Curbing Problem Drinking with Personalized-Feedback Interventions

Review and Special Articles Curbing Problem Drinking with Personalized-Feedback Interventions A Meta-Analysis Heleen Riper, PhD, MSc, Annemieke van Straten, PhD, Max Keuken, BSc, Filip Smit, PhD, Gerard Schippers, PhD, Pim Cuijpers, PhD Context: The effectiveness of personalized-feedback interventions to reduce problem drinking has been evaluated in several RCTs and systematic reviews. A meta-analysis was performed to examine the overall effectiveness of brief, single-session personalized-feedback interventions without therapeutic guidance. Evidence acquisition: The selection and analyses of studies were conducted in 2008. Fourteen RCTs of single-session personalized-feedback interventions without therapeutic guidance were identified, and their combined effectiveness on the reduction of problematic alcohol consumption was evaluated in a meta-analysis. Alcohol consumption was the primary outcome measure. Evidence synthesis: The pooled standardized-effect size (14 studies, 15 comparisons) for reduced alcohol consumption at post-intervention was d⫽0.22 (95% CI⫽0.16, 0.29; the number needed to treat⫽8.06; areas under the curve⫽0.562). No heterogeneity existed among the studies (Q⫽10.962; p⫽0.69; I2⫽0). Conclusions: The use of single-session personalized-feedback interventions without therapeutic guidance appears to be a viable and probably cost-effective option for reducing problem drinking in student and general populations. The Internet offers ample opportunities to deliver personalized-feedback interventions on a broad scale, and problem drinkers are known to be amenable to Internet-based interventions. More research is needed on the long-term effectiveness of personalized-feedback interventions for problem drinking, on its potential as a first step in a stepped-care approach, and on its effectiveness with other groups (such as youth obliged to use judicial service programs because of violations of minimum-age drinking laws) and in other settings (such as primary care). (Am J Prev Med 2009;36(3):247–255) © 2009 American Journal of Preventive Medicine Introduction P roblem drinking is a major public health issue, particularly due to its high prevalence in adult1,2 and student populations.3,4 It is these groups of problem drinkers—and not those with severe alcohol dependence, as is often thought—who account for the bulk of the alcohol-related harm in the general popuFrom the Innovation Centre of Mental Health and Technology (Riper, Smit), Trimbos Institute (Netherlands Institute of Mental Health and Addiction), Utrecht; the Department of Clinical Psychology and Institute for Research in Extramural Medicine (EMGO) Institute (Riper, van Straten, Smit, Cuijpers), Vrije Universiteit; Amsterdam Medical Centre (Schippers), Cognitive Science (Keuken), Institute for Interdisciplinary Studies, University of Amsterdam, Amsterdam, The Netherlands Address correspondence and reprint requests to: Heleen Riper, PhD, MSc, Trimbos Institute, P.O. Box 725, 3500 AS Utrecht, The Netherlands. E-mail: [email protected]. The full text of this article is available via AJPM Online at www.ajpm-online.net; 1 unit of Category-1 CME credit is also available, with details on the website. lation.5,6 Problem drinking causes a formidable array of serious health problems7 and a heavy social and economic burden.8,9 Besides short-term and long-term morbidity1,2 and mortality,10 the consequences include acute unintended injuries, sexual and physical assault, violence-related trauma, vandalism, and poor academic or work performance.11,12 Early identification and brief interventions have been increasingly advocated as cost-effective strategies to curb problem drinking.8,13–15 Evidence is strongest for brief interventions in primary and secondary care,6,16 –18 but effectiveness has also been shown in settings such as the general population19 –21 or student communities.22–25 Less encouraging is the fact that the implementation of brief interventions is still hampered by constraints such as a limited number of professionals who administer them, the difficulty of contacting problem drinkers, and the high costs of implementation and delivery.21,26,27 As a Am J Prev Med 2009;36(3) © 2009 American Journal of Preventive Medicine • Published by Elsevier Inc. 0749-3797/09/$–see front matter doi:10.1016/j.amepre.2008.10.016 247 consequence, as many as 80% of problem drinkers are not yet receiving help.28 –30 Innovative ways are needed for reaching out to them. Brief personalized feedback (i.e., personalizedfeedback interventions) could be one such strategy,31 providing personal feedback regarding an individual’s alcohol-consumption patterns. This feedback may consist of different components, such as an overview of mean weekly alcohol consumption; blood alcohol concentration levels (BAC); associated health and social risks of problem drinking; or self-help guidelines to change problematic alcohol consumption. Normative feedback is another important component of many personalized-feedback interventions. It enables problem drinkers to compare their own alcohol consumption (in terms of frequency, quantity, or other measures) to the level of their own cohort—such as the average man or woman in the general population or their student peer group32–34—as well as to the recommended guidelines for sensible drinking. The rationale of normative feedback is that such comparisons trigger an awareness in problem drinkers of their own drinking patterns and the risks they are taking, thus motivating them to reduce their alcohol use.33 One underlying explanation for such behavioral change is that many problem drinkers overestimate the alcohol consumption of others while underestimating their own.35–37 Personalized-feedback interventions may consist only of normative feedback.38 Personalized feedback began as a component of evidencebased, face-to-face individual or group motivationalenhancement interventions.17,39 Today, personalizedfeedback interventions are being successfully provided as autonomous, face-to-face self-help interventions in both individual and group formats. Technologic advances also now enable the delivery of automated mail, computer, and Internet-based personalized feedback. This includes individual single-session interventions without therapeutic guidance, provided in various settings and to various populations.22,24,31,40 The systematic review by White25 on personalized feedback for college students has shown that mail- or web-based personalized-feedback interventions without professional guidance were as effective in student populations as brief face-to-face interventions. Studies38,41 on needs assessment in problem-drinking populations also suggest that personalized feedback is a highly practical method for the target groups concerned. Both adults and students often prefer to use self-help interventions without therapeutic involvement to address their problem drinking instead of more-intensive individual or group treatments.41,42 Given these promising results with single-session, stand-alone, personalized-feedback interventions, it was decided to assess their effectiveness in a meta-analysis. To the best of our knowledge, this is the first meta-analysis to focus on brief personalized-feedback interventions without professional guidance for young and adult problem drinkers. The expectation was that 248 personalized-feedback interventions for problem drinkers would be more effective than non-intervention in reducing problem drinking. Evidence Acquisition Identification and Selection of Studies The relevant studies were identified in 2008, using several systematic search strategies: 1. Systematic searches were carried out in the following bibliographical databases: MEDLINE; PsycINFO (1985 to present); Science Citation Index Expanded; Social Sciences Citation Index; Arts & Humanities Citation Index® (1988 to present); CINAHL®; EMBASE; the Cochrane Drug and Alcohol Group Specialised Register; the Cochrane Effective Practice and Organisation of Care Group; the Alcohol and Alcohol Problems Science Database; and ETOH (etoh.niaaa.nih.gov; 1972 to 2003). Text and key words indicative of personalized-feedback interventions for problem drinking (personalized feedback, personalized normative feedback, self-help, brief intervention, brief psychotherapy, bibliotherapy) were combined with terms referring to the content of the problem (problem drinking, binge drinking, hazardous drinking, alcohol abuse, alcoholism— both MeSH terms and free text words); the setting (primary care, general population, community, Internet, adults, students, mail, web-based); and the study design (RCTs). These search strategies were combined with the optimal search strategy for RCTs designed by the United Kingdom’s Cochrane Centre.43 2. References were examined relating to earlier meta-analyses and systematic reviews on brief interventions, self-help interventions, and personalized-feedback interventions for problem drinking.6,16,18 –21,31,44 –53 3. Unpublished literature was searched by scanning Dissertation Abstracts and Digital Dissertations. 4. Reference lists of retrieved papers were screened, and papers that possibly met inclusion criteria were retrieved and studied (Figure 1). No language restrictions were applied. Selection of Primary Studies For inclusion in the meta-analysis, studies on personalized feedback for problem drinkers were selected that (1) applied Figure 1. Flow chart of study selection resulting in inclusion of 14 studies (15 comparisons) American Journal of Preventive Medicine, Volume 36, Number 3 www.ajpm-online.net a randomized– controlled design (including control groups with assessment only and no treatment, with wait-listing, and with a semi-placebo in the form of an alcohol-information brochure); (2) reported data that were usable for metaanalytic procedures; (3) assessed alcohol-drinking behavior (e.g., frequency or quantity) as a primary outcome measure; (4) applied individually focused personalized-feedback interventions; and (5) delivered the interventions without therapeutic support, with a maximum duration of 15 minutes per participant (Table 138,54 – 64). The assessment of studies for inclusion in the review was undertaken by two independent raters. Preselection from the initial search was based on information derived from titles, abstracts, and key words; if they yielded insufficient information to assess the inclusion criteria, then the full paper was retrieved. All papers excluded at this stage were re-checked to ensure that all potentially relevant papers had been retrieved. All retrieved papers were assessed for inclusion using the above criteria (Table 138,54 – 64); any disagreement was resolved by discussion and consensus. Using SPSS version 15, Cohen’s ␬ was used to assess the agreement on inclusion between the two raters (␬⫽0.72, which reflects a substantial agreement). Methodologic Quality Assessment of Primary Studies At least 25 scales are available to assess the validity and quality of RCTs.43 As there is no evidence that the more elaborate scales give more reliable assessments of validity than simpler ones, an approach was used like the one suggested by Higgins and Green43 and like the ones applied in several reviews of brief interventions for problem drinking in primary care.6,12 This resulted in four basic criteria for assessing the validity and quality of the studies analyzed: (1) allocation to condition by an independent third party, (2) the adequacy of random-allocation concealment to respondents, (3) the blinding of assessors of outcomes, and (4) attrition in follow-up data. Meta-Analysis Effect sizes (d) were calculated by subtracting (at post-test) the average score of the control condition (Mc) from the average score of the experimental condition (Me) and dividing the result by the average of the standard deviations of the experimental and control conditions (SDec). An effect size of ⱕ0.15 can be regarded as small, 0.45 as moderate, and ⱖ0.90 as large.65 Effect-size calculations were restricted to instruments that explicitly measured alcohol consumption (Table 138,54 – 64). If a study used more than one alcohol measure, the mean of the effect sizes was calculated, giving each study (or contrast group) a single effect size. In one study38 where more than one experimental condition was compared to a control condition, the number of participants in the control condition was divided evenly over the experimental conditions so that each participant was used only once in the meta-analysis. The pooled mean effect sizes were calculated using both random- and fixed-effects models. A fixed-effects model assumes that all studies in the meta-analysis are considered to have been conducted under similar conditions with similar subjects. The only difference among studies is their power to detect the outcome of interest. In a random-effects model, March 2009 studies are regarded as having been drawn from a population of studies. Effect sizes may vary due to error across studies. This allows for more uncertainty in the meta-analytical data, does not make the (possibly too restrictive) assumption that all studies are exact replications, and generally produces wider CIs around the pooled estimates. By implication, the random-effects model is more conservative in flagging significant results. In the absence of heterogeneity (see below), the fixed- and the random-effects models produce the same results. In that case, the more simple fixed-effects model is usually preferred. As the analyzed studies used different measures (both continuous and dichotomous) to indicate effectiveness, some ORs had to be converted into effect sizes in terms of Cohen’s d. This was done using the formula provided by the Comprehensive Meta-Analysis program version 2.2.021. Next, the mean effect sizes were converted into the number needed to treat (NNT) and area under the curve (AUC).66 The NNT estimates how many people must receive the intervention to achieve a good clinical outcome in one person; hence, a smaller NNT is better than a large one. The AUC is a measure of an intervention’s effectiveness; a score ⬎0.50 indicates that its outcome is superior to that in the control condition, and a score ⬍0.50 indicates that it is inferior. This analysis also tested whether genuine differences underlay the results of the studies (heterogeneity) or whether variations in findings were attributable to chance alone (homogeneity).67 The Q statistic was calculated as an indicator of homogeneity. A significant Q rejects the null hypothesis of homogeneity and shows that the variability among effect sizes is greater than what would likely have resulted from sampling error alone in the primary studies. Additionally, the I2 statistic, an indicator of heterogeneity, was calculated; 0% indicates no observed heterogeneity, and larger values show increasing heterogeneity, with 25% regarded as low, 50% as moderate, and 75% as high.67 As heterogeneity was nonexistent in all analyses and the differences between the fixedand random-effects results were negligible, only the results from the fixed-effects model are reported here. Meta-regression analyses were performed to assess whether effect sizes decayed over time and whether multicomponent personalized feedback differed in impact in comparison to personalized normative feedback alone. To assess and adjust for any publication bias, a fail-safe analysis was conducted, a funnel plot was constructed, and Duval and Tweedie’s trimand-fill analysis was performed. The Comprehensive MetaAnalysis program version 2.2.021 was used for all such operations. Evidence Synthesis Description of the Primary Analyzed Studies The combined literature search generated 406 abstracts and yielded 14 studies (Boon B, Institute for Addiction Research, unpublished findings, 2006; Boon B, Institute for Addiction Research, unpublished findings, 2008)38,54 – 64 (15 comparisons) that met the inclusion criteria (Table 138,54 – 64). The analysis involved a total of 3682 participants (1904 in personalized-feedback conditions and 1778 in conAm J Prev Med 2009;36(3) 249 250 Table 1. Principal characteristics of RCTs on personalized feedback for problem drinking American Journal of Preventive Medicine, Volume 36, Number 3 Study C Target population Agostinelli (1995)54 U.S. CSt Boon B, Institute for Addiction Research, unpublished findings, 2006 Boon B, Institute for Addiction Research, unpublished findings, 2008 Collins (2002)55 NL Gpop aged ⱖ18 NL Gpop/men only aged ⱖ18 ⱖ21/14 SU m/f wk; ⱖ6/4 in row ⱖonce past mo Gpop/Internet/V 1. PF/N: web 2. Ctrl: PBA U.S. CA Doumas (2008)57 U.S. Juarez (2006)58 U.S. Kypri (2004)59 AUS CSt ⱖ5/4 SU in row m/f ⱖtwice past mo ⱖ5 drinks at least once per mo ⱖ5/4 SU in row m/f past 2 wks ⱖ5/4 SU in row m/f ⱖonce past 2 wks ⱖ8 AUDIT CC/V Cunningham (2002)56 CSt aged 17–20 Gpop aged ⱖ18 Empl aged 18–24 Cst Lewis (2007)38 U.S. 1st-yr CSt ⱖ5/4 SU in row m/f ⱖonce past mo Neighbors (2004)60 U.S. CSt Neighbors (2006)61 U.S. 1st-yr CSt Walters (2000)62 U.S. CSt ⱖ5/4 SU in row m/f ⱖonce past mo ⱖ5/4 SU m/f ⱖ1 drinking session past mo ⱖ40 SU standard drinks past mo 1. 2. 1. 2. 1. 2. 1. 2. 1. 2. 1. 2. 3. 1. 2. 1. 2. Walters (2007)63 U.S. 1st-yr CSt Wild (2007)64 CA Gpop aged ⱖ18 Follow-up LFU (%) Outcome instruments ITT/CO 1.5 mo 11.5 DDQ, BAC CO 9 mo 31.6 WR, QFV, RCQ CO 1. 230 2. 220 1. 1 mo 2. 6 mo 1. 10.5 2. 15 WR, QFV, RCQ ITT 1. 2. 1. 2. 1. 2. 1. 2. 1. 2. 1. 2. 3. 1. 2. 1. 2. 1. 1.5 mo 2. 6 mo 6 mo 1. 1 2. 28.5 21 DDQ, DNRF ITT ITT 1 mo 36.7 2 mo 27 AUDIT, WRC, PI QFV, WRC, DDQ DDQ, BAC CO 1. 6 wks 2. 6 mo 5 mo 1. 20.2 2. 10.4 14.7 AUDIT, QF CO DDQ, DNRF ITT 1. 3 mo 2. 6 mo 2 mo 1. 21 2. 18 13.5 ACI, DDQ, PEAK, RAPI DNRF, DDQ; RAPI ITT SIP, AUDIT, CHUG, AEFQ WRC, DDQ, RAPI AUDIT Diagnosis Recruitment Conditions N ⬍40 oz ethanol past mo ⱖ21/14 SU m/f wk; ⱖ6/4 in row ⱖonce past mo CC/V 1. 2. 1. 2. 1. 2. 1. 2. ⱖ5/4 SU m/f ⱖ1 occ. past mo ⱖ8/6 m/f AUDIT Gpop/Internet/V Gpop/V Workplace/V CC/V Student health service/V CC/V CC/V CC/V PF/N: mail Ctrl: WLC PF/N: web Ctrl: PBA PF/N: mail Ctrl: PBA PF/N: mail Ctrl: AO PF/N: web/IS Ctrl: AO PF/N: mail Ctrl: AO PF/N: web/IS Ctrl: PBA PNF: web/IS PNF/gend: T Ctrl: AO PNF: web/IS Ctrl: AO PNF: web/IS Ctrl: AO 13 13 102 89 47 48 21 26 38 23 20 21 51 53 82 75 88 126 126 108 106 CC/V 1. PF/N: mail 2. Ctrl: AO 1. 11 2. 12 1.5 mo 14 CC/V 1. 2. 1. 2. 1. 2. 1. 2. 1. 2 mo 2. 4 mo 6 mo 1. 28.3 2. 22.6 24.4 Gpop/Com/V PF/N: web Ctrl: WLC PNF: mail Ctrl: WLC 103 103 877 850 CO CO CO ITT CO www.ajpm-online.net Note: For the abbreviations of the measurement instruments, the reader is referred to the references of the included studies. ACI, alcohol consumption index; AEFQ, alcohol-effects questionnaire; AO, assessment only; AUDIT, alcohol use disorder identification test; AUS, Australia; BAC, blood alcohol concentration; C, country; CA, Canada; CC, college community; CHUG, check-up to go; CO, completers only; Com, community; CSt, college students; ctrl, control; DDQ, daily-drinking questionnaire; DNRF, drinking-norms rating form; Empl, employees; gend, gender; Gpop, general population; ip, in preparation; ITT, intention to treat; LFU, lost to follow-up; mail, PF delivered by mail; mo, month(s); m/f, males/females; NL, The Netherlands; occ., occasion; oz, ounce(s); PBA, psycho-educational alcohol information brochure; PEAK, peak quantity (highest number of drinks consumed on one occasion over the last month; PF, personal feedback; PF/N, personal feedback, including normative feedback; PI, problem index, a scale to measure alcohol-related problems; PNF, personalized normative feedback; QF, quantity and frequency; QFV, quantity, frequency, variability; RAPI, Rutgers alcohol-problem index; RCQ, readiness-to-change questionnaire; SIP, shortened inventory of problems; SU, standard unit of alcohol; T, tailormade; V, voluntarily; Web/IS, PF web-based but completed in situ (research lab, healthcare setting, or workplace); web, PF web-based delivery; wks, weeks; WLC, wait-list control; WR, weekly recall; WRC, the weekly recall method asks about actual alcohol consumption on the previous 7 days; yr, year trol conditions). The number of participants ranged from 11 to 877 per condition per comparison. All38,54–58,60–64 but one study59 used nonclinical samples from settings either in the community, in higher education, or at work. Nine studies38,54,55,58–63 recruited their participants, aged 17–24 years, from higher-education institutions. Four studies (Boon B, Institute for Addiction Research, unpublished findings, 2006; Boon B, Institute for Addiction Research, unpublished findings, 2008)56,64 recruited from the general adult population, and one study57 targeted employees in a work setting. Six studies54–56,58,62,64 delivered the personalized-feedback interventions by mail, and the other eight did so via the Internet (Boon B, Institute for Addiction Research, unpublished findings, 2006; Boon B, Institute for Addiction Research, unpublished findings, 2008).38,57,59 – 61,63 Five studies38,57,59 – 61 delivered the personalized-feedback interventions in situ (i.e., a research laboratory, health service clinic, or at work). Six54 –56,58,62,64 of the remaining nine studies delivered the intervention by mail; three (Boon B, Institute for Addiction Research, unpublished findings, 2006; Boon B, Institute for Addiction Research, unpublished findings, 2008)63 enabled participants to access the intervention via the Internet at their venue of preference. Eight studies38,55–58,60,61,63 used binge drinking as the primary inclusion criterion; two (Boon B, Institute for Addiction Research, unpublished findings, 2006; Boon B, Institute for Addiction Research, unpublished findings, 2008) used drinking in excess of a low-risk drinking guideline (including binge drinking; two59,64 used the AUDIT68 screening test (with a score of ⱖ8 indicating problem drinking); and two studies used the amount of alcohol intake (⬎40 oz. ethanol in the past month54 and ⱖ40 standard drinks in the past month62). There were different types of control conditions: seven studies38,56 –58,60 – 62 used an assessment-only format, three54,63,64 used a wait-list condition, and four (Boon B, Institute for Addiction Research, unpublished findings, 2006; Boon B, Institute for Addiction Research, unpublished findings, 2008)55,59 gave control participants a short psychoeducational leaflet on alcohol use. Six of the studies (Boon B, Institute for Addiction Research, unpublished findings, 2008)38,55,56,60,63 were based on intention-to-treat analysis, and eight (Boon B, Institute for Addiction Research, unpublished findings, 2006)54,57–59,61,62,64 on completers-only analysis. The studies were conducted in various Western countries. The quality of the studies varied. All used randomized– controlled designs, well-validated alcohol-consumption measures, and well-described, theoretically based interventions. Only three studies, however (Boon B, Institute for Addiction Research, unpublished findings, 2008)59,60—reported the independent allocation of participants, the concealment of random allocation to participants, and the blinding of assessors; such conditions were not possible in all studies. Loss to follow-up ranged from 1% to 37%. Effects of Personalized Feedback on Alcohol Consumption at Follow-Up Fourteen studies (Boon B, Institute for Addiction Research, unpublished findings, 2006; Boon B, Institute for Addiction Research, unpublished findings, 2008)38,54–64 with 15 comparison groups assessed the effects of personalized feedback on alcohol use at post-intervention (Figure 2). The overall mean effect size was 0.22 (95% CI⫽0.16, 0.29) in the fixed model. Outliers were not excluded, as there was no heterogeneity among the studies (Q⫽10.962, p⫽0.69, I2⫽0). These results correspond to an NNT of 8.06, indicating that about eight people need to be recipients of the intervention in order to generate one good clinical outcome (AUC⫽0.562). Sensitivity Analyses Figure 2. Pooled estimates for the effectiveness of personalized-feedback intervention a Boon B, Institute for Addiction Research, unpublished findings, 2006 b Boon B, Institute for Addiction Research, unpublished findings, 2008 March 2009 The overall mean effect size was maintained even when the largest study (n⫽172764) was excluded. Without this study, the overall effect size rose from d⫽0.22 (95% CI⫽0.16, 0.29) to d⫽0.28 (95% CI⫽ 0.19, 0.37), which is not significant, as evidenced by the overlapping CIs. Separate analyses were conducted to correct for small-sample bias; however, results showed an identical overall effect size and corresponding CIs, as Am J Prev Med 2009;36(3) 251 reported for the pooled standardized difference in means. Meta-regression analyses did not establish significant differences in the effects of personalizedfeedback interventions over time (␤⫽ – 0.006, 95% CI⫽ – 0.014, 0.0015, p⫽0.12), nor could a significant difference be established for personalized-feedback interventions inclusive of normative feedback and personalized-feedback interventions based solely on this normative feedback (␤⫽0.09, CI⫽ – 0.05, 0.24, p⫽0.22). Duval and Tweedie’s trim-and-fill analysis did not detect any publication bias (observed d⫽0.22, 95% CI⫽0.16, 0.29; adjusted d⫽0.22, 95% CI⫽0.15, 0.29); neither did the funnel plot analysis. The fail-safe analysis indicated that 122 studies with null effects have to be added to the database before the pooled effect size would be no longer significant (p⬎0.05). In view of these results, we believe that our results are robust.69,70 Discussion Main Findings This meta-analysis shows that single-session, individually personalized feedback without professional guidance can be an effective intervention for reducing risky alcohol consumption in young and adult problem drinkers. Adverse consequences in terms of increased alcohol use among participants resulting from their exposure to personalized-feedback interventions were not identified.12 These results may indicate that personalized feedback is an effective intervention for different target groups across different settings, using a variety of delivery modes. Despite the modest effect sizes overall, personalized feedback could have a major health impact at the population level, in view of the high percentage of problem drinkers who potentially could benefit.49 The effect sizes reported here are comparable to those from several other meta-analyses of brief interventions to curb problem drinking. Those were in the small-to-moderate range, both for college students22 and the adult population in both nonclinical and primary care samples.6,17,19,20,71 The NNT of 8.06 found in this meta-analysis for the overall effect of personalized feedback seems appreciable, given the brief and unguided nature of the intervention. It is in the range of those NNTs (i.e., from 772 to 823) reported for brief, face-to-face alcohol-reduction advice in primary care. The meta-analysis by Carey et al.22 indicated that individual brief interventions are more effective than group interventions for college drinkers (with d ranging from 0.11 to 0.41 on several alcohol-related outcome measures). This study also showed that brief interventions based on motivational interviewing and normative feedback are more effective than those that 252 do not include these features. In a review dedicated to personalized-feedback interventions for students, White25 found more favorable results for written and computer-based personalized-feedback interventions than for face-to-face individual or group interventions; Walters and Neighbors31 found similar results. The effectiveness of personalized feedback may therefore depend not on personal contact but on the content of the feedback, such as normative feedback and the mode of mail and web-based delivery.24,25 For example, this meta-analysis showed that personalized normative feedback was as effective as multi-component personalized feedback. Further research into the effective components is, however, required to evaluate the robustness of this observation. Limitations Several limitations need to be taken into account when interpreting the results of this study. First, this meta-analysis is based on 14 primary studies; the findings can be generalized only to the groups studied,73 who were at-risk drinkers in student and general populations. Second, some studies had methodologic drawbacks such as small samples54,55,62 or dropout rates above 30% (Table 138,54–64). Third, all studies relied on self-reported alcohol-consumption measures. Although there is some concern about the reliability and validity of such measures,38,74 they are currently the best option available.12,75 Indeed, their validity actually improves in interventions delivered online, which facilitate self-disclosure in comparison to pen-and-paper questionnaires.38,42,76 Conclusion Personalized feedback without therapeutic guidance may be cost effective in view of the minimal time and financial investments needed to make it widely available. It is expected that this potential cost effectiveness, the attractiveness of personalized feedback for participants,77 and the diffusion potential could be greatly expanded by delivering it over the Internet. Many of the traditional impediments to implementing brief interventions could be overcome thereby.27,57 The advantages of web-based delivery include the widespread availability of personalized feedback to underserved or difficult-to-reach groups such as college students, female problem drinkers,22,40 and those in geographically dispersed areas,27,42 many of whom now have Internet access.78 Brief web-based personalizedfeedback interventions appear to be more readily accepted by both young and mature risky drinkers, as the unobtrusive nature of the intervention allays fears of stigmatization and violation of privacy.79,80 The constant availability of these web-based interventions makes it more convenient to take part.40 Information and Communication Technologies (ICT) also now facilitates personal- American Journal of Preventive Medicine, Volume 36, Number 3 www.ajpm-online.net needs assessments, including those necessary for public health interventions.81 A further advantage is that the Internet—in particular, the advent of Web 2.0 technologies82—facilitates the gathering of knowledge about targeted groups as well as their active participation in the interventions. The use of ICT devices in group settings is yet another promising avenue, as shown by the recent RCT of LaBrie et al.83 among college students. This study evaluated the effectiveness of a professionally guided, interactive, group-specific, personalized, normative-feedback intervention by means of personal digital handhelds. Results demonstrated the effectiveness of this intervention on reducing alcohol consumption at 2-month follow-up. All such features of web-based personalized feedback could increase the effectiveness of intervening, and could make it possible—as Neighbors et al.61 have pointed out—to reach out with preventive interventions to people in different settings and on a large scale. In a public health approach to problem drinking, it would therefore seem beneficial to integrate personalized feedback into the first stage of a stepped-care model for problem drinking. A stepped-care approach is based on the rationale that problem drinkers first receive a brief, low-threshold intervention that has a reasonable chance of success. For those for whom personalized feedback, for example, does not work, the level of treatment can be stepped up in terms of more-intensive interventions with increasing levels of therapeutic involvement and related cost. It therefore is recommended that personalized feedback be further investigated. However, many of the expected advantages still lack an empirical basis, while potential disadvantages have not yet been fully investigated. Alternative strategies to reach out to high-risk drinkers are also required, because with effect sizes in the small-tomedium range, not all high-risk drinkers benefit from personalized feedback. In addition, research into the cost effectiveness of personalized-feedback interventions is required. While it is expected that personalized feedback can be delivered at low cost, the empirical evidence for this is not yet available. Cost-effectiveness studies should also include evaluations of effective recruitment strategies for single-session personalized-feedback interventions, as the latter could involve higher costs than the actual personalized-feedback intervention itself. Future studies should focus on factors that influence the long-term effectiveness of personalized feedback, as well as examining the groups (e.g., youth obliged to use judicial service programs because of violations of minimum-age drinking laws and adults obliged to use these programs because of convictions for alcohol-impaired driving) for which it might have greater or lesser effectiveness. Research should also explore its applicability to other settings, such as primary care. As is the case in college settings, a whole range of barriers exist to implementing interventions in primary care. These involve motiMarch 2009 vating and training general practitioners to use the intervention, a lack of pragmatic screening interventions,26,27,44,45 and the costs of implementation.21,84 Future research should investigate the role that single-session personalized feedback could play in primary care with and without the involvement of the general practitioner. As the effects of personalized feedback appear comparable to those of more intensive (and hence costly and intrusive) brief interventions, it could be of interest to a range of stakeholders, including university officials, public health planners, insurance companies, and employers. In addition, it is worthwhile to further investigate the potential applicability of a single session for altering lifestyle behaviors such as overeating or common mental health disorders such as depression. We thank Karin Mutsaers; this study would not have been possible without her contribution. We are grateful to Michael Dallas for English-language edits. No financial disclosures were reported by the authors of this paper. References 1. Murray CJ, Lopez AD. The global burden of disease: a comprehensive assessment of mortality and disability from diseases, injuries, and risk factors in 1990 and projected to 2020. Cambridge MA: Harvard School of Public Health, 1996. 2. WHO. WHO expert committee on problems related to alcohol consumption. Second report. World Health Organ Tech Rep Ser 2007;1–7. 3. Knight JR, Wechsler H, Kuo M, Seibring M, Weitzman ER, Schuckit MA. Alcohol abuse and dependence among U.S. college students. J Stud Alcohol 2002;63:263–70. 4. Wechsler H, Lee JE, Kuo M, Seibring M, Nelson TF, Lee H. Trends in college binge drinking during a period of increased prevention efforts. Findings from 4 Harvard School of Public Health college alcohol study surveys: 1993–2001. J Am Coll Health 2002;50:203–17. 5. Heather N, Kaner E. Brief interventions against excessive alcohol consumption. In: Warell DA, Cox TM, Firth JD, et al, eds. Oxford textbook of medicine. Oxford: Oxford University Press, 2003. 6. Kaner EF, Beyer F, Dickinson HO, et al. Effectiveness of brief alcohol interventions in primary care populations. Cochrane Database Syst Rev 2007;CD004148. 7. White HR, Jackson KM. Social and psychological influences on emerging adult drinking. Alcohol Res Health 2004;28:182–90. 8. Anderson P, Baumberg B. Alcohol in Europe. A public health perspective. London: Institute of Alcohol Studies, 2006. 9. Smit F, Cuijpers P, Oostenbrink J, Batelaan N, de Graaf R, Beekman A. Costs of nine common mental disorders: implications for curative and preventive psychiatry. J Ment Health Policy Econ 2006;9:193–200. 10. Holman CD, English DR, Milne E, Winter MG. Meta-analysis of alcohol and all-cause mortality: a validation of NHMRC recommendations. Med J Aust 1996;164:141–5. 11. Hingson R, Heeren T, Winter M, Wechsler H. Magnitude of alcohol-related mortality and morbidity among U.S. college students ages 18 –24: changes from 1998 to 2001. Annu Rev Public Health 2005;26:259 –79. 12. Whitlock EP, Polen MR, Green CA, Orleans T, Klein J. 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. 13. Freemantle N, Gill P, Godfrey C, et al. Brief interventions and alcohol use. Qual Health Care 1993;2:267–73. 14. IOM. Broadening the base of treatment of alcohol problems: report of a study by a committee of the Institute of Medicine. Washington DC: National Academy Press, 1990. Am J Prev Med 2009;36(3) 253 15. USDHHS. The Surgeon General’s call to action to prevent and reduce underage drinking. Washington DC: Office of the Surgeon General, 2007. 16. 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. 17. Burke BL, Arkowitz H, Menchola M. The efficacy of motivational interviewing: a meta-analysis of controlled clinical trials. J Consult Clin Psychol 2003;71:843– 61. 18. Cuijpers P, Riper H, Lemmers L. The effects on mortality of brief interventions for problem drinking: a meta-analysis. Addiction 2004;99:839 – 45. 19. Apodaca TR, Miller WR. A meta-analysis of the effectiveness of bibliotherapy for alcohol problems. J Clin Psychol 2003;59:289 –304. 20. Moyer A, Finney JW, Swearingen CE, Vergun P. Brief interventions for alcohol problems: a meta-analytic review of controlled investigations in treatmentseeking and non-treatment-seeking populations. Addiction 2002;97:279 –92. 21. Raistrick D, Heather N, Godfrey C. Review of the effectiveness of treatment for alcohol problems. London: National Treatment Agency for Substance Misuse, 2006. 22. Carey KB, Scott-Sheldon LA, Carey MP, DeMartini KS. Individual-level interventions to reduce college student drinking: a meta-analytic review. Addict Behav 2007;32:2469 –94. 23. Ehrlich PF, Haque A, Swisher-McClure S, Helmkamp J. Screening and brief intervention for alcohol problems in a university student health clinic. J Am Coll Health 2006;54:279 – 87. 24. Larimer ME, Cronce JM. Identification, prevention, and treatment revisited: individual-focused college drinking prevention strategies 1999 –2006. Addict Behav 2007;32:2439 – 68. 25. White HR. Reduction of alcohol-related harm on U.S. college campuses: the use of personal feedback interventions. Int J Drug Policy 2006;17:310 –9. 26. Kaner EF, Lock CA, McAvoy BR, Heather N, Gilvarry E. A RCT of three training and support strategies to encourage implementation of screening and brief alcohol intervention by general practitioners. Br J Gen Pract 1999;49:699 –703. 27. Moyer A, Finney JW. Brief interventions for alcohol problems: factors that facilitate implementation. Alcohol Res Health 2005;28:44 –50. 28. Anderson P. The risk of alcohol [dissertation]. Nijmegen, The Netherlands: Radboud University, 2003. 29. Andrews G, Issakidis C, Sanderson K, Corry J, Lapsley H. Utilising survey data to inform public policy: comparison of the cost-effectiveness of treatment of ten mental disorders. Br J Psychiatry 2004;184:526 –33. 30. Cunningham JA, Breslin FC. Only one in three people with alcohol abuse or dependence ever seek treatment. Addict Behav 2004;29:221–3. 31. Walters ST, Neighbors C. Feedback interventions for college alcohol misuse: what, why and for whom? Addict Behav 2005;30:1168 – 82. 32. Chan KK, Neighbors C, Gilson M, Larimer ME, Alan MG. Epidemiological trends in drinking by age and gender: providing normative feedback to adults. Addict Behav 2007;32:967–76. 33. Larimer ME, Turner AP, Anderson BK, et al. Evaluating a brief alcohol intervention with fraternities. J Stud Alcohol 2001;62:370 – 80. 34. Marlatt GA, Baer JS, Kivlahan DR, et al. Screening and brief intervention for high-risk college student drinkers: results from a 2-year follow-up assessment. J Consult Clin Psychol 1998;66:604 –15. 35. Borsari B, Carey KB. Descriptive and injunctive norms in college drinking: a meta-analytic integration. J Stud Alcohol 2003;64:331– 41. 36. Borsari B, Carey KB. Peer influences on college drinking: a review of the research. J Subst Abuse 2001;13:391– 424. 37. Miller WR, Toscova RT, Miller JH, Sanchez V. A theory-based motivational approach for reducing alcohol/drug problems in college. Health Educ Behav 2000;27:744 –59. 38. Lewis MA, Neighbors C, Oster-Aaland L, Kirkeby BS, Larimer ME. Indicated prevention for incoming freshmen: personalized normative feedback and high-risk drinking. Addict Behav 2007;32:2495–508. 39. Miller WR, Rollnick S. Motivational interviewing: preparing people for change. New York: Guilford Press, 2002. 40. Cunningham JA, Humphreys K, Koski-Jannes A, Cordingley J. Internet and paper self-help materials for problem drinking: is there an additive effect? Addict Behav 2005;30:1517–23. 41. Cunningham JA, Wild TC, Bondy SJ, Lin E. Impact of normative feedback on problem drinkers: a small-area population study. J Stud Alcohol 2001;62:228 –33. 42. Chiauzzi E, Green TC, 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. 43. Cochrane Collaboration. Cochrane handbook for systematic reviews of interventions. Oxford: Cochrane Collaboration, 2008. 254 44. Beich A, Thorsen T, Rollnick S. Screening in brief intervention trials targeting excessive drinkers in general practice: systematic review and meta-analysis. BMJ 2003;327:536 – 42. 45. Bertholet N, Daeppen JB, 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. 46. Bien TH, Miller WR, Tonigan JS. Brief interventions for alcohol problems: a review. Addiction 1993;88:315–35. 47. Emmen MJ, Schippers GM, Bleijenberg G, Wollersheim H. Effectiveness of opportunistic brief interventions for problem drinking in a general hospital setting: systematic review. BMJ 2004;328:318. 48. Gould RA, Clum GA. A meta-analysis of self-help treatment approaches. Clin Psychol Rev 1993;13:169 – 86. 49. Kahan M, Wilson L, Becker L. Effectiveness of physician-based interventions with problem drinkers: a review. CMAJ 1995;152:851–9. 50. Mains JA, Scogin FR. The effectiveness of self-administered treatments: a practice-friendly review of the research. J Clin Psychol 2003;59:237– 46. 51. Miller WR, Wilbourne PL. Mesa Grande: a methodological analysis of clinical trials of treatments for alcohol use disorders. Addiction 2002;97:265–77. 52. Poikolainen K. Effectiveness of brief interventions to reduce alcohol intake in primary health care populations: a meta-analysis. Prev Med 1999;28:503–9. 53. Wilk AI, Jensen NM, Havighurst TC. Meta-analysis of randomized control trials addressing brief interventions in heavy alcohol drinkers. J Gen Intern Med 1997;12:274 – 83. 54. Agostinelli G, Brown JM, Miller WR. Effects of normative feedback on consumption among heavy drinking college students. J Drug Educ 1995;25:31– 40. 55. Collins SE, Carey KB, Sliwinski MJ. Mailed personalized normative feedback as a brief intervention for at-risk college drinkers. J Stud Alcohol 2002;63:559 – 67. 56. Cunningham JA, Koski-Jannes A, Wild TC, Cordingley J. Treating alcohol problems with self-help materials: a population study. J Stud Alcohol 2002;63:649 –54. 57. Doumas DM, 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. 58. Juarez P, Walters ST, Daugherty M, Radi C. A randomized trial of motivational interviewing and feedback with heavy drinking college students. J Drug Educ 2006;36:233– 46. 59. Kypri K, Saunders JB, Williams SM, et al. Web-based screening and brief intervention for hazardous drinking: a double-blind randomized controlled trial. Addiction 2004;99:1410 –7. 62. Neighbors C, Larimer ME, Lewis MA. Targeting misperceptions of descriptive drinking norms: efficacy of a computer-delivered personalized normative feedback intervention. J Consult Clin Psychol 2004;72:434 – 47. 61. Neighbors C, Lewis MA, Bergstrom RL, Larimer ME. Being controlled by normative influences: self-determination as a moderator of a normative feedback alcohol intervention. Health Psychol 2006;25:571–9. 62. Walters ST, Bennett ME, Miller JH. Reducing alcohol use in college students: a controlled trial of two brief interventions. J Drug Educ 2000;30:361–72. 63. Walters ST, Vader AM, Harris TR. A controlled trial of web-based feedback for heavy drinking college students. Prev Sci 2007;8:83– 8. 64. Wild TC, Cunningham JA, Roberts AB. Controlled study of brief personalized assessment-feedback for drinkers interested in self-help. Addiction 2007;102:241–50. 65. Lipsey MW, Wilson DB. Practical meta-analysis. Thousand Oaks CA: Sage Publications, 2001. 66. Kraemer HC, Morgan GA, Leech NL, Gliner JA, Vaske JJ, Harmon RJ. Measures of clinical significance. J Am Acad Child Adolesc Psychiatry 2003;42:1524 –9. 67. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327:557– 60. 68. Saunders JB, Aasland OG, Babor TF, Defuentes-Merillas L, 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. 69. Hedges LV, Waddington T. From evidence to knowledge to policy: research synthesis for policy formation. Rev Educ Res 1993;63:345–52. 70. Wilson DB. Meta-analyses in alcohol and other drug abuse treatment research. Addiction 2000;95(3S):419 –38. 71. Vasilaki EI, Hosier SG, Cox WM. The efficacy of motivational interviewing as a brief intervention for excessive drinking: a meta-analytic review. Alcohol 2006;41:328 –35. American Journal of Preventive Medicine, Volume 36, Number 3 www.ajpm-online.net 72. National Health Committee. Guidelines for recognising, assessing and treating alcohol and cannabis abuse in primary care. Wellington, New Zealand: National Health Committee, 1999. 73. Heather N. Interpreting the evidence on brief interventions for excessive drinkers: the need for caution. Alcohol 1995;30:287–96. 74. Hustad JT, Carey KB. Using calculations to estimate blood alcohol concentrations for naturally occurring drinking episodes: a validity study. J Stud Alcohol 2005;66:130 – 8. 75. Laforge RG, Borsari B, Baer JS. The utility of collateral informant assessment in college alcohol research: results from a longitudinal prevention trial. J Stud Alcohol 2005;66:479 – 87. 76. Kypri K, Gallagher SJ, Cashell-Smith ML. An Internet-based survey method for college student drinking research. Drug Alcohol Depend 2004;76:45–53. 77. Kypri K, Saunders JB, Gallagher SJ. Acceptability of various brief intervention approaches for hazardous drinking among university students. Alcohol 2003;38:626 – 8. 78. Internet World Stats. World Internet users and population stats. 2008. www.internetworldstats.com./stats.htm. 79. Curry SJ. eHealth research and healthcare delivery beyond intervention effectiveness. Am J Prev Med 2007;32:S127–S130. 80. Simon-Arndt CM, Hurtado SL, Patriarca-Troyk LA. Acceptance of webbased personalized feedback: user ratings of an alcohol misuse prevention program targeting U.S. Marines. Health Commun 2006;20:13–22. 81. Copeland J, Martin G. Web-based interventions for substance use disorders: a qualitative review. J Subst Abuse Treat 2004;26:109 –16. 82. Musser J, O’Reilly T, the O’Reilly Radar Team. Web 2.0 principles and best practices: an O’Reilly radar report. 2006. radar.oreilly.com/research/web2report.html. 83. LaBrie JW, Hummer JF, Neighbors C, Pedersen ER. Live interactive group-specific normative feedback reduces misperceptions and drinking in college students: a randomized cluster trial. Psychol Addict Behav 2008;22:141– 8. 84. Boyd GM, Faden V. Overview. J Stud Alcohol 2002;14:6 –14. Did you know? You can personalize the American Journal of Preventive Medicine website to meet your individual needs. Visit www.ajpm-online.net today to see what else is new online! March 2009 Am J Prev Med 2009;36(3) 255