Why is this review important?
Harmful alcohol use is one of the main contributors to the global burden of disease. In low- and middle-income countries, harmful alcohol use is increasing. However, services to prevent and treat harmful alcohol use are limited. One contributor to the lack of available services is the limited information about which intervention approaches are effective in reducing harmful alcohol use and whether these approaches are feasible and acceptable in low-resource settings. In order to prevent the physical, psychological, and societal burden of harmful alcohol use, it is important that effective interventions are available to reduce alcohol-related harm.
What is the goal of this review?
The goal of this review is to summarize evidence on whether psychosocial and pharmacologic interventions are able to reduce harmful alcohol use in low- and middle-income countries. We also aim to evaluate the safety of treatments and how many people remain in treatment until its completion.
What does the research say?
We identified 66 randomized controlled trials that evaluated the effect of interventions on reducing harmful alcohol use. Most of these studies assessed psychosocial interventions (n = 52 studies), six assessed pharmacologic interventions alone, and eight assessed combined pharmacologic and psychosocial interventions.
The majority of included trials were funded by government agencies (36 trials) followed by multiple public and private funders (8 trials) or private foundations (5 trials). Seventeen trials did not report the funding source.
We are uncertain whether brief psychosocial interventions and other psychosocial interventions reduce harmful alcohol use. Combined pharmacologic and psychosocial interventions may reduce harmful alcohol use relative to psychosocial interventions combined with placebo, yet the certainty of the evidence has been assessed as low. No studies were found that looked at the effect of pharmacologic interventions alone on harmful alcohol use. We did not find evidence that rates of retention differed between study conditions for any of the intervention types.
Certainty of evidence: The certainty of evidence on the effect of brief and other psychosocial interventions on harmful alcohol use was very low due to lack of masking, differential attrition and missing data, selective outcome reporting, high heterogeneity, and differences in patient populations. The certainty of evidence on the effect of combined pharmacologic and psychosocial interventions on harmful alcohol use was low due to lack of blinding, incomplete outcome data, and heterogeneity in the pharmacologic interventions under investigation. No studies evaluated the effect of pharmacologic interventions alone on harmful alcohol use.
The evidence is current to 12 December 2021.
What are the next steps?
Future research on interventions to reduce harmful alcohol use is needed to address some of the limitations described in this review. Studies evaluating similar interventions across studies and contexts are needed to improve generalizability and understand the necessary characteristics of and conditions for effective intervention. Studies must utilize measurement tools that are valid in the population they are studying. Lastly, studies must be designed to reduce the risk of bias leading to the high degree of uncertainty in the results reported in this review.
In LMICs there is low-certainty evidence supporting the efficacy of combined psychosocial and pharmacologic interventions on reducing harmful alcohol use relative to psychosocial interventions alone. There is insufficient evidence to determine the efficacy of pharmacologic or psychosocial interventions on reducing harmful alcohol use largely due to the substantial heterogeneity in outcomes, comparisons, and interventions that precluded pooling of these data in meta-analyses. The majority of studies are brief interventions, primarily among men, and using measures that have not been validated in the target population. Confidence in these results is reduced by the risk of bias and significant heterogeneity among studies as well as the heterogeneity of results on different outcome measures within studies. More evidence on the efficacy of pharmacologic interventions, specific types of psychosocial interventions are needed to increase the certainty of these results.
Harmful alcohol use is defined as unhealthy alcohol use that results in adverse physical, psychological, social, or societal consequences and is among the leading risk factors for disease, disability and premature mortality globally. The burden of harmful alcohol use is increasing in low- and middle-income countries (LMICs) and there remains a large unmet need for indicated prevention and treatment interventions to reduce harmful alcohol use in these settings. Evidence regarding which interventions are effective and feasible for addressing harmful and other patterns of unhealthy alcohol use in LMICs is limited, which contributes to this gap in services.
To assess the efficacy and safety of psychosocial and pharmacologic treatment and indicated prevention interventions compared with control conditions (wait list, placebo, no treatment, standard care, or active control condition) aimed at reducing harmful alcohol use in LMICs.
We searched for randomized controlled trials (RCTs) indexed in the Cochrane Drugs and Alcohol Group (CDAG) Specialized Register, the Cochrane Clinical Register of Controlled Trials (CENTRAL) in the Cochrane Library, PubMed, Embase, PsycINFO, CINAHL, and the Latin American and Caribbean Health Sciences Literature (LILACS) through 12 December 2021. We searched clinicaltrials.gov, the World Health Organization International Clinical Trials Registry Platform, Web of Science, and Opengrey database to identify unpublished or ongoing studies. We searched the reference lists of included studies and relevant review articles for eligible studies.
All RCTs comparing an indicated prevention or treatment intervention (pharmacologic or psychosocial) versus a control condition for people with harmful alcohol use in LMICs were included.
We used standard methodological procedures expected by Cochrane.
We included 66 RCTs with 17,626 participants. Sixty-two of these trials contributed to the meta-analysis. Sixty-three studies were conducted in middle-income countries (MICs), and the remaining three studies were conducted in low-income countries (LICs). Twenty-five trials exclusively enrolled participants with alcohol use disorder. The remaining 51 trials enrolled participants with harmful alcohol use, some of which included both cases of alcohol use disorder and people reporting hazardous alcohol use patterns that did not meet criteria for disorder.
Fifty-two RCTs assessed the efficacy of psychosocial interventions; 27 were brief interventions primarily based on motivational interviewing and were compared to brief advice, information, or assessment only. We are uncertain whether a reduction in harmful alcohol use is attributable to brief interventions given the high levels of heterogeneity among included studies (Studies reporting continuous outcomes: Tau² = 0.15, Q =139.64, df =16, P<.001, I² = 89%, 3913 participants, 17 trials, very low certainty; Studies reporting dichotomous outcomes: Tau²=0.18, Q=58.26, df=3, P<.001, I² =95%, 1349 participants, 4 trials, very low certainty). The other types of psychosocial interventions included a range of therapeutic approaches such as behavioral risk reduction, cognitive-behavioral therapy, contingency management, rational emotive therapy, and relapse prevention. These interventions were most commonly compared to usual care involving varying combinations of psychoeducation, counseling, and pharmacotherapy. We are uncertain whether a reduction in harmful alcohol use is attributable to psychosocial treatments due to high levels of heterogeneity among included studies (Heterogeneity: Tau² = 1.15; Q = 444.32, df = 11, P<.001; I²=98%, 2106 participants, 12 trials, very low certainty).
Eight trials compared combined pharmacologic and psychosocial interventions with placebo, psychosocial intervention alone, or another pharmacologic treatment. The active pharmacologic study conditions included disulfiram, naltrexone, ondansetron, or topiramate. The psychosocial components of these interventions included counseling, encouragement to attend Alcoholics Anonymous, motivational interviewing, brief cognitive-behavioral therapy, or other psychotherapy (not specified). Analysis of studies comparing a combined pharmacologic and psychosocial intervention to psychosocial intervention alone found that the combined approach may be associated with a greater reduction in harmful alcohol use (standardized mean difference (standardized mean difference (SMD))=-0.43, 95% confidence interval (CI): -0.61 to -0.24; 475 participants; 4 trials; low certainty).
Four trials compared pharmacologic intervention alone with placebo and three with another pharmacotherapy. Drugs assessed were: acamprosate, amitriptyline, baclofen disulfiram, gabapentin, mirtazapine, and naltrexone. None of these trials evaluated the primary clinical outcome of interest, harmful alcohol use.
Thirty-one trials reported rates of retention in the intervention. Meta-analyses revealed that rates of retention between study conditions did not differ in any of the comparisons (pharmacologic risk ratio (RR) = 1.13, 95% CI: 0.89 to 1.44, 247 participants, 3 trials, low certainty; pharmacologic in addition to psychosocial intervention: RR = 1.15, 95% CI: 0.95 to 1.40, 363 participants, 3 trials, moderate certainty). Due to high levels of heterogeneity, we did not calculate pooled estimates comparing retention in brief (Heterogeneity: Tau² = 0.00; Q = 172.59, df = 11, P<.001; I2 = 94%; 5380 participants; 12 trials, very low certainty) or other psychosocial interventions (Heterogeneity: Tau² = 0.01; Q = 34.07, df = 8, P<.001; I2 = 77%; 1664 participants; 9 trials, very low certainty). Two pharmacologic trials and three combined pharmacologic and psychosocial trials reported on side effects. These studies found more side effects attributable to amitriptyline relative to mirtazapine, naltrexone and topiramate relative to placebo, yet no differences in side effects between placebo and either acamprosate or ondansetron.
Across all intervention types there was substantial risk of bias. Primary threats to validity included lack of blinding and differential/high rates of attrition.