Comparative Effectiveness of Naltrexone Formulations in Alcohol Use Disorder: An Updated Meta-Analysis

Abstract

Purpose:

We aimed to synthesize and provide an updated meta-analysis of the current data on extended-release naltrexone (XR-NTX) and oral naltrexone (NTX) for alcohol use disorder (AUD).

Methods:

A comprehensive database search was conducted and updated on October 1, 2025. This meta-analysis encompassed randomized controlled trials (RCTs) and observational studies that compared XR-NTX and oral NTX in adults with AUD. Included studies reported outcomes on treatment persistence and healthcare utilization, specifically ED visits and inpatient hospitalizations during follow-up. Odds ratios (ORs) were computed using the Mantel-Haenszel random-effects model.

Results:

Out of 101 studies screened, seven studies (N=42,268; n=2,103 for XR-NTX, n=40,165 for NTX) met the inclusion criteria for the meta-analysis. Overall, treatment persistence was significantly higher with XR-NTX compared with oral NTX (OR=1.94 [95% CI, 1.22–3.10]), with benefits observed at both three months (pooled OR=2.24 [95% CI, 1.14–4.41], p=0.019) and six months (OR=1.59 [95% CI, 1.04–2.43], p=0.032). No significant differences were observed in healthcare utilization, including all-cause ED visits (OR=1.33 [95% CI, 0.93–1.90]) or all-cause inpatient admissions (OR=0.84 [95% CI, 0.49–1.43]).

Conclusion:

This meta-analysis suggests higher treatment persistence among patients with AUD receiving XR-NTX versus oral NTX, underscoring formulation as a key factor in adherence and retention. However, alcohol-related healthcare utilization metrics, including ED visits and hospitalizations, showed no significant between-group differences. Further pragmatic trials accounting for patient preferences, sex-specific outcomes, and barriers to care accessibility are needed.

INTRODUCTION

Alcohol use disorder (AUD) remains a significant public health concern worldwide, affecting approximately 27.9 million individuals in the past year (Substance Abuse and Mental Health Services Administration [SAMHSA], Center for Behavioral Health Statistics and Quality [CBHSQ],¹ AUD is frequently comorbid with anxiety and mood disorders, contributing to a worse prognosis and poorer treatment outcomes.^2,3 The U.S. Food and Drug Administration (FDA) has approved four medications for AUD: naltrexone (NTX) (available in oral and extended-release injectable [XR-NTX] forms), acamprosate, and disulfiram.^4–7 NTX is the most commonly used treatment for AUD among these and primarily acts as a μ-opioid antagonist, reducing craving and modulating reward-related neurobiological pathways.^8,9 XR-NTX has received particular attention due to its pharmacokinetic properties, which may improve treatment persistence and healthcare utilization. Studies have shown that XR-NTX demonstrates up to four times greater persistence at 3 months post-hospitalization while significantly reducing heavy drinking days.^10,11 Despite this evidence, NTX remains underutilized, with only up to 3% of patients with AUD receiving any FDA-approved medication.^12,13

Our previous systematic review and meta-analysis examined XR-NTX and oral NTX for alcohol and opioid use disorders which found significantly higher treatment persistence with XR-NTX at both 3 and 6 months compared to NTX.¹⁴ Since our prior review, two important studies have emerged: the Alcohol Disorder Hospital Treatment (ADOPT) trial, a randomized controlled trial (RCT) of 248 hospitalized adults with AUD concluded no significant differences between XR-NTX and NTX in reducing heavy drinking days or emergency department (ED) visits, although persistence remained significantly higher in the XR-NTX arm at the 3-month follow-up (40.7% vs. 26.6%, OR = 1.89 [95% CI, 1.08–3.32]).¹⁵ Additionally, a retrospective study of U.S. veterans (n = 31,384) found that XR-NTX was associated with longer average treatment duration (~92 days) with reduced hospitalization rates and greater outpatient engagement compared to oral agents (~55–59 days).16 However, these studies highlight a clinically important uncertainty for AUD: whether XR-NTX confers benefits beyond adherence, particularly for acute care utilization, and whether observed effects differ by study design and healthcare setting.

Given the mixed findings across naturalistic studies and randomized trials regarding long-term adherence, treatment outcomes and variability across care settings^15–18 we conducted an updated narrow meta-analysis focusing on AUD to quantify differences between oral NTX and XR-NTX in (1) short-term and long-term treatment persistence, (2) ED visits, and (3) inpatient hospitalizations during follow-up. While incorporating new available large-scale data with the prior evidence we aim to provide an updated and clinically actionable synthesis of comparative effectiveness for oral NTX vs. XR-NTX.

METHODS

Data sources and search strategies

A comprehensive literature search was conducted by an expert medical librarian (LH) on April 23, 2025, and updated on October 1, 2025. Databases searched included: Ovid MEDLINE(R) (1946+), Ovid Embase (1974+), Ovid Cochrane Central Register of Controlled Trials (1991+), Ovid Cochrane Database of Systematic Reviews (2005+), APA PsycInfo (1987+), and Scopus (1970+). Controlled vocabulary supplemented with keywords was used, with no date or language limits applied. Complete search strategies are available in the appendix (Table S1). We followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines,¹⁹ and the study protocol from which this meta-analysis focusing only on AUD (https://osf.io/x4pmk).

This updated meta-analysis included only comparative studies of oral NTX and XR-NTX in AUD. Inclusion criteria were: 1) adults (≥ 18 years) with AUD diagnosis; 2) intervention: XR-NTX; 3) comparator: oral NTX; 4) outcomes: treatment persistence and/or healthcare utilization (inpatient hospitalizations or ED visits). Outcomes reporting AUD-related ED visits or hospitalizations were prioritized. We excluded studies with small sample sizes (n < 10), case series, case reports, and reviews. Articles were screened using an online spreadsheet, with disagreements resolved by the senior author (BS).

Methodological quality and risk of bias assessment

We used Cochrane Collaboration’s tool for assessing the risk of bias to evaluate the methodological quality of RCTs.²⁰ We assessed the risk of bias for random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting, and other biases. For the non-randomized/observational studies, the methodological quality was assessed using Methodological Index for Non-Randomized Studies (MINORS).²¹

Statistical analysis

We conducted separate meta-analysis for our main outcomes: a) short-term (up to 3 month) and long-term treatment persistence at 6 months, b) ED visits, and c) inpatient hospitalizations applying a random-effect Mantel–Haenszel models with inverse-variance weighting. Between-study variance was estimated using the restricted maximum-likelihood approach. Between-study heterogeneity was quantified using Cochran’s Q, τ², and the I² statistic, with I² values of 25%, 50%, and 75% interpreted as low, moderate, and high heterogeneity, respectively. We conducted a sensitivity analysis (e.g., leave one out) by study design excluding influential datasets to assess the pooled estimates and overall heterogeneity. Standardized residuals were analyzed to identify outliers. Analysis were performed using the “meta” and “metafor” package of R software for statistical computing (version 4.2.2).^22,23 An alpha level of 0.05 indicated statistical significance.

RESULTS

Study Selection and Characteristics

A total of 101 studies screened, 13 were reviewed in full, of which 7 met our inclusion criteria for this updated meta-analysis (Figure S1). Seven studies^{15–18,24–26} using studies reporting both mean and standard deviation, the pooled mean age was 46.4 years for XR-NTX [95% CI, 40.8–52.0] and 47.5 years for oral NTX [95% CI, 42.7–52.3]. Only one of these studies was an RCT,¹⁵ and the remaining six were observational studies.^{16–18,24–26} One of the studies examined the use of different treatments for AUD: XR-NTX, NTX, disulfiram, and acamprosate in a large US Veteran’s population.¹⁶ Table 1 summarizes the included studies for this meta-analysis.

Table 1.

Summary of the Comparative Studies of Extended-Release (XR) vs Oral Naltrexone in the Treatment of Alcohol Use Disorder (AUD).

Study (Year)	Study Design	Duration (months)	N (XR-NTX)	N (NTX)	Mean Age ± SD (XR-NTX)	Mean Age ± SD (NTX)	Conclusions
Mark et al. (2010)	Retrospective claims database	6	295	2064	N/A	N/A	XR-NTX reduced detox days vs oral NTX; no difference in coverage or utilization.
Baser et al. (2011a)	Retrospective claims analysis	6	661	2391	44.2	45.9	XR-NTX showed better persistence and lower relapse; utilization similar.
Bryson et al. (2011)	Retrospective cohort	6	211	1408	40.5 ± 13.3	42.0 ± 11.2	XR-NTX linked to lower relapse and better persistence, but effect diminished after adjustment.
Grebla et al. (2025)	Retrospective VA cohort	12	431	25082	46.0 ± 12.5	47.9 ± 13.5	Long-term XR-NTX associated with lower ED/hospital use; early discontinuation common.
Magane et al. (2025)	Randomized clinical trial (ADOPT Trial)	3	108	109	49.4 ± 10.3	49.3 ± 10.5	XR-NTX showed greater reduction in heavy drinking days than PO-NTX.
Harris et al. (2012)	Retrospective claims analysis	6	356	8560	N/A	N/A	XR-NTX improved persistence over oral NTX and acamprosate.
Walker et al. (2019)	Retrospective chart review of Veterans Integrated Service Networks (VISN-7)	6	41	551	52.1 ± N/A	52.4 ± N/A	XR-NTX demonstrated the highest mean adherence (54.6%) compared with oral NTX (49.8%) and other agents. XR-NTX more likely to achieve ≥80% adherence than oral NTX

Treatment Persistence

Six studies reported treatment persistence at 45 days, 3 months, or 6 months. Overall persistence, based on the last time point for each study, was significantly higher with XR-NTX compared to oral-NTX (OR = 1.94 [95% CI, 1.22–3.1], p = 0.005), but with high heterogeneity (I² = 88.9%) (Figure 1A). Sensitivity analysis excluding the16 study as an outlier yielded consistent results, with XR-NTX retaining statistical significance over oral-NTX (OR = 1.56 [95% CI, 1.12–2.19], p < 0.05).

Figure 1.

A. Forest plot of pooled rates for treatment overall persistence for XR-NTX and oral NTX. B. Forest plot of pooled rates for short-term treatment persistence (up to 3 months) for XR-NTX and oral NTX. C. Forest plot of pooled rates for long-term treatment persistence (at 6 months) for XR-NTX and oral NTX.

For short-term persistence (up to 3 months), our meta-analytic random effects data included 4 studies15–17,25 reaching a total sample size of N = 36,265 (XR-NTX = 1,106; NTX = 35,159). Grebla et al. (2025)¹⁶ provided persistence estimate at 45 days. Our data showed a statistically significant difference in short-term treatment persistence between XR-NTX and NTX, favoring XR-NTX (pooled OR = 2.24 [95% CI, 1.14–4.41], p = 0.019) with substantial heterogeneity (I² = 92.4%; τ2 = 0.432) (Figure 1B). Leave-one-out sensitivity analysis showed that with exclusion of the outlier, Grebla et al. (2025)¹⁶ yielded a pooled OR of 1.58 ([95% CI, 1.08–2.31], p = 0.018) and a decrease in heterogeneity (I² = 69.3%, τ2 = 0.075) (Figure S5A).

For long-term persistence (at 6 months), pooled data showed a statistically significant result still favoring XR-NTX compared to oral-NTX (OR = 1.59 [95% CI, 1.04–2.43], p = 0.032) with a moderate heterogeneity I² = 67.5% (Figure 1C).

ED Visits During Follow-Up

Five studies^{15–17,24,25} were included in the pooled analysis of all-cause ED visits comparing XR-NTX with NTX (n = 1,706 for XR-NTX; n = 31,054 for NTX), showing no statistically significant difference between groups (OR = 1.33 [95% CI, 0.93–1.90], p = 0.11) and with substantial heterogeneity (I² = 83.0%) (Figure S3A). When we examined only AUD-related ED visits, only two studies^15,16 provided data directly comparing XR-NTX with oral NTX. Neither study reported a significant difference between the treatments (OR = 1.21 [95% CI, 0.93–1.59]), with no apparent heterogeneity (I² = 0.0%) (Figure S3B). Leave-one-out sensitivity analysis showed that excluding the only RCT,¹⁵ the pooled analysis for all-cause ED visits showed no statistically significant difference (OR = 1.39 [95% CI, 0.92–2.09]), with substantial heterogeneity (I² = 86.9%) (Figure S5B).

Inpatient Admission During Follow-Up

A total of four studies^15,16,24,25 reported data on all-cause inpatient admissions during the follow-up period. The pooled analysis showed no statistically significant differences between XR-NTX and NTX in terms of inpatient hospitalizations (OR = 0.84 [95% CI, 0.49–1.43]) and substantial heterogeneity (I² = 91.7%) (Figure S4A). Leave-one-out sensitivity analysis showed that excluding the only RCT, Magane et al. (2025)¹⁵, yielded a pooled OR of 0.78 [95% CI, 0.41–1.50] and heterogeneity remained substantial (I² = 94.3%) (Figure S5C). Although, while removing the Baser et al. (2011)²⁴ study from the model, there was a decrease in heterogeneity (I² = 74.6%), no statistically significant differences were found between both formulations (OR = 1.09 [95% CI, 0.11–11.26]). AUD-related inpatient admissions were reported by only two studies,^15,16 which also showed no significant differences between XR-NTX and oral NTX (pooled OR = 0.97 [95% CI, 0.79–1.20]) with no heterogeneity (I² = 0.0%) (Figure S4B).

Methodological Quality and Risk of Bias Assessment

The quality assessment of non-randomized studies meets criteria for moderate quality (Table S2). For the only RCT, the Magane et al., (2025)¹⁵ study indicated an overall low risk across most domains (Figure S2); however, there was insufficient information regarding blinding of participants and personnel which may have introduced some bias.

DISCUSSION

This updated meta-analysis provides additional evidence of higher treatment persistence with XR-NTX compared to NTX in AUD. However, the ED visits and inpatient admissions did not differ significantly between groups. The plausible pharmacological advantages of long-acting injectable (LAI) formulations, including reduced dosing burden and improved bioavailability, may contribute to superior adherence-treatment persistence observed in our meta-analysis and in previous studies.^14,15 Although heterogeneity in our study ranged from moderate to substantial, the direction of association remained consistent across all models, suggesting that XR-NTX’s may have pharmacokinetic advantages which translate into higher retention rates compared with NTX.

When examining healthcare utilization outcomes, our results align with the ADOPT trial, which found no statistically significant difference in AUD-related ED visits between XR-NTX and oral-NTX. In our meta-analysis, there was a trend towards higher ED utilization in the XR-NTX group, but differences in ED visits, AUD-related ED visits, and overall inpatient admissions did not reach statistical significance.

Several factors may explain the absence of significant differences in healthcare utilization despite the observed higher persistence rates with XR-NTX, including: limited statistical power due to smaller sample sizes in some studies, heterogeneity in study populations, and methodological variations across studies. These findings could also reflect the broader influence of environmental and clinical factors, such as medical comorbidities, crisis management, and access to care. A recent population-based study underscores the clinical relevance of short-term treatment persistence, reporting a median treatment duration of 31 days, with only 15.6% of patients remaining in treatment after 6 months.²⁷ These findings highlight the strong association between poor treatment persistence and diminished treatment effectiveness. Similarly, earlier U.S. claims data also showed that >85% of cases discontinued within 6 months, and non-persistence was linked to increased acute-care utilization.²⁸ These observations explain why poor persistence is a leading cause of diminished treatment efficacy, as clinical benefit ceases regardless of prior adherence once treatment is discontinued. Persistence is the duration of time from treatment initiation to discontinuation—generally considered more important than adherence, the proportion of prescribed doses actually taken during the period of treatment, for achieving optimal outcomes with naltrexone treatment for AUD. This is because the clinical benefit of naltrexone is closely linked to ongoing exposure to the medication over time, and discontinuation (i.e., poor persistence) is a major driver of relapse and reduced efficacy in real-world settings. Interventions that promote treatment persistence—such as XR-NTX, integrated outpatient follow-up, and psychoeducation—may improve outcomes.

Beyond the overall small number of studies included in this meta-analysis, several additional limitations merit consideration. The absence of rigorous RCT data—with only one RCT included in this analysis—underscores the significant challenges in clinical trial design for addressing these questions and providing robust clinical recommendations. Additionally, the lack of sex-stratified data limits our ability to assess potential sex-based differences in treatment outcomes, which may be clinically significant and warrant future investigation.

Overall, our findings reinforce the pivotal role of XR-NTX in the management of AUD and underscore the need for timely interventions to improve adherence. These data also inform future research efforts, particularly the development of pragmatic clinical trials that reflect real-world scenarios and account for patient medication preferences, sex-specific outcomes, socioeconomic factors, and barriers to care accessibility.

Highlights.

• Our findings suggest that patients with alcohol use disorder (AUD) receiving extended-release naltrexone (XR-NTX) exhibited a longer treatment persistence compared to oral NTX.

• Treatment persistence was superior for XR-NTX both at 3 months and 6 month follow up periods.

• No significant differences were found in ED visits or inpatient hospitalization rates between the two formulations.

• Overall, these findings suggest that XR-NTX may improve treatment retention in AUD, even in the context of substantial between-study heterogeneity.