Introduction

Alcohol is a widely used psychoactive substance, with 47% of individuals over the age of 15 reporting alcohol use in the past year, globally [1]. Alcohol consumption is associated with a myriad of poor health outcomes for the individual consuming alcohol, as well as for others around them (e.g. driving while intoxicated, violence while under the influence). In 2016, the Global Burden of Disease Collaboration identified alcohol as the seventh leading risk factor for death, globally [2]. The Global Burden of Disease 2017 study further found that among young persons aged 15–49 years, alcohol use is the leading risk factor for premature death and burden of disease, which has been the case since 1990 [3]. More recently, a comparative risk assessment has affirmed alcohol use continues to be a leading risk factor for death [4]. The impact of alcohol consumption on health outcomes is often dependent on two factors: the total volume of alcohol consumed and the pattern of alcohol use [5]. Harmful alcohol use is defined by the World Health Organization (WHO) as “drinking that causes detrimental health and social consequences for the drinker, the people around the drinker and society at large, as well as patterns of drinking that are associated with increased risk of adverse health outcomes” [6]. Harmful alcohol use is associated with increased risk of morbidity and mortality and has been causally linked to hundreds of diseases and injuries [2, 5, 7]. Globally, patterns of alcohol use, the volume of alcohol per capita consumed (APC), and the subsequent health and social consequences vary greatly by region and country.

The WHO Africa Region has an APC of 6.3 liters per person, which is comparable to the global average of 6.4 liters per person. However, this is attributed to a large proportion of the population abstaining from drinking altogether, especially women [5]. Among men and women (15 years of age or older) who do use alcohol, total APC is high at 18.4 liters per person (compared to 15.1 liters among alcohol users, globally), suggesting that those who drink alcohol in Africa consume more than drinkers in other parts of the world [5]. One pattern of harmful alcohol use that is particularly hazardous to health is heavy episodic drinking (HED), also known as binge drinking, although the two have slightly different definitions.

The WHO defines HED as consuming at least 60 grams of pure alcohol on at least one occasion in the past 30 days [5]. The U.S. National Institute on Alcohol Abuse and Alcoholism (NIAAA) defines binge drinking as drinking until one’s blood alcohol level exceeds 0.08 g/dL, which is typically achieved by four drinks within two hours for women or five drinks within two hours for men (based on the United States’ definition of a standard drink, i.e., 14 grams of pure alcohol) [8]. HED is prevalent in sub-Saharan Africa with over half of drinkers (50.2%) in the WHO Africa Region engaging in this behavior (compared to 39.5% of drinkers globally) [5]. While there has been a slight decline in the prevalence of HED among drinkers in the Africa Region since 2000 (from 55.5% to 50.2%), sub-Saharan Africa continues to experience the highest prevalence of HED among alcohol users globally [5]. In the Africa Region, HED is most prevalent among persons aged 20 to 24 years (50.7% among drinkers) and men are more than twice as likely to engage in HED than women (60.5% of drinkers compared to 28.2%) [5]. Prevalence of HED varies substantially throughout sub-Saharan Africa. In the Republic of Congo, Gabon and Equatorial Guinea, prevalence among drinkers exceeds 80%, while prevalence in Niger, Senegal, Chad and Guinea hovers just below 35% [5].

HED has been linked to increased risk of injury and cardiovascular disease [5, 9], and harmful alcohol consumption generally in sub-Saharan Africa is associated with an array of health and social problems, including infectious diseases [5, 9, 10]. Given the heavy burden of both HIV and tuberculosis (TB) in sub-Saharan Africa, HED is of special concern in the region [11, 12]. A meta-analysis of experimental studies found a dose response relationship between alcohol use and intention to engage in condomless sex, which increases risk of HIV acquisition [13]. Another meta-analysis found that individuals that drink heavily had a three-fold greater risk of TB infection (pooled relative risk 2.94, 95% CI: 1.89–4.59) [14]. Heavy alcohol use is also associated with poor engagement and retention in HIV and TB care, as well as accelerated disease progression, which is especially problematic in a generalized HIV epidemic; adherence to treatment not only improves health outcomes but also reduces the risk of further transmission [14–16]. Similarly, poor adherence to TB treatment can lead to complications, such as the development of drug resistance.

Pharmacological interventions involve the use of pharmacotherapy in the treatment of alcohol misuse. Pharmaceuticals can be used to aid in the withdrawal process (e.g. benzodiazepines, phenobarbital, anticonvulsants) among persons who are physically dependent as well as to promote abstinence (e.g. naltrexone, disulfiram) [17]. They can be used in conjunction with non-pharmacological approaches or as a stand-alone intervention. Despite the pervasiveness of HED in sub-Saharan Africa, there are limited resources available to address this health issue, including limited availability of pharmacological alcohol treatment options [18]. When available, medications (especially newer ones such as naltrexone) tend to be expensive and many individuals lack health insurance to subsidize costs [17]. Furthermore, for those that do have access to insurance, treatment of alcohol use disorders is often not covered, making finances a significant barrier to pharmacological treatment access in low income countries [19].

Non-pharmacological interventions can involve the use of psychosocial and structural approaches to address alcohol misuse but they do not include a pharmaceutical/medication component. Psychosocial interventions are defined as “psychologically-based interventions aimed at reducing consumption behavior or alcohol-related problems” [20], while structural interventions go beyond the individual level to change the environments in which risk behavior occurs, such as alcohol regulation. Non-pharmacological interventions to reduce harmful alcohol consumption have been piloted and implemented in numerous settings throughout sub-Saharan Africa, but this evidence has not been systematically reviewed and synthesized. Understanding the effect of non-pharmacological interventions on HED is particularly important in the context of sub-Saharan Africa, given elevated rates of this pattern of drinking, and its harmful effects on health.

In order to address this gap, we reviewed the existing literature for non-pharmacological interventions to address alcohol use in sub-Saharan African settings that reported HED outcomes. This review is part of a companion review and meta-analysis [21] with the same search criteria that looked at different alcohol use outcomes (e.g. Alcohol Use Disorders Identification Test [AUDIT] score). We report HED outcomes in the present manuscript separate from the larger meta-analysis [21] as they could not be quantitatively pooled due to heterogeneity in their measurement.

Methods

This systematic review was guided by both the Cochrane Handbook recommendations for the review of interventions [22] and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (see “S1 Appendix”) [23]. This review was registered with the PROSPERO online registry (registration number CRD42019094509).

In this systematic review, we searched Embase, Medline, PsycINFO, EBSCO, CINAHL, and Cochrane CENTRAL on December 21, 2017 for published reports in English from the earliest available date per database. This search was rerun on March 14, 2019 and again on August 19, 2020. The search protocol is provided in “S2 Appendix”. We also hand-searched reports and included supplementary data sent by study authors. As discussed, this review was part of the search for a larger systematic review inclusive of other alcohol consumption outcomes. The parameters of the search were as follows: randomized or nonrandomized controlled trial design, conducted in sub-Saharan Africa, assessing a non-pharmacological intervention aimed at alcohol reduction, and measuring at least one alcohol consumption outcome.

For this paper, eligible alcohol consumption outcomes were those measuring HED outcomes. Given the wide variability of measurement of these outcomes, we included studies using variations of established measures of HED (e.g., at least 60 grams of pure alcohol on at least one occasion in the past 30 days) and binge drinking (e.g., four drinks for women or five drinks for men on a given occasion). In addition, we included measures indicative of this pattern of drinking (i.e., high consumption of alcohol on at least one occasion), such as high elevated Blood Alcohol Concentration (BAC) or frequency of getting drunk. We determined whether outcomes met these criteria by discussion and consensus among study authors.

Reasons for exclusion included: alcohol reduction not being a primary goal of the intervention; alcohol reduction only being addressed in the context of sex; not having a comparator group; or having a comparator that was another evidence-based or ‘bona-fide’ alcohol intervention (i.e., non-inferiority trial) as the aims and effect size would differ from that of an efficacy/effectiveness trial.

Eligible comparator groups included interventions unrelated to alcohol, usual care for alcohol or other services, brief feedback on an alcohol screening tool, alcohol or other informational materials, wait-list, and nothing.

One author (KS) screened all titles and abstracts. A second author (AM) did a targeted review of the screened titles and abstracts. All authors and three research assistants reviewed full-text reports and assessed their eligibility for inclusion in pairs. The standardized rubric that was used for the review of full-text articles is provided in “S3 Appendix.” Disagreements between pairs of reviewers were resolved by discussion and consensus was reached between the reviewers, or by a third author.

Results

A total of 1508 unique citations were identified through the database search after the exclusion of duplicates. Six additional studies were identified through hand-searching and correspondence with study authors. Of the 101 reports that underwent full-text screening, 88 were excluded for reasons outlined in Fig 1. See “S6 Appendix” for a list of ineligible studies reviewed as full-text with reasons for exclusion. In total, 13 studies met criteria for inclusion in this review [27–39].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Studies included in systematic review.

Adapted from the 2009 PRISMA Flow Diagram. Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group. Preferred Reporting Items for Systematic Reviews and Meta Analyses: The PRISMA Statement. PLoS Med 2009;6(7): e1000097. doi:10.1371/journal.pmed1000097.

https://doi.org/10.1371/journal.pone.0242678.g001

Characteristics of included studies

Characteristics of the study design and samples for each included study are summarized in Table 1. The thirteen studies included in this review were conducted in four countries: South Africa (n = 9) [27, 31, 33–39], Kenya (n = 2) [30, 32], Nigeria (n = 1) [29], and Zimbabwe (n = 1) [28]. Studies included randomized control trials (n = 6) [30–32, 35–37], cluster randomized control trials (n = 5), [27, 28, 34, 38, 39], one non-randomized controlled trial [29], and one quasi-experimental pretest/posttest design [33]. The majority of studies included general adult populations, whereas one was with university students [36], two with young adults [31], and one with women 15 years or older [39]. Three studies included patient populations, including general outpatients [35], HIV outpatients [32], and TB patients [34]. Two workplace-based studies were identified, one with safety and security employees [27] and another with employees of alcohol establishments [33]. Another study included market traders [29]. One study was with female sex workers [30], and two were focused on fetal alcohol syndrome prevention among pregnant women [38] and women at-risk for alcohol-exposed pregnancy [37]. Of the thirteen studies included, seven included alcohol misuse at baseline as part of their eligibility criteria [30–32, 34–37].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Summary of study and population characteristics of non-pharmacological alcohol interventions in sub-Saharan Africa.

https://doi.org/10.1371/journal.pone.0242678.t001

Description and effects of interventions

A description of the included interventions, comparator groups, and the reported intervention effects on HED outcomes are reported in Table 2. Table 2 also includes outcome definitions, as measured by each study. Only one study included a biological outcome, which was BAC [33]; all other studies included self-reported drinking outcomes.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Intervention details and summary of the intervention effect on reduction in heavy episodic drinking (HED) outcomes.

https://doi.org/10.1371/journal.pone.0242678.t002

Interventions were primarily individual-level psychosocial interventions, including seven utilizing Motivational Interviewing (MI) and/or Brief Intervention (BI) in single or multi-sessions [30, 31, 34–38]. Other psychosocial intervention approaches identified include: three multi-component risk reduction interventions, including two alcohol/HIV risk reduction interventions [27, 39] and one focused on risk factors for hypertension [29], as well as one cognitive-behavioral therapy (CBT) intervention [32]. Two studies were identified that targeted change beyond the individual-level. At the interpersonal and community-levels, one intervention employed a social-network intervention approach based on the Diffusion of Innovation Theory [28], and another organizational-level intervention provided training to employees of alcohol establishments in the responsible sale and serving of alcohol [33]. The details of the intervention effects are presented in the next section organized by intervention approach.

Comparator groups most commonly included an educational leaflet or general information (n = 4) [34–37], standard-of-care health services (n = 3) [31, 38, 39], or nutrition or lifestyle interventions (n = 3) [27, 30]. Three studies did not provide any treatment to the comparator group [28, 29, 33].

In general, results from six of the thirteen trials showed at least one statistically significant change in a HED outcome in the expected direction among intervention compared to those in comparator groups. Only three of these studies were deemed clinically significant by review authors. The other seven trials reported no intervention effect of HED at a level of statistical significance.

Description of interventions and intervention effects on HED outcomes.

Individually-based Motivational-Interviewing (MI), Brief Interventions (BI). Eleven out of the thirteen interventions identified in this review were individual-level psychosocial interventions. Of these interventions, the most commonly used intervention approach was MI interventions, which was the focus of seven interventions [27, 30–34, 36], six of which were described as BI [27, 30–32, 34, 36] with four based on the WHO BI for Alcohol Use [27, 31, 32, 37]. Despite commonalities in the core approach, implementation across studies varied. The total intervention dose ranged from single 20-minute sessions to a 120-minute 6-session intervention. MI/BI interventions were most commonly held in health facilities, but also took place in participants’ homes and other community venues. Only two out of the seven MI/BI studies reported statistically significant changes in one of the review’s specified HED outcomes.

One of the two interventions reporting statistically significant results was L’Engle et al.’s [30] study with Kenyan female sex workers, which included six 20-minute individually-delivered BI sessions using MI to reduce alcohol use with dual focus on HIV risk reduction. Those receiving the intervention reported drinking 3 or more drinks on the same occasion less frequently in the prior month at 6 months compared to participants receiving a time-matched nutrition intervention (Adjusted Odds Ratio [AOR] = 0.13, 90% Confidence Interval [CI] = 0.10, 0.17, p < 0.0001) and 12 months follow-up (AOR = 0.18, 90% CI = 0.13, 0.23, p < 0.0001). The review authors deemed these findings clinically significant, given the effect size and sustained change. The second was Pengpid et al.’s [36] study focused on South African university students who drink at hazardous levels. The intervention was based on the WHO BI for Alcohol Use [38], with additional content informed by the Information-Motivation-Behavioral (IMB) Skills Model [39]. Similar to L’Engle et al. [30], the intervention was delivered individually; however, it included only a single 20 minute session as opposed to six. Pengpid et al.’s [36] reported statistically significant reductions in HED (≥ 5 standard drinks on one occasion if male and ≥ 4 standard drinks if female) over time across both treatment groups. The respondents who received BI showed a higher decline in HED during the follow-up compared to control participants at the 12-month follow-up (β = −0.44; 95% CI = −0.76, −0.12; p = 0.007); however, these small changes were not judged as clinically significant by the review authors.

The other five interventions incorporating MI did not report statistically significant reductions in the review’s alcohol outcomes [31, 34, 35, 38, 41]. Three of these studies employed one to two (15–20 minute) BI sessions with patient populations in clinic settings in South Africa, including young adults in primary care [27], TB patients [34] and hospital outpatients [35]. Two of the trials included the same intervention reported in Pengpid et al. [36] (WHO BI for Alcohol Use with IMB model modifications) using a health education leaflet as control [34, 35]. In both trials, reductions in HED (i.e., drinking ≥ 5 standard drinks for men and ≥ 4 standard drinks for women on one occasion) were observed in both intervention and control. However, unlike Pengpid et al., [32] there was not a statistically significant difference between intervention and control.

The two remaining MI interventions had commonalities in their goals to reduce drinking during pregnancy among South African women [37, 38]. Rotheram-Borus et al. [38] reported on a single-session alcohol BI integrated into a home visiting intervention delivered by locally trained “Mentor Mothers” aimed at fetal alcohol syndrome prevention in South Africa. Rendall-Mkosi and colleagues [37] tested the effectiveness of a five-session, individually-based MI intervention to reduce the risk of alcohol exposed pregnancy among South African women of reproductive age screened as at-risk for alcohol exposed pregnancy. Both saw trends towards improvement in HED measures. However, Rendall-Mkosi and colleagues [34] did not reach statistical significance. Rotheram-Borus et al. [30] reported drinking increased over time in both study arms, and that the intervention attenuated an increase in problem drinking. The study authors report that the intervention effect became “substantial” at the 5-year time point for problem drinkers, with the proportion of mothers classified as problem drinkers approximately 6% in the intervention arm and 12% in the control arm. Despite this difference, we classify this study with other studies that show no statistical significance, as the authors did not explicitly test the statistical significance of this effect.

Integrated risk reduction interventions. Three interventions, including two group and one individually-based, included a risk reduction approach focused on alcohol within a broader multi-component intervention package, including two focused on HIV [27, 39] and one on risk factors for hypertension [29]. All three reported statistically significant reductions in HED outcomes. Among the alcohol/HIV risk reduction interventions was “Team Awareness” and “Women’s Health CoOP Plus” (WHC+) [27, 39]. Employees receiving the Team Awareness intervention reported a statistically significant reduction in frequency of binge drinking in the prior 30 days [F(1,117) = 25.16, p <0.0001)]. The mean number of days participants in the intervention condition reported having five or more drinks in one sitting in the prior 30 days reduced from 2.1 days to 1.4 days, in the predicted direction. The review authors did not deem this change clinically significant; drinking five or more drinks on at least one day in the past month still fits within the definition of binge drinking. WHC+ was delivered over two 1-hour one-on-one sessions aimed to reduce alcohol and other drug use among women living with HIV in Cape Town. WHC+ participants were significantly less likely to engage in frequent heavy drinking at 6 months follow-up (4 or more drinks on 11 or more days in the prior 30 days) (AOR = 0.45; 95% CI = 0.28, 0.73; p = 0.02), but not at 12-month follow-up (AOR = 0.71, 95% CI = 0.43, 1.18; p = 0.19). Participants in the WHC+ also reported less days of heavy drinking (4 or more drinks) in the prior 30 days at 6-month follow-up (p = 0.01), but not 12-month follow-up (p = 0.36). The review author deemed the reduction in the proportion of women engaging in frequent heavy drinking as clinically significant, reducing from nearly 40% to 20% at 12-months follow-up. While there was also considerable change in the control condition, we still deemed the difference meaningful.

Eze et al. [29]’s risk reduction intervention focused on the reduction of alcohol use as a risk factor of hypertension, while simultaneously aiming to reduce other risk factors (e.g., poor diet, physical inactivity). The intervention included two 5-hour group sessions with market traders in Nigeria, as well as the establishment of health posts in the region for reinforcement of intervention content and blood pressure screening. Compared to the control, intervention participants saw a statistically significant reduction in binge/excessive drinking (5 standard drinks/day if male, and 4 standard drinks if female on 5 days in the prior 30 days) (χ² = 15.09, p <0.001). This change was considered clinically significant by the review authors, with the proportion of binge alcohol drinkers reduced by 9.5% among the intervention group.

Cognitive behavioral therapy. One study employed group-delivered cognitive behavior therapy (CBT). The Kenya Health Behavior Study assessed the efficacy of a six-session gender stratified group cognitive behavioral therapy (CBT) intervention delivered by counselors to reduce alcohol use among HIV outpatients in Kenya [32]. Compared to a healthy lifestyle control, intervention participants reported statistically significant reductions in the number of heavy drinking days (≥ 4 standard drinks if male and ≥ 5 standard drinks if female) in the prior 60 days at 7–30 weeks follow-up (mean difference [MD] = -0.21; 95% CI = -0.35, -0.08; p = 0.002). However, these effects were not maintained at the 31–46 week follow-up (MD = -0.02; 95% CI = -0.09, 0.06; p = 0.67). The average change in heavy drinking days from baseline to 7–30 weeks follow-up changed from approximately 6 days on average to 0 days, which the review authors deemed a clinically meaningful change. Although this was a statistically significant change compared to control, similarly meaningful change was observed in the control group. The review authors did not view this difference between intervention and control as clinically significant.

Interventions targeted change beyond the individual-level. Two studies, Cubbins et al. [28] and Peltzer et al. [33], implemented interventions at a community or organizational-level, but did not report any statistically significant change in alcohol outcomes. Cubbin et al.’s [28] community approach was based on Diffusion of Innovation Theory, employing a popular opinion leader model to spread culturally-specific health related messages across social networks in rural Zimbabwe. The intervention was targeted at the individual level but was expected to diffuse throughout a community via word of mouth and social normative influence. However, no statistically significant differences were found between communities receiving the CPOL intervention compared to those that did not in the number of days respondents reported getting drunk in the prior 30 days at either time point, nor was there significant individual-level change in this outcome. Peltzer et al.’s [33] structural intervention targeted change at the organizational-level, conducting a quasi-experimental controlled trial to assess the effects of a responsible alcohol beverage sales and servicing training intervention with owners and servers from alcohol serving establishments in Gugulethu, South Africa. A cross-sectional design was employed to assess pre/post changes of BAC level of bar patrons. Although tests of statistical significance were not reported, changes in the intervention sites and the control sites were not in the expected direction.

Risk of bias

In general, included studies evaluated with the Cochrane risk of bias tool were of moderate quality (see Fig 2). Just under half of studies were judged as low risk for random sequence generation [30, 32, 34, 35, 37, 39] and allocation concealment [30–32, 34, 35, 37], with the other half judged as high risk. None of the studies blinded both participants and study personnel, and only one third of studies blinded outcome assessment [29, 30, 34–36]–potential sources of performance and detection bias. Other methodological weaknesses included a lack of published study protocols resulting in high risk for selective reporting bias in half of studies [29–31, 33, 37, 39], and attrition bias due to loss to follow-up being greater than 20% [27, 33–35, 37]. Just under 25% of studies were judged as high-risk for failure to control for potential confounders [33, 37, 38]. The full risk of bias assessment per study is included in “S7 Appendix” and a graphic depiction of the assessment results per study is provided in “S8 Appendix.”

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Risk of bias graph: Review authors’ judgements about each risk of bias item presented as percentages across included studies.

https://doi.org/10.1371/journal.pone.0242678.g002

Discussion

This systematic review of non-pharmacological interventions aimed to reduce alcohol consumption in sub-Saharan Africa identified 13 trials that measured change in HED outcomes. Although sub-Saharan Africa has among the highest occurrence of HED in the world and high rates of alcohol-related morbidity and mortality, this is the only review to synthesize the effect of non-pharmacological alcohol interventions on HED outcomes in sub-Saharan Africa to-date. The majority of studies evaluated individual-level psychosocial interventions, such as BI with MI and other individual or group psychosocial approaches. Six of the trials showed statistically significant reductions in HED, three of which the review authors judged as clinically significant. Seven of the trials showed no effect. Alcohol interventions achieving statistical and/or clinical significance were conducted across diverse settings and populations, using a range of psychosocial approaches, including MI integrated in BI, CBT, integrated risk reduction interventions, delivered both individually and in groups. Taken together, the picture remains unclear regarding which interventions show the most promise for reducing HED outcomes in sub-Saharan African settings, pointing to the need for more research. However, this review can shed light on approaches for future investigation, as well as gaps in the literature.

The most commonly employed intervention approach included in this study was MI, commonly integrated into single or multi-session BI, employed in seven studies. Despite the wide use of this intervention approach, this review provides limited evidence supporting its effectiveness at reducing HED in sSA, with only two studies showing statistically significant effects, one of which was not considered clinically meaningful. This finding coincides with the companion meta-analysis we conducted as part of this same search; the meta-analysis similarly found limited evidence for MI and the WHO SBI guidelines in changing AUDIT scores in sub-Saharan Africa [21]. However, several of the MI-based studies in this review were trending towards significance or saw an improvement in both treatment arms but no significant differences between intervention participants compared to control. The latter could be an effect of the standard comparator for this intervention approach being minimal alcohol intervention (e.g., feedback on AUDIT screening and an alcohol leaflet), which could drive null effects. Nevertheless, these findings are in contrast to a large body of research in high-income country settings that reports moderate effects at alcohol reduction achieved through MI-based BI [42], warranting further investigation.

Based on a qualitative comparison of interventions reporting statistically significant changes in HED outcomes vs. those that did not, there were no clear patterns in differences in success between individual vs. group format of intervention delivery. However, interventions with greater dosage (2 hours+) tended to be more successful than BIs. This is another potential explanation for the underwhelming effects of MI-based BI studies in this review. While this points researchers towards the use interventions with a greater dose and intensity, the scale-up of more intensive approaches is challenged by time and resource constraints common to resource-limited settings. A cost-benefit analysis associated with the CBT Kenya study [32] reported CBT can be effectively and economically task-shifted to paraprofessionals in Kenya [43]. More costing research like this, as well as hybrid implementation studies aimed to assess implementation and effectiveness, are needed in order to better understand the appropriate intervention dosage that could be feasibly scaled up in sub-Saharan African clinical and community settings.

Of the studies included in this review, the overwhelming majority tested psychologically driven interventions focused on individual-level change, such as through MI or other forms of individual or group counseling and education. Only two studies went beyond an individual-level approach, taking an organizational or community-level approach to alcohol reduction [25, 29]. Despite the lack of support found for these interventions, they represent important attempts to address influences of alcohol use beyond individual-level knowledge, motivation, and self-efficacy. Research has established the importance of higher-level factors on alcohol consumption in sub-Saharan African settings, including social and cultural norms [44], alcohol outlet density [45], alcohol marketing [46], and a lack of alcohol regulation and policy enforcement [47]. More research is needed that tests structural intervention approaches in sub-Saharan African settings, such as policy interventions described by the WHO as “best buys” including increases in taxes on alcoholic beverages, bans and restrictions on alcohol advertising, and reductions in retail alcohol availability through reduced hours of sale [44]. These approaches, although difficult to implement, have the largest potential effect in low and middle-income country settings per disability-adjusted life year (DALY) averted.

Although sub-Saharan Africa has among the highest occurrence of HED in the world, no studies were identified in this review that were solely focused on the reduction of this pattern of drinking. The lack of interventions tailored to explicitly reduce HED may in part explain the lack of effect reported on this outcome. A larger literature exists with this aim in high-income country settings, which has demonstrated effects among mainly psychosocial approaches in reducing HED across settings and populations, particularly with adolescents and college students [48–50]. These interventions include a range of psychologically-based interventions, such as BI, personalized normative feedback, and protective behavioral strategies tailored to the dangers of binge drinking specifically [48]. Given strong social and environmental influences of binge drinking, studies targeting HED in high-income country settings tend to be implemented in places where HED occurs (e.g., fraternities, birthday parties), where widespread social norms need to change (e.g., schools), or situations conducive to a teachable moment related to binge drinking (e.g., emergency rooms) [48]. This again highlights the dearth of social and environmental approaches identified in this review; focusing on changing social relationships and environments where HED occurs may strengthen interventions in sub-Saharan Africa. This approach has been used with success in HIV interventions aimed to reduce alcohol-related HIV risk by altering the relationships, norms, and environment of alcohol venues where people engage in alcohol-related sexual risk, including multilevel approaches such as altering proximity to venues, the physical characteristics of venues, and social norms [51].

Supporting information

Acknowledgments

Janene Batten, Yale School of Nursing Librarian, helped to develop the search protocol and ran the search. Dr. Elizabeth Reed, Jessica Sibal, and Alice Wong at San Diego State University and Dr. Jennifer Wagman at the University of California, Los Angeles assisted us with the review of full-text articles. We thank these individuals for their contributions.