Introduction
Methods
Search Strategy
Selection Criteria
Review Screening and Data Extraction
First author | Year | Purpose | Inclusion criteria | No of trials | Population/Setting | Control/comparator conditions | Follow up | Meta-analysis | Calculation of effect size | Moderator 1: Therapeutic Orientation | Moderator 2: Length of intervention | Moderator 3: Guidance | Moderator 4: Trial Engagement | Main findings |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Bewick | 2008 | To review published literature on the effectiveness of web-based interventions designed to decrease consumption of alcohol and/or prevent alcohol abuse | 1. Online intervention 2. Alcohol consumption as focus 3. Evaluation of intervention | 5 | 4 Student 1 General population | 1 No intervention 2 Alcohol website 2 Printed material | No mention | No | No | SBI (personalized feedback) | Includes both single-session and multiple session interventions | No mention | No mention | Heterogeneous outcome measures Small sample size at follow-up Large SD’s indicating skew Lack of pure controls Large CI’s around effect size |
Carey | 2009 | To evaluate efficacy and moderators of computer delivered interventions to reduce alcohol use among college students | 1. Electronic intervention 2. Undergraduate students 3. Behavioural Outcomes 4. Enough info to calculate effect sizes | 35 | Student | 18 waiting list/no intervention 17 alcohol relevant content | 0 to 26 weeks | Yes | Yes Short-term (≤5 weeks) and long-term (≥6 weeks) d = 0.02–0.28 on several different alcohol-related outcomes | SBI: feedback, norm comparison, alcohol education, tailored material | Includes mostly single-session interventions but some multiple session interventions Length is used as moderator in analysis | Mentions human interaction in some studies and uses this in moderator analysis | Not analysed | Interventions effective at reducing specific intervals/drink-days and max quantity, but not other quantity and frequency measures or problems (short-term). Interventions effective at reducing quantity, frequency of drinking days and problems, but not other quantity and frequency measures (long-term). Significant moderators of effect: earlier publications, fewer participants, not a commercial program, include human interaction, provide education materials, no feedback on problems. |
Khadjesari | 2010 | To determine the effects of computer based interventions aimed at reducing alcohol consumption in adult populations | 1. RCT 2. Adults drinking alcohol at any level 3. Behavioural interventions 4. Stand-alone interventions only 5. Measures change in alcohol consumption | 24 | 18 Student 3 General population 2 Employee 1 Emergency department | 22 Assessment only/waiting list 3 Active comparison | 2 weeks to 12 months | Yes | Yes Grammes of ethanol: −25.9 g/week (−41 to −11) Binge frequency: −0.23 days/week (−0.47 to 0) | 15 SBI (personalized feedback) 5 virtual campus 1 expectancy challenge 3 extensive behaviour change intervention | Includes mostly single-session interventions but some multiple session interventions | Not eligible | Not analysed | Interventions more effective than minimally active comparators at reducing alcohol consumed per week, and binge frequency among students. Interventions as effective as active comparator groups. Less pronounced difference among students, possibly due to baseline risk, i.e. students consumed less at baseline. Future research must use suitable measures of central tendency to summarize data |
Rooke | 2010 | 1. To quantify overall effectiveness of computer-delivered interventions for alcohol and tobacco 2. To determine whether effectiveness is associated with treatment characteristics (moderators) | 1. RCT with control group 2. Computer-based 3. Substance use (only tobacco and alcohol included) | 34 studies, 42 articles | 28 Young adult 9 Adult 30+ 3 Adolescent 3 No info | 35 attention/placebo 7 active comparison | 1 to 156 weeks | Yes | Yes d = 0.20 for both alcohol and tobacco Alcohol more effective than tobacco d = 0.26 vs. d = 0.12 | Monitoring, normative feedback, chat, entertainment, relapse prevention These are used as moderators | Includes mostly single-session interventions but some multiple session interventions Number of treatment sessions used as moderator | Therapist involvement used as moderator | Intention to Treat (ITT) used as moderator | Effect sizes were higher for studies with control group relative to studies in which the comparison condition was an active comparison Lower effect sizes associated with studies that include entertainment |
Tait | 2010 | To conduct a systematic review of randomized trials of web-based interventions for problematic substance use for young adults and adolescents | 1. RCT 2. No-treatment control 3. Adolescents or young adults (aged ≤25) 4. Consumption measure 5. Only fully automated interventions 6. No CDROM/stand-alone computer | 14 | 13 student 1 employee | 4 assessment only 3 No intervention 3 printed material 2 web education 2 other | 30 days to 12 months | Yes | Yes Overall effect d = −0.22 Level of alcohol consumption, d = − 0.12 Binge or heavy drinking frequency, d = − 0.35 Alcohol-related soc. problems, d = − 0.57 | SBI: personalized feedback, norm comparisons, alcohol education | Includes mostly single-session interventions but some multiple session interventions | Not eligible | Not analysed | Small significant effect on drinking outcomes Web interventions targeting alcohol-related problems have an effect about equivalent to brief in-person interventions Interventions to prevent the development of alcohol-related problems in those who do not currently drink appear to have minimal impact |
White | 2010 | To review the efficacy of online interventions for alcohol problems | 1. Intervention delivered via the Internet 2. Focus on moderating or stopping drinking 3. RCT 4. screening assessment or intervention | 17 | 12 student 3 community 2 employee | 10 Psycho-education 7 assessment only | 1 week to 12 months | Yes, but only based on 5 studies | Yes Alcohol units/ week or month Cohens d = 0.42 | SBI, information, alcohol education | Includes both single-session interventions and multiple session interventions | No mention | Retention ranged from 33.4 % to 100 %, Median reported retention in treatment condition 83.4 % at 1 month, 74.5 % at 3 months, and 74.5 % at 6 months. In controls, median retention was 80 %, 70.4 %, and 74.9 %. | Overall, online alcohol interventions appear to bring small reductions in alcohol use. However, this is based on synthesis of the results of only 5/17 included studies |
Riper | 2011 | To assess overall effectiveness of self-help alcohol e-health interventions and to perform subgroup analyses. | 1. digital self-help interventions (online or offline) 2. RCT 3. Data usable in meta-analytic procedures 4. Alcohol-drinking behaviour as primary outcome variable 5. Only results from problem drinkers 6. Adult problem drinkers (18+) 7. Control group | 9 (5 single session 4 extended self-help) | General | 4 Alcohol leaflet 3 Waiting list 3 Assessment only | 4 weeks to 9 months | Yes | Yes Hedges g Fixed Effects =0.39 (0.29–0.50); Random Effects =0.44 (0.17–0.71) | SBI Cognitive behaviour therapy | Includes both single-session interventions and multiple session interventions Extended vs single session treatment used as moderator: Single-session Mixed Effects =0.27 (0.11–0.43) Multisession Mixed Effects =0.61 (0.33–0.90) | Not included, but its relevance is mentioned in discussion | 0–42 % Loss to follow-up is discussed as a limitation | Alcohol self-help interventions are effective in general Earlier reviews showing smaller effects may be due to including studies in student populations Larger effects for multisession interventions in comparison to single session interventions |
Riper | 2014 | To conduct a meta-analysis including both guided and unguided alcohol Internet interventions | 1. RCT with control group 2. Low intensity interventions on computer or mobile device, with or without guidance 3. Alcohol consumption primary outcome measure 4. age 18 or over 5. Drinkers who exceed local guidelines for low-risk drinking | 16 | 9 community 2 workplace 1 emergency department 1 women and infant services 1 general practice 1 military 1 alcohol website | 6 Assessment only 3 Waiting list 7 Alcohol info | 4 weeks to 12 months | Yes | Yes Hedges g = 0.20 (0.13–0.27), 22 g/week (2.2 drinks) less at post-intervention | 7 personalized normative feedback (NF) 9 Combination of NF, Motivational Interviewing and, CBT 1 generic health education approach | Includes both single-session interventions and multiple session interventions | Yes, moderator analysis performed displaying a non significant difference: 0.23 for guided, 0.20 for unguided | Substantial drop out rate mentioned as a limitation | Supports the use of both guided and unguided interventions More studies on guided interventions are needed No significant differences between single session interventions and extended ones Studies with longer follow-up needed The brevity of intervention descriptions, intervention uptake, completion rates and heterogeneity of outcome measures and follow up measures impede the ability to generalize the efficacy |
Donoghue | 2014 | To determine effectiveness of alcohol Internet interventions over time in nontreatment-seeking hazardous or harmful drinkers | 1. RCT 3. eSBI vs control group 2. Hazardous drinking 3. At least 10 participants 4. No alcohol dependence | 23 | 16 student 7 other | 19 Assessment only 2 Alcohol leaflet 2 Other | 4 weeks to 12 months | Yes, but only based on 17 studies | Yes Grammes of ethanol: - <3 months −32.74 g/wk. 3–6 months −14.91 g/wk. 6–12 months −7.46 g /wk. | Brief intervention with information and advice | Mostly single-session interventions but some multiple session interventions | No mention | Drop-out in included studies range 7.6–44.9 % The possibility of using incentives to reduce drop-out is mentioned in the discussion | Review found a significant difference between intervention and control group at less than 3 months and between 3 and 6 months follow up. However, there were no significant differences at follow ups 12 months or longer |
Balhara | 2014 | To evaluate the available evidence for the effectiveness of web based interventions for reducing alcohol use | 1.Internet intervention 2. fully automated 3. RCT 4. RCT with at least one no-treatment control group | 35 studies 41 articles | 21student/ school/ adolescent 14 adult | 29 no-intervention 7 active control | 1 week to 12 months | No | No | Mostly normative personalized feedback | Mostly single-session interventions but some multiple session interventions | Not eligible | No mention | Among adults, interventions were found to be more effective in reducing alcohol consumption than a control group only in three out of 14 studies. Among students, reduced alcohol consumption was found in 10 out of 21 studies at different time points. |
Bhochiboya | 2015 | To review Internet interventions targeting binge drinking among students | 1. Binge drinking outcome 2. Internet 3. Studies on multiple health behaviours included as well 4. Studies combining face-to-face with e-intervention included 4. Student population | 14 | Student | 10 control groups 4 experimental groups | 4 weeks to 2 years | No | No | Mostly personalized feedback. Social cognitive theory, social norms feedback theory etc. are mentioned | Mostly single-session interventions but some multiple session interventions | No mention | No mention | Periodic interventions lead to larger effects than one-time interventions. Changing perceived/subjective norms seems to be effective in reducing risky drinking behaviour The reliance on self-report is a limitation to the field because of potential social desirable answering tendencies |
Dedert | 2015 | To characterize treatment intensity and review evidence for efficacy of e-interventions for reducing alcohol consumption and alcohol-related impairment among adults and students, | 1. RCT with active or inactive controls 2. Follow-up 6 months or longer 3. Online, CD-ROM, mobile applications or IVR | 28 | 14 adult 14 student | Student population: 8 waiting list 5 attention control 1 treatment as usual Adult population: 5 waiting list 6 attention control 3 treatment as usual | 6 to 12 months | Yes | Yes Grammes of ethanol: Adults: 6 months- - 25.0 g/wk. (95 % CI −51.9 to 1.9) 12 months −8.6 g/wk. (95 % CI −53.7 to 36.5) Students:, 6 months −11.7 g/wk. (95 % CI −19.3 to _4.1) 12 months −4.7 g/wk. (95 % CI −24.5 to 15.1) | Mostly brief intervention, personalized feedback and psycho-education | Mostly single-session interventions but some multiple session interventions | Yes, the review classifies studies by level of human support | No mention | General reduction by 1 drink/week, at 6 months but no difference at 12 months, Diagnostic assessment of participants would be valuable in future research More studies on intensive interventions, preferably with some form of human interaction, are needed |
Leeman | 2015 | To evaluate the efficacy of very brief, web-based alcohol interventions for college students | 1. data collection entirely web-based 2. intervention takes no more than 15 min to complete 3. intervention concerns alcohol use in general (as opposed to event-specific drinking) 4. primary focus on alcohol consumption 5. students 6. random assignment | 15 | Student | 9 assessment only 3 attention control 2 alternate personalized feedback 1 education | 1 to 24 months | No | No | All interventions were based on personalized normative feedback, and most of these were multi- component interventions with for example behavioural techniques and facts about alcohol | All included interventions were single- session | No mention | Yes, reports on attrition bias for each individual study | Nine of the 15 RCT’s included reported an ES of at least 0.20 or greater in alcohol reduction for intervention, while 2 reported an ES in the opposite direction (control group fared better) |
Dotson | 2015 | To summarize available research and to perform a meta-analytic review of computer-delivered stand-alone personalized normative feedback (PNF) interventions for college student drinking | 1. examines a stand-alone PNF drinking intervention 2. student sample 3. includes a control condition 4. reports outcomes for drinking norms and actual drinking behaviour 5. uses a pre-post experimental design with at least 28 days between baseline and follow-up 6. provides adequate information for effect size calculation | 8 | Student | 5 assessment only 3 attention control | 4 to 20 weeks | Yes | Yes Gender neutral d = 0.291 [95 % CI, 0.16 to 0.42] Gender-specific d = 0.284 [95 % CI 0.12 to 0.45] Number of drinks: Gender neutral −3.027 [95 % CI, 2.17 to 3.88] Gender-specific −3.089 [95 % CI −0.99 to 0.59] | Personalized normative feedback | Not specified, but probably mostly single session interventions | No mention | Specifies attrition rates for each included study (4.2–21 %) | Both gender-neutral and gender-specific PNF interventions reported greater reductions when compared to controls. Results were consistent regardless of intervention setting, suggesting that computer-delivered PNF is equally effective when completed in a structured setting or when completed in a non-structured setting. Researchers should assess outcomes appropriately for the population of interest, rather than selecting a measure solely because it is a common metric. |
Results
Question 1: Are Computer-Based Interventions for Problematic Alcohol Use Effective?
Reviews on Mixed Populations
Reviews on Student Populations1
Reviews on Adult (Non-Student) Populations2
Question 2. What Is Known about Moderators of Effectiveness and Trial Engagement?
Theme | Studies |
---|---|
1. Therapeutic orientation—outcome | |
2. Length of intervention—outcome | |
3. Guidance—outcome | |
4. Trial engagement—outcome | |
5. Therapeutic orientation—trial engagement | – |
6. Length of intervention—trial engagement | – |
7. Guidance—trial engagement | – |