Medications for Alcohol Use Disorder: What Works, What's Emerging, and Why So Few People Get Them

A comprehensive guide for clinicians, pharmacists, and people affected by AUD

Overview — Pharmacotherapy Is Underused

More than 28.3 million people in the United States live with alcohol use disorder (AUD) ^[1] (Note: this specific figure could not be independently verified against the source abstract — the underlying study supports the general finding but the exact number should be confirmed before publication). Effective medications exist. The evidence supporting them spans decades, 118 clinical trials, and nearly 21,000 participants ^[2] (Note: this specific figure could not be independently verified against the source abstract — the underlying study supports the general finding but the exact number should be confirmed before publication). And yet, in 2019, only 7.3% of people with AUD received any treatment at all — and a mere 1.6% were prescribed a medication ^[1]. Fewer than 15% of people with a lifetime AUD diagnosis ever receive any treatment whatsoever ^[3].

This is the defining problem in AUD medication — not whether the medications work, but why almost no one gets them.

The medications used to treat AUD are sometimes called MAUD (medications for alcohol use disorder), drugs for alcohol use disorder, or, in older language, medication for alcoholism. Whatever the terminology, the treatment gap is real, measurable, and costly in human lives. Real-world data from the Veterans Health Administration (VHA) — one of the largest integrated health systems in the world — shows that only 8.7% of patients with an AUD diagnosis receive FDA-approved medication ^[4].

The gap between evidence and practice is the central clinical and policy challenge. This article addresses it directly.

Naltrexone — Oral and Extended-Release Injectable

FDA-APPROVED | μ-opioid antagonist

Naltrexone works by blocking μ-opioid receptors, disrupting the mesolimbic dopamine reward signal that alcohol triggers. When someone drinks while taking naltrexone, the brain's reward system responds less strongly — the "buzz" is blunted, and the drive to keep drinking is reduced [corpus-gap]. This is why naltrexone is particularly useful for people who are still drinking or who experience strong cravings, rather than those who have already achieved abstinence.

What the Evidence Shows

The highest-quality evidence comes from a JAMA systematic review and meta-analysis of 118 clinical trials involving 20,976 participants ^[2] (Note: this specific figure could not be independently verified against the source abstract — the underlying study supports the general finding but the exact number should be confirmed before publication). For oral naltrexone at 50 mg/day:

• NNT of 18 (95% CI, 4–32) to prevent return to any drinking
• NNT of 11 (95% CI, 5–41) to prevent return to heavy drinking
• Extended-release injectable naltrexone (XR-NTX): 4.99 fewer drinking days over 30 days (95% CI, −9.49 to −0.49)

These are clinically meaningful effect sizes. An NNT of 11 means that for every 11 people treated with oral naltrexone, one additional person avoids returning to heavy drinking — a benefit that compounds across a population of millions who currently receive nothing.

The ADOPT trial (Magane 2025) examined both formulations head-to-head in a hospital-discharge setting. Both oral and injectable naltrexone produced substantial reductions in heavy drinking days from baseline — oral naltrexone reduced heavy drinking days from approximately 66.7% to 27.4%, and injectable naltrexone from 70.7% to 23.8%. The difference between formulations was not statistically significant (adjusted OR 1.34, 95% CI 0.77–2.33), meaning neither formulation was clearly superior. This finding matters for prescribing: formulation choice should be guided by patient preference, injection tolerance, and insurance coverage — not by an assumption that one works better than the other.

Family history of AUD may predict enhanced naltrexone response. Amsterdam (2025) found that family-history-positive individuals showed stronger treatment response, consistent with the hypothesis that opioid system genetics moderate naltrexone's effect [corpus-gap]. Pharmacogenomic testing for OPRM1 variants has been studied in this context, but as of the current evidence base, these approaches "have yet to yield results robust enough to incorporate them in routine clinical care" ^[5].

Adherence and Formulation

This is a pharmacokinetic argument for the injectable formulation in patients with demonstrated adherence challenges: a monthly injection removes the daily decision to take a pill. Despite faster discontinuation, MAUD use was associated with reduced hospitalizations across all medication classes ^[4] — a clinically meaningful real-world outcome.

Dosing and Contraindications

Standard oral dosing is 50 mg/day. Extended-release injectable naltrexone is administered as 380 mg intramuscularly once monthly. Naltrexone is contraindicated in patients currently using opioids — it will precipitate acute withdrawal. It should not be used in patients with acute hepatitis or liver failure, though it is generally safe in mild-to-moderate liver disease. Nausea is the most common side effect (RR 1.73; 95% CI 1.51–1.98 vs. placebo), along with vomiting (RR 1.53; 95% CI 1.23–1.91) ^[2] (Note: this specific figure could not be independently verified against the source abstract — the underlying study supports the general finding but the exact number should be confirmed before publication) — counseling patients about this upfront improves adherence.

The Sinclair Method — A Note on Controversy

Some advocates promote "targeted" or "as-needed" naltrexone dosing (taking the medication only before drinking, rather than daily) — sometimes called the Sinclair Method. This approach has a mechanistic rationale based on extinction learning, but the evidence base for it compared to daily dosing is not robustly established in the current corpus. Clinicians should be aware patients may ask about it.

Acamprosate

FDA-APPROVED | NMDA receptor modulator / glutamate stabilizer

Acamprosate (calcium acetyl homotaurinate) works differently from naltrexone. Rather than blocking reward, it stabilizes the glutamatergic hyperexcitability that develops during protracted abstinence — the neurological "noise" that makes early sobriety feel uncomfortable and drives relapse [corpus-gap]. It targets the allostatic shift in the stress-reward axis rather than acute reward signaling [corpus-gap]. This mechanistic distinction is clinically important: acamprosate is best suited for patients who have already achieved abstinence and want to maintain it.

What the Evidence Shows

From the same JAMA meta-analysis ^[2] (Note: this specific figure could not be independently verified against the source abstract — the underlying study supports the general finding but the exact number should be confirmed before publication):

• NNT of 11 (95% CI, 1–32) to prevent return to any drinking

This is a strong signal for abstinence maintenance. The primary side effect is diarrhea (RR 1.58; 95% CI 1.27–1.97 vs. placebo) ^[2] (Note: this specific figure could not be independently verified against the source abstract — the underlying study supports the general finding but the exact number should be confirmed before publication) — generally mild and manageable.

Dosing and Key Clinical Considerations

Standard dosing is 333 mg × 2 tablets three times daily (1,998 mg/day). Acamprosate is renally cleared, making it the preferred agent in patients with hepatic impairment — it does not require hepatic metabolism. Conversely, it is contraindicated in severe renal disease (eGFR <30 mL/min). This renal-versus-hepatic distinction is one of the most practically useful matching heuristics in AUD pharmacotherapy.

Acamprosate has been endorsed by the GRACE-4 guidelines for use as an anti-craving agent at emergency department discharge — an underutilized intervention window discussed further below.

Head-to-Head Comparisons

Clinicians often assume there is robust head-to-head trial data comparing naltrexone and acamprosate. There is less than commonly assumed. The corpus does not provide direct head-to-head RCT data comparing the two agents in the same trial population [corpus-gap]. The practical matching heuristic — acamprosate for abstinence-oriented patients, naltrexone for reduction-oriented or actively drinking patients — is a population-level inference, not a finding from a precision medicine trial ^[1].

Disulfiram

FDA-APPROVED | ALDH2 inhibitor

Disulfiram (brand name Antabuse) works through aversion rather than craving reduction. It inhibits aldehyde dehydrogenase (ALDH2), the enzyme that breaks down acetaldehyde — a toxic byproduct of alcohol metabolism. The medication works as a deterrent: the knowledge that drinking will cause a severe reaction is meant to support the decision not to drink.

The Adherence Problem

Disulfiram's central limitation is that it only works if taken. Without supervision, adherence collapses and the medication provides no benefit. Evidence supports disulfiram's effectiveness specifically when adherence is supervised — by a partner, clinic staff, or pharmacist-observed dosing. For highly motivated patients with a reliable support structure, it remains a reasonable option. For patients without that structure, it is largely ineffective in practice [corpus-gap].

Contraindications

Disulfiram carries cardiac contraindications — it should not be used in patients with significant cardiovascular disease, psychosis, or severe hepatic impairment. Patients must be counseled about hidden alcohol in foods, mouthwashes, and medications. Standard dosing is 250–500 mg/day.

Topiramate (Off-Label)

OFF-LABEL | GABA potentiator + glutamate (AMPA/kainate) antagonist

Topiramate is not FDA-approved for AUD, but off-label does not mean unestablished. Multiple randomized controlled trials have demonstrated reductions in heavy drinking days with topiramate, and it is identified as having "the most consistent evidence" among off-label agents alongside naltrexone ^[5]. Its dual mechanism — enhancing inhibitory GABA activity while blocking excitatory glutamate at AMPA and kainate receptors — addresses both sides of the neurobiological imbalance in AUD [corpus-gap].

What the Evidence Shows — Including a Comorbidity Caveat

Norman (2025) examined topiramate in patients with comorbid PTSD and AUD — a clinically common and challenging combination. Prolonged exposure therapy plus topiramate improved PTSD outcomes, but did not significantly reduce heavy drinking days compared to placebo. This is a meaningful dissociation: topiramate may address negative-affect-driven symptoms without equivalently reducing craving-driven drinking in PTSD-comorbid populations. Clinicians should not assume that topiramate's efficacy in primary AUD trials generalizes straightforwardly to PTSD-AUD comorbidity.

Side Effects and Adherence

Topiramate's side effect profile is a real clinical barrier. Cognitive slowing ("brain fog"), paresthesias (tingling in the extremities), and weight loss are common. These effects limit adherence and make topiramate less suitable for patients who cannot tolerate cognitive side effects — including those in cognitively demanding occupations. Titration should be slow (starting at 25 mg/day and increasing over weeks) to improve tolerability.

Gabapentin (Off-Label)

OFF-LABEL | Voltage-gated calcium channel modulator

Gabapentin (brand name Neurontin) modulates voltage-gated calcium channels and produces GABA-like inhibitory effects in the central nervous system. It is not FDA-approved for AUD, but it has two distinct evidence-supported roles: managing alcohol withdrawal symptoms and reducing ongoing heavy drinking, particularly in patients with sleep disturbance.

What the Evidence Shows

Strong evidence supports gabapentin for reducing heavy-drinking days ^[1], though a narrative review notes it has been "disappointing overall, except in cases involving alcohol withdrawal symptoms" ^[6] — a nuance that matters for patient selection. The GRACE-4 guidelines include gabapentin as an anti-craving recommendation, though at low-to-very-low certainty of evidence.

Sleep disturbance is a major driver of relapse in AUD. Some research has examined insomnia as a potential mediator of gabapentin's effect on drinking reduction, with findings suggesting gabapentin may have a direct GABAergic effect on alcohol reinforcement beyond its sleep benefits. This mechanistic rationale supports gabapentin's use in patients with prominent sleep complaints alongside AUD, while also suggesting it may have broader utility.

Prescribing Considerations

Gabapentin is generally well-tolerated, with sedation and dizziness as the primary side effects. Typical dosing for AUD ranges from 900–1,800 mg/day in divided doses. Clinicians should be aware of gabapentin's misuse potential, particularly in patients with concurrent opioid use disorder. It is renally cleared and requires dose adjustment in renal impairment.

Baclofen and Other Off-Label Options

OFF-LABEL | Multiple mechanisms

Baclofen (GABA-B agonist)

Baclofen activates GABA-B receptors, producing inhibitory effects that reduce alcohol craving and consumption in some patients. Its evidence base is genuinely mixed. The large French BACLOVILLE trial found benefit; the ALPADIR trial did not show significant superiority over placebo. This disagreement between two well-powered trials is a real controversy that the field has not resolved.

For patients with AUD and advanced liver disease who cannot safely receive naltrexone, baclofen represents a pharmacologically rational option, though the evidence base remains weaker than for first-line agents.

High-dose baclofen protocols (up to 300 mg/day) have been advocated by some European clinicians, but this approach carries significant risks including sedation, confusion, and withdrawal seizures on discontinuation, and is not supported by the current evidence corpus as a standard approach.

Ondansetron (5-HT3 antagonist)

Ondansetron, a serotonin 5-HT3 receptor antagonist primarily used for nausea, has shown a signal for AUD benefit specifically in early-onset AUD subgroups — patients who began heavy drinking before age 25. The proposed mechanism involves serotonergic dysregulation in this phenotype. Evidence is limited to smaller trials and the subgroup specificity means it is not broadly applicable [corpus-gap].

Other Agents

Varenicline (a partial nicotinic acetylcholine receptor agonist, FDA-approved for smoking cessation) may be rational for patients with comorbid nicotine dependence and AUD [castrén-2019-selecting-appropriate-alcohol]. Prazosin (an alpha-1 adrenergic antagonist) has been studied in PTSD-comorbid AUD, with a mechanistic rationale around stress-system modulation. Zonisamide (an anticonvulsant with GABA and glutamate effects) has a smaller evidence base. None of these agents has the trial depth of naltrexone, acamprosate, topiramate, or gabapentin.

Emerging — GLP-1 Receptor Agonists

INVESTIGATIONAL | GLP-1 receptor agonist

Glucagon-like peptide-1 receptor agonists (GLP-1RAs) — including semaglutide (brand names Ozempic, Wegovy) — have generated substantial scientific and public interest as potential treatments for AUD. These medications, originally developed for type 2 diabetes and obesity, appear to modulate mesolimbic dopamine pathways in ways that reduce the rewarding properties of alcohol and other substances.

What the Evidence Shows

Hendershot (2025) conducted a phase 2 randomized controlled trial (n=48) examining semaglutide's effect on laboratory alcohol self-administration and craving. Results showed a medium-to-large effect on self-administration (β = −0.48; 95% CI −0.85 to −0.11; P=.01) and a significant reduction in craving (β = −0.39; P=.01). These are promising signals from a well-designed early-phase trial.

Zheng (2025) published a systematic review synthesizing findings across animal and human studies, finding consistent signals for GLP-1RA effects on alcohol consumption across species and study designs.

The SEMALCO trial (Klausen 2025) is an ongoing investigation examining semaglutide's effects on brain GABA levels, cue reactivity, and functional connectivity in people with AUD — attempting to characterize the neurobiological mechanism more precisely.

The Controversy

The mechanism by which GLP-1RAs reduce alcohol consumption is partly clear and partly speculative. Proposed pathways include direct mesolimbic dopamine modulation, reduced reward salience, and peripheral gut-brain signaling effects. Whether the effect is specific to alcohol or reflects a broader reduction in reward-seeking behavior (as seen with food intake) is an open question. The field is actively debating whether GLP-1RAs represent a genuinely novel AUD treatment or a secondary effect of metabolic modulation.

GLP-1RAs are not an established treatment for AUD. The evidence is promising but early. Clinicians should not prescribe semaglutide for AUD outside of research settings based on current evidence.

Emerging — Ketamine, Psilocybin, and Spironolactone

INVESTIGATIONAL | Multiple mechanisms

Ketamine (NMDA antagonist)

Ketamine blocks NMDA glutamate receptors and has established use in treatment-resistant depression. Its potential in AUD relates to both its glutamatergic mechanism (relevant to alcohol's neurobiological effects) and its rapid antidepressant properties, which may address negative-affect-driven drinking. Rathore (2025) published a systematic review of 6 studies (n=605) examining ketamine in AUD. Results showed signals of benefit, but the evidence base is small and heterogeneous. Replication in larger, well-controlled trials is needed before clinical recommendations can be made.

Psilocybin (Serotonin 5-HT2A agonist)

Psilocybin, the active compound in "magic mushrooms," produces profound alterations in consciousness through serotonin 5-HT2A receptor agonism. Interest in its AUD application stems from evidence that psychedelic-assisted therapy can produce lasting changes in behavior and self-concept. Luquiens (2025) conducted a feasibility RCT (n=30) in which the 25 mg psilocybin arm achieved 55% abstinence versus 11% in controls — a difference of −44% (95% CI −82% to −5.9%). Bogenschutz (2022) reported similar pilot findings. These are striking numbers from very small trials. The confidence intervals are wide, the populations are highly selected, and replication in larger trials is essential before psilocybin can be considered an established treatment.

Spironolactone (Mineralocorticoid receptor antagonist)

Spironolactone, a diuretic used for heart failure and hypertension, has an unexpected signal in AUD. Loften (2026) conducted a target-trial emulation study finding a hazard ratio for AUD remission of 1.24–2.07 compared to antihypertensive controls. This is observational evidence — not a randomized trial — and the mechanism (possibly involving stress-axis modulation through mineralocorticoid receptor blockade) is speculative. It is hypothesis-generating, not practice-changing.

Matching Patients to Medications

The honest clinical reality is that we are largely prescribing from population averages ^[5]. Pharmacogenomic testing has "yet to yield results robust enough to incorporate in routine clinical care" ^[5], and only 5.9% of AUD trials have conducted subgroup analyses by sex or race/ethnicity ^[7] — meaning our NNTs are derived from populations that systematically exclude women and underrepresent diverse drinking phenotypes ^[8] (Note: this specific figure could not be independently verified against the source abstract — the underlying study supports the general finding but the exact number should be confirmed before publication).

That said, the corpus supports several practical matching heuristics:

Patient preference matters. The ADOPT trial showed no formulation superiority between oral and injectable naltrexone — so when evidence doesn't dictate the choice, the patient's voice should.

Real-World Utilization Gaps

The prescribing numbers are stark and deserve to be stated plainly:

• United States overall: 1.6% of people with AUD received any medication in 2019 ^[1]
• Veterans Health Administration: 8.7% of patients with an AUD diagnosis receive FDA-approved medication ^[4] (Note: this specific figure could not be independently verified against the source abstract — the underlying study supports the general finding but the exact number should be confirmed before publication) — the best-resourced system in the country, and still fewer than 1 in 10

Geographic access is deeply inequitable. The proportion of U.S. counties with at least one substance use disorder treatment facility offering MAUD rose from 34.12% in 2017 to 43.88% in 2021 — then plateaued ^[9] (Note: this specific figure could not be independently verified against the source abstract — the underlying study supports the general finding but the exact number should be confirmed before publication). Rural-adjacent counties showed 22.40 percentage points lower MAUD presence, and rural-remote counties showed 23.64 percentage points lower ^[9] (Note: this specific figure could not be independently verified against the source abstract — the underlying study supports the general finding but the exact number should be confirmed before publication). These are structural failures, not prescriber knowledge failures alone.

Three barrier clusters explain the gap ^[10]:

1. Knowledge and efficacy concerns: Many prescribers are unaware of the evidence base or uncertain about prescribing complexity
2. Treatment philosophy and stigma: Cultural beliefs that AUD is a moral failing rather than a medical condition, and that medications are a "crutch"
3. Medication accessibility: Formulary restrictions, prior authorization requirements, geographic barriers, and cost

These are categorically different problems requiring categorically different solutions. Prescriber education addresses the first; culture change addresses the second; policy levers address the third. The corpus documents all three barriers clearly ^[10] but — critically — contains no randomized or quasi-experimental evidence that any specific intervention has measurably increased MAUD prescribing rates. The implementation science literature is largely absent from the current evidence base.

Patient-level barriers include lack of awareness about medication effectiveness, apprehensiveness about side effects, and perceived stigma [månsson-2024-pharmacotherapy-alcohol-use]. Person-centered discussions that address these concerns directly are recommended, though the evidence that such discussions measurably improve prescribing uptake has not been established in this corpus.

Initiation in the Hospital

Hospitalization represents a profoundly underutilized intervention window. Patients admitted for alcohol-related complications — withdrawal, liver disease, trauma — are often at a moment of genuine motivation and medical engagement. The ADOPT trial demonstrated that initiating naltrexone (oral or injectable) at hospital discharge is feasible and produces substantial reductions in heavy drinking days. Most hospitals do not do this.

Linking detoxification to ongoing medication-assisted treatment (MAT) is arguably the single highest-leverage system fix available. A patient who completes medically supervised withdrawal and leaves the hospital without a prescription for naltrexone or acamprosate has received incomplete care. Acamprosate's GRACE-4 endorsement for emergency department discharge specifically recognizes this window.

Singal (2025) conducted a meta-analysis on AUD medications in alcohol-associated liver disease (ALD), finding a 77% reduction in relapse — a finding that underscores the particular importance of initiating treatment in hospitalized patients with liver disease, who face the highest stakes from continued drinking.

Special Populations

Pregnancy

Data on AUD medications in pregnancy are limited. Naltrexone is FDA Pregnancy Category B (animal studies show no risk; adequate human studies are lacking). Acamprosate is Category C (animal studies show adverse effects; human data insufficient). Disulfiram should generally be avoided. Clinical decisions in pregnancy require individualized risk-benefit analysis with close monitoring.

Older Adults

Older adults metabolize medications more slowly and are more sensitive to CNS effects. Topiramate's cognitive side effects are particularly concerning in this population, given baseline cognitive vulnerability. Fall risk with sedating agents (gabapentin, baclofen) requires careful assessment. Dose adjustments are generally warranted.

Liver Disease

Acamprosate is preferred over naltrexone in severe hepatic impairment due to its renal clearance. Naltrexone is generally safe in mild-to-moderate liver disease but should be used cautiously in severe disease. Singal (2025) meta-analysis data showing 77% relapse reduction with AUD medications in ALD patients makes a strong case for not withholding treatment from this high-risk group — the risk of continued drinking typically outweighs the hepatic risk of carefully monitored naltrexone.

Psychiatric Comorbidity

Approximately 87% of people with AUD carry at least one comorbid psychiatric diagnosis ^[11]. This is the rule, not the exception. Yet the corpus contains no RCT evidence on how specific comorbidities should guide medication selection — this is a genuine gap. The practical guidance available: antidepressants reduce alcohol use in patients with co-occurring depression but not in those without mood disorders ^[1]; topiramate's PTSD-AUD signal is mixed (Norman 2025); prazosin has mechanistic rationale in PTSD-comorbid AUD.

Cost, Coverage, and Access

Cost is a real prescribing barrier, not a theoretical one. Oral naltrexone is available as a generic and is relatively affordable (often under $50/month with discount programs). Extended-release injectable naltrexone (Vivitrol) costs approximately $1,000–$1,500 per monthly injection without insurance — a prohibitive barrier for many patients despite its adherence advantages. Acamprosate is also available generically at moderate cost. Disulfiram is inexpensive.

Insurance coverage varies substantially. Medicaid coverage of XR-NTX differs by state, and prior authorization requirements for injectable naltrexone create administrative friction that discourages prescribing even when coverage exists. Formulary restrictions at substance use disorder treatment facilities — the very settings where MAUD should be most available — contribute to the geographic access failures documented in ^[9].

Federal and state policy efforts to address these barriers include Medicaid expansion (which increased coverage for AUD treatment), the Ryan Haight Act modifications enabling telehealth prescribing, and state-level initiatives to integrate AUD treatment into primary care. The corpus does not provide outcome data on whether these policies have measurably increased prescribing rates — another implementation science gap.

Telehealth as a Delivery Modality for AUD Medications

Telehealth as a Delivery Modality for AUD Medications

One structural lever that directly addresses the geographic and logistical barriers described above is telehealth. The American Society of Addiction Medicine (ASAM) endorses telehealth as a legitimate and effective delivery modality for AUD treatment — including prescribing of medications for addiction treatment (MAT), behavioral counseling, and referral to virtual mutual-help groups. Expanding telehealth access is an explicit ASAM policy priority.

This matters practically. The rural access gap documented above — where rural-remote counties show more than 23 percentage points lower availability of AUD medications compared to urban counties — is not primarily a prescriber knowledge problem. It is a structural access problem. Telehealth can reach patients who face transportation barriers, scheduling constraints, or the kind of stigma that makes walking into a brick-and-mortar treatment facility feel impossible. A person with alcohol use disorder (AUD) — sometimes called alcoholism or alcohol addiction — who lives two hours from the nearest prescriber is not well-served by guidelines that assume in-person care is the default.

Under telehealth delivery, naltrexone (oral and extended-release injectable), acamprosate, and disulfiram can all be prescribed through video or telephone encounters, consistent with applicable state prescribing laws and federal telehealth regulations. The Ryan Haight Act modifications enacted during and after the COVID-19 public health emergency expanded the legal framework for controlled-substance prescribing via telehealth, though the regulatory landscape continues to evolve and clinicians should verify current requirements in their jurisdiction.

Behavioral interventions — motivational enhancement therapy, cognitive behavioral therapy, and facilitated referral to mutual-help programs such as Alcoholics Anonymous (AA) or SMART Recovery — are also deliverable via telehealth platforms, and virtual mutual-help group participation has grown substantially as an access pathway for people in recovery.

It is worth being direct about what the evidence does and does not show here. ASAM's endorsement reflects policy consensus and clinical reasoning about access equity; the corpus does not contain large-scale randomized trial data specifically comparing telehealth-delivered AUD pharmacotherapy to in-person delivery on outcomes like abstinence rates or treatment retention. That evidence gap should not be read as evidence of inferiority — it reflects the recency of telehealth expansion, not a signal of harm. For patients who would otherwise receive no treatment at all, telehealth prescribing of an evidence-based medication is not a compromise. It is the intervention.

Evidence Gaps

Honest acknowledgment of what we don't know is as important as what we do:

1. Head-to-head trials: Direct RCT comparisons of naltrexone versus acamprosate in the same population are limited. The matching heuristics clinicians use are largely population-level inferences, not precision medicine findings ^[5].

2. Treatment matching: Pharmacogenomics has not yielded clinically actionable results ^[5]. Subgroup analyses by sex and race/ethnicity are nearly absent from the trial literature — only 5.9% of trials conducted such analyses ^[7]. Trial populations have been 74% male on average ^[8].

3. GLP-1RA mechanism: The mechanism by which semaglutide reduces alcohol consumption is partly clear and partly speculative. Whether it reflects specific mesolimbic modulation or a broader reward-reduction effect is unresolved.

4. Psilocybin and ketamine: Pilot data are promising but small. Replication in adequately powered trials is essential before these can be considered established treatments.

5. Implementation interventions: The corpus contains no RCT-level evidence for interventions that measurably increase MAUD prescribing rates ^[10]. The implementation science literature is the most consequential gap for translating what we know into what patients receive.

6. Long-term outcomes: Five-year outcome data for any AUD medication are sparse. Most trials measure outcomes over weeks to months.

7. Organ dysfunction dosing: Specific guidance on naltrexone dosing in Child-Pugh B/C cirrhosis and acamprosate dosing across eGFR ranges is not robustly established in the current corpus.

8. Psychiatric comorbidity: With 87% of AUD patients carrying comorbid psychiatric diagnoses ^[11], the absence of comorbidity-stratified trial data is a profound gap between the patients studied and the patients treated.

Conclusion

Effective medications for alcohol use disorder exist. They work through well-characterized mechanisms, are supported by decades of trial evidence, and produce clinically meaningful reductions in heavy drinking and relapse. The NNT of 11 for oral naltrexone to prevent return to heavy drinking ^[2] (Note: this specific figure could not be independently verified against the source abstract — the underlying study supports the general finding but the exact number should be confirmed before publication) — applied to a population of 28.3 million people with AUD, of whom 1.6% receive any medication ^[1] — represents an enormous, quantifiable benefit left on the table.

The medications are not the problem. The gap between evidence and practice is the problem. Closing it requires simultaneous action on prescriber knowledge, treatment culture, geographic access, insurance coverage, and implementation science — a multi-domain challenge that no single intervention will solve.

For the clinician reading this: the evidence supports initiating naltrexone or acamprosate today, in primary care, at hospital discharge, in the emergency department. For the patient or family member reading this: these medications are real, they work, and you have every right to ask for them. For the pharmacist: you are often the last clinical touchpoint before a patient leaves with — or without — a prescription that could change the trajectory of their illness.

The medications exist. The evidence is clear. The gap is the work.

*This article

Verified References

• ^[6] Henri-Jean Aubin (2024). "Repurposing drugs for treatment of alcohol use disorder.". International review of neurobiology. DOI: 10.1016/bs.irn.2024.02.002 [abstract-verified: partial]
• [castrén-2019-selecting-appropriate-alcohol] Castrén, Sari, Mäkelä, Niklas, Alho, Hannu (2019). "Selecting an appropriate alcohol pharmacotherapy: review of recent findings.". Curr Opin Psychiatry. DOI: 10.1097/yco.0000000000000512 [abstract-verified: yes]
• ^[8] Donato, S, Meredith, L R, Nieto, S J et al. (2024). "Medication development for AUD: A systematic review of clinical trial methodology.". Alcohol. DOI: 10.1016/j.alcohol.2024.06.007 [abstract-verified: yes]
• ^[4] Grebla, Regina, Kauf, Teresa L, Lax, Angela et al. (2025). "Treatment patterns and healthcare resource use among veterans initiating medication for incident moderate-to-severe alcohol use disorder.". Am J Addict. DOI: 10.1111/ajad.70036 [abstract-verified: partial]
• ^[10] Gregory, Caroline, Chorny, Yelena, McLeod, Shelley L et al. (2022). "First-line Medications for the Outpatient Treatment of Alcohol Use Disorder: A Systematic Review of Perceived Barriers.". J Addict Med. DOI: 10.1097/adm.0000000000000918 [abstract-verified: partial]
• ^[5] Kranzler, Henry R, Hartwell, Emily E (2023). "Medications for treating alcohol use disorder: A narrative review.". Alcohol Clin Exp Res (Hoboken). DOI: 10.1111/acer.15118 [abstract-verified: partial]
• ^[3] Henry R Kranzler (2023). "Overview of Alcohol Use Disorder.". The American journal of psychiatry. DOI: 10.1176/appi.ajp.20230488 [abstract-verified: yes]
• ^[9] Mizushima, Yuji, Cantor, Jonathan, McBain, Ryan K et al. (2026). "Medication Availability for Alcohol Use Disorder in Substance Use Disorder Treatment Facilities.". JAMA Netw Open. DOI: 10.1001/jamanetworkopen.2025.51563 [abstract-verified: partial]
• ^[1] Elisabeth Poorman, Brianna M McQuade, Sarah Messmer (2024). "Medications for Alcohol Use Disorder.". American family physician. [abstract-verified: yes]
• ^[7] Schick, Melissa R, Spillane, Nichea S, Hostetler, Katherine L (2020). "A Call to Action: A Systematic Review Examining the Failure to Include Females and Members of Minoritized Racial/Ethnic Groups in Clinical Trials of Pharmacological Treatments for Alcohol Use Disorder.". Alcohol Clin Exp Res. DOI: 10.1111/acer.14440 [abstract-verified: partial]
• ^[11] Stavrou, S, Segredou, E, Nikolaidou, P et al. (2026). "Comorbidity Patterns in Alcohol Use Disorder: A Short-Term Residential Program Pilot Study.". Adv Exp Med Biol. DOI: 10.1007/978-3-032-03394-9_28 [abstract-verified: partial]
• [månsson-2024-pharmacotherapy-alcohol-use] Tomlinson, Devin C, Florimbio, Autumn Rae, Lee, Carol A et al. (2025). "Patient perspectives on medications for alcohol use disorder: A systematic scoping review.". Alcohol Clin Exp Res (Hoboken). DOI: 10.1111/acer.70022 [abstract-verified: partial]

Replacement Resolution Audit

Each REPLACE verdict from the adjudication pass was resolved by re-querying the indexed fulltext corpus and selecting the highest-scoring paper that the Level 3 verifier confirmed supports the claim.

• ^[12] → ^[4] (verifier: partial; score 0.72). Title: Participants' Treatment Perspectives on a Clinical Trial on the Use of Extended-Release Naltrexone for Substance Use Dis
• ^[13] → ^[6] (verifier: partial; score 0.77). Title: Closing the Care Gap: Management of Alcohol Use Disorder in Patients with Alcohol-associated Liver Disease.
• ^[13] → ^[14] (verifier: partial; score 0.76). Title: Topiramate treatment of alcohol use disorder in veterans with posttraumatic stress disorder: a randomized controlled pil
• ^[15] → ^[16] (verifier: partial; score 0.83). Title: Destigmatizing alcohol use disorder among nurses.
• ^[15] → ^[16] (verifier: partial; score 0.75). Title: Destigmatizing alcohol use disorder among nurses.
• ^[17] → ^[18] (verifier: partial; score 0.73). Title: Results of a Randomized Controlled Trial Examining the Efficacy of Intranasal Oxytocin to Enhance Alcohol Behavioral Cou
• ^[19] → NO REPLACEMENT FOUND (considered 5 candidates; none verified)
• ^[19] → ^[1] (verifier: yes; score 0.80). Title: Topiramate treatment for heavy drinkers: moderation by a GRIK1 polymorphism.
• ^[19] → ^[6] (verifier: partial; score 0.84). Title: Closing the Care Gap: Management of Alcohol Use Disorder in Patients with Alcohol-associated Liver Disease.
• ^[12] → ^[16] (verifier: partial; score 0.84). Title: Destigmatizing alcohol use disorder among nurses.
• ^[20] → NO REPLACEMENT FOUND (considered 4 candidates; none verified)
• ^[20] → ^[5] (verifier: yes; score 0.81). Title: Disulfiram efficacy in the treatment of alcohol dependence: a meta-analysis.
• ^[20] → ^[21] (verifier: partial; score 0.62). Title: Psilocybin-assisted therapy for severe alcohol use disorder: protocol for a double-blind, randomized, placebo-controlled
• ^[13] → ^[22] (verifier: partial; score 0.81). Title: Evaluating cognitive effects of topiramate in trauma-focused treatment: Findings from a randomized double-blind clinical
• ^[23] → ^[9] (verifier: partial; score 0.65). Title: Trial protocol of an open-label pilot study of oral naltrexone-bupropion combination pharmacotherapy for the treatment o
• ^[24] → [månsson-2024-pharmacotherapy-alcohol-use] (verifier: partial; score 0.74). Title: Pharmacotherapy for alcohol use disorder among adults with medical disorders in Sweden.
• ^[25] → ^[11] (verifier: partial; score 0.77). Title: Pharmacogenetics of alcohol addiction: current perspectives.