Reductions in World Health Organization Risk Drinking Levels as a Primary Efficacy Endpoint for Alcohol Clinical Trials: A Review

Abstract

Importance:

Alcohol use disorder (AUD) is a highly prevalent and costly psychiatric disorder. Abstinence has been considered the optimal outcome of treatment for AUD. Yet, most individuals with AUD do not seek treatment because they do not have a goal of abstinence. The Food and Drug Administration (FDA) has recently qualified reductions in drinking, defined by at least a two level reduction in the World Health Organization risk drinking levels (WHO RDLs), as another primary endpoint for alcohol pharmacotherapy trials. The approval of drinking reductions as an endpoint for alcohol clinical trials aligns with an accumulating literature on drinking reductions in the alcohol field. The current paper provides a narrative review of 34 papers that have examined WHO RDLs as a “surrogate marker” of how people with AUD feel and function.

Observations:

Results from epidemiological studies, community samples, and clinical trials indicate that drinking reductions are associated with improvements in how patients feel and function, including reduced risk of substance use disorder, medical and psychiatric diseases, and reductions in alcohol-related consequences, craving, and health care costs. Drinking reductions are also associated with improvements in functioning and quality of life. Drinking reductions are also achieved by most clinical trials participants, and effect sizes for the WHO RDL reductions for active medications versus placebo are similar to alternative endpoints.

Conclusions and Relevance:

The FDA acceptance of reduction in WHO RDLs as another primary endpoint for alcohol clinical trials may increase opportunities for AUD medications development, encourage patients to seek treatments that target drinking reductions, and engage providers in prescribing medications shown to be effective in supporting drinking reductions. The WHO RDLs may be particularly useful for targeted drinking reductions in clinical practice. Qualification of the WHO RDL endpoint facilitates a paradigm shift towards a harm reduction approach in AUD treatment.

Alcohol use disorder (AUD) is a prevalent and costly psychiatric disorder, with estimates of lifetime rates of AUD of 29% in the United States and at least 8% globally¹, and economic costs of alcohol consumption estimated at 2.6% of the global gross domestic product². Most individuals with AUD never seek treatment³, one reason for which is reluctance to fully abstain from alcohol. Although abstinence has typically been used to define success in AUD treatment, fewer than one-third of individuals achieve alcohol abstinence in AUD treatment⁴.

The Food and Drug Administration (FDA) has long utilized abstinence as a responder-based primary surrogate endpoint for improvement in how a patient “feels and functions” after treatment in alcohol clinical trials. In 2006, FDA approved long acting naltrexone (Vivitrol^®) based on the event rate of heavy drinking days (defined as ≥4 drinks for females and 5 drinks for males per occasion). In 2015, FDA guidance⁵ recommended two primary endpoints: abstinence or no heavy drinking days. However, using these drinking endpoints, relatively few patients are defined as “successfully treated.” The validity of abstinence and heavy drinking cutoffs has been questioned^6–8, especially since neither drinking endpoint captures the large reductions in drinking often observed in alcohol treatment trials^4,6,9,10.

In 2025, using the World Health Organization (WHO) Risk Drinking Levels (RDLs) metric (Box 1), the FDA qualified a reduction of drinking by at least two levels¹¹ as a primary endpoint for alcohol clinical trials¹², which is also an approved intermediate drinking harm reduction endpoint in the European Medicines Agency (EMA) regulatory guidance¹³. These approvals of drinking reductions as a drinking endpoint that validly indicates successful treatment align with an accumulating literature on drinking reductions in the alcohol field for which this article provides a brief overview. We then provide a narrative review of papers that examined WHO RDLs as a surrogate marker of how people with AUD feel and function and studies that examined reductions in WHO RDLs as an endpoint for alcohol clinical trials.

Box 1: World Health Organization risk drinking levels

	World Health Organization Risk Drinking Levels (for males)
	No risk	Low risk	Medium risk	High risk	Very high risk
Drinks per day (in grams)	0	1–40	41–60	61–100	≥ 101
Drinks per day (approximate U.S. standard drinks, 14 g)	0	<2.9	2.9-<4.3	4.3–7.1	≥ 7.1
Drinks per week (in grams)	0	1–280	281–420	421–700	≥701
Drinks per week (approximate U.S. standard drinks, 14 g)	0	<=20	21–30	31–50	≥51
	World Health Organization Risk Drinking Levels (for females)
	No risk	Low risk	Medium risk	High risk	Very high risk
Drinks per day (in grams)	0	1–20	21–40	41–60	≥61
Drinks per day (approximate U.S. standard drinks, 14 g)	0	<1.4	1.4-<2.9	2.9-<4.3	≥4.3
Drinks per week (in grams)	0	1–140	141–280	281–420	>420
Drinks per week (approximate U.S. standard drinks, 14 g)	0	<10	11–20	21–30	≥31

Drinking Reductions as an Outcome and Target in Alcohol Clinical Trials

Drinking reductions have been studied in the alcohol field for over 50 years¹⁴, pioneered by the development of a behavioral treatment targeting reduced drinking^15,16, with drinking reductions assessed using the Timeline Follow-Back method¹⁷. The Timeline Follow-Back is a calendar-based interview that quantifies the amount of alcohol consumed and the frequency of drinking on a daily basis, which can also be assessed using daily diaries or other mobile health tools^18,19. The Timeline Follow-Back remains a widely-used standard assessment of self-reported alcohol use, accepted by the FDA, that can be used to measure a range of different drinking endpoints for alcohol clinical trials, including the WHO risk drinking levels.

Many researchers and clinicians have since targeted drinking reductions²⁰ and examined continuous drinking reductions (e.g., reductions in percent heavy drinking days) as alternatives to the FDA binary responder definitions of abstinence or no heavy drinking days²¹. The European EMA guidance for alcohol clinical trials¹³ recommends several harm reduction endpoints, including reductions in total alcohol consumption, reductions in heavy drinking days, and a reduction in WHO RDL of at least two levels as an intermediate endpoint. The regulatory approvals of nalmefene for AUD in both Europe^22,23 and Japan^24,25 were based on reductions in drinking as primary endpoints.

Despite the inclusion of the WHO RDLs in the EMA guidance¹³, until recently, very few studies^23,26,27 have validated the WHO RDLs, related them to improvement in social or physical functioning, or examined them in the context of alcohol clinical trials. Thus, in 2015, the question of whether the WHO RDLs could validly be used as an endpoint for alcohol clinical trials became the focus of the Alcohol Clinical Trials Initiative (ACTIVE)^28,29, a public-private partnership of academics, the FDA, the National Institute on Alcohol Abuse and Alcoholism (NIAAA), and pharmaceutical companies. The ACTIVE group has since published 16 papers examining the WHO RDLs. This article reviews these papers and additional research that has examined reductions in WHO RDLs in association with how individuals feel and function, or as an endpoint in alcohol clinical trials.

METHODS

This report presents a narrative review of papers that examined WHO RDLs published by the ACTIVE group or identified based on a literature search conducted on February 19, 2025, for empirical articles using the PubMed database with the search terms “World Health Organization,” risk, alcohol, drinking, and level. The search identified a total of 229 papers. We focused on empirical papers in English, not including reviews, case reports, or dissertations, that accomplished one of two goals: (1) examined the validity of the WHO RDLs as a surrogate marker of how people feel and function; and (2) examined the WHO RDLs as an endpoint (i.e., primary or secondary outcome) in an alcohol clinical trial examining a pharmacotherapy. Of the 229 papers identified for the current review, we excluded 195 (Figure 1). Thus, the current review focused on 34 papers that examined the WHO RDLs in association with how individuals feel and function (n=27) or examined the WHO RDLs as an outcome in alcohol clinical trials (n=7).

Figure 1.

PRISMA flowchart for review

RESULTS

WHO Risk Drinking Level Associations with Patient Functioning

Studies that examined the WHO RDLs with respect to how individuals feel and function (n=27) were further subdivided into epidemiological studies (n=8) and studies that examined the WHO RDLs in community or clinical settings (n=19). A summary of the constructs studied and findings is included in Box 2.

Box 2: Validation of WHO risk drinking levels in epidemiological and clinical trial data, where “X” indicates significant associations or non-significant (“NS”) findings between external validators in column headers and WHO risk drinking levels

Reference (sample size)	AUD criteria / diagnosis	Alcohol-related consequences	Drug diagnosis	Psychiatric symptoms / diagnosis	Medical symptoms / disease	Functioning / Quality of life	Biomarkers / Brain-based metrics
Hasin (n=22,005)32	X
Knox (n=21,925)33					X
Knox (n=1,058)34					X
Knox (n=22,005)35			X
Knox (n=22,005)36				X
Shmulewitz (n=25,659)37	X		X	X	X	X
Barbosa (n=36,309)38						X
Niemelä (n=22,327)39							X
Conigrave (n=1863)26							X
Witkiewitz (n=1142)44							X
Durazzo (n=75)40							X
May (n=68)41							X
May (n=85)42							X
Witkiewitz (n=1383)45		X		X		X
Witkiewitz (n=1226)46		X		X		X	X
Witkiewitz (n=2125)47		X		X		X	X
Witkiewitz (n=1500)48		X		X		X
Aldridge (n=964)49						X (health costs)
Garcia (n=346)55						X (sleep)
Tuchman (n=1873)53						X (craving)
Richards (n=346)54	X
Mejldal (n=693)56	X	X				X
Reid (n=1436)57		X		NS		X
Roerecke (n=641)50						X (mortality)
Evon (n=162)58						X (motivation)
Mitchell (n=297)59				X		X
Stevenson (n=241)60	X	NS		NS		NS

Note. NS = non-significant finding, X = significant association between WHO risk-level reductions and column header

Epidemiological studies.

Seven of the eight epidemiological studies were conducted using data from the U.S. National Epidemiological Survey of Alcohol and Related Conditions (NESARC) surveys^30,31, four of which focused on change in RDL and change in disease prevalence. Hasin and the ACTIVE group³² examined WHO RDLs in the NESARC wave 1 (2001–2002) to wave 2 (2004–2005) surveys, and found reductions in WHO RDLs across waves were associated with significantly lower odds of alcohol dependence diagnosis at wave 2. Knox and the ACTIVE group followed up these analyses by showing that reductions in WHO RDLs were also associated with lower odds of liver disease³³, cardiovascular disease³⁴, drug use disorder diagnosis³⁵, and depression and anxiety disorder diagnoses³⁶ at wave 2.

Shmulewitz and the ACTIVE group conducted an analysis of the NESARC-III data (2012 2013) and found that prevalence of AUD, drug use disorder, liver disease, and any depression or anxiety disorder diagnosis were significantly higher among individuals drinking at higher WHO RDLs³⁷. Shmulewitz and the ACTIVE group also examined functional impairment in NESARC wave 1 and NESARC-III and found that functional impairment was significantly greater among high and very high risk drinkers, as compared to low risk drinkers, in both surveys³⁷. Using the NESARC-III data, Barbosa and colleagues³⁸ also examined continuous health-related quality of life and found lower average health-related quality of life at higher WHO RDLs. A large nationally representative study³⁹ conducted in Finland found that individuals with higher WHO RDLs had greater gamma glutamyltransferase (GGT), alanine aminotransferase (ALT), C-reactive protein (CRP), and lower serum lipid levels than abstainers.

Community/Clinical studies.

In the first study of WHO RDLs, Conigrave and colleagues²⁶ found that WHO RDLs were significantly associated with alcohol biomarkers in a large sample of individuals recruited from community treatment programs in five countries (Australia, Brazil, Canada, Finland, Japan). Three studies that examined WHO RDLs and neuroimaging markers of brain health or function found drinking at higher WHO RDLs was associated with smaller cortical volume^40,41 and lower N-acetylaspartate concentration levels⁴², as compared to abstainers and those drinking at lower risk drinking levels.

Several clinical studies used data from the COMBINE study⁴³, a large randomized clinical trial of medications for AUD conducted in the United States, to examine reductions in WHO RDLs and improvements in how individuals feel and function. Witkiewitz and the ACTIVE group⁴⁴ found that reductions in WHO RDLs were associated with significant improvements in alcohol biomarkers, including carbohydrate-deficient transferrin (CDT), GGT, aspartate aminotransferase (AST), as well as ALT and systolic blood pressure from baseline to the end of treatment in the COMBINE study. Additional analyses of COMBINE study data found that reductions in WHO RDLs were also associated with significant improvements in quality of life⁴⁴, mental health⁴⁵, and physical health⁴⁶, as well as reductions in alcohol-related consequences⁴⁵ at the end of treatment and up to one year following treatment. All of these studies found clinically meaningful improvements with at least 1- or 2-level reductions in WHO RDLs among those at high or very high-risk levels prior to treatment. These findings were replicated in data from a clinical trial conducted in the United Kingdom⁴⁷ and across three years following treatment in two separate studies conducted in the U.S.⁴⁸. Secondary analyses of the COMBINE study data by Aldridge and the ACTIVE group⁴⁹ found reductions in WHO RDLs during treatment were associated with 52% lower health care costs in the year following treatment and 44% lower health care costs over three years following treatment. Roerecke and colleagues⁵⁰ found reductions in WHO RDLs were associated with reduction in mortality risk.

The FDA qualification package that led to approval of the WHO RDLs as a primary endpoint for alcohol clinical trials¹² included several validation studies using data from the COMBINE study⁴³ and NIAAA sponsored trials examining varenicline⁵¹ and extended-release gabapentin⁵². The qualification package¹² provided further evidence that reductions in WHO RDLs were associated with improvements in mental health, quality of life, and alcohol-related consequences, as well as reductions in alcohol craving⁵³ and AUD criteria⁵⁴, as well as lower risk of AUD diagnosis⁵⁴. Using the extended release gabapentin dataset, an additional paper led by the ACTIVE group found reductions in WHO RDLs were associated with improvements in sleep quality⁵⁵. Consistent with prior work, these studies also found clinically meaningful improvements with at least 1- or at least 2-level reductions in WHO RDLs among those at high or very high-risk levels prior to treatment.

Several studies have examined the correspondence between WHO RDLs and functioning in the context of unique clinical trial populations. Mejldal and colleagues⁵⁶ studied reductions in WHO RDLs among older adults (60 years old+) who received psychotherapy and found reductions in WHO RDLs were significantly associated with reductions in alcohol-related consequences and AUD criteria, as well as significant improvements in quality of life. Reid and colleagues⁵⁷ examined WHO RDLs among college students receiving brief alcohol interventions and found that individuals at higher WHO RDLs had greater alcohol-related consequences and those at the very high-risk level had significantly worse school functioning.

Three studies examined the WHO RDLs in the context of nonpharmacological alcohol treatment trials for individuals with comorbid medical or psychiatric diseases. In the context of alcohol treatment for individuals with hepatitis C, Evon⁵⁸ found reductions in WHO RDLs were associated with significant improvements in motivation to change alcohol use, greater self-efficacy, and greater use of alcohol reduction strategies. Among adolescents with comorbid ADHD, Mitchell⁵⁹ showed that reductions in WHO RDLs were associated with improvements in global functioning and reductions in ADHD symptoms. Finally, in a study of individuals receiving treatment for AUD and comorbid anxiety disorder⁶⁰, achieving reductions in WHO RDLs predicted lower alcohol dependence symptom severity at the end of treatment.

WHO Risk Drinking Levels in Alcohol Pharmacotherapy Clinical Trials

Seven papers examined reductions in WHO risk drinking level as an endpoint in alcohol clinical trials testing a pharmacological treatment. For each trial we provide the original statistics reported in the paper, when reported, including the percent of individuals who achieved WHO risk level reductions, and differences between active medication and placebo using odds ratios (OR) or Cohen’s h, 95% confidence intervals (CI), and number needed to treat (NNT). A summary of the findings from each study is included in Box 3.

Box 3: Examining the WHO Risk Drinking Levels in Alcohol Clinical Trials

Medication
Reference (sample size)	Cohen’s h or Odds Ratio or % achieving reduction	NNT (vs placebo)
Nalmefene
Aubin (n=667)27	One-level: Not reported	NA
	Two-level: OR (95% CI) =1.87 (1.35, 20.59)	7
Castera (n=262)62	One-level: Not reported	NA
	Two-level: 55% achieved drinking reductions	NA
Naltrexone
Falk (n=590)63	One-level: Cohen’s h (95% CI)=0.12 (−0.05, 0.28)	23
	Two-level: Cohen’s h (95% CI)=0.21 (0.05, 0.38)	11
Topiramate
Falk (n=364)63	One-level: Cohen’s h (95% CI)=0.01 (−0.19, 0.22)	144
	Two-level: Cohen’s h (95% CI)=0.23 (0.02, 0.44)	9
Hartwell (n=308)65	One-level: (Cohen’s h (95% CI)=0.26 (0.03, 0.50)	7.8
	Two-level: Cohen’s h (95% CI)=0.19 (−0.04, 0.42)	14.5
Varenicline
Falk (n=200)63	One-level: Cohen’s h (95% CI)=0.34 (0.05, 0.62)	7
	Two-level: Cohen’s h (95% CI)=0.27 (−0.006, 0.55)	8
Nieto (n=165)66	One-level: Cohen’s h (95% CI)=0.24 (−0.06, 0.55)	8.6
	Two-level: Cohen’s h (95% CI)=−0.03 (−0.34, 0.27)	--
Gabapentin
Falk (n=346)52	One-level: OR (95% CI)=0.87 (0.48, 1.58)	90.9
	Two-level: OR (95% CI)=1.11 (0.69, 1.80)	30.3
Samidorphan
O’Malley (n=406)68	One-level: OR = 1.60 (95% CI Not reported)	5.6
	Two-level OR = 2.22 (95% CI Not reported)	5.6

Note. NNT = Number needed to treat; NA = information not reported to estimate NNT, -- = the placebo condition had a better response than medication.

Nalmefene.

Aubin and colleagues²⁷ conducted a secondary analysis of nalmefene trials^22,61 that examined nalmefene versus placebo among the subgroup of patients at high or very high risk WHO levels at baseline²³. A significantly greater odds of achieving a two-level reduction in WHO RDLs was found in the nalmefene condition (57% achieved), compared to placebo (42% achieved). Castera and colleagues⁶² also examined nalmefene in the context of an open-label trial in primary care, finding that 55% of patients reduced their drinking by at least two WHO RDLs over a three month study period.

Naltrexone, Topiramate, and Varenicline.

Falk and the ACTIVE group⁶³ led a secondary analysis study of reductions in WHO RDLs in three randomized clinical trials of naltrexone⁴³, topiramate⁶⁴, and varenicline⁵¹. In this study, reductions in WHO RDLs were examined as a responder endpoint compared with the FDA-recommended endpoints of abstinence and no heavy drinking days. A significantly greater proportion of patients on naltrexone (75%) versus placebo (65%) achieved at least a two-level reduction, a significantly greater proportion of patients on varenicline (70%) versus placebo (54%) achieved at least a one-level reduction, and a significantly greater proportion of patients on topiramate (45%) versus placebo (34%) achieved at least a two-level reduction. Importantly, reductions in WHO RDLs were as sensitive or more sensitive than FDA-recommended endpoints of abstinence and no heavy drinking days in detecting medication effects versus placebo. Response rates for abstinence and no heavy drinking days were much lower across conditions (naltrexone: abstinence 34.7%, no heavy drinking days 51%; varenicline: abstinence 7.3%, no heavy drinking days 24%; topiramate: abstinence 11.7%, no heavy drinking days 20.7%).

Two additional secondary analysis papers examined the effects of topiramate⁶⁵ and varenicline with or without naltrexone among heavy-drinking smokers⁶⁶ in predicting reductions in WHO RDLs following treatment. Results from the topiramate⁶⁵ study indicated significant effects of topiramate versus placebo in predicting at least a one-level reduction (topiramate: 50%, placebo: 37%), although topiramate was not significantly associated with at least a two-level reduction (topiramate: 18%, placebo: 12%). Results from the varenicline/naltrexone study⁶⁶ indicated that heavy-drinking smokers who received varenicline and naltrexone had greater, although not statistically significant, reductions in WHO RDLs over time relative to the varenicline and placebo conditions (varenicline/naltrexone: 42%, placebo: 30%). Importantly, the original report⁶⁷ also did not find any significant effects of varenicline/naltrexone versus varenicline and placebo on drinking outcomes, in essence confirming a negative study.

Gabapentin.

The FDA qualification package¹² included findings from one study of extended-release gabapentin⁵², which did not find effects of gabapentin versus placebo on reductions in WHO RDLs, or any of the other drinking endpoints. In that study, 52% achieved at least a two-level reduction on placebo, and 55% achieved the reduction on gabapentin.

Samidorphan.

Samidorphan, an opioid antagonist, was tested in a phase 2 multisite randomized clinical trial⁶⁸ for individuals with AUD. Results from an analysis of WHO RDL reductions indicated that generally, the samidorphan-treated groups had a greater proportion of patients who achieved at least a one- or two-level reduction across doses versus placebo. Patients with the 1 mg dose of samidorphan were significantly more likely to achieve at least a two-level reduction at a 12-week follow-up (samidorphan: 44%, placebo 26%).

DISCUSSION

Some data suggest that no safe level of drinking exists with respect to morbidity for most diseases and mortality, and that risk increases with higher levels of consumption^69–71. Likewise, decreased levels of drinking are monotonically associated with decreased levels of risk^10,72. The literature reviewed above indicates that drinking reductions are associated with improvements in how people feel and function across a range of physical, psychological, and functional domains. Specifically, drinking reductions are associated with reduced AUD criteria, lower risk of AUD, other drug use disorder, and psychiatric diagnoses, and reductions in alcohol-related consequences, psychiatric symptoms, medical symptoms, craving, and health care costs. There are also reported improvements in functioning, quality of life, motivation to change drinking, and sleep quality. Among those individuals entering treatment who were drinking at high or very high-risk levels, reducing by at least one- or two-WHO risk levels was associated with clinically meaningful improvements.

For individuals with AUD who seek alcohol treatment, the results from seven clinical trials indicate that WHO risk level reductions are achieved by most patients and a much larger proportion of patients achieved WHO risk level reductions than alternative FDA approved endpoints of abstinence and no heavy drinking days. Further, for many of the medications studied in this review, individuals who received active medications were more likely to achieve at least one- or two-risk level reductions than those randomized to placebo, and those studies that did not find significant medication effects using WHO risk level reductions also did not demonstrate significant effects of medications for other drinking outcomes (e.g., percent heavy drinking days, percent days abstinent). Across most trials, achieving at least a two-level reduction was more sensitive to medication effects than a one-level reduction.

A substantial treatment gap exists for AUD, whereby most individuals with AUD do not receive treatment and few receive medications⁴. Multiple reasons may underlie the treatment gap, including general perceptions of treatment as being solely abstinence-based, provider perceptions that medications are not effective, and the general lack of FDA-approved medications to treat AUD^4,10,28. The FDA acceptance of reduction in WHO RDLs as another primary endpoint for alcohol clinical trials¹⁰ is a paradigm shift that may open more opportunities for medications development. Further, more patients with AUD might become interested in treatments that target drinking reductions, and providers may become more interested in prescribing novel medications that have been shown to be highly effective in achieving drinking reductions. As such, these opportunities parallel medical conditions such as treatments for high blood pressure, excessive weight, diabetes, and many others where reductions, rather than cure, lead to long-lasting health and social benefits.

The WHO RDLs may be particularly useful in clinical practice. Patients’ drinking can be assessed and then categorized to provide clear feedback on current level of risk. This can be an opening to a conversation with patients about their preferred level of risk. Providers can also discuss the benefits of achieving even a one-level reduction, with potentially even greater benefits for reducing by at least two or more levels. WHO RDLs can be assessed and monitored using paper diaries, calendars, or mobile applications^73,74.

Future research examining reductions in WHO RDLs as an efficacy endpoint for clinical trials should consider several unresolved issues. First, the papers reviewed in this study and the FDA qualified endpoint assess the WHO RDLs based on grams per day (averaged across days). Taking an average across days acknowledges that drinking patterns are likely more indicative of how people feel and function over time than the amount of drinking on any one occasion. Subsequent research has examined WHO RDLs using both grams per day and per occasion drinking⁵⁷ and future studies should consider the population and outcomes of interest in examining WHO RDLs using grams per day versus per occasion. The second issue involves individuals who are already drinking at low or medium risk levels and cannot reduce by two or more levels. Future studies should study the validity of the one-level reduction endpoint for individuals who enter treatment at medium risk. The third issue is how the WHO RDLs compare to other continuous drinking outcome measures commonly reported in alcohol clinical trials (e.g., percent heavy drinking days) and how these measures may be more or less sensitive than WHO RDLs in relation to how patients feel and function, as well as medication effect sizes. It is often the case that continuous measures have more statistical power than binary outcomes⁷⁵, however it is unclear whether that is the case in alcohol clinical trials and, thus, more research is needed. A fourth issue is that future research should examine WHO RDL reductions in trials of the FDA approved AUD medications, disulfiram and acamprosate, which were not examined in the studies reviewed in this paper.

Finally, the FDA qualification of the WHO risk drinking maintains the FDA guidance⁵ for clinical trial length of 6 months to demonstrate medication efficacy. However, evidence regarding appropriate trial length is lacking. Several studies in the current review provide evidence of the durability of WHO risk drinking level reductions over time^46,56 with trial lengths of 4 months or less. Further, reductions achieved in a 3- or 4-month trial are stable even up to three years following treatment⁴⁸. Future research should examine whether trials of 6 month duration are necessary to establish efficacy and safety of medications in improving how patients feel and function. Additionally, work is needed to study the implementation of WHO RDLs in clinical practice settings, particularly with respect to patient and provider perspectives on targeting reductions in WHO RDLs.

This review highlights the evidence that achieving drinking reductions as defined by the WHO RDLs are associated with substantial health and social benefits. FDA acceptance of reductions in the WHO RDLs as a valid clinical endpoint could encourage the development of medications for AUD by providing an endpoint that captures improvement among more patients. Additionally, the WHO RDLs are used in Europe and elsewhere, and thus has the potential to promote international consistency in regulatory guidance on accepted endpoints. Finally, more individuals might seek treatment for excessive alcohol use, while providing clinicians a tool to communicate treatment success and health benefit to their patients.