Pharmacotherapy for Alcohol Use Disorder: A Systematic Review and Meta-Analysis

Key Points

Question  Which pharmacotherapies are associated with improved outcomes for people with alcohol use disorder?

Findings  In this systematic review and meta-analysis that included 118 clinical trials and 20 976 participants, 50 mg/d of oral naltrexone and acamprosate were each associated with significantly improved alcohol consumption-related outcomes compared with placebo.

Meaning  These findings support oral naltrexone at 50 mg/d and acamprosate as first-line therapies for alcohol use disorder.

Importance  Alcohol use disorder affects more than 28.3 million people in the United States and is associated with increased rates of morbidity and mortality.

Objective  To compare efficacy and comparative efficacy of therapies for alcohol use disorder.

Data Sources  PubMed, the Cochrane Library, the Cochrane Central Trials Registry, PsycINFO, CINAHL, and EMBASE were searched from November 2012 to September 9, 2022 Literature was subsequently systematically monitored to identify relevant articles up to August 14, 2023, and the PubMed search was updated on August 14, 2023.

Study Selection  For efficacy outcomes, randomized clinical trials of at least 12 weeks’ duration were included. For adverse effects, randomized clinical trials and prospective cohort studies that compared drug therapies and reported health outcomes or harms were included.

Data Extraction and Synthesis  Two reviewers evaluated each study, assessed risk of bias, and graded strength of evidence. Meta-analyses used random-effects models. Numbers needed to treat were calculated for medications with at least moderate strength of evidence for benefit.

Main Outcomes and Measures  The primary outcome was alcohol consumption. Secondary outcomes were motor vehicle crashes, injuries, quality of life, function, mortality, and harms.

Results  Data from 118 clinical trials and 20 976 participants were included. The numbers needed to treat to prevent 1 person from returning to any drinking were 11 (95% CI, 1-32) for acamprosate and 18 (95% CI, 4-32) for oral naltrexone at a dose of 50 mg/d. Compared with placebo, oral naltrexone (50 mg/d) was associated with lower rates of return to heavy drinking, with a number needed to treat of 11 (95% CI, 5-41). Injectable naltrexone was associated with fewer drinking days over the 30-day treatment period (weighted mean difference, −4.99 days; 95% CI, −9.49 to −0.49 days) Adverse effects included higher gastrointestinal distress for acamprosate (diarrhea: risk ratio, 1.58; 95% CI, 1.27-1.97) and naltrexone (nausea: risk ratio, 1.73; 95% CI, 1.51-1.98; vomiting: risk ratio, 1.53; 95% CI, 1.23-1.91) compared with placebo.

Conclusions and Relevance  In conjunction with psychosocial interventions, these findings support the use of oral naltrexone at 50 mg/d and acamprosate as first-line pharmacotherapies for alcohol use disorder.

Unhealthy alcohol use is the third leading preventable cause of death in the United States, accounting for 145 000 deaths annually.¹ Data from the 2020 National Survey on Drug Use and Health suggested that more than 28.3 million people aged 12 years or older in the United States met Diagnostic and Statistical Manual of Mental Disorders (Fifth Edition) (DSM-5) criteria for alcohol use disorder (eTable 1 in Supplement 1) in the past year.^2,3 The COVID-19 pandemic may have been associated with increased numbers of people with alcohol use disorder.^2,3 Among the 29.5 million people reporting a past-year alcohol use disorder in 2021, an estimated 0.9%, or 265 000 people, received pharmacotherapy for alcohol use disorder.⁴

This systematic review and meta-analysis evaluated efficacy and comparative efficacy of 9 therapies for alcohol use disorder that are either approved by the US Food and Drug Administration (FDA) (eTable 2 in Supplement 1) or more commonly used in the United States for alcohol use disorder.

The protocol was registered with PROSPERO (CRD42022324376). A full technical report that addressed 5 questions (eTable 3 in Supplement 1) details methods, search strategies, and additional information.

PubMed, the Cochrane Library, the Cochrane Central Trials Registry, PsycINFO, CINAHL, and EMBASE were searched for English-language studies of adults aged 18 years or older from November 1, 2012, to September 9, 2022; eligible articles published before these searches were obtained from a previously published (2014) systematic review on this topic.^5,6 A librarian (C.V.) performed all searches. A second librarian peer-reviewed the searches using the validated Peer Review of Electronic Search Strategies (PRESS) checklist.⁷ Reference lists of pertinent reviews and trials were manually searched for additional relevant citations. After September 9, 2022, an ongoing systematic monitoring of the literature was conducted through article alerts. An updated search of PubMed was conducted on August 14, 2023, to identify studies published since that may affect the conclusions or understanding of the evidence; those searches did not identify new studies for inclusion.

Studies that enrolled adults with alcohol use disorder and evaluated an FDA-approved medication (acamprosate, disulfiram, or naltrexone) or any of 6 off-label medications (baclofen, gabapentin, varenicline, topiramate, prazosin, and ondansetron) for at least 12 weeks of treatment in an outpatient setting were eligible for inclusion. Quiz Ref IDTwelve weeks of treatment were required because longitudinal studies reported that shorter treatment may yield misleading conclusions about efficacy due to fluctuations in drinking behavior. Eligible studies were required to assess 1 of the following outcomes: (1) alcohol consumption, consisting of return to any drinking, return to heavy drinking, percentage of drinking days, percentage of heavy drinking days (≥4 drinks per day for women; ≥5 drinks per day for men), or number of drinks per drinking day; (2) health outcomes—motor vehicle crashes, injuries, quality of life, function, or mortality; or (3) adverse events.

For efficacy outcomes, double-blind randomized clinical trials (RCTs) that compared 1 of the FDA-approved or off-label medications listed above with placebo or with another medication were eligible for inclusion. For adverse effects, in addition to the double-blind RCTs included for efficacy, studies with the following designs were eligible if they compared 2 drugs of interest: nonrandomized or open-label trials, subgroup analyses from trials, prospective cohort studies, and case-control studies. Nonrandomized and observational studies were included to address harms because RCTs had insufficient sample sizes and duration to identify rare harms.

Two investigators independently reviewed each title and abstract. Studies marked for possible inclusion by either reviewer underwent independent full-text review by 2 reviewers. If the reviewers disagreed, they resolved conflicts by discussion and consensus or by consulting a third, senior member of the team.

Structured data extraction forms were used to gather relevant data from each article. At least 2 investigators reviewed all data extractions for completeness and accuracy.

To assess the risk of bias of studies, the investigators used predefined criteria based on established guidance.^8-10 The studies were rated as having low, medium, high, or unclear risk of bias.^8,9 Questions were included about adequacy of randomization, allocation concealment, similarity of groups at baseline, masking, attrition, validity and reliability of measures, approaches to analyses, and methods of handling missing data. Two independent reviewers assessed risk of bias for each study. Disagreements were resolved by consensus.

The strength of evidence was graded as high, moderate, low, or insufficient based on established guidance.¹¹ The approach incorporated 4 key domains: risk of bias, consistency, directness, and precision. Two reviewers assessed each domain for each outcome and determined an overall grade. Differences were resolved by consensus.

In these analyses, results are presented for medications for which there was at least low strength of evidence for benefit for some outcomes.

The primary outcome was alcohol consumption, defined as any alcohol use, return to heavy drinking, and number of drinks per week. Meta-analyses of RCTs were performed using random-effects models.¹² We used the DerSimonian and Laird estimator for our primary analyses, with sensitivity analyses using a restricted maximum likelihood model when the pooled effects were statistically significant. For continuous outcomes, weighted mean differences (WMDs) and 95% CIs were calculated. For binary outcomes, risk ratios (RRs) between groups and 95% CIs were calculated. The I² statistic was calculated to assess statistical heterogeneity.^13,14 Potential sources of heterogeneity were examined by analyzing subgroups defined by patient population (eg, US vs non-US studies). Analyses were conducted using Stata version BE-17 (StataCorp). Statistical significance was assumed when 95% CIs of pooled results did not cross 0. All testing was 2-sided. Numbers needed to treat were calculated when pooled RRs for binary outcomes found a statistically significant result and there was at least moderate strength of evidence for benefit. When quantitative synthesis was not appropriate (eg, <2 similar studies), the data were synthesized qualitatively.

The database search identified 2860 citations, and 2543 citations were excluded during title and abstract review. Of 317 full-text articles included after title and abstract review, 267 were excluded, leaving 156 articles that described results of 118 RCTs (Figure 1). Of these, 81 RCTs (106 articles) were included in the 2014 systematic review on this topic,⁵ and 37 RCTs (50 articles) were new. No observational studies providing data on adverse effects were identified, and therefore all data on adverse events were obtained from RCTs.

Characteristics of the 37 RCTs that were new since 2104 are shown in eTable 4 in Supplement 1. Sample sizes ranged from 12 to 921. Treatment duration ranged from 12 to 52 weeks. All participants met criteria for alcohol dependence in 103 of 118 of the clinical trials. Recruitment methods varied and included treatment programs, advertisements, referrals, or a combination. Eighty-seven (73.7%) of 118 studies included psychosocial co-interventions. For these studies, effect sizes reflect the benefits of medications added to psychosocial interventions compared with placebo added to psychosocial interventions. Of 23 studies that assessed efficacy of acamprosate, 16 were conducted in Europe and 4 were conducted in the United States. Of 49 studies of naltrexone, 32 were conducted in the United States and 8 were conducted in Europe. Of the 118 included studies, 100 included a co-intervention such as medical management, specific harm reduction, or counseling approaches.

Three medications (ondansetron, varenicline, and prazosin) had either low strength of evidence suggesting benefit or insufficient evidence and are not further discussed (eTable 5 in Supplement 1).

Quiz Ref IDAmong the medications with an FDA indication for alcohol use disorder, acamprosate and naltrexone were associated with statistically significant improvement in alcohol consumption outcomes (Table, Figure 2, Figure 3, Figure 4, Figure 5, and Figure 6; eAppendix in Supplement 1).^15-66 Compared with placebo, numbers needed to treat to prevent 1 person from returning to any drinking were 11 (95% CI, 1-32; 20 trials; n = 6380) for acamprosate and 18 (95% CI, 4-32; 16 trials; n = 2347) for oral naltrexone (50 mg/d), respectively. There was no significant difference in return to heavy drinking between acamprosate and placebo (RR, 0.99; 95% CI, 0.94-1.05; P = .69; range, 41.9%-81.5% with acamprosate, 45.8%-82.9% with placebo). Compared with placebo, oral naltrexone (50 mg/d) was associated with a statistically significant improvement in return to heavy drinking (RR, 0.81; 95% CI, 0.72-0.90; P < .001; range, 14.3%-94.6% with naltrexone, 29.7%-93.5% with placebo) with a number needed to treat of 11 (95% CI, 5-41; 19 trials; n = 2875). Compared with placebo, injectable naltrexone was not associated with lower rates of return to any drinking (RR, 0.96; 95% CI, 0.90-1.03; P = .14; 2 trials; n = 939; range, 82.3%-93.5% with naltrexone, 89.8%-94.7% with placebo) or return to heavy drinking (RR, 1.00; 95% CI, 0.82-1.21; P = .09; 2 trials; n = 615; range, 59.2%-77.2% with naltrexone, 52.7%-84.1% with placebo). Compared with placebo, injectable naltrexone was associated with greater reduction in percentage of drinking days (WMD, −4.99; 95% CI, −9.49 to −0.49; P = .23; 2 trials; n = 467) and percentage of heavy drinking days (WMD, −4.7; 95% CI, −8.6 to −0.73; P = .80; 3 trials; n = 956). Data from 3 RCTs that included 622 participants did not show an association of disulfiram compared with placebo for preventing return to any drinking (RR, 1.03; 95% CI, 0.90-1.17; P = .28; range, 22.7%-81.2% with disulfiram, 34.4%-88.1% with placebo) (Table).

Among medications without an FDA indication for alcohol use disorder treatment, compared with placebo, topiramate was associated with statistically significant improvement in the weighted mean of absolute percentage of drinking days (WMD, −7.2; 95% CI, −14.3 to −0.1; P = .14; range, 5.5%-62.4% with topiramate, 6.4%-70.9%), percentage of heavy drinking days (WMD, −6.2; 95% CI, −10.9 to −1.4; P = .32; range, 2.3%-43.8% with topiramate, 5.3%-51.8% with placebo), and number of drinks per drinking day (WMD, −2.0; 95% CI, −3.1 to −1.0; P = .19; range, 1.2-6.5 with topiramate, 4.0-8.8 with placebo). These findings were associated with moderate strength of evidence. Of 13 double-blind placebo-controlled RCTs that included 1607 participants, compared with placebo, baclofen was associated with significantly lower rates of return to any drinking (RR, 0.83; 95% CI, 0.70-0.98; P < .001; range, 28.6%-92.4% with baclofen, 53.2%-89.9% with placebo). Because of imprecision of the effect estimate and inconsistency of results, baclofen data were graded as having low strength of evidence. Compared with placebo, gabapentin was not significantly associated with lower rates of return to any drinking (RR, 0.92; 95% CI, 0.83-1.02: P = .08; range, 79.5-86.1 with gabapentin, 88.2-95.9 with placebo) or with significant reduction in return to heavy drinking (RR, 0.90; 95% CI, 0.82-0.98; P = .75; range, 63.4-75.9 with gabapentin, 77.6-87.0 with placebo), but both results had low strength of evidence and only 3 clinical trials reported these outcomes.

A meta-analysis of 4 RCTs including 1141 participants that directly compared acamprosate with naltrexone^19,37,45,48 found no statistically significant difference between the 2 medications for improvement in alcohol use outcomes consisting of return to any drinking (RR, 1.03; 95% CI, 0.96-1.10; P = .57; range, 75.0-80.5 with acamprosate, 65.0-83.0 with naltrexone; 3 trials; n = 800) or return to heavy drinking (RR, 1.02; 95% CI, 0.93-1.11; P = .65; range, 50.0-72.7 with acamprosate, 50.9-73.6 with naltrexone; 4 trials; n = 1141).

Quiz Ref IDThere was insufficient evidence from RCTs to assess whether treatment with most medications was associated with improved health outcomes. Outcomes such as motor vehicle crashes, injuries, quality of life, function, and mortality were infrequently reported in the included studies (Table).

Adverse event data were often not collected using standardized measures, and methods for systematically capturing adverse events were often not reported (Figure 7).

Among medications with at least some (low) strength of evidence for benefit in any outcome, compared with placebo, dizziness was the most common mild adverse effect across medications and was reported with naltrexone (RR, 1.99; 95% CI, 1.47-2.69; P = .37; range, 2.9%-34.8% with naltrexone, 0.0%-20.6% with placebo), baclofen (RR, 1.89; 95% CI, 1.40-2.55; P = .40; range, 4.8%-30.2% with baclofen, 0.0%-22.8% with placebo), topiramate (RR, 2.29; 95% CI, 1.39-3.78; P = .65; range, 0.0%-28.0% with topiramate, 1.9%-10.7% with placebo), and gabapentin (RR, 1.70; 95% CI, 1.24-2.32; P = .83; range, 6.5%-7.8% with gabapentin, 3.8%-6.0% with placebo). Compared with placebo, any gastrointestinal distress was more common for acamprosate (diarrhea: RR, 1.58; 95% CI, 1.27-1.97; P = .03; range, 3.0%-63.7% with acamprosate, 1.6%-64.9% with placebo) and naltrexone (nausea: RR, 1.73; 95% CI, 1.51-1.98; P = .19; range, 2.5%-57.6% with naltrexone, 0.0%-47.1% with placebo; vomiting: RR, 1.53; 95% CI, 1.23-1.91; P = .79; range, 0.0%-25.6% with naltrexone, 0.0%-23.4% with placebo). Compared with placebo, baclofen was associated with higher rates of drowsiness (RR, 1.46; 95% CI, 1.15-1.86; P = .28; range, 6.3%-50.0% with baclofen, 9.4%-32.6% with placebo), numbness (RR, 7.78; 95% CI, 1.42-42.56; P = .48; range, 7.1%-12.6% with baclofen, 0.0%-1.1% with placebo), and sleepiness (RR, 1.81; 95% CI, 1.11-2.97; P = .77; range, 2.4%-36.2% with baclofen, 0.0%-17.7% with placebo). Compared with placebo, topiramate was associated with higher risks of many adverse events, including paresthesias (RR, 3.08; 95% CI, 2.11-4.49; P = .06; range, 0.0%-57.3% with topiramate, 1.9%-29.4% with placebo), taste abnormalities (RR, 3.01; 95% CI, 1.70-5.34; P = .04; range, 15.1%-53.3% with topiramate, 4.8%-31.3% with placebo), and cognitive dysfunction (RR, 2.37; 95% CI, 1.58-3.55; P = .48; range, 12.6%-23.9% with topiramate, 5.4%-11.3% with placebo). Compared with placebo, gabapentin was associated with cognitive dysfunction (RR, 2.76; 95% CI, 1.51-5.06; P = .37; range, 5.9%-25.5% with gabapentin, 5.7%-17% with placebo) and dizziness (RR, 1.70; 95% CI, 1.24-2.32; P = .83; range, 21.2%-56.8% with gabapentin, 13.7%-32.6% with placebo). In direct comparisons of acamprosate and oral naltrexone in RCTs, patients treated with acamprosate had lower rates of nausea (RR, 0.56; 95% CI, 0.35-0.88; P = .11; range, 3.8%-23.8% with acamprosate, 2.5%-55.6% with naltrexone) and vomiting (RR, 0.60; 95% CI, 0.39-0.93; P = .88; range, 8.9%-11.1% with acamprosate, 14.6%-22.2% with naltrexone) compared with those treated with naltrexone.

In this systematic review and meta-analysis that included 118 clinical trials, the highest strength of evidence for treatment of alcohol use disorder was available for acamprosate and oral naltrexone (50 mg/d). Randomized clinical trials that directly compared naltrexone, 50 mg/d, with acamprosate did not consistently established superiority of either medication. Studies of naltrexone had moderate strength of evidence for reducing return to any drinking, return to heavy drinking, percentage of drinking days, and percentage of heavy drinking days at the 50-mg/d oral dose compared with placebo. Fewer data were available for the 100-mg/d oral and injectable doses. Studies of acamprosate showed moderate strength of evidence for significant reduction in return to any drinking and reduction in drinking days compared with placebo. Acamprosate was not associated with benefit for return to heavy drinking (moderate strength of evidence).

Oral naltrexone is more convenient than acamprosate, requiring a single daily dose, whereas acamprosate is typically prescribed as 2 tablets administered 3 times daily. Acamprosate is contraindicated for people with severe kidney impairment and requires dose adjustments for moderate kidney impairment. Oral naltrexone is contraindicated for patients with acute hepatitis or liver failure and for those using opioids or who have anticipated need for opioids. Naltrexone can precipitate severe withdrawal for patients dependent on opioid medications.

Disulfiram has been FDA approved for alcohol use disorder since the 1950s. However, relatively limited evidence exists to support the efficacy of disulfiram compared with placebo for preventing return to any drinking or other alcohol consumption outcomes. Four RCTs of disulfiram have been published that were not eligible for this review because of their trial designs and comparisons.^67-70 These small trials (with 15 or fewer disulfiram-treated patients in each) had limitations that included a small sample size and inability to distinguish between benefits from disulfiram and benefits of counseling or benefits from therapeutic relationships with the investigative team.^71,72

Among medications without FDA approval for alcohol use disorder, studies of topiramate compared with placebo had moderate strength of evidence for significant reductions in percentage of drinking days, percentage of heavy drinking days, and drinks per drinking days. However, topiramate was associated with adverse effects that included cognitive dysfunction, dizziness, numbness and/or tingling, and taste abnormalities. Studies of baclofen and gabapentin had low strength of evidence for benefit in at least 1 outcome. Evidence was largely insufficient or low for benefit on health outcomes, including quality of life, motor vehicle crashes, and mortality.

Alcohol use disorder is associated with numerous health problems, including but not limited to hypertension, heart disease, stroke, cognitive impairment, sleep problems, depression, anxiety, peripheral neuropathy, gastritis and gastric ulcers, liver disease including cirrhosis, pancreatitis, osteoporosis, anemia, fetal alcohol spectrum disorders, and several types of cancer.^73,74 Excessive alcohol consumption is also associated with higher rates of homicide, suicide, motor vehicle crashes and deaths, sexual violence, domestic violence, and drownings.⁷⁵

Using DSM-5 criteria, most participants in the included studies likely had moderate to severe alcohol use disorder. Thus, applicability of the findings to people with mild alcohol use disorder is uncertain. The mean age of participants was typically between 40 and 49 years, with only 21 studies enrolling younger or older populations. Thus, it is uncertain whether the medications have similar efficacy for older (eg, aged ≥65 years) or younger (eg, aged in their 20s) people. Of the 70 studies that provided data on race and sex, most (n = 63) included a majority of White male participants, and none specified sex other than male or female. Because 100 of 118 clinical trials studied drug therapy combined with a nonmedication treatment (such as counseling), results reflect benefits from a combination of medication and cotherapy compared with placebo and cotherapy.

Of the 5 studies of acamprosate that were conducted in the United States, most reported no significant benefit either for return to any drinking or return to heavy drinking. Clinical trials conducted in the United States recruited patients largely through advertisements, while 15 of 22 clinical trials in other countries recruited participants from inpatient settings, where patients may have undergone alcohol withdrawal and medications may have been initiated before discharge. Patients recruited in the clinical trials conducted in the United States may have represented a more general population with a larger range of alcohol use at baseline. Thus, the lack of efficacy in US-based trials for acamprosate may reflect differences in patient characteristics and differences in the health care systems compared with clinical trials from other countries.

Most studies required patients to abstain for at least a few days before initiating medication, and the medications were generally recommended for maintenance of abstinence. Acamprosate and injectable naltrexone are FDA approved only for use in patients who have established abstinence, although the duration of required abstinence is not established. Three studies enrolling patients who were not yet abstinent reported reduction in heavy drinking with naltrexone compared with placebo^30,76 or acamprosate compared with placebo.³³

This review has several limitations. First, clinical trials with less than 12 weeks of follow-up from the time of medication initiation were excluded. Second, the meta-analysis combined studies of participants with diagnoses of both alcohol dependence and depression and studies of participants without both alcohol dependence and depression. Third, studies may have selectively reported outcomes. Fourth, long-term information about adverse effects was not available. Fifth, for adverse event outcomes, due to small sample sizes and relatively small numbers of events, evidence was often insufficient to determine whether adverse event outcomes were increased. Sixth, in some included studies, less than 100% of participants had alcohol use disorder. Specifically, 3 studies reported that less than 90% of participants had alcohol use disorder.^24,77,78

In conjunction with psychosocial interventions, these findings support the use of oral naltrexone, 50 mg/d, and acamprosate as first-line pharmacotherapies for alcohol use disorder.

Corresponding Author: Melissa McPheeters, PhD, MPH, RTI International, 3040 E Cornwallis Rd, Research Triangle Park, NC 27709 (mmcpheeters@rti.org).

Accepted for Publication: September 12, 2023.

Author Contributions: Dr McPheeters had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: McPheeters, O’Connor, Edlund, Bobashev, Jonas.

Acquisition, analysis, or interpretation of data: McPheeters, O’Connor, Riley, Kennedy, Voisin, Kuznacic, Coffey, Bobashev, Jonas.

Drafting of the manuscript: McPheeters, O’Connor, Riley, Kennedy, Kuznacic, Coffey, Jonas.

Critical review of the manuscript for important intellectual content: McPheeters, O’Connor, Voisin, Edlund, Bobashev, Jonas.

Statistical analysis: McPheeters, O’Connor, Riley, Jonas.

Obtained funding: McPheeters, Jonas.

Administrative, technical, or material support: Kennedy, Voisin, Bobashev, Jonas.

Supervision: McPheeters, Jonas.

Conflict of Interest Disclosures: None reported.

Funding/Support: This project was funded under contract 75Q80120D00007, task order 75Q80122F32004 from the Agency for Healthcare Research and Quality (AHRQ) of the US Department of Health and Human Services.

Role of the Funder/Sponsor: This topic was nominated by the AHRQ program official for EvidenceNow: Managing Unhealthy Alcohol Use Initiative and selected by AHRQ for systematic review by an evidence-based practice center. A representative from AHRQ served as a contracting officer’s representative and provide technical assistance during the conduct of the full evidence report and provided comments on draft versions of the full evidence report. AHRQ did not directly participate in the literature search, determination of study eligibility criteria, data analysis or interpretation, or preparation, review, or approval of the manuscript for publication.

Disclaimer: The authors of this article are responsible for its content. Statements in the article do not necessarily represent the official views of or imply endorsement by AHRQ or the US Department of Health and Human Services. AHRQ retains a license to display, reproduce, and distribute the data and the report from which this manuscript was derived under the terms of the agency’s contract with the author.

Data Sharing Statement: See Supplement 2.

Additional Contributions: We gratefully acknowledge the following individuals for their contributions to this project, none of whom received compensation: from AHRQ: Meghan Wagner, PharmD, AHRQ Task Order Officer Elisabeth Kato, MD, MRP, and Cleo Alford, MPS, MSc; from the American Psychiatric Association: Laura Fochtmann, MD, and Jennifer Medicus; from the American Society of Addiction Medicine: Anna Pagano, PhD, and Ray Denny, PhD; from RTI International–University of North Carolina at Chapel Hill Evidence-Based Practice Center (for administrative support, review, and/or editing): Roberta Wines, MPH, Carol Woodell, BSPH, Nila Sathe, MA, MLIS, Sharon Barrell, MA, Mary Gendron, and Teyonna Downing.