcohol and other Drugs
