Treatments for Alcohol-Use Disorders

Treatments for AUDs continue to evolve as research on the effectiveness of various treatments is published, and as new treatments are introduced and used more frequently. No single best approach has yet proven superior among the variety of available treatment options. Some common treatments for AUDs include cognitive behavioral therapy, motivational enhancement therapy, 12-step programs (e.g., Alcoholics Anonymous), and pharmacotherapy. Treatment may be delivered via individual outpatient counseling, intensive outpatient programs using group or individual methods, alcoholism treatment centers, or other approaches. Most treatment is currently delivered in specialty settings rather than in primary care settings. Primary care providers are typically trained to refer patients with AUDs for specialized treatment, and primary care providers are generally unfamiliar with medications for treating AUDs.²⁰

Over the past 15 to 20 years, awareness has grown that treatment may be beneficial even if complete abstinence is not achieved. As a result, research has used other outcomes to measure the effectiveness of treatment, which can be subsumed under the concept of harm reduction.²¹ These measures include significant increases in abstinent days or decreases in heavy drinking episodes, improved physical health, and improvements in psychosocial functioning.

Pharmacological Interventions

Beginning in the 1950s, the pharmacotherapy for AUDs consisted only of disulfiram, an aversive deterrent that produces very uncomfortable symptoms when alcohol is consumed. Since the 1990s, two oral medications (naltrexone and acamprosate) and one long-acting intramuscular formulation (of naltrexone) have been approved by the U.S. Food and Drug Administration (FDA) for alcohol dependence. Table B describes the medications available in the United States that are FDA approved, their mechanism of action, and dosing. Many additional medications have been used off label or studied for treatment of AUDs. These include antidepressants, mood stabilizers, anticonvulsants, alpha-adrenergic blockers, antipsychotics, and anxiolytics.

Table B

Medications that are FDA approved for treating adults with alcohol dependence.

Despite ongoing developments and advancements in treatment approaches, AUDs are among the most undertreated disorders in the U.S. health care system; it is estimated that fewer than one in three individuals with AUDs receive treatment.¹⁰ Furthermore, data from the Veterans Health Administration show that, of those patients who receive treatment, fewer than 1 in 10 receive medication as part of treatment.^27,28 Therefore, expanding awareness and access to this relatively new treatment modality has the potential to improve outcomes and reduce the burden of this devastating illness that affects millions.

Scope and Key Questions

The use of medications for AUDs is associated with uncertainty and variation across providers and settings. Since the last report commissioned by the Agency for Healthcare Research and Quality (AHRQ) on medications for alcohol dependence (1999),^32,33 there has been more than a tenfold increase in the number of individuals studied in controlled clinical trials of naltrexone and acamprosate, and many trials of medications that are not FDA approved. Other reasons for conducting a new review on this topic include the following: (1) to assess the comparative effectiveness of the FDA-approved medications; (2) to determine whether any agents that are not FDA approved have evidence supporting their efficacy; (3) to evaluate the evidence on intramuscular naltrexone (Vivitrol^®), a fairly recently approved medication; (4) to evaluate whether trials provide evidence of effectiveness in primary care settings; (5) to assess whether some medications are more or less effective for adults with specific genotypes; and (6) to provide a comprehensive review of medications for AUDs that is relevant for clinicians, researchers, and policymakers.

Our report focuses on clinically relevant medications—those commonly used, those with sufficient literature for systematic review, and those of greatest interest to clinicians and to the developers of guidelines. Our report is limited to people with AUDs; it does not address those with risky or hazardous alcohol use (for whom medications are likely not an appropriate intervention).

The main objective of this report is to conduct a systematic review and meta-analysis of the comparative effectiveness and harms of medications for adults with AUDs. In this review, we address the following Key Questions (KQs):

KQ 1a: Which medications are efficacious for improving consumption outcomes for adults with AUDs in outpatient settings?

KQ 1b: How do medications for adults with AUDs compare for improving consumption outcomes in outpatient settings?

KQ 2a: Which medications are efficacious for improving health outcomes for adults with AUDs in outpatient settings?

KQ 2b: How do medications for adults with AUDs compare for improving health outcomes in outpatient settings?

KQ3a: What adverse effects are associated with medications for adults with AUDs in outpatient settings?

KQ 3b: How do medications for adults with AUDs compare for adverse effects in outpatient settings?

KQ 4: Are medications for treating adults with AUDs effective in primary care settings?

KQ 5: Are any of the medications more or less effective than other medications for men or women, older adults, young adults, racial or ethnic minorities, smokers, or those with co-occurring disorders?

KQ 6: Are any of the medications more or less effective for adults with specific genotypes (e.g., related to polymorphisms of the mu-opioid receptor gene [OPRM1])?

Analytic Framework

We developed an analytic framework to guide the systematic review process (Figure A).

Figure A

Analytic framework for pharmacotherapy for adults with alcohol-use disorders in outpatient settings. KQ = Key Question.

Literature Search Strategy

To identify articles relevant to each KQ, we searched PubMed^®, the Cochrane Library, PsycINFO^®, CINAHL^®, and Embase^® for English-language and human-only studies published from January 1, 1970, to October 11, 2013. Searches were run by an experienced Evidence-based Practice Center (EPC) librarian and were peer reviewed by another EPC librarian. We manually searched reference lists of pertinent reviews, trials, and background articles on this topic to look for any relevant citations that our searches might have missed.

We searched for unpublished studies relevant to this review using ClinicalTrials.gov, the World Health Organization International Clinical Trials Registry Platform, and the FDA Web site. In addition, AHRQ's Scientific Resource Center requested any unpublished studies and pertinent data from relevant pharmaceutical companies. We also retrieved and assessed references suggested by our peer reviewers and the public.

Eligibility Criteria

We developed inclusion and exclusion criteria with respect to populations, interventions, comparators, outcomes, timing, and setting (PICOTS) and study designs. We included studies enrolling adults with AUDs that evaluated one or more of the following medications: acamprosate, disulfiram, naltrexone, amitriptyline, aripiprazole, atomoxetine, baclofen, buspirone, citalopram, desipramine, escitalopram, fluoxetine, fluvoxamine, gabapentin, imipramine, nalmefene, olanzapine, ondansetron, paroxetine, prazosin, quetiapine, sertraline, topiramate, valproate, varenicline, and viloxazine.

Studies were required to assess at least one of the following outcomes: return to any drinking (lapse), return to heavy drinking (relapse), drinking days, heavy drinking days, drinks per drinking day, time to lapse or relapse, accidents, injuries, quality of life (QoL), function, mortality, or adverse effects. Studies were required to treat patients with a medication for a minimum of 12 weeks in an outpatient setting.

For KQs 1, 2, and 4, double-blind randomized controlled trials (RCTs) that compared one of the medications with placebo or another medication and recent systematic reviews (searches ending no earlier than 2007) were eligible. For KQ 2b, prospective cohort studies were also eligible. For KQ 3 (harms), double-blind RCTs and recent systematic reviews that compared one of the medications with placebo or with another medication were eligible. The following designs were also eligible for KQ 3 if they compared two or more drugs of interest: nonrandomized controlled trials, open-label trials, secondary analyses or subgroup analyses from trials, prospective cohort studies, and case-control studies. For KQ 5 (subgroups), double-blind RCTs, recent systematic reviews, nonrandomized controlled trials, open-label trials, secondary analyses or subgroup analyses from trials, prospective cohort studies, and case-control studies were eligible as long as the studies compared two or more drugs. For KQ 6 (specific genotypes), double-blind RCTs, analyses of subjects enrolled in trials, and prospective cohort studies comparing people with different genotypes were eligible.

Study Selection

Two members of the research team independently reviewed each title and abstract (identified through searches) to determine eligibility. Studies marked for possible inclusion by either reviewer and those that lacked adequate information to determine eligibility underwent a full-text review. Two members of the team independently reviewed each full-text article to determine eligibility. If the reviewers disagreed, they resolved conflicts by discussion and consensus or by consulting a senior member of the team.

Data Extraction

We designed and used structured data extraction forms to gather pertinent information from each article; this included characteristics of study populations, settings, interventions, comparators, study designs, methods, and results. Trained reviewers extracted the relevant data from each included article. All data extractions were reviewed for completeness and accuracy by a second member of the team.

Risk-of-Bias Assessment of Individual Studies

To assess the risk of bias (internal validity) of studies for major outcomes of interest, we used predefined criteria based on guidance from the AHRQ “Methods Guide for Effectiveness and Comparative Effectiveness Reviews.”³⁴ We assessed selection bias, confounding, performance bias, detection bias, and attrition bias; we included questions about adequacy of randomization, allocation concealment, similarity of groups at baseline, blinding, attrition, whether intention-to-treat analysis was used, methods of handling missing data, and fidelity. We rated the studies as low, medium, high, or unclear risk of bias.³⁵ Two independent reviewers assessed the risk of bias for each study. Disagreements between the two reviewers were resolved by discussion and consensus or by a third member of the team.

Data Synthesis

We conducted meta-analyses using random-effects models to estimate pooled effects.³⁶ For continuous outcomes, we used weighted mean differences (WMDs). For binary outcomes, we calculated risk differences (RDs) between groups. We did not include studies rated as high or unclear risk of bias in our main analyses but did include them in sensitivity analyses. We calculated the chi-squared statistic and the I² statistic to assess statistical heterogeneity in effects between studies.^37,38 We also examined potential sources of heterogeneity by analysis of subgroups defined by patient population (e.g., U.S. vs. non-U.S. studies) and variation in interventions (e.g., dose). When quantitative synthesis was not appropriate (e.g., because of clinical heterogeneity, insufficient numbers of similar studies, or insufficiency or variation in outcome reporting), we synthesized the data qualitatively.

Strength of the Body of Evidence

We graded the strength of evidence (SOE) as high, moderate, low, or insufficient based on established guidance.³⁹ Developed to grade the overall strength of a body of evidence, the approach incorporates four key domains: risk of bias (includes study design and aggregate quality), consistency, directness, and precision of the evidence. It also considers optional domains. Two reviewers assessed each domain for each key outcome and determined an overall SOE grade based on domain ratings. In the event of disagreements on the domain or overall grade, they resolved differences by discussion or by consulting an experienced investigator. We graded the SOE for the following outcomes: return to any drinking, return to heavy drinking, drinking days, heavy drinking days, drinks per drinking day, accidents, injuries, QoL or function, mortality, and adverse events.

Applicability

We assessed applicability of the evidence following guidance from the “Methods Guide for Effectiveness and Comparative Effectiveness Reviews.”⁴⁰ We used the PICOTS framework to explore factors that affect applicability.

Key Question 1. Consumption Outcomes

We found moderate SOE that both acamprosate and oral naltrexone (50 mg/day) are effective for improving alcohol consumption outcomes (Table C). Numbers needed to treat (NNTs) to prevent 1 person from returning to any drinking were 12 and 20, respectively. For return to heavy drinking, evidence did not support the efficacy of acamprosate, whereas oral naltrexone (50 mg/day) was efficacious, with an NNT of 12. We found low SOE that injectable naltrexone is efficacious for reducing percentage of heavy drinking days. Evidence from well-controlled trials does not adequately support the efficacy of disulfiram compared with placebo for preventing return to any drinking or for other alcohol consumption outcomes. Some disulfiram trials reported fewer drinking days for subjects who returned to any drinking and who had a complete set of assessment interviews, and suggest that disulfiram may have a role in the treatment of alcohol dependence for some individuals.

Table C

Summary of findings and strength of evidence for efficacy of FDA-approved medications for alcohol dependence.

Our meta-analyses of three head-to-head RCTs comparing acamprosate with naltrexone,^49-51 all rated as low risk of bias, found no statistically significant difference between the two medications for return to any drinking, and our meta-analysis of four head-to-head RCTs^49-52 similarly found no statistically significant difference between the two medications for return to heavy drinking (Table D). The COMBINE study was one of the head-to-head RCTs.⁴⁹ It found that “patients receiving medical management with naltrexone, combined behavioral intervention (CBI), or both had better drinking outcomes than those who received placebo, but acamprosate showed no evidence of efficacy, with or without CBI.”

Table D

Summary of findings and strength of evidence for comparative effectiveness of acamprosate and naltrexone.

For the vast majority of medications used off label and those under investigation, either evidence was insufficient to determine whether they are efficacious for reducing alcohol consumption or evidence suggested that they are not efficacious for people with AUDs. We found some exceptions. First, for topiramate, we found moderate SOE supporting efficacy for reducing drinking days, heavy drinking days (WMD, -11.5; 95% CI [confidence interval], -18.3 to -4.8), and drinks per drinking day (WMD, -1.1; 95% CI, -1.7 to -0.4) based on the results of two RCTs rated as low or medium risk of bias (total N=521).^53,54 The included RCTs did not report data for return to any drinking or return to heavy drinking. Second, for nalmefene, we found moderate SOE supporting efficacy for reducing heavy drinking days per month (WMD, -2.0; 95% CI, -3.0 to -1.0) and drinks per drinking day (WMD, -1.0; 95% CI, -1.8 to -0.3).^55,56 Finally, limited evidence from two small RCTs (total N=88), one enrolling people with bipolar disorder, supports efficacy of valproic acid for reducing return to heavy drinking, heavy drinking days, and drinks per drinking day (low SOE).

Key Question 2. Health Outcomes

We found insufficient direct evidence from trials to conclude that treatment with acamprosate or naltrexone leads to improvement in health outcomes—i.e., accidents, injuries, QoL, function, or mortality (Table C). Very few trials reported any health outcomes, and the included trials were not designed or powered to assess impact on health outcomes; they typically focused on alcohol consumption outcomes. The largest pharmacotherapy trial, COMBINE, reported some evidence of improvement in QoL with naltrexone plus behavioral intervention (on the physical health scale from the 12-item Short Form health survey, version 2), but the difference between groups did not reach a clinically meaningful threshold.⁴⁶

Key Question 3. Harms

Of the included studies, 114 provided information on harms. Evidence for many potential adverse events was insufficient to determine whether the risk was increased or not, often primarily because of lack of precision. For most of the specific adverse events, point estimates favored placebo (i.e., there were more adverse events with medications), but differences were not statistically significant. In head-to-head studies, the risk of withdrawal due to adverse events was not significantly different between acamprosate and naltrexone, whereas the risks of headache and vomiting were higher for those treated with naltrexone. Compared with placebo, patients treated with acamprosate had a higher risk of anxiety, diarrhea, and vomiting; those treated with naltrexone had a higher risk of dizziness, nausea, and vomiting; and those treated with nalmefene had a higher risk of dizziness, headache, insomnia, nausea, and vomiting. Individual trials of topiramate reported increased risk of many adverse events, including paresthesias, taste perversion, anorexia, difficulty with concentration/attention, nervousness, dizziness, pruritis, psychomotor slowing, and weight loss.^53,54 A single trial that reported adverse effects for valproic acid compared with placebo found a higher rate of nausea for patients treated with valproic acid.

Key Question 4. Evidence From Primary Care Settings

Evidence from primary care settings was scant. One trial (N=100) that recruited subjects primarily by advertisement in two family medicine settings in the United States found no significant treatment effect when comparing acamprosate with placebo.⁵⁷ The only other trial meeting our inclusion criteria that was conducted partly in primary care settings compared nalmefene with placebo in 15 sites (about half were primary care settings) in Finland.⁵⁸ See the Discussion section below (under Primary Care) for more information about studies that may have applicability to primary care settings.

Key Question 5. Subgroups

We did not find any compelling evidence that naltrexone, acamprosate, topiramate, nalmefene, or valproic acid are more or less effective (compared with each other) for men or women, older adults, young adults, racial or ethnic minorities, smokers, or those with co-occurring disorders.

Key Question 6. Genetic Polymorphisms

We found no studies that assessed the clinical utility of genotype-guided dosing strategies or genotype-guided medication selection and none that randomized by genotype. All included studies were either subgroup analyses of trials or prospective cohort studies of people treated with a medication, and all assessed the association between genotype and response to medication (i.e., clinical validity). For most polymorphism-medication pairs, we found just one eligible study, and we graded the SOE as insufficient.

Seven eligible studies assessed variation in naltrexone response related to mu-opioid receptor gene (OPRM1) polymorphisms. Our meta-analyses for return to any drinking and return to heavy drinking found no significant difference between A-allele homozygotes and those with at least one G allele, both without inclusion of studies rated as high or unclear risk of bias (RD, -0.03; 95% CI, -0.6 to 0.5, and RD, 0.26; 95% CI, -0.01 to 0.53, respectively) and with them (RD, 0.01; 95% CI, -0.2 to 0.2, and RD, 0.14; 95% CI, -0.03 to 0.3, respectively). Point estimates for return to heavy drinking suggest it is possible that patients with at least one G allele of A118G polymorphism of OPRM1 might be more likely to respond to naltrexone compared with patients without a G allele, but CIs were wide; additional studies are needed to improve confidence in the estimate of the effect.

Primary Care

Direct evidence in primary care settings was scant. One included trial (N=100) conducted completely in primary care settings found no significant treatment effect when comparing acamprosate with placebo.⁵⁷ The only other included trial was conducted partly in primary care settings and compared nalmefene with placebo in 15 sites (about half were primary care settings) in Finland.⁵⁸

Some included studies conducted in non–primary-care settings used interventions that may be adaptable for delivery in primary care. For example, in the COMBINE study,⁴⁹ providers delivered a medical management intervention comprised of up to nine manual-guided counseling visits. The first visit was approximately 45 minutes and followup visits were about 20 minutes each. Medical management included advice for reducing drinking, inquiries about medication side effects, and emphasis on the importance of taking medications as prescribed. Another trial (included in KQ 1 but not in KQ 4), which compared naltrexone with placebo for 12 weeks in the United States, described the use of a “primary care model.”⁶⁶ Although the trial did not take place in a primary care setting (it was a treatment research center) and the investigators were from a department of psychiatry, the psychosocial cointervention was delivered by a nurse practitioner with a primary care background, and the trial may have implications for how psychosocial cointerventions could be provided in primary care settings.

In terms of implementing treatment programs for AUDs in primary care, we identified four other publications that did not meet our inclusion criteria (due to the study design or comparators) but may have important implications for primary care settings.^67-70 While these studies found conflicting results, they demonstrated the feasibility of managing AUDs in primary care. In general, these interventions involved formal clinic structure, staffing, and protocols. They used variations of chronic care management, multidisciplinary team-based care, and care coordination between primary care providers and mental health providers (e.g., physicians coordinating with social workers to connect patients to community resources or provide counseling).

First, a nested sequence of three RCTs based in the United States compared naltrexone plus “primary care management” (PCM) with naltrexone plus cognitive behavioral therapy.⁶⁷ They found no difference in avoiding persistent heavy drinking between those who received naltrexone plus PCM and those who received naltrexone plus cognitive behavioral therapy. Among responders enrolled in a maintenance trial, those who received naltrexone and PCM had significantly better response than those who received placebo and PCM. Second, a pragmatic trial with 149 general practitioners in France randomized patients (N=422) to acamprosate plus standard care or standard care alone.⁶⁸ The trial reported better outcomes for the acamprosate group for alcohol-related health, personal, and social problems, and quality of life. Third, an RCT based in the United States (N=163) found that participants in a primary care–based alcohol care management program were more likely to receive naltrexone (65.9% vs. 11.5%), to be engaged in treatment (OR [odds ratio], 5.36; 95% CI, 2.99 to 9.59), and to have a lower percentage of heavy drinking days (OR, 2.16; 95% CI, 1.27 to 3.66) than participants in a specialty treatment program.⁶⁹ Fourth, the Alcohol Health Evaluation and Disease Management (AHEAD) study, based in the United States (N=563), compared chronic care management (CCM) that included longitudinal care coordinated by a primary care clinician with no CCM for people with alcohol or drug dependence who were not currently engaged in primary care.⁷⁰ Of those enrolled, 12 percent had alcohol dependence without also meeting criteria for other drug dependence. CCM included motivational enhancement therapy; relapse prevention counseling; onsite medical, addiction, and psychiatric treatment; social work assistance; and referrals. The no-CCM group received a primary care appointment and a list of treatment resources, including a telephone number to arrange counseling. The trial found no difference between groups for the primary outcome of abstinence over 12 months.

Barriers to prescribing medications for AUDs in primary care may include lack of familiarity with the medications, lack of confidence in their effectiveness, or inability to provide suitable psychosocial cointerventions (e.g., due to competing demands or insufficient practice resources, personnel, or training). Like behavioral counseling interventions for risky drinking delivered in primary care, implementing the use of medications and psychosocial cointerventions for AUDs in primary care might require development of support systems and additional provider and staff training.^1,3 Further, primary care providers are typically trained to refer patients with AUDs for specialized treatment. O'Malley and O'Connor recently reviewed the issues surrounding the use of medications for alcohol dependence in primary care settings.²⁰ They concluded that “the implementation and widespread use of medications to treat alcohol problems faces a unique set of barriers in primary care. Although primary care providers are proficient at prescribing a wide variety of medications, they generally are unfamiliar with medications for treating alcohol problems other than those used to treat alcohol withdrawal.” They referenced a body of research to support basic screening methods, brief interventions, and medication therapy that has yet to have a major impact on how primary care providers care for individuals at risk for or with alcohol problems.⁷¹

Applicability

Most studies reported that all subjects met criteria for alcohol dependence. We did not identify any studies that evaluated medications and reported them to be efficacious for people with AUDs who did not meet criteria for alcohol dependence (i.e., people with alcohol abuse or harmful alcohol use). The included literature used definitions from DSM-III or DSM-IV. DSM-5 (2013) describes a single AUD category measured on a continuum from mild to severe and no longer has separate categories for alcohol abuse and dependence.²⁵ Using DSM-5 terminology, most participants in the included studies likely had moderate to severe AUD. Thus, applicability of our findings to people with mild AUDs is uncertain.The mean age of subjects was generally in the 40s, with very few studies enrolling slightly younger or older populations. Thus, it is uncertain whether the medications have similar efficacy for older (e.g., 65 and older) or younger (e.g., in the 20s) subgroups. We did not find evidence to confirm or refute whether treatments are more or less efficacious for gender groups, racial or ethnic minorities, smokers or nonsmokers, and those with certain coexisting conditions.

Although the majority of included trials assessing the efficacy of acamprosate were conducted in Europe (16 of 22) and a minority were conducted in the United States (4 of 22), the opposite was true for naltrexone (27 of 44 in the United States and 8 of 44 in Europe). Further, the few studies of acamprosate conducted in the United States did not find it to be efficacious. It is unclear whether the different results were due to population differences or other factors. The European trials of acamprosate typically identified patients from inpatient settings or treatment programs, whereas the trials of acamprosate based in the United States relied on advertisements and referrals. It is possible that this resulted in populations with differing AUDs severity and differing potential for benefit. For example, studies of subjects recruited via advertisements may enroll people who have less severe disorders.

Most studies required patients to abstain for at least a few days prior to initiating medication, and the medications are generally recommended for maintenance of abstinence. Acamprosate and injectable naltrexone are approved only for use in patients who have established abstinence, although the duration of required abstinence is not set. However, some studies enrolling patients who were not yet abstinent reported reduction in heavy drinking with naltrexone^72,73 or acamprosate.⁷⁴

Limitations of the Comparative Effectiveness Review Process

The scope of this review was focused on medications. We did not evaluate the effectiveness or comparative effectiveness of other interventions for AUDs (e.g., 12-step programs). We required that trials have at least 12 weeks of followup from the time of medication initiation, excluding trials of shorter duration. Some might consider this approach to omit potentially important information. However, longitudinal studies have found that shorter treatment periods may yield misleading conclusions about treatment efficacy due to fluctuations in drinking behavior that are typical of the course of AUDs,^75,76 suggesting that longer durations of followup might more accurately reflect the outcomes of greatest interest and importance.

Our review focused on benefits and harms of medications and how they compare with other medications, and our findings generally reflect the added benefits of medications beyond those of psychosocial cointerventions. However, studies used a variety of different psychosocial cointerventions, and this heterogeneity limits our certainty about the effect of medications when used alone (with no psychosocial cointervention) or when added to a particular psychosocial intervention. Reporting of previous and ongoing psychosocial interventions was variable across the included studies, and we were unable to determine whether subjects actually received some cointerventions; for example, Alcoholics Anonymous was recommended, but no information was reported about how many subjects adhered to the recommendation.

Finally, publication bias and selective reporting are potential limitations. Although we searched for unpublished studies and unpublished outcomes, and did not find direct evidence of either of these biases, many of the included trials were published prior to the availability of trial registries (e.g., clinicaltrials.gov) that would allow for greater certainty in determining the potential for either type of bias.

Limitations of the Evidence Base

The evidence base was inadequate to draw conclusions for some of our questions or subquestions of interest. In particular, as described above, we found insufficient direct evidence on health outcomes, limited and varying reporting on harms, few trials conducted in primary care settings, and scant head-to-head evidence on differences for population subgroups.

We found insufficient direct evidence to determine whether medications are efficacious for improving health outcomes. Although evidence from epidemiologic literature consistently relates high average and heavy per-occasion alcohol use to an increased risk of health problems, it is challenging to estimate the magnitude of reduction in the risk of health problems that is derived from a reduction in consumption. For example, it is unclear how much benefit for health outcomes is derived from 10 percent fewer patients returning to any drinking.

Many included trials had methodological limitations introducing some risk of bias. Some had high proportions of subjects lost to followup. High attrition rates are not uncommon in studies of psychiatric conditions, but methods of handling missing data varied, and some trials did nothing to address missing data (i.e., analyzing only completers). However, many trials conducted true intention-to-treat analyses and used appropriate methods of handling missing data, such as imputing return to heavy drinking for subjects lost to followup or using multiple imputations.

Future Research

We identified numerous gaps in the evidence that future research could address. Many of these gaps are highlighted in the previous sections of this Discussion. Of note, these gaps relate only to the KQs addressed by this report, and they should not eliminate a wide range of potentially important research that falls outside of our scope. Table E summarizes the key gaps and potential future research that could address the gaps.

Table E

Evidence gaps for future research by Key Question.