Can standalone computerbased interventions ... - Wiley Online Library

REVIEW

doi:10.1111/j.1360-0443.2010.03214.x

Can stand-alone computer-based interventions reduce alcohol consumption? A systematic review add_3214

267..282

Zarnie Khadjesari1, Elizabeth Murray1, Catherine Hewitt2, Suzanne Hartley3 & Christine Godfrey2 E-health Unit, Research Department of Primary Care and Population Health, University College London, Royal Free Hospital, London,1 Department of Health Sciences and HYMS, Seebohm Rowntree Building, University of York, Heslington, York2 and Clinical Trials Research Unit, University of Leeds, Leeds, UK3

ABSTRACT Aim To determine the effects of computer-based interventions aimed at reducing alcohol consumption in adult populations. Methods The review was undertaken following standard Cochrane and Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidance for systematic reviews. The literature was searched until December 2008, with no restrictions on language. Randomized trials with parallel comparator groups were identified in the form of published and unpublished data. Two authors independently screened abstracts and papers for inclusion. Data extraction and bias assessment was undertaken by one author and checked by a second author. Studies that measured total alcohol consumption and frequency of binge drinking episodes were eligible for inclusion in metaanalyses. A random-effects model was used to pool mean differences. Results Twenty-four studies were included in the review (19 combined in meta-analyses). The meta-analyses suggested that computer-based interventions were more effective than minimally active comparator groups (e.g. assessment-only) at reducing alcohol consumed per week in student and non-student populations. However, most studies used the mean to summarize skewed data, which could be misleading in small samples. A sensitivity analysis of those studies that used suitable measures of central tendency found that there was no difference between intervention and minimally active comparator groups in alcohol consumed per week by students. Few studies investigated non-student populations or compared interventions with active comparator groups. Conclusion Computer-based interventions may reduce alcohol consumption compared with assessment-only; the conclusion remains tentative because of methodological weaknesses in the studies. Future research should consider that the distribution of alcohol consumption data is likely to be skewed and that appropriate measures of central tendency are reported. Keywords

Alcohol consumption, computer-based intervention, meta-analysis, systematic review.

Correspondence to: Zarnie Khadjesari, E-health Unit, Research Department of Primary Care and Population Health, University College London, Upper Floor 3, Royal Free Hospital, Rowland Hill Street, London NW3 2PF, UK. E-mail: [email protected] Submitted 10 May 2010; initial review completed 28 July 2010; final version accepted 17 September 2010

INTRODUCTION There is strong international evidence for the use of brief interventions to reduce hazardous and harmful alcohol consumption, particularly in the primary care setting [1–3]. The World Health Organization provides a manual for their implementation in primary care and a dissemination strategy for developing countries is under way [4]. Brief interventions provide a means to fill the apparent gap between primary prevention and intensive treatment approaches [5]; however, hazardous and harmful drink-

ing are rarely identified in family practice and so the opportunity for early identification and brief intervention is often missed [6,7]. In addition, health-care professionals report similar barriers to implementation across the world. These include lack of financial incentive, time constraints, lack of training and support [8,9] and a fear of offending patients by discussing their alcohol consumption [8]. Delivering brief interventions over the internet may address some of the barriers to implementing the conventional face-to-face approach. In 2009, 76% of adults in

Conflict of Interest: None. © 2010 The Authors, Addiction © 2010 Society for the Study of Addiction

Addiction, 106, 267–282

268

Zarnie Khadjesari et al.

the United Kingdom (37.4 million people) were accessing the internet [10], with a slightly lower proportion in Europe as a whole (52%), but similar proportions in the United States (74%) and Australia (80%) [11]. The internet provides a means of combining the scalability of a public health intervention, with the capacity to deliver an individualized approach [3,12]. The internet setting allows for increased access to the intervention and flexibility of use. There are also cost advantages of internetbased interventions delivered on this scale, in that the marginal cost per additional user is low, unlike conventional face-to-face interventions [13,14]. Internet-based interventions may be integrated into health-care and other settings such as the work-place or higher education, but are also available from any location with internet access. The possibility of accessing these interventions autonomously via the internet allows for anonymity, which is a major advantage with sensitive or stigmatized behaviours such as alcohol consumption [15,16]. Internet-based interventions have demonstrated their ability to attract large numbers of people interested in reducing their drinking [17–21]. There has been a notable increase in recent years in the number of trials assessing the effectiveness of these interventions. In 2004, Copeland & Martin conducted a qualitative review of web-based interventions for substance use disorders in all adult populations, concluding that there was limited research on the efficacy of these interventions in changing substance use behaviour [12]. Four years later, Elliot et al. (2008) identified 17 trials of computer-based interventions (on- and offline) for college drinkers, finding them to be more effective than no treatment and as effective as alternative treatment approaches [22]. The first avowedly systematic review in this field, conducted by Bewick et al. (2008), concluded that there was inconsistent evidence for the use of web-based electronic screening and brief intervention for reducing alcohol intake based on five trials in all adult populations [23]. A recent meta-analysis by Carey et al. (2009) supported those findings of Elliot et al. and found computer-delivered interventions to reduce the quantity and frequency of drinking in student populations when compared with assessment-only controls, and found them as effective as other alcohol-related interventions [24]. The most recent meta-analysis (Rooke et al. 2010) pooled computer-based interventions (both stand-alone and therapeutically guided) for alcohol and tobacco use in all populations, and reported a significant reduction in substance use [25]. One limitation of the studies in this field, highlighted by the Bewick review, is the lack of appropriate statistics to account for the skewed distribution of the data [23]. The Cochrane handbook states that ‘analyses based on © 2010 The Authors, Addiction © 2010 Society for the Study of Addiction

means are appropriate for data that are at least approximately normally distributed, and for data from very large trials’ [26]. The distribution of alcohol consumption data in the population is thought to be positively skewed, where most people are abstinent or drinking relatively low levels of alcohol, while fewer people are drinking very large quantities of alcohol [27–29]. In a skewed distribution, where sample sizes are small and the data contain extreme outliers, study data should be characterized by non-parametric methods or by transformation [30]. The extent to which different measures of central tendency (e.g. mean versus median) impact upon the results are unknown, but any differences may have clinical significance and compromise the robustness of the outcomes. Furthermore, it has been highlighted that alcohol consumption is measured typically as count data, e.g. number of drinks consumed within a given time-frame [27,28], and as such the data are not continuous and a normal distribution cannot be assumed. This review builds on those studies conducted previously in this field. It is the first to include meta-analyses of mean differences in grams of alcohol and frequency of binges, giving the findings immediate clinical relevance. It is also novel in comparing the findings of those studies that presented appropriate measures of central tendency, given the distribution of the data, with those that did not (i.e. means in the presence of skew). Of the aforementioned reviews conducted in this field, two were restricted to student samples [22,24], two included web-based interventions alone (as opposed to computer-based) [12,23] and the most recent meta-analysis included smoking and alcohol interventions as both stand-alone and therapeutically guided interventions [25]. This review adds to the field by including stand-alone computer-based interventions (available on- and offline) in all adult populations. Students represent a highly selective population who lack generalizability to the general adult population drinking at hazardous and harmful levels. Interventions that are computer-based have the potential to be made available online, with online interventions often evaluated on computers in a fixed location. Finally, it is important to gauge the effectiveness of stand-alone interventions as they carry the benefits of reach, availability, anonymity and cost savings.

METHODS Search strategy The following databases were searched from inception to December 2008 with no restrictions on language: the Cochrane Library (2008, issue 4), MEDLINE, EMBASE, CINAHL, PsychINFO, ERIC, Web of Science and Addiction, 106, 267–282

Computer-based interventions for alcohol use

International Bibliography of the Social Sciences (IBSS). Unpublished data were sought in the form of conference proceedings (Conference Proceedings Citation Index, formally ISI Proceedings) and theses (Index to Theses). Search terms were selected through discussions with an information specialist and the research team by considering the inclusion criteria, scanning the background literature and by browsing the MEDLINE Thesaurus (MeSH) (see Appendix 1 for MEDLINE search strategy). The thesaurus terms were redefined for each database. The included studies were citation-tracked through Web of Science. The reference lists of relevant reviews and included studies were hand-searched. Selection criteria Randomized controlled trials were eligible for inclusion. All adult populations (aged 18 years and over) with any level of alcohol consumption were included. Eligible computer-based interventions were those considered behavioural interventions, aimed at bringing about positive behaviour change, adapted for a computer-based format [31]. Inclusion was restricted to stand-alone (nonguided) computer-based interventions. Eligible studies compared computer-based interventions with either a minimally active (e.g. assessment-only, usual care, generic non-tailored information or educational materials) or an active comparator group (e.g. brief intervention). This review included studies that measured a change in alcohol consumption. A reduction in alcohol consumption was considered a positive behaviour change. Study screening and data extraction Study references identified by the search strategy were screened by two independent reviewers trained in systematic review methodology (Z.K. and S.H.). Full papers were ordered for all potentially relevant studies and screened in duplicate. Discrepancies were resolved by a third party (E.M.). One reviewer (Z.K.) extracted data from the included studies into pre-designed forms (Microsoft Excel), which were piloted on three studies for suitability. The data extraction was verified for accuracy by a second reviewer (E.M.). Authors were contacted for missing data. Bias assessment The risk of bias associated with allocation concealment was assessed in each of the included studies, as it is shown to have the greatest impact on treatment effect compared with other potential sources of bias [32–34]. Bias assessment, as advocated by the Cochrane handbook [35], considers the likelihood that a particular aspect of © 2010 The Authors, Addiction © 2010 Society for the Study of Addiction

269

trial quality would have biased the findings, given the design of the trial. Studies were classified by one author (Z.K.) and checked by a second (E.M.) as having high, low or unclear risk of bias. A third party helped resolve any discrepancies (C.G.).

Data synthesis There is no gold-standard measure of alcohol consumption, therefore two outcomes that represent different patterns of drinking were chosen for inclusion in metaanalyses. These were total alcohol consumption and number of binge drinking episodes (‘binge’ defined by the authors of the primary studies). Mean weekly alcohol intake (measured in grams) or number of binges per week, corresponding standard deviation and number of participants in the intervention and comparator groups at follow-up were entered into Review Manager software version 5. Where outcomes were not presented per week, data were adjusted to represent this time-frame [2]. Where studies did not detail the number of grams included in a standard drink, information on countryspecific standard units was obtained from an established source [36]. Furthest point of follow-up was used unless a primary time-point was specified. The distribution of alcohol consumption is often skewed [27,28]: ‘when the data are skewed we can either use a non-parametric method, or try a transformation of the raw data’ ([30], p. 199). A preliminary look at the data found the majority of studies reported the mean, while a few reported the median and transformed data. To allow for pooling of all data in meta-analyses, medians were used as the best estimate of the sample mean and an estimated standard deviation was generated from the range, using a method that makes no assumption on the distribution of the underlying data [37]. Transformed data were back-transformed. Studies were pooled using the inverse variance method with a random effects model; all analyses were two-tailed. Studies comparing a computer-based intervention with a minimally active comparator group were pooled separately to those with an active comparator. Heterogeneity was examined through use of forest plots, c2 test and I2 test. A subgroup analysis by population (student versus non-student) was planned a priori. The data included in the meta-analyses were assessed for skew. The test for normality, advocated by Altman & Bland, was applied by dividing the mean by the standard deviation; where the ratio was less than two this indicated a skewed distribution [38]. A sensitivity analysis was conducted of those studies that used appropriate measures of central tendency, given the distribution of the data. Addiction, 106, 267–282

270


RESULTS Study description A total of 24 studies were included in the review (see Fig. 1). The earliest study was published in 1997, with most studies published recently in 2007 and 2008. The majority were conducted in the United States (n = 18). Students were the most commonly studied population group (n = 18) [39–56], with three studies of adult problem drinkers from the general population [57–59], two of work-place employees [60,61] and one of emergency department attendees [62]. Eight studies appeared to screen for hazardous drinking, either in the form of binge drinking, total number of drinks per week, Alcohol Use Disorders Identification Test (AUDIT) cut-off score (generally reported as ⱖ8) or some combination of these. The other studies used either a lower cut-off score or did not restrict inclusion based on alcohol intake (see Table 1). The majority of studies (n = 22) compared a computer-based intervention with a minimally active comparator group. Minimally active comparators consisted mainly of assessment with some factual information about the harms of excess alcohol consumption, or a waiting-list design. Three studies compared a

Records identified through database searching (n = 10 973)

Records after duplicates removed (n = 8084)

Records screened (n = 8084)

Records excluded (n = 7930)

Full-text articles assessed for eligibility (n = 154)

Full-text articles excluded, with reasons (available on request) (n = 130) Common reasons for exclusion: • no parallel comparator group • no measure of change in alcohol consumption • intervention not computerised

Studies included in review (n = 24)

Studies excluded from meta-analyses (n = 5)

Studies included in quantitative synthesis (meta-analyses) (n = 19)

• No measure of total alcohol consumption or binge frequency (n = 3) • Measured proportion of binge days (not frequency) and no standard deviation for total alcohol consumption (n = 1) • Measured frequency of binging as a categorical variable (n = 1)

Figure 1 Flow-chart of study selection © 2010 The Authors, Addiction © 2010 Society for the Study of Addiction

computer-based intervention with an active comparator group. Active comparator groups consisted of an in-person motivational interview [39], cognitive behaviour therapy [42] and an expectancy challenge [49] (see Table 2). Intervention—delivery mode Most studies delivered the intervention via the internet (n = 14). One study sent tailored text-messages to handheld computers [56], while the others were available from a computer in a fixed location. Most interventions were accessed from computers at a location determined by the researchers (n = 16); the remainder were able to access the intervention online at a location and time convenient to them [40,41,44,54–56,59,61]. Intervention—content Fifteen studies consisted of personalized feedback on current levels of drinking and comparison with safe drinking limits. This was often accompanied with normative feedback, associated health risk, information on calculating units and support services. Five studies investigated interventions designed to resemble the campus setting. These included a variety of interactive games and assignments, motivational feedback and information on risk taking and refusal skills [39,41,42,49,54]. One study presented a video of people undergoing an alcohol/placebo expectancy–disconfirming experience. This aimed to increase awareness of how participants expected alcohol to affect them, and how these expectancies can lead to detrimental effects. It was followed by a description of the alcohol expectancy concept and the effect of alcohol expectancies on behaviour. The intervention also included games and questions requiring interaction [45]. Three studies based on adult problem drinkers from the general population provided a more extensive intervention, featuring common elements from behaviour change interventions. They included components such as readiness to change, decisional-balance, goalsetting, self-monitoring, strategies for behaviour change, behavioural contracting with rewards and penalties, maintenance of change and relapse prevention [57–59]. One of these studies also provided access to a peer-to-peer discussion forum [59] (see Table 2 for more information). Intervention—theoretical basis The studies cite different theoretical foundations of their interventions. The authors of the primary studies provided different justifications for using personalized feedback, reporting it to have originated from: Motivational Interviewing [63], FRAMES (Feedback, Responsibility, Advice, Menu, Empathy and Self-efficacy—illustrates Addiction, 106, 267–282

56

73

42

58 40

48

(Donohue et al. 2004)

(Doumas & Hannah 2008)

(Doumas & Haustveit 2008)

(Hedman 2007) (Hester & Delaney 1997)

(Hester et al. 2005)

© 2010 The Authors, Addiction © 2010 Society for the Study of Addiction

49 Intervention: 52; control: 52 57 52 55 78

(Kypri & McAnally 2005) (Kypri et al. 2008)

55

(Weitzel et al. 2007)

19.2

18.5 19.7 Intervention: median 30; control: median 31 18.1 Intervention: 45.9; control: 46.2 –

Intervention: 19.9; control: 20.4 20.2 Intervention: 20.1; control: 20.1 20 19 20 40

21.2

46.1씹; 45.2씸

19.5 36.3

18.1

Range: 18–24

18.8 21.3 Intervention: 20; control: 19.8 20.6

Age (mean years)

75

72.7

30.3 –

79.5 98 –

76 99.6 97 83

75 –

Presented separately for each site –

79

93.8 70

54

87

62.6

75.6 – 73.2

White (%)

AUDIT AUDIT score ⱖ8; >6씹/4씸 drinks on at least one occasion in preceding 4 weeks Not used AUDIT AUDIT score ⱖ8 ⱖ2 heavy episodic drinking occasions (in past 30 days), or ⱖ5 weekly standard drinks (but fewer than 40) ⱖ1 heavy drinking episode in past month (ⱖ5씹/4씸 drinks at one setting) ⱖ1 heavy drinking episode (ⱖ5씹/4씸 drinks at one sitting) in past month AUDIT and CAGE Participants separated into low and moderate risk for analysis. High-risk participants were excluded ⱖ1 heavy drinking episode in past month (ⱖ5씹/4씸 drinks) ⱖ1 heavy drinking episode in past month (ⱖ5씹/4씸 drinks) AUDIT AUDIT score ⱖ5 Not used Weekly recall and quantity–frequency variability index of alcohol intake >21씹/14씸 units per week or ⱖ6씹/4씸 units at least 1 day per week for past 3 months (1 unit = 10 g ethanol) Not used Participants with at least one heavy drinking episode in past month (ⱖ5 drinks 씹, ⱖ4 drinks 씸) were included in the analysis Drinking more than once a week

6 drinks per week and fewer than 6 drinks per day

Not used (mandated students) Not used Daily drinking questionnaire Binge drinkers: ⱖ5씹/4씸 drinks, per drinking occasion in the past week Time-line follow-back ⱖ1 alcoholic drink in past 30 days Binge drinking (ⱖ5씹/4씸 drinks in row, in past 2 weeks) All participants included but separated into low and high risk for analysis Binge drinking (ⱖ5씹/4씸 drinks in row, in past 2 weeks) All participants included but separated into low and high risk for analysis Binge drinking: ⱖ5씹/4씸 drinks in row, at least once in 2 weeks preceding survey MAST and AUDIT AUDIT score ⱖ8; ⱖ120씹/70씸 drinks per month; weekly drinking with ⱖ6 drinks per episode; drinking at least once per week AUDIT AUDIT score ⱖ8 Time-line follow-back

Screening test and cut-off score

AUDIT: Alcohol Use Disorders Identification Test [74]; MAST: Michigan Alcohol Screening Test [75]; CAGE: mnemonic for cut-down, annoyed, guilty, eye-opener [76].

(Walters et al. 2007)

(Paschall et al. 2006) (Riper et al. 2008)

59 56 Intervention: 20; control: 22 52 Intervention: 49; control: 49 48

(Neighbors et al. 2004) (Neighbors et al. 2006) (Neumann et al. 2006)

(Lau-Barraco & Dunn 2008) (Lewis et al. 2007) (Lewis & Neighbors 2007) (Matano et al. 2007)

50

(Kypri et al. 2004)

0

51 69 54

(Barnett et al. 2007) (Bewick et al. 2008) (Chiauzzi et al. 2005)

(Hunt 2004)

Female (%)

Study

Table 1 Characteristics of study participants.

Computer-based interventions for alcohol use 271

Addiction, 106, 267–282

United States: university students

United States: work-place employees

Respondents of student experience survey

Newspaper advertisements, flyers and in-person at busy campus locations and events

Not stated

Human resource departments of local companies were contacted for participation

(Bewick et al. 2008)

(Chiauzzi et al. 2005)

(Donohue et al. 2004)

(Doumas & Hannah 2008)a


United Kingdom: university students

United States: mandated students

Voluntary alternative to individual session with university health educator after mandated health education session

(Barnett et al. 2007)

Population

Recruitment

Study

Table 2 Characteristics of included studies.

AC: brief motivational interview (n = 112)

Alcohol 101 (n = 113): interactive computer-delivered intervention that features a virtual party where participants can observe the effects of gender, weight, drink type and speed of consumption on BAC. Information on alcohol refusal skills, consequences of unsafe sex, multiple choice games and stories of actual campus tragedies involving alcohol. Personalized normative feedback was provided Single session ⫾ booster session Guiding principles not stated Intervention on CD-ROM; location determined by researcher Personalized feedback (n = 234): Feedback on level of alcohol consumption and associated health risk, social norms information and generic information, such as calculating units, sensible drinking guidelines, support services Single session with continued access to the website for the study duration Based on social norms approach Intervention online; location determined by participant MyStudentBody.com: alcohol (n = 131) interactive website that includes: Rate Myself, based on BASICS model and consists of four sets of questions regarding (i) alcohol beliefs, (ii) life-style issues, (iii) risk-taking while drinking, and (iv) consequences resulting from drinking. Responses used to tailor feedback. The site also includes: articles, interactive tools, peer stories, ask the expert, participants and campus health news Four weekly 20-minute sessions. Each session needed to be completed before advancing to the next Based on BASICS model Intervention online; location determined by participant Alcohol 101 (n = 40): interactive computer-delivered intervention that features a virtual party where participants can observe the effects of gender, weight, drink type, and speed of consumption on BAC. There is information on alcohol refusal skills, consequences of unsafe sex, multiple choice games and stories of actual campus tragedies involving alcohol. Personalized normative feedback was provided Single session lasting approx. 45 minutes Guiding principles not stated Intervention on CD-ROM; location determined by researcher Check Your Drinking (n = 60): personalized normative feedback on drinking and associated risks. Also feedback on cost and calories associated with drinking, the rate at which the body processes alcohol, risk status for negative drinking-related consequences and problematic drinking based on AUDIT score Single session Based on social norms approach and motivational enhancement models Intervention online www.checkyourdrinking.net Location determined by researcher

© 2010 The Authors, Addiction © 2010 Society for the Study of Addiction 1. MAC: control—assumed assessment-only (n = 73) 2. 3rd arm excluded: Check Your Drinking plus motivational interview

AC: cognitive behaviour therapy (n = 39)

MAC: Alcohol and You: text-based, educationonly website containing articles on high-risk drinking (n = 134)

MAC: assessment-only (n = 272)

Comparator

Intervention No. of drinking days No. of heavy drinking days Average no. of drinks/drinking day Average estimated BAC

1. Weekend drinking 2. Peak consumption (quantity) 3. Frequency of drinking to intoxication

1. No. alcoholic beverages/past month 2. No. days alcohol consumed/past month 3. No. alcoholic beverages consumed/drinking occasion in past month

Binge drinking days/week Max. no. drinks/drinking day Drinks/week Drinking days/week Average consumption/drinking day Alcohol composite score Total consumption during special occasion drinking 8. Peak consumption during special occasion drinking

1. 2. 3. 4. 5. 6. 7.

1. Units/occasion 2. Units/week

1. 2. 3. 4.

Drinking outcomes

Total: 63

Total: 92

1 month

30 days

Intervention: 80 Control: 82



Follow-up at furthest time-point %

1, 3 months

12 weeks

3, 12 months

Follow-up time-points

272 Zarnie Khadjesari et al.

Addiction, 106, 267–282

University athletics department

Health, sport and exercise science department

Local health centre, other health/mental health care providers, screening program for driving while intoxicated, and through media advertisements

Media advertisements

Online participant pools from psychology departments across three sites

(Doumas & Haustveit 2008)

(Hedman 2007)

(Hester & Delaney 1997)

(Hester et al. 2005)


(Hunt 2004)b United States: university students

United States: adult problem drinkers in the general population

United States: adult problem drinkers in the general population


United States: collegiate athletes

Check Your Drinking (n = 28): personalized normative feedback on drinking and associated risks. Also feedback on cost and calories associated with drinking, the rate at which the body processes alcohol, risk status for negative drinking-related consequences and problematic drinking based on AUDIT score Single session lasting 15 minutes Based on social norms approach and motivational enhancement models Intervention online http://notes.camh.net/efeed.nsf/newform Location determined by researcher Personalized feedback (n = 68): personalized feedback consisted of: peak blood alcohol level, time to alcohol oxidation, dollars spent on alcohol, caloric intake, alcohol-related risks, information on sensible drinking behaviours. Feedback was supplemented with health communication messages on risks and consequences associated with heavy alcohol consumption Viewed feedback via e-mail, followed by health communication messages twice a week for 6 weeks Based on Health Belief Model, Cognitive Dissonance Theory, Elaboration Likelihood Model Intervention via e-mail; location determined by participant Behavioural Self-Control Program for Windows (n = 20): teaches the following skills: goal-setting, self-monitoring, rate control and drink refusal, behavioural contracting, evaluating triggers and problem solving, functional analysis of drinking, and relapse prevention. Also provided normative feedback Eight weekly sessions over 10 weeks Based on Miller & Munoz (1982) protocol for self-control training [70] Intervention on disk; location determined by researcher (except for 2 participants who used their home PCs) Drinker’s check-up (n = 35): consisted of assessment (including decisional balance exercise), feedback, and decision-making (including Rollnick’s ‘Readiness Ruler’, negotiating goals of change and developing alternatives and a change plan) modules Approx. 90 minutes to complete. Summary of worksheets and feedback from completed assessments were printed Based on FRAMES and MI approach Intervention online www.drinkerscheckup.com Location determined by researcher Expectancy challenge (n = 52): video of people undergoing an alcohol/placebo expectancy–disconfirming experience followed by description of alcohol expectancy concept and effect of alcohol expectancies on behaviour. The program had audiovisual elements, including games and questions requiring interaction Approx. 20 minutes Based on the expectancy concept [71] Intervention computer-based; location determined by researcher 1. MAC: PowerPoint presentation on safe driving practices (n = 54) 2. 3rd arm excluded: non-interactive power-point presentation of expectancy challenge

MAC: waiting-list control (n = 26)

MAC: waiting-list control (n = 20)

MAC: alcohol facts received via e-mail twice a week for 6 weeks (n = 63)

MAC: educational website containing alcohol facts and consumption guidelines http://www. radford.edu/kcastleb/toc.html (n = 24)

1. Mean drinks consumed per day 2. Quantity–frequency 3. Proportion of binge days

1. Average drinks per day 2. Drinks per drinking day 3. Average peak BAC

1. Total drinks per week 2. Estimated peak BAC per week 3. No. of drinking days per week

1. 30-day frequency of alcohol use (>1 drink) 2. No. of typical drinks reported at one setting in past 30 days 3. 30-day frequency of binge drinking 4. 14-day frequency of binge drinking

1. Weekly drinking quantity 2. Peak consumption (quantity) 3. Frequency of drinking to intoxication

1 month

4 weeks

10 weeks

6 weeks

6 weeks, 3 months

Not reported

Not reported at 4 weeks





Addiction, 106, 267–282

New Zealand: university students

University health centre


Psychology classes

Orientation course

(Kypri & McAnally 2005)a

(Kypri et al. 2008)

(Lau-Barraco & Dunn 2008)


(Lewis et al. 2007)






(Kypri et al. 2004)

Population

Recruitment

Study

Table 2 Cont.

MAC: participants received a paper-based leaflet on alcohol facts and effects (n = 53)

Personalized feedback (n = 51): feedback consisted of a summary of recent consumption and comparison with recommended limits, estimate of BAC for heaviest drinking session (criterion feedback), normative feedback and correction of norm misperceptions. Participants also received the leaflet provided in the control condition Single session 10–15 minutes Feedback component of brief intervention and motivational interviewing [63] Intervention online; location determined by researcher Personalized feedback (n = 72): feedback consisted of health authority recommendations, social norms and selfcomparison. Blood pressure and demographic details were also taken Single session Feedback component of brief intervention and motivational interviewing [63] Intervention online; location determined by researcher Personalized feedback plus information pamphlet on health effects of alcohol consumption (single and multi-dose groups combined) (n = 283): feedback consisted of: risk status, summary of recent consumption, comparison of consumption with recommended limits, estimate of blood alcohol concentration for heaviest drinking occasion in past 4 weeks, comparison of consumption with national and university norms and correction of misperceptions of norms Single dose: single session of assessment and feedback at baseline. Multi-dose: assessment and feedback at baseline, 1 and 6 months Feedback component of brief intervention and motivational interviewing [63] Intervention online; location determined by researcher Alcohol 101 (n = 39): information on the effects of alcohol misuse and drinking behaviour among peers (also see Barnett et al. 2007 and Donohue et al. 2004) Single session lasting 90–120 minutes Guiding principles not stated Intervention on CD-ROM; location determined by researcher Normative feedback (gender-specific and gender-neutral groups combined) (n = 157): feedback on personal drinking behaviour, personal perceptions of typical student drinking behaviour, information on actual norms for typical student drinking behaviour Feedback viewed on screen then provided as print-out Based on social norms approach Intervention online; location determined by researcher MAC: assessment-only (n = 88)

1. AC: expectancy challenge (n = 114) 2. MAC: assessment-only (n = 64)

1. MAC: assessment-only (comprising of blood pressure, demographic data and assessment) (n = 74) 2. 3rd arm excluded: minimal contact (blood pressure and demographic data) MAC: screening and information pamphlet on health effects of alcohol consumption (n = 146)

Comparator

Intervention Frequency of drinking Typical occasion quantity Total volume Frequency of heavy episodes

Frequency of drinking Typical occasion quantity Total volume Frequency of heavy episodes

1. Drinks/week 2. Drinking frequency

1. Average drinks/week 2. Heavy episodic drinking frequency

1. 2. 3. 4.

1. Percentage compliance with recommendations (alcohol consumed per occasion) 2. Peak estimated BAC

1. 2. 3. 4.

Drinking outcomes

5 months

Post-test, 1 month

6, 12 months

6 weeks

6 weeks, 6 months


Intervention (combined): 83 Control: 89

Intervention: 89 Control (MAC): 93 Control (AC): 91






Addiction, 106, 267–282

United States: work-place employees

Mailed recruitment flyer

Psychology classes

Psychology classes

Emergency department after initial care

(Matano et al. 2007)a

(Neighbors et al. 2004)

(Neighbors et al. 2006)


(Neumann et al. 2006)

Germany: emergency department attendees




Psychology classes

(Lewis & Neighbors 2007)

Normative feedback (n = 126): consisted of perceived drinking norms compared with actual drinking norms, and summary of reported consumption compared with average college drinking behaviour. Also feedback on percentile ranking compared with other college student drinking Viewed feedback on screen for approx. 1 minute while printing Based on social norms approach Intervention computer-based; location determined by researcher Normative feedback (n = 108): consisted of perceived drinking norms for quantity and frequency of alcohol intake compared with actual quantity and frequency norms, and summary of reported consumption compared with actual norms. Also feedback on percentile ranking compared with other college student drinking Feedback viewed on screen for 1–2 minutes then provided as print-out Based on social norms approach. Intervention computer-based; location determined by researcher Brief intervention (n = 561): feedback on current drinking status based on AUDIT and Readiness to Change responses. The intervention contained feedback on: comparison of consumption with safe drinking levels, personal responsibility for change, advice on need to change drinking and on developing goals for change. Alternative strategies for changing consumption were provided (treatment-assisted or self-change). Alcohol-related feedback was imbedded with information about other lifestyle risks. Participants also had access to usual care Results were presented on screen, printed and provided to participant Based on FRAMES model Intervention computer-based; location determined by researcher

Normative feedback (gender-specific and gender neutral groups combined) (n = 125): feedback on personal drinking behaviour, perceptions of typical student drinking behaviour, information on actual norms for typical student drinking behaviour Feedback viewed on screen for 1–2 minutes then provided as print-out Based on social norms approach Intervention computer-based; location determined by researcher Coping matters (n = not reported, total sample = 145): provided individualized feedback on risk of alcohol-related problems, recommendations, mini-workshops, drinking journal and links to online resources. Feedback was also given on stress level and use of coping strategies Participants had access to the website for 90 days Based on concepts derived from social learning perspective Intervention online; location determined by participant

1. No. of drinks/week

1. Proportion of at-risk drinking 2. Alcohol intake (g/day)

MAC: usual care (n = 575)

1. Frequency of drinking 2. Usual no. of beers consumed when drinking 3. Usual no. of glasses of wine consumed when drinking 4. Usual no. of shots of hard liquor when drinking 5. Most no. of beers consumed when drinking 6. Most no. of glasses of wine consumed when drinking 7. Most no. of shots of hard liquor when drinking 8. Frequency of beer binges 9. Frequency of wine binges 10. Frequency of hard liquor binges 1. Overall consumption (Alcohol Consumption Index) 2. Typical weekly drinking 3. Peak quantity

1. Overall consumption (Alcohol Consumption Inventory) 2. Typical weekly drinking 3. Typical no. drinks consumed/drinking occasion



MAC: computer-based individualized feedback on stress level and coping strategies but not alcohol consumption (n = not reported, total sample = 145)


6, 12 months

2 months

3, 6 months

3 months

1 month



Total: 82

Total: 84

Total: 89


Addiction, 106, 267–282


Campus orientation sessions and by letter and e-mail

Newspaper advertisements and via health-related websites

Not stated

Flyers, e-mails and advertisements

(Paschall et al. 2006)c

(Riper et al. 2008)

(Walters et al. 2007)

(Weitzel et al. 2007)

© 2010 The Authors, Addiction © 2010 Society for the Study of Addiction MAC: assessment-only (n = 312)

College Alc (n = 310): alcohol misuse and harm prevention course consisting of 5 units: college alcohol use, harm prevention, how it works, risky business and practical solutions. Encourages development of a harm prevention plan. The program includes interactive animation and assignments, challenges normative misconceptions and alcohol expectancies Approx. 3 hours (participants given 6 weeks for completion) Theories of problem and health-related behaviour [72,73] Intervention online; location determined by participant Drinking less (n = 130): consists of four stages: (i) preparing for action, (ii) goal setting, (iii) behavioural change and (iv) maintenance of gains and relapse prevention. Access to peer-to-peer discussion forum Recommended treatment period of 6 weeks. Measured actual use of the intervention Based on cognitive behavioural and self-control principles Intervention online http://www.minderdrinken.nl Location determined by participant e-CHUG (n = 50): personalized feedback consisted of: (i) quantity/frequency drinking summary (including caloric ‘cheeseburger’ equivalent); (ii) comparison to US adult and college norms; (iii) estimated level of risk; (iv) money spent on alcohol per year; (v) no. cigarettes smoked per month; and (vi) advice and local services. Feedback was derived from responses to AUDIT, questions on genetic risk of alcoholism, weight and expenditure on alcohol Single session where feedback was viewed on screen Feedback based on motivational interviewing and social psychology approaches. Followed FRAMES model Intervention online www.e-chug.com Location determined by participant Hand-held computer with messaging (n = 20): tailored text messages sent to hand-held computer daily on avoiding alcohol-related consequences. Messages addressed three situations: (i) drinking with negative consequence, (ii) drinking without consequence and (iii) not drinking. Messages were tailored to behaviour, self-efficacy and outcome expectancies regarding alcohol-related consequences Messages were sent daily to those participants providing consumption data. Number of messages sent to and read by participants was recorded Guiding principles not stated Intervention computer-based; location determined by participant 1. MAC: hand-held computer without messaging (n = 20) 2. 3rd arm excluded from publication


MAC: web-based psychoeducational brochure describing impact of alcohol use on physical and social functioning (n = 131)

Comparator

Intervention

1. Total drinks consumed in study period 2. Drinking days 3. Drinks/drinking day

1. Typical drinks/week 2. Peak BAC

1. No. of problem drinkers 2. Mean weekly alcohol consumption

1. Frequency of alcohol use in past month 2. Frequency of heavy drinking in past month 3. Frequency of feeling drunk in past month

Drinking outcomes

2 weeks

8, 16 weeks

6 months

30 days



Total: 77




AC: active comparator group; MAC: minimally active comparator group; FRAMES: Feedback, Responsibility, Advice, Menu, Empathy and Self-efficacy [64]; MI: motivational interview [63]; BASICS: Brief Alcohol Screening and Intervention for College Students [65]; BAC: blood alcohol concentration; AUDIT: Alcohol Use Disorders Identification Test; aexcluded from meta-analyses as no measure of total alcohol consumption or binge frequency; bexcluded from meta-analyses for providing proportion of binge days and no standard deviation for total alcohol consumption; cexcluded from meta-analyses for providing frequency of heavy drinking as a categorical variable.



Netherlands: adult problem drinkers in the general population

Population

Recruitment

Study

Table 2 Cont.


Addiction, 106, 267–282


effective components from brief intervention) [64], BASICS (Brief Alcohol Screening and Intervention for College Students) [65] and the social norms approach [66–68]. In those studies that used a more extensive range of behaviour change techniques, self-control training and cognitive behaviour therapy were referenced [69,70]. Three studies did not state any guiding principles, possibly because the computer-based intervention was used in a comparator arm [39,42,49]. Intervention—intensity of intervention In many studies personalized feedback was made available on screen for a few minutes, and in some cases it was possible to print and take away. The campus-based interventions comprised longer sessions of up to 3 hours. Some studies allowed participants access to the intervention over a period of time [54,61], while others recommended revisiting the website to complete different sessions [41,59]. Two studies investigated multiple exposures to the intervention as part of their study design [39,48]. Bias assessment Three studies made explicit reference to randomization sequence generation and the procedure for allocating participants to groups. These studies were classified as having low risk of bias associated with allocation concealment [46–48]. The remainder of studies were assessed as having unclear risk of bias, meaning that there was insufficient information in the publication to judge this aspect of trial quality. Study outcomes A variety of different self-reported outcomes were used to measure alcohol consumption. Most of the studies reported between one and four different drinking outcomes, while one study reported eight [41] and another reported 10 [61] (see Table 2). Twelve studies measured short-term outcomes (less than 3 months), nine measured medium-term outcomes (3–6 months) and three measured long-term outcomes (longer than 6 months). The shortest length of follow-up was 2 weeks [56] and the longest was 12 months [39,48,62]. Twenty studies reported a sample size of fewer than 300 participants, six of which had fewer than 100 participants. The smallest sample size was 40, reported in two studies [56,57], while the largest comprised more than 1000 [62]. Total alcohol consumption (quantity measure) Nineteen studies measured the quantity of alcohol as actual or average drinks/units consumed within a given time-frame. One study was excluded from the meta© 2010 The Authors, Addiction © 2010 Society for the Study of Addiction

277

analyses as it did not provide standard deviations [45]. Fifteen studies appeared to have skewed data. Five of the 15 studies presented appropriate measures of central tendency given the skewed distribution of the data: two provided transformed data [55,58] and three reported medians [46,48,62]. Hence, a total of 18 studies (10 of which were unadjusted for skewed data) were included in the meta-analyses for this outcome (analyses 1 and 2). Analysis 1: computer-based intervention versus minimally active comparator—g/week The primary meta-analysis compared computer-based interventions with a minimally active comparator. It included 16 trials (nine unadjusted for skewed data) with a total of 3118 participants. Participants receiving the computer-based intervention reduced the amount of alcohol consumed per week significantly more than those receiving the minimally active comparator (mean difference = -25.9 g per week; 95% confidence interval (CI): -41 to -11). The mean difference was equal to 3.24 UK units of alcohol (1 UK unit = 8 g ethanol). However, there was substantial heterogeneity between the findings of the trials, with an I2 value of 62%. This suggests that although participants in most studies appeared to benefit from the computer-based intervention, the estimated benefit varied substantially between the trials.

Analysis 1.1: subgroup analysis: students versus nonstudents—g/week. This heterogeneity was explored in a subgroup analysis by population. The studies were separated into two groups: students and non-students (three studies in adult problem drinkers from the general population and one in emergency department attendees). The two groups were found to differ significantly from each other (P < 0.001), suggesting a more pronounced effect in the non-student adult population (see Fig. 2). The heterogeneity was reduced substantially within the student subgroup (I2 = 28% for students, I2 = 77% for non-students).

Analysis 1.2: sensitivity analysis (within analysis 1.1): studies presenting appropriate measures of central tendency given the distribution of the data—g/week. This analysis included two studies presenting medians [46,48], one study that presented back-transformed data [55] and two studies that reported no evidence of skew [51,52]. These five studies in student populations (994 participants) found no significant difference between computer-based interventions and minimally active comparator groups in alcohol consumed per week. This analysis was not possible in the non-student adult population due to the small number of studies. Addiction, 106, 267–282

278


Study or Subgroup 1.1.1 students

Weight

Bewick 2008 8.3% Chiauzzi 2005 6.0% Doumas & Haustveit 2008 7.0% Kypri 2004 7.3% Kypri 2008 8.0% Lau-Barraco 2008 6.2% Lewis & Neighbors 2007 9.5% Lewis et al. 2007 10.3% Neighbors 2004 8.7% Neighbors 2006 7.4% Walters 2007 8.4% Weitzel 2007 1.7% Subtotal (95% CI) 88.9%

Mean Difference IV, Random, 95% CI

Mean Difference IV, Random, 95% CI

-22.64 [-51.05, 5.77] -15.17 [-58.09, 27.75] -7.47 [-43.90, 28.96] 15.00 [-19.69, 49.69] -32.93 [-63.08, -2.78] -14.47 [-56.21, 27.27] -28.51 [-50.39, -6.63] -43.44 [-61.00, -25.88] -17.28 [-43.50, 8.94] -10.04 [-43.69, 23.61] 2.22 [-26.07, 30.51] -13.77 [-118.73, 91.19] -19.42 [-29.83, -9.00]

Heterogeneity: Tau² = 89.12; Chi² = 15.23, df = 11 (P = 0.17); I² = 28% Test for overall effect: Z = 3.65 (P = 0.0003) 1.1.2 non-students Hester 1997 Hester 2005 Neumann 2006 Riper 2008 Subtotal (95% CI)

1.1% -242.56 [-376.51, -108.61] 2.8% -132.30 [-210.14, -54.46] 4.2% -16.03 [-74.47, 42.41] 2.9% -119.00 [-194.39, -43.61] 11.1% -114.94 [-198.60, -31.29]

Heterogeneity: Tau² = 5374.82; Chi² = 12.83, df = 3 (P = 0.005); I² = 77% Test for overall effect: Z = 2.69 (P = 0.007) Total (95% CI)

100.0%

-25.88 [-40.78, -10.98]

Heterogeneity: Tau² = 481.91; Chi² = 39.26, df = 15 (P = 0.0006); I² = 62% -500 -250 0 250 500 Test for overall effect: Z = 3.40 (P = 0.0007) Favours experimental Favours control Figure 2 Forest plot of subgroup analysis by population—computer-based interventions versus minimally active comparator groups (g/week)

Analysis 2: computer-based intervention versus active comparator—g/week

Analysis 3: computer-based intervention versus minimally active comparator—binge frequency/week

Three studies (two unadjusted for skewed data), including 457 student participants, compared a computer-based intervention with an active comparator [39,42,49]. There was no significant difference between participants receiving a computer-based intervention and an active comparator group in alcohol consumed per week. There was no heterogeneity observed between the findings of the trials (I2 = 0%). However, the analysis was heavily weighted by one particular study [39].

This analysis included five trials (three unadjusted for skewed data) with a total of 848 student participants [41,44,46,48,49]. Participants receiving a computer-based intervention appeared to reduce their frequency of binge drinking compared with those receiving a minimally active comparator (mean difference = -0.23 days per week; 95% CI: -0.47, 0.00; P = 0.05). There was no heterogeneity observed between the findings of the trials (I2 = 0%).

Binge drinking (frequency measure) Eight studies measured frequency of heavy/binge drinking days or episodes within a given time-frame. Two studies were excluded from analyses for reporting the proportion of binge days [45] and frequency of heavy drinking as a categorical variable [54]. All the studies reporting this outcome demonstrated a skewed distribution at furthest point of follow-up. Two studies accounted for this by presenting medians [46,48]. © 2010 The Authors, Addiction © 2010 Society for the Study of Addiction

Analysis 4: computer-based intervention versus active comparator—binge frequency/week Only two studies made this comparison [39,49], and so the findings were not pooled in a meta-analysis. Both studies reported no significant difference in binge frequency between the intervention and an active comparator group. Addiction, 106, 267–282


DISCUSSION The data identified by this review suggest that computerbased interventions were more effective than minimally active comparator groups at reducing alcohol consumed per week (in both student and non-student adult populations) and binge frequency (in student populations). A small number of studies found no difference between alcohol consumed per week in those receiving the intervention or an active comparator. However, most studies reported skewed data, which was summarized using the mean. A sensitivity analysis of those studies that presented suitable measures of central tendency for the distribution of the data found that in student populations there was no difference between intervention and minimally active comparator groups in alcohol consumed per week. These findings should therefore be interpreted with caution. Notwithstanding the limitations of the data in the current review, a mean difference of 26 g of alcohol per week was found between computer-based interventions and minimally active comparator groups. This difference is of similar magnitude to that reported in a Cochrane review of hazardous and harmful drinkers in primary care, where participants receiving conventional face-toface brief interventions reduced their alcohol intake significantly more than those receiving a control (difference of 38 g per week) [2]. The effectiveness of computerbased interventions in student populations was less pronounced than in non-student populations and diluted the overall reduction in alcohol consumption (see Fig. 2). These differences in the size of effect between the population groups may be due to baseline risk, where nonstudent drinkers were consuming greater amounts of alcohol than students and therefore had greater capacity for reducing their intake. The theoretical basis of the intervention may also have influenced differences in effect, where non-student populations received more extensive brief interventions. Such factors, along with impact of length of follow-up, could have been considered in further analyses. However, this was not deemed appropriate given the limitations with the data. The initial finding that computer-based interventions are effective in student populations supports findings from previous research [22,24,25]. However, this review has highlighted that most studies use an inappropriate measure of central tendency (i.e. mean) given the skewed distribution of the data and the small sample sizes. A problem then lies in constructing meta-analyses of mean differences. In order not to exclude studies that used appropriate measures of central tendency (three studies that reported medians [46,48,62]), we used the median to estimate the mean and the range to generate an estimated standard deviation [37]. Estimating the sample © 2010 The Authors, Addiction © 2010 Society for the Study of Addiction

279

mean in this way may have introduced errors; however, an estimation of a correct statistic was considered preferable to the exclusion of these studies from meta-analyses. At present, there is no consensus on how best to pool different measures of central tendency in meta-analyses. An ideal analysis of the current data on computer-based interventions for reducing alcohol intake would include the individual patient data from all eligible studies. This would allow the pooling of rate ratios from negative binomial models, as advocated when using count data [27]. This review investigated the effectiveness of computer-based interventions with two specific measures of alcohol consumption: total consumption and binge frequency. It is possible that the selection of another consumption measure may have resulted in different findings. The strength of this approach is that it provides a meaningful interpretation of the pooled data, i.e. grams of alcohol consumed per week and frequency of binge drinking episodes per week. It also acknowledges that different measures of alcohol consumption reflect different patterns of drinking. This review considered allocation concealment as a potential source of bias. Only three studies were assessed as having low risk of bias [46–48], while the other studies provided insufficient information to pass judgement. It is likely that many studies assessed as unclear were poorly reported rather than poorly designed; for example, those conducted over the internet in their entirety would consequently have concealed allocation to randomized group. Other potential sources of bias in trial design include: inadequate sequence generation and blinding, incomplete outcome data and selective reporting. These features of valid trial design are most applicable to conventional drug trials and problems occur when applying them to trials of computer-based behavioural interventions, particularly those conducted online. In an online trial it is likely that sequence generation and allocation concealment will have been performed by a computer in a fully automated process. Blinding of participants and study personnel is not truly possible with behavioural interventions where some participants receive access to an intervention and others do not. Also, blinding of outcome assessors may not be relevant in an online trial where participants complete follow-up questionnaires from a remote location over the internet. Future trial designs and publications would benefit from explicit reference to these factors, and further attempts to identify other sources of bias unique to online trials and computer-based interventions, such as re-registration. At present, it is not possible to interpret the evidence with any degree of certainty. It is vital that future research in this area considers that alcohol consumption data are likely to be skewed, and that appropriate measures of central tendency are reported in trial Addiction, 106, 267–282

280


publications. The current literature is also limited by small sample sizes, short-term follow-up, insufficient information to judge potential sources of bias, few studies in non-student adult populations and few comparisons with active comparator groups. However, the volume of research is encouraging and the potential benefits of computer-based interventions for reducing alcohol consumption should continue to drive interest in this area. Declarations of interest None. Acknowledgements With special thanks to Angela Young for help in designing the search strategy and searching the databases, and Giancarlo Manzi and Simon Thompson for involvement in protocol development. We are also grateful to the reviewers for their helpful suggestions on improving the paper. References 1. Bien T. H., Miller W. R., Tonigan J. S. Brief interventions for alcohol problems: a review. Addiction 1993; 88: 315–35. 2. Kaner E. F., Beyer F., Dickinson H. O., Pienaar E., Campbell F., Schlesinger C. et al. Effectiveness of brief alcohol interventions in primary care populations. Cochrane Database Syst Rev 2007; issue 2: CD004148. 3. Moyer A., Finney J. W., Swearingen C. E., 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. 4. World Health Organization. Screening and Brief Intervention for Alcohol Problems in Primary Health Care. 2010. Available at: http://www.who.int/substance_abuse/activities/sbi/ en/index.html (accessed 26 October 2010; archived by WebCite at http://www.webcitation.org/5tlXHduaA). 5. Babor T. F., Higgins-Biddle J. C. Brief Intervention for Hazardous and Harmful Drinking: A Manual for use in Primary Care. Geneva: World Health Organization, Department of Mental Health and Substance Dependence; 2001. 6. Seppa K., Aalto M., Raevaara L., Perakyla A. A brief intervention for risky drinking—analysis of videotaped consultations in primary health care. Drug Alcohol Rev 2004; 23: 167–70. 7. Denny C. H., Serdula M. K., Holtzman D., Nelson D. E. Physician advice about smoking and drinking: are U.S. adults being informed? Am J Prev Med 2003; 24: 71–4. 8. McAvoy B. R., Donovan R. J., Jalleh G., Saunders J. B., Wutzke S. E., Lee N. et al. General practitioners, prevention and alcohol—a powerful cocktail? Facilitators and inhibitors of practising preventive medicine in general and early intervention for alcohol in particular: a 12-nation key informant and general practitioner study. Drugs Educ Prev Policy 2001; 8: 103–17. 9. Wutzke S. E., Gomel M. K., Donovan R. J. Enhancing the delivery of brief interventions for hazardous alcohol use in the general practice setting: a role for both general practitioners and medical receptionists. Health Promot J Austr 1998; 8: 105–8. © 2010 The Authors, Addiction © 2010 Society for the Study of Addiction

10. Office of National Statistics. Internet Access: Households and Individuals. Statistical Bulletin; 2009. Available at http://www.statistics.gov.uk/pdfdir/iahi0809.pdf (accessed 26 October 2010; archived by WebCite at http://www. webcitation.org/5tlbWhDP8) 11. Internet World Stats. Internet World Stats: Usage and Population Statistics. 2009. Available at: http://www. internetworldstats.com/stats.htm (accessed 15 February 2010). 12. Copeland J., Martin G. Web-based interventions for substance use disorders: a qualitative review. J Subst Abuse Treat 2004; 26: 109–16. 13. Linke S., Murray E., Butler C., Wallace P. Internet-based interactive health intervention for the promotion of sensible drinking: patterns of use and potential impact on members of the general public. J Med Int Res 2007; 9: e10. 14. Murray E. The role of internet-delivered interventions in self-care. In: Edwards A., Elwyn G., editors. Shared DecisionMaking in Health Care. Achieving Evidence-Based Patient Choice, 2nd edn. Oxford: Oxford University Press; 2009, p. 223–30. 15. Gerbert B., Bronstone A., Pantilat S., McPhee S., Allerton M., Moe J. When asked, patients tell: disclosure of sensitive health-risk behaviors. Med Care 1999; 37: 104–11. 16. Turner C. F., Ku L., Rogers S. M., Lindberg L. D., Pleck J. H., Sonenstein F. L. Adolescent sexual behavior, drug use, and violence: increased reporting with computer survey technology. Science 1998; 280: 867–73. 17. Cloud R. N., Peacock P. L. Internet screening and interventions for problem drinking: results from the www. carebetter.com pilot study. Alcohol Treat Q 2001; 19: 23–44. 18. Cunningham J. A., Humphreys K., Koski-Jannes A. Providing personalized assessment feedback for problem drinking on the internet: a pilot project. J Stud Alcohol 2000; 61: 794–8. 19. Lieberman D. Z. Determinants of satisfaction with an automated alcohol evaluation program. Cyberpsychol Behav 2003; 6: 677–82. 20. Linke S., Brown A., Wallace P. Down your drink: a webbased intervention for people with excessive alcohol consumption. Alcohol Alcohol 2004; 39: 29–32. 21. Saitz R., Helmuth E. D., Aromaa S. E., Guard A., Belanger M., Rosenbloom D. L. Web-based screening and brief intervention for the spectrum of alcohol problems. Prev Med 2004; 39: 969–75. 22. Elliott J. C., Carey K. B., Bolles J. R. Computer-based interventions for college drinking: a qualitative review. Addict Behav 2008; 33: 994–1005. 23. Bewick B. M., Trusler K., Barkham M., Hill A. J., Cahill J., Mulhern B. The effectiveness of web-based interventions designed to decrease alcohol consumption—a systematic review. Prev Med 2008; 47: 17–26. 24. Carey K. B., Scott-Sheldon L. A., Elliott J. C., Bolles J. R., Carey M. P. Computer-delivered interventions to reduce college student drinking: a meta-analysis. Addiction 2009; 104: 1807–19. 25. Rooke S., Thorsteinsson E., Karpin A., Copeland J., Allsop D. Computer-delivered interventions for alcohol and tobacco use: a meta-analysis. Addiction 2010; 105: 1381–90. 26. Deeks J. J., Higgins J. P. T., Altman D. G. Analysing data and undertaking meta-analyses. 9.4.5.3 Meta-analysis of skewed data. In: Higgins J. P. T., Green S., editors. Cochrane Handbook for Systematic Reviews of Interventions, version 5.0.2 [updated September 2009]. The Cochrane Collaboration; 2009. Addiction, 106, 267–282


27. Horton N. J., Kim E., Saitz R. A cautionary note regarding count models of alcohol consumption in randomized controlled trials. BMC Med Res Methodol 2007; 7: 9. 28. Kypri K. Methodological issues in alcohol screening and brief intervention research. Subst Abuse 2007; 28: 31–42. 29. Orford J. Addiction as excessive appetite. Addiction 2001; 96: 15–31. 30. Altman D. G. Practical Statistics for Medical Research. London: Chapman and Hall; 1991. 31. Ritterband L. M. Internet interventions: in review, in use, and into the future. Prof Psychol Res Pr 2003; 34: 527–34. 32. Hewitt C., Hahn S., Torgerson D. J., Watson J., Bland J. M. Adequacy and reporting of allocation concealment: review of recent trials published in four general medical journals. BMJ 2005; 330: 1057–8. 33. Schulz K. F., Chalmers I., Hayes R. J., Altman D. G. Empirical evidence of bias. Dimensions of methodological quality associated with estimates of treatment effects in controlled trials. JAMA 1995; 273: 408–12. 34. Kjaergard L. L., Villumsen J., Gluud C. Reported methodologic quality and discrepancies between large and small randomized trials in meta-analyses. Ann Intern Med 2001; 135: 982–9. 35. Higgins J. P. T., Altman D. G. Assessing risk of bias in included studies. In: Higgins J. P. T., Green S., editors. Cochrane Handbook for Systematic Reviews of Interventions, version 5.0.2 [updated September 2009]. The Cochrane Collaboration; 2009. 36. Miller W. R., Heather N., Hall W. Calculating standard drink units: international comparisons. Br J Addict 1991; 86: 43–7. 37. Hozo S. P., Djulbegovic B., Hozo I. Estimating the mean and variance from the median, range, and the size of a sample. BMC Med Res Methodol 2005; 5: 13. 38. Altman D. G., Bland J. M. Detecting skewness from summary information. BMJ 1996; 313: 1200. 39. Barnett N. P., Murphy J. G., Colby S. M., Monti P. M. Efficacy of counselor versus computer-delivered intervention with mandated college students. Addict Behav 2007; 32: 2529– 48. 40. Bewick B. M., Trusler K., Mulhern B., Barkham M., Hill A. J. The feasibility and effectiveness of a web-based personalised feedback and social norms alcohol intervention in UK university students: a randomised control trial. Addict Behav 2008; 33: 1192–8. 41. Chiauzzi E., Green T. C., Lord S., Thum C., Goldstein M. My student body: a high-risk drinking prevention web site for college students. J Am Coll Health 2005; 53: 263–74. 42. Donohue B., Allen D. N., Maurer A., Ozols J., DeStefano G. A controlled evaluation of two prevention programs in reducing alcohol use among college students at low and high risk for alcohol related problems. J Alcohol Drug Educ 2004; 48: 13–33. 43. Doumas D. M., Haustveit T. Reducing heavy drinking in intercollegiate athletes: evaluation of a web-based personalized feedback program. Sport Psychol 2008; 22: 212–28. 44. Hedman A. S. Effects of personalized feedback and tailored health communication on alcohol consumption, alcoholrelated behaviors, and attitude among binge drinking college students. Diss Abstr Int A Humanit Soc Sci 2007; 68: 891. 45. Hunt W. M. Effects of participant engagement on alcohol expectancies and drinking outcomes for a computerized expectancy challenge intervention. Diss Abstr Int B Sci Eng 2004; 65: 6655. © 2010 The Authors, Addiction © 2010 Society for the Study of Addiction

281

46. Kypri K., Saunders J. B., Williams S. M., McGee R. O., Langley J. D., Cashell-Smith M. L. et al. Web-based screening and brief intervention for hazardous drinking: a doubleblind randomized controlled trial. Addiction 2004; 99: 1410–7. 47. Kypri K., McAnally H. M. Randomized controlled trial of a web-based primary care intervention for multiple health risk behaviors. Prev Med 2005; 41: 761–6. 48. Kypri K., Langley J. D., Saunders J. B., Cashell-Smith M. L., Herbison P. Randomized controlled trial of web-based alcohol screening and brief intervention in primary care. Arch Int Med 2008; 168: 530–6. 49. Lau-Barraco C., Dunn M. E. Evaluation of a single-session expectancy challenge intervention to reduce alcohol use among college students. Psychol Addict Behav 2008; 22: 168–75. 50. Lewis M. A., Neighbors C., Oster-Aaland L., Kirkeby B. S., Larimer M. E. Indicated prevention for incoming freshmen: personalized normative feedback and high-risk drinking. Addict Behav 2007; 32: 2495–508. 51. Lewis M. A., Neighbors C. Optimizing personalized normative feedback: the use of gender-specific referents. J Stud Alcohol Drugs 2007; 68: 228–37. 52. Neighbors C., Larimer M. E., Lewis M. A. Targeting misperceptions of descriptive drinking norms: efficacy of a computer-delivered personalized normative feedback intervention. J Consult Clin Psychol 2004; 72: 434–47. 53. Neighbors C., Lewis M. A., Bergstrom R. L., Larimer M. E. Being controlled by normative influences: selfdetermination as a moderator of a normative feedback alcohol intervention. Health Psychol Off J Div Health Psychol Am Psychol Assoc 2006; 25: 571–9. 54. Paschall M. J., Bersamin M., Fearnow-Kenney M., Wyrick D., Currey D. Short-term evaluation of a web-based college alcohol misuse and harm prevention course (College Alc). J Alcohol Drug Educ 2006; 50: 49–65. 55. Walters S. T., Vader A. M., Harris T. R. A controlled trial of web-based feedback for heavy drinking college students. Prev Sci Off J Soc Pre Res 2007; 8: 83–8. 56. Weitzel J. A., Bernhardt J. M., Usdan S., Mays D., Glanz K. Using wireless handheld computers and tailored text messaging to reduce negative consequences of drinking alcohol. J Stud Alcohol Drugs 2007; 68: 534–7. 57. Hester R. K., Delaney H. D. Behavioral Self-Control Program for Windows: results of a controlled clinical trial. J Consult Clin Psychol 1997; 65: 686–93. 58. Hester R. K., Squires D. D., Delaney H. D. The Drinker’s Check-up: 12-month outcomes of a controlled clinical trial of a stand-alone software program for problem drinkers. J Subst Abuse Treat 2005; 28: 159–69. 59. Riper H., Kramer J., Smit F., Conijn B., Schippers G., Cuijpers P. Web-based self-help for problem drinkers: a pragmatic randomized trial. Addiction 2008; 103: 218–27. 60. Doumas D. M., Hannah E. Preventing high-risk drinking in youth in the workplace: a web-based normative feedback program. J Subst Abuse Treat 2008; 34: 263–71. 61. Matano R. A., Koopman C., Wanat S. F., Winzelberg A. J., Whitsell S. D., Westrup D. et al. A pilot study of an interactive web site in the workplace for reducing alcohol consumption. J Subst Abuse Treat 2007; 32: 71–80. 62. Neumann T., Neuner B., Weiss-Gerlach E., Tonnesen H., Gentilello L. M., Wernecke K. D. et al. The effect of computerized tailored brief advice on at-risk drinking in subcritically injured trauma patients. J Trauma 2006; 61: 805–14. Addiction, 106, 267–282

282


63. Miller W. R., Rollnick S. Motivational Interviewing: Preparing People for Change, 2nd edn. New York: Guilford Press; 2002. 64. Miller W. R., Sanchez V. C. Motivating young adults for treatment and lifestyle change. In: Howard G., Nathan P. E., editors. Alcohol Use and Misuse by Young Adults. Notre Dame, IN: University of Notre Dame Press; 1994. 65. Dimeff L. A., Baer J. S., Kivlahan D. R., Marlatt G. A. Brief Alcohol Screening and Intervention for College Students: A Harm Reduction Approach. New York: Guilford Press; 1999. 66. Perkins H. W., Berkowitz A. D. Perceiving the community norms of alcohol use among students: some research implications for campus alcohol education programming. Int J Addict 1986; 21: 961–76. 67. Perkins H. W. Social norms and the prevention of alcohol misuse in collegiate contexts. J Stud Alcohol Suppl 2002; 14: 164–72. 68. Borsari B., Carey K. B. Peer influences on college drinking: a review of the research. J Subst Abuse 2001; 13: 391–424. 69. Hester R. K. Behavioral self-control training. In: Hester R. K., Miller W. R., editors. Handbook of Alcoholism Treatment Approaches: Effective Alternatives, 2nd edn. Needham Heights, MA: Allyn and Bacon; 1995, p. 148–59.

70. Miller W. R., Munoz R. F. How to Control Your Drinking. Albuquerque, NM: University of New Mexico Press; 1982. 71. Goldman M. S. Expectancy operation: cognitive/neural models and architectures. In: Kirsch I. E., editor. Expectancy, Experience, and Behavior. Washington, DC: APA Books; 1999, p. 41–63. 72. Ajzen I. Attitudes, Personality and Behavior. Milton Keynes, UK: Open University Press; 1988. 73. Jessor R., Jessor S. L. Problem Behavior and Psychosocial Development. New York: Academic Press; 1977. 74. Saunders J. B., Aasland O. G., Babor T. F., de l. F., Jr, Grant M. Development of the alcohol use disorders identification test (AUDIT): WHO collaborative project on early detection of persons with harmful alcohol consumption—II. Addiction 1993; 88: 791–804. 75. Selzer M. L. The Michigan Alcoholism Screening Test: the quest for a new diagnostic instrument. Am J Psychiatry 1971; 127: 1653–8. 76. Mayfield D. G., McLeod G., Hall P. The CAGE questionnaire: validation of a new alcoholism screening instrument. Am J Psychiatry 1974; 131: 1121–3.

APPENDIX 1 MEDLINE search strategy Search strategy used for MEDLINE database Computer-related terms: #33 ((personal adj digital adj assistant) or pda) in ti,ab,kw 3325 #32 (surf* near4 internet*) in ti,ab,kw 60 #31 (surf* near4 web*) in ti,ab,kw 92 #30 (virtual adj reality) in ti,ab,kw 2096 #29 (consumer adj health adj informatic*) in ti,ab,kw 49 #28 ((e adj health) or e-health or (electronic adj health)) in ti,ab,kw 1463 #27 (interactive near ((health adj communicat*) or televis* or video* or technolog* or multimedia)) in ti,ab,kw 1420 #26 ((bulletin adj board*) or bulletinboard* or messageboard* or (message adj board*)) in ti,ab,kw 280 #25 (blog* or web-log* or weblog*) in ti,ab,kw 149 #24 ((chat adj room*) or chatroom*) in ti,ab,kw 144 #23 (online or on-line) in ti,ab,kw 27137 #22 ((internet adj based) or internet-based) in ti,ab,kw 1690 #21 ((web adj based) or web-based) in ti,ab,kw 5154 #20 ((world adj wide adj web) or (world-wide-web) or www or (world-wide adj web) or (worldwide adj web) or website*) in ti,ab,kw 6587 #19 ((electronic adj mail) or e-mail* or email*) in ti,ab,kw 3683 #18 ((mobile or cellular or cell) adj (phone* or telephone*)) in ti,ab,kw 1860 #17 ((CD adj ROM) or cd-rom or cdrom or (compact adj dis*)) in ti,ab,kw 1238 #16 (decision adj (tree* or aid*)) in ti,ab,kw 2693 #15 (internet or (local adj area adj network*)) in ti,ab,kw 15034 #14 (computer* or microcomputer* or laptop) in ti,ab,kw 175387 #13 explode ‘Software-’ / all SUBHEADINGS in MIME,MJME,PT 66293 #12 explode ‘Computer-Graphics’ / all SUBHEADINGS in MIME,MJME,PT 11752 #11 explode ‘Public-Health-Informatics’ / all SUBHEADINGS in MIME,MJME,PT 679 #10 explode ‘Computer-Assisted-Instruction’ / all SUBHEADINGS in MIME,MJME,PT 6187 #9 explode ‘Audiovisual-Aids’ / all SUBHEADINGS in MIME,MJME,PT 61028 #8 explode ‘Decision-Support-Techniques’ / WITHOUT SUBHEADINGS in MIME,MJME,PT 51993 #7 explode ‘Medical-Informatics’ / all SUBHEADINGS in MIME,MJME,PT 147451 #6 explode ‘Computer-Systems’ / all SUBHEADINGS in MIME,MJME,PT 103304 Alcohol-related terms: #5 (alcohol* near (abuse or related disorder* or drink* or excessive or consum* or intake or reduction or misuse* or dependen*)) in ti,ab,kw 57068 #4 ((heavy or hazardous or harmful or excessive or problem or binge or controlled) adj drink*) in ti,ab,kw 6605 #3 explode ‘Alcoholic-Beverages’ / all SUBHEADINGS in MIME,MJME,PT 9255 #2 explode ‘Alcohol-Drinking’ / all SUBHEADINGS in MIME,MJME,PT 35790 #1 explode ‘Alcohol-Related-Disorders’ / all SUBHEADINGS in MIME,MJME,PT 80278


Addiction, 106, 267–282