Key Points
Question
What is the comparative effectiveness of buprenorphine-naloxone vs extended-release (XR) naltrexone for all-cause mortality and nonfatal opioid overdose among adults after discharge from medically managed opioid withdrawal treatment?
Findings
In this comparative effectiveness study using a target trial emulation framework of 106 052 discharge episodes from opioid withdrawal treatment followed up for 24 weeks, the cumulative incidence of nonfatal overdose was lower for buprenorphine-naloxone than for XR naltrexone, but risk of all-cause mortality was similar between the 2 treatment groups.
Meaning
Findings from this study using observational data provide further evidence of improved outcomes for individuals treated with buprenorphine-naloxone compared with XR naltrexone.
This comparative effectiveness trial compares the association of buprenorphine-naloxone vs extended-release (XR) naltrexone treatment with reduction of all-cause mortality and nonfatal opioid overdose among patients who completed medically managed opioid withdrawal.
Abstract
Importance
Buprenorphine-naloxone and extended-release (XR) naltrexone are both efficacious medications for opioid use disorder that reduce unregulated opioid use and improve opioid cravings and treatment retention. While prior comparative effectiveness studies demonstrated superiority of buprenorphine-naloxone over XR naltrexone for relapse and treatment retention, the comparative effectiveness of both treatments for all-cause mortality and nonfatal opioid overdose outcomes has not been rigorously tested.
Objective
To compare the effectiveness of buprenorphine-naloxone vs XR naltrexone after medically managed opioid withdrawal (MMOW) treatment.
Design, Setting, and Participants
This observational comparative effectiveness study using a target trial emulation framework leveraged individually linked administrative data from the Massachusetts Public Health Data Warehouse to emulate the protocol of the Extended-Release Naltrexone vs Buprenorphine for Opioid Treatment randomized clinical trial. Participants included adults 18 years or older who were discharged from MMOW in Massachusetts between January 1, 2014, and December 31, 2018. Individuals could meet eligibility criteria more than once. Data were analyzed from December 22, 2023, to January 30, 2025.
Exposures
Initiation of buprenorphine-naloxone or XR naltrexone therapy within 28 days of MMOW discharge.
Main Outcomes and Measures
Estimated 24-week cumulative incidence of all-cause mortality and nonfatal opioid overdose using inverse probability weighting to adjust for baseline and time-varying confounding.
Results
A total of 36 752 unique individuals who met eligibility criteria were identified and contributed 106 052 MMOW discharge episodes; 79 089 (74.6%) episodes were among individuals who were male and 44 463 (42.1%) were among individuals aged 18 to 29 years. By 28 days after MMOW discharge, initiation of buprenorphine-naloxone was observed among 14 194 discharge episodes (13.4%) and initiation of XR naltrexone was observed among 4734 (4.5%) episodes. The adjusted 24-week cumulative incidence of all-cause mortality after MMOW discharge was 1.4% (95% CI, 1.2%-1.7%) for buprenorphine-naloxone and 1.4% (95% CI, 1.1%-1.8%) for XR naltrexone, corresponding to a risk difference of 0.0 percentage points (pp) (95% CI, −0.4 to 0.4 pp). The adjusted 24-week cumulative incidence of nonfatal opioid overdose after MMOW discharge was 11.6% (95% CI, 10.8%-12.3%) for buprenorphine-naloxone and 13.9% (95% CI, 12.9%-15.0%) for XR naltrexone, corresponding to a risk difference of 2.3 pp (95% CI, 1.2-3.6 pp).
Conclusions and Relevance
In this comparative effectiveness study of buprenorphine-naloxone vs XR naltrexone, initiating buprenorphine-naloxone after MMOW discharge was associated with a lower risk of nonfatal opioid overdose at 24 weeks than initiating XR naltrexone, but a similar risk of all-cause mortality.
Introduction
Methadone, buprenorphine-naloxone, and extended-release (XR) naltrexone are US Food and Drug Administration–approved medications for opioid use disorder (MOUD) shown to be efficacious across a range of outcomes, including treatment retention and reducing unregulated opioid use.1,2,3,4,5,6,7,8,9,10 Prior studies suggest methadone and buprenorphine-naloxone may reduce fatal and nonfatal opioid overdose and all-cause mortality,4,11,12,13 with observational data also demonstrating lower overdose rates among individuals receiving methadone or buprenorphine-naloxone than those who receive XR naltrexone.4,14 However, comparative effectiveness trials of different MOUD that examine all-cause mortality and nonfatal overdose outcomes and rigorously account for differences in individuals who receive these MOUD have not been conducted.
Although randomized clinical trials (RCTs) represent an essential step in determining the safety and efficacy of MOUD treatments, trials are typically underpowered to study critical but less frequent public health outcomes such as all-cause mortality or overdose.15 For example, the Extended-Release Naltrexone vs Buprenorphine for Opioid Treatment (X:BOT) trial was a randomized comparative effectiveness trial of 24 weeks of treatment with buprenorphine-naloxone vs XR naltrexone following acute admission to a community-based medically managed opioid withdrawal (MMOW) program. The primary outcome was time to relapse to regular opioid use by 24 weeks, while secondary safety outcomes were reported to 36 weeks and included all-cause mortality and overdose (combined fatal and nonfatal, involving any substance).16 Neither the original X:BOT trial report nor any of its reanalyses performed any statistical analysis to assess mortality differences. For overdose, both the original publication16 and a follow-up analysis by X:BOT trial authors17 using additional, more inclusive classifications of overdose events found no significant differences in overdose rates between the treatment groups. However, a separate reanalysis of X:BOT trial data by Ajazi et al18,19 found a significantly higher overdose risk in the XR naltrexone group, although this conclusion was sensitive to overdose ascertainment method and the analytic approach.
While powering RCTs such as the X:BOT trial to examine either mortality or overdose as outcomes may not be feasible, observational studies with administrative data offer the potential for larger sample sizes that enable robust evaluation of these infrequent outcomes. Target trial emulation is a framework that can help minimize bias and improve causal inference in analyses of observational data.20 In prior work, several authors of the present study used administrative data from the Massachusetts Public Health Data Warehouse (PHD) to emulate the X:BOT RCT protocol and bench mark this observational analysis against X:BOT trial data for the outcome of treatment discontinuation.20,21 In the present study, we use target trial emulation and data from the PHD to compare the effectiveness of initiating buprenorphine-naloxone vs XR naltrexone for reduction of all-cause mortality and nonfatal opioid overdose.
Methods
Study Design and Data Source
We conducted an observational comparative effectiveness study using a target trial emulation framework and deidentified data from the Massachusetts PHD. The PHD includes all Massachusetts residents with at least 1 claim in the Massachusetts All Payer Claims Database and longitudinally links claims to records from 32 other government datasets, 8 of which were used in this study (eTables 1 and 2 in Supplement 1 provide data sources and variable specifications). The Boston University Medical Campus Institutional Review Board determined this study was not human participants research and therefore waived the need for study approval and informed consent. We report study procedures and results in accordance with the Transparent Reporting of Observational Studies Emulating a Target Trial (TARGET) statement.
Target Trial Emulation
We adapted a target trial emulation framework from a prior study comparing a reanalysis of X:BOT data with observational data from the PHD for the outcome of treatment interruption.20 In that study, X:BOT was the index trial, and we harmonized a hypothetical target trial protocol with each of X:BOT’s components: eligibility criteria, treatment strategies and assignment, outcome definitions, follow-up, and causal contrasts.20 We summarize each of these components for our present study and include a comparison between the our study’s target trial and observational emulation in eTable 3 in Supplement 1. eFigure 1 in Supplement 1 depicts the study design, including the eligibility and treatment assignment time windows. We preregistered the study protocol prior to examining outcomes.22
Eligibility Criteria
To emulate X:BOT eligibility criteria within the PHD, we included individuals 18 years or older discharged from voluntary MMOW between January 1, 2014, through December 31, 2018. Data were analyzed from December 22, 2023, to January 30, 2025. We excluded individuals with diagnoses of psychosis or suicidal ideation or attempt in the 6 months before MMOW admission. We excluded individuals with receipt of methadone in the 30 days prior to MMOW admission and those with long-term opioid analgesic therapy, defined as at least 3 opioid prescriptions at least 21 days apart with at least 84 days supplied in the past 180 days. We identified and excluded pregnant individuals using vital statistics records for live births or fetal deaths within 42 weeks of MMOW discharge. Individuals could be included more than once if they met eligibility criteria at multiple MMOW discharges. Race and ethnicity data were self-reported by individuals at the time of intake into MMOW as Hispanic, non-Hispanic Black, non-Hispanic White, and non-Hispanic other (including American Indian or Alaska Native, Asian, Pacific Islander, or other). These data were included to enable adjustment for race and ethnicity, as structural racism is known to impact access to and receipt of medications for opioid use disorder and risk of adverse outcomes.
Treatment Strategies and Assignment
The treatment strategies were initiation of XR naltrexone or buprenorphine-naloxone within 28 days of MMOW discharge. We assigned episodes of MMOW discharge to either the XR naltrexone or the buprenorphine-naloxone treatment group based on pharmacy and medical claims.
Outcomes
Outcomes were all-cause mortality and nonfatal opioid overdose within 24 weeks of MMOW discharge. We ascertained all-cause mortality using death certificate data and nonfatal opioid-specific overdose using data from ambulance encounters and acute care hospital discharge records (emergency department, outpatient observation, and inpatient) from the PHD. The ambulance and hospital discharge opioid overdose records were deduplicated and then linked with the death certificate data to confirm that the opioid overdose was not fatal.
Follow-Up and Causal Contrasts
Follow-up for each episode began on the day of MMOW discharge and ended at the earliest date of the outcome, 24 weeks, or death (for the nonfatal opioid overdose outcome only). We estimated the observational analog of a per-protocol effect, which is the effect of receiving either XR naltrexone or buprenorphine-naloxone at least once within 28 days of MMOW discharge.
Statistical Analysis
We calculated standardized mean differences to assess for balance in baseline characteristics between discharge episodes assigned to either the XR naltrexone or the buprenorphine-naloxone treatment group. We considered a standardized mean difference less than 0.1 to indicate good balance between the 2 groups.23
We used the cloning-censoring-weighting approach to adjust for confounding and selection bias. More specifically, we copied each discharge episode into 2 clones, with 1 clone assigned to each treatment strategy.24,25 We censored the clones if and when they initiated the nonassigned treatment strategy, methadone, an opioid analgesic, or buprenorphine for pain management during the 28-day grace period after MMOW discharge or at day 28 if they did not initiate the assigned treatment. Receipt of oral naltrexone during the grace period did not affect treatment group eligibility. Because the risk of all-cause mortality was low and did not differ between treatment groups, when the outcome was nonfatal opioid overdose, we also censored follow-up at death. Outcomes were counted beginning on the day of MMOW discharge, including during the 28-day treatment assignment window.
We then fit weighted pooled logistic regression models to estimate the 24-week risk of all-cause mortality and nonfatal opioid overdose under each treatment strategy, as well as the between-group 24-week absolute risk difference for both outcomes. Outcome models included an indicator of treatment group, time in months, restricted cubic splines for time since baseline, and the interaction between treatment and time.
To adjust outcome models for selection bias induced by censoring, we applied a daily time-varying weight inversely proportional to the probability of not deviating from the assigned treatment strategy during the grace period. These weights were estimated using a pooled logistic regression model for the daily probability of being uncensored and included baseline confounders (age; gender; race or Hispanic ethnicity; homelessness; educational level; employment status; marital status; use of stimulants, sedatives, or cannabis; insurance type; counts of recent nonfatal opioid overdoses and MMOW admissions; history of incarceration; diagnoses of anxiety, depression, and bipolar disorder; prior MOUD, opioid analgesic, or benzodiazepine use; and Elixhauser score) and daily residential treatment status as a postbaseline confounder. We used a bootstrap with 500 repetitions to compute percentile-based 95% CIs. We used SAS Enterprise, release 3.81 (SAS Institute Inc), for all analyses.
Subgroup and Sensitivity Analyses
We conducted several prespecified subgroup and sensitivity analyses to assess the robustness of our estimates. First, because many individuals did not start either buprenorphine-naloxone or XR naltrexone in the 28-day grace period after MMOW discharge, we restricted the analysis to only MMOW discharge episodes with initiation of the assigned treatment. Second, we conducted 3 analyses assessing whether the effect of buprenorphine-naloxone or XR naltrexone on outcomes differed by educational level, employment status, or housing status.21 For each of these analyses, we added an interaction term between treatment group and an indicator of educational level (some college or more vs high school, General Educational Development credential, or less), employment status (full- or part-time employment vs not employed), or housing status (those reporting homelessness vs housed status at MMOW admission) into outcome models. Finally, we conducted additional sensitivity analyses restricting to the first MMOW discharge per individual (the first ever observed episode) and allowing a 14-day rather than a 28-day grace period for assigned treatment initiation.
Medication Possession
As prior work identified poor retention to buprenorphine-naloxone or XR naltrexone following MMOW,21 we conducted an exploratory analysis to determine whether outcomes during follow-up occurred on days with or without MOUD possession as a proxy for treatment adherence. We computed crude rates of all-cause mortality and first nonfatal opioid overdose based on daily medication possession status, restricting to the first MMOW discharge after which an individual initiated their assigned treatment. We calculated outcome rates as events per 100 person-years for each of 5 non–mutually exclusive medication possession status categories: (1) any status, (2) possession of the assigned treatment on the day of the event, (3) possession of any MOUD on the day of the event, (4) no MOUD on the day of the event regardless of time since last possession, and (5) no MOUD on the day of the event with last possession within the previous 28 days (ie, recent discontinuation).
Results
In the emulated trial, 36 752 unique individuals met eligibility criteria, contributing 106 052 MMOW discharge episodes (Figure 1). By 28 days after MMOW discharge, initiation of buprenorphine-naloxone was observed among 14 194 discharge episodes (13.4%) and initiation of XR naltrexone was observed among 4734 (4.5%) episodes. Individuals represented in 79 089 (74.6%) of these discharge episodes were male and 26 963 (25.4%) were female, and 44 463 (42.1%) were aged 18 to 29 years (Table 1). In terms of race and ethnicity, individuals represented in 17 589 discharge episodes (16.6%) identified as Hispanic, 5052 (4.8%) as non-Hispanic Black, 79 918 (75.4%) as non-Hispanic White, and 3493 (3.3%) as another race and/or ethnicity. Compared with discharge episodes after which buprenorphine-naloxone was initiated, those with XR naltrexone initiation were more likely to be among individuals who were younger, non-Hispanic White, employed full or part time, or commercially insured and less likely to be among individuals who reported homelessness or previous incarceration at MMOW intake. In addition, discharge episodes with XR naltrexone initiation were more likely to be among individuals with recent naltrexone use, while those with buprenorphine-naloxone initiation were more likely to be among individuals with prior buprenorphine use.
Figure 1. Flowchart of Eligible Medically Managed Opioid Withdrawal (MMOW) Discharge Episodes Included in the Emulated Trial.
aReasons for exclusion may include more than 1 criterion.
Table 1. Baseline Characteristics of Medically Managed Opioid Withdrawal Discharge Episodes Meeting Eligibility Criteria, 2014-2018.
| Characteristic | MMOW discharge episodes, No. (%) | Standardized mean differencea | ||
|---|---|---|---|---|
| Overall (n = 106 052) | Initiated buprenorphine-naloxone (n = 14 194) | Initiated XR naltrexone (n = 4734) | ||
| Age, y | ||||
| 18-29 | 44 643 (42.1) | 4801 (33.8) | 2462 (52.0) | 0.41 |
| 30-44 | 46 955 (44.3) | 7131 (50.2) | 1963 (41.5) | |
| ≥45 | 14 454 (13.6) | 2262 (15.9) | 309 (6.5) | |
| Gender | ||||
| Female | 26 963 (25.4) | 3388 (23.9) | 1202 (25.4) | 0.04 |
| Male | 79 089 (74.6) | 10 806 (76.1) | 3532 (74.6) | |
| Race and ethnicityb | ||||
| Non-Hispanic Black | 5052 (4.8) | 581 (4.1) | 119 (2.5) | 0.23 |
| Hispanic | 17 589 (16.6) | 2469 (17.4) | 466 (9.8) | |
| Non-Hispanic White | 79 918 (75.4) | 10 766 (75.8) | 4034 (85.2) | |
| Non-Hispanic otherc | 3493 (3.3) | 378 (2.7) | 115 (2.4) | |
| Educational attainment | ||||
| Less than high school | 29 012 (27.4) | 3845 (27.1) | 1043 (22.0) | 0.16 |
| High school or GED | 49 352 (46.5) | 6607 (46.5) | 2197 (46.4) | |
| Some college and above | 23 683 (22.3) | 3284 (23.1) | 1278 (27.0) | |
| Other, unknown, or missing | 4005 (3.8) | 458 (3.2) | 216 (4.6) | |
| Employment | ||||
| Full time | 10 478 (9.9) | 1333 (9.4) | 834 (17.6) | 0.40 |
| Part time | 3648 (3.4) | 480 (3.4) | 241 (5.1) | |
| Unemployed | 82 201 (77.5) | 10 730 (75.6) | 3451 (72.9) | |
| Other | 9725 (9.2) | 1651 (11.6) | 208 (4.4) | |
| Marital status | ||||
| Never married | 86 150 (81.2) | 11 364 (80.1) | 3941 (83.2) | 0.10 |
| Married or partnership | 7842 (7.4) | 1098 (7.7) | 394 (8.3) | |
| Separated, divorced, or widowed | 11 388 (10.7) | 1626 (11.5) | 381 (8.0) | |
| Missing | 672 (0.6) | 106 (0.7) | 18 (0.4) | |
| No. of prior drug overdoses reported at MMOW intake | ||||
| 0 | 60 065 (56.6) | 7297 (51.4) | 2517 (53.2) | 0.12 |
| 1 | 13 714 (12.9) | 1878 (13.2) | 694 (14.7) | |
| 2 | 9105 (8.6) | 1275 (9.0) | 470 (9.9) | |
| ≥3 | 23 139 (21.8) | 3741 (26.4) | 1052 (22.2) | |
| Missing | 29 (0.03) | NRd | NRd | |
| Self-reported homelessness | 24 106 (22.7) | 3508 (24.7) | 605 (12.8) | 0.31 |
| Primary drug used | ||||
| Heroin or fentanyl | 100 486 (94.8) | 13 465 (94.9) | 4371 (92.3) | 0.06 |
| Prescription opioids | 892 (0.8) | 224 (1.6) | 34 (0.7) | |
| Nonprescription opioids | 4674 (4.4) | 505 (3.6) | 329 (6.9) | |
| Heavy alcohol usee | 16 491 (15.5) | 2451 (17.3) | 600 (12.7) | 0.13 |
| Stimulant usee | 13 717 (12.9) | 2021 (14.2) | 497 (10.5) | 0.11 |
| Sedative usee | 12 919 (12.2) | 2293 (16.2) | 553 (11.7) | 0.13 |
| Cannabis usee | 3015 (2.8) | 348 (2.5) | 180 (3.8) | −0.08 |
| Previous opioid MMOWf | ||||
| 0 | 51 138 (48.2) | 6870 (48.4) | 2265 (47.8) | 0.12 |
| 1 | 23 353 (22.0) | 3073 (21.6) | 1195 (25.2) | |
| 2 | 12 546 (11.8) | 1658 (11.7) | 598 (12.6) | |
| ≥3 | 19 015 (17.9) | 2593 (18.3) | 676 (14.3) | |
| Insurance | ||||
| Commercial | 6856 (6.5) | 667 (4.7) | 547 (11.6) | 0.30 |
| Medicaid | 77 269 (72.9) | 11 603 (81.7) | 3655 (77.2) | |
| Medicare | 653 (0.6) | 136 (1.0) | NRd | |
| Other | 2516 (2.4) | 219 (1.5) | 66 (1.4) | |
| Missing | 18 758 (17.7) | 1569 (11.1) | 457 (9.7) | |
| Previous incarcerationf | 18 380 (17.3) | 2700 (19.0) | 588 (12.4) | 0.18 |
| Previous methadone usef | 6146 (5.8) | 717 (5.1) | 211 (4.5) | 0.03 |
| Previous buprenorphine usef | 26 829 (25.3) | 9353 (65.9) | 1142 (24.1) | 0.93 |
| Previous naltrexone usef | 8022 (7.6) | 863 (6.1) | 1437 (30.4) | −0.66 |
| Anxietyf | 29 247 (27.6) | 5670 (39.9) | 1583 (33.4) | 0.14 |
| Depressionf | 27 292 (25.7) | 5087 (35.8) | 1337 (28.2) | 0.16 |
| Bipolar disorderf | 8808 (8.3) | 1598 (11.3) | 380 (8.0) | 0.11 |
| Opioid analgesic usef | 16 756 (15.8) | 2127 (15.0) | 639 (13.5) | 0.04 |
| Benzodiazepine usef | 12 942 (12.2) | 2591 (18.3) | 723 (15.3) | 0.08 |
| Elixhauser scoref | ||||
| 0 | 70 009 (66.0) | 7895 (55.6) | 3160 (66.8) | 0.28 |
| 1 | 20 825 (19.6) | 3398 (23.9) | 1030 (21.8) | |
| 2 | 8431 (7.9) | 1535 (10.8) | 341 (7.2) | |
| ≥3 | 6787 (6.4) | 1366 (9.6) | 203 (4.3) | |
Abbreviations: GED, General Educational Development; MMOW, medically managed opioid withdrawal; NR, not reported; XR, extended release.
Used to evaluate the difference in patients’ characteristics between 2 groups. A standardized mean difference less than 0.1 is considered indicative of good balance between the 2 groups.23
Race and ethnicity data are self-reported by individuals at the time of intake into MMOW.
Includes non-Hispanic American Indian or Alaska Native, Asian, Pacific Islander, or other.
Cells suppressed cells due to counts less than 11.
Substance identified as a secondary or tertiary substance by an individual upon MMOW intake.
Evidence from administrative data of exposure in the 6 months prior to MMOW intake.
Primary Outcomes
There were 184 deaths and 1511 nonfatal opioid overdoses among MMOW discharge episodes with buprenorphine-naloxone initiation and 68 deaths and 615 nonfatal overdoses among those with XR naltrexone initiation. Opioid overdose was the cause of 199 of 252 deaths (79.0%). The adjusted 24-week incidence of all-cause mortality was 1.4% (95% CI, 1.2%-1.7%) for buprenorphine-naloxone and 1.4% (95% CI, 1.1%-1.8%) for XR naltrexone, corresponding to an absolute risk difference of 0.0 percentage points (pp) (95% CI, −0.4 to 0.4 pp) (Figure 2 and Figure 3A). The adjusted 24-week incidence of nonfatal overdose was 11.6% (95% CI, 10.8%-12.3%) for buprenorphine-naloxone and 13.9% (95% CI, 12.9%-15.0%) for XR naltrexone, corresponding to an absolute risk difference of 2.3 pp (95% CI, 1.2-3.6 pp) (Figure 2 and Figure 3B).
Figure 2. Estimated 24-Week Risks and Dot Plots of Risk Differences for All-Cause Mortality and Nonfatal Opioid Overdose.
Comparison groups include medically managed opioid withdrawal (MMOW) discharge episodes with initiation of buprenorphine-naloxone (BUP-NX) versus extended-release naltrexone (XR-NTX) in 2014 to 2018 in main and subgroup and sensitivity analyses.
Figure 3. Cumulative Incidence Curves of All-Cause Mortality and Nonfatal Opioid Overdose.
MMOW indicates medically managed opioid withdrawal.
Subgroup and Sensitivity Analysis Outcomes
In general, results for both primary outcomes were similar across subgroup and sensitivity analyses (Figure 2). One exception was the analysis restricted to the first MMOW discharge per individual for the nonfatal opioid overdose outcome; rates were lower in each treatment group and buprenorphine-naloxone was no longer favored over XR naltrexone (risk difference, 1.2 pp [95% CI, −0.2 to 2.5 pp) (Figure 2).
Medication Possession
Beyond the first 28 days following MMOW discharge, a higher percentage of individuals in episodes with initiation of buprenorphine-naloxone possessed either their assigned MOUD or any MOUD over time than those who initiated XR naltrexone (eFigure 2 in Supplement 1). In analyses describing unadjusted outcome rates based on medication possession status on the day of the event among individuals initiating treatment, rates of both all-cause mortality and first nonfatal opioid overdose were higher for those without possession of any MOUD compared with those with possession of either the assigned treatment or any MOUD (Table 2).
Table 2. Unadjusted Risk of All-Cause Mortality and Nonfatal Overdose Based on Possession of Assigned Treatment or Any MOUD on the Day of the Event.
| Variable | Any possession status | Medication possession status on the day of the event | |||
|---|---|---|---|---|---|
| Assigned treatment | Any MOUD | No MOUD, regardless of time since last possession | No MOUD, with last possession ≤28 d prior | ||
| MMOW discharge episodes with initiation of buprenorphine-naloxone | |||||
| All-cause mortality | |||||
| No. of events | 184 | 72 | 72 | 112 | 46 |
| Person-days at risk | 2 370 421 | 1 281 833 | 1 388 973 | 981 448 | 453 385 |
| Events/100 person-years (95% CI) | 2.8 (0.5-8.5) | 2.1 (0.3-7.4) | 1.9 (0.2-7.1) | 4.2 (1.2-10.5) | 3.7 (0.9-9.8) |
| First nonfatal overdose | |||||
| No. of events | 1511 | 724 | 752 | 759 | 407 |
| Person-days at risk | 2 217 466 | 1 218 802 | 1 315 373 | 902 093 | 417 816 |
| Events/100 person-years (95% CI) | 24.9 (16.1-36.8) | 21.7 (13.6-32.9) | 20.9 (12.9-32.0) | 30.7 (20.8-43.6) | 35.6 (24.9-49.4) |
| MMOW discharge episodes with initiation of XR naltrexone | |||||
| All-cause mortality | |||||
| No. of events | 68 | 23 | 25 | 43 | 16 |
| Person-days at risk | 789 156 | 353 233 | 393 459 | 395 697 | 153 676 |
| Events/100 person-years (95% CI) | 3.1 (0.7-8.9) | 2.4 (0.4-7.9) | 2.3 (0.3-7.7) | 4.0 (1.1-10.2) | 3.8 (1.0-10.0) |
| First nonfatal overdose | |||||
| No. of events | 615 | 181 | 200 | 415 | 192 |
| Person-days at risk | 726 606 | 336 249 | 370 864 | 355 742 | 141 450 |
| Events/100 person-years (95% CI) | 30.9 (20.9-43.9) | 19.6 (11.9-30.4) | 19.7 (12.0-30.5) | 42.6 (30.8-57.5) | 49.5 (36.7-65.3) |
Abbreviations: MMOW, medically managed opioid withdrawal; MOUD, medication for opioid use disorder.
Discussion
Using a target trial framework to examine a statewide sample in Massachusetts, we found that initiating buprenorphine-naloxone vs XR naltrexone was not associated with an absolute risk difference in all-cause mortality but was associated with a 2.4-pp risk reduction in nonfatal opioid overdose in the 24 weeks following MMOW discharge. Study findings were generally robust across subgroup and sensitivity analyses. One exception was the lack of significant difference in this outcome between buprenorphine-naloxone and XR naltrexone when restricting to the first MMOW discharge per individual. In this sensitivity analysis, nonfatal overdose rates were lower in each group. Although the effect estimate was in the same direction as the main analysis, the 95% CI included the null, likely reflecting lower power to detect a difference. While restricting to the first MMOW discharge may more closely mirror clinical trial enrollment procedures with a single enrollment allowed per individual, the findings from the main analysis allowing repeated eligibility events likely more closely reflect population-level impacts in non-RCT settings where MMOW readmission is common, especially among individuals with more severe opioid use disorder.
Comparing our findings with 2 prospective RCTs of buprenorphine-naloxone vs XR naltrexone may help contextualize our results. A 12-week trial in Norway with 232 participants found no deaths and only 1 overdose event.26 X:BOT, which served as the target trial for our emulation, was not powered to study all-cause mortality or nonfatal overdose and ascertained overdose events differently than in our study. In X:BOT, only 3 of 570 individuals (0.5%) died by the end of 24 weeks of follow-up (P. Novo, MPA, MPH, electronic communication, December 2, 2025), a rate approximately one-third of that observed in our study. Overdose rates were also substantially lower in X:BOT compared with our study, despite use of more inclusive overdose definitions in both the original trial publication and its reanalyses. These more inclusive definitions counted not just nonfatal opioid-specific overdoses as in our study, but also fatal and self-reported nonfatal events involving any substance.16,17,18 In their reanalysis of X:BOT data, Ajazi et al18,19 report 14 of 283 (4.9%) individuals in the XR naltrexone group and 4 of 287 (1.4%) of those in the buprenorphine-naloxone group had a fatal or nonfatal overdose involving any substance by 24 weeks. In contrast, the overdose rates from our analysis, which were ascertained from hospital and ambulance encounters and restricted to only nonfatal opioid overdose, were more than 2 times higher (13.9% for XR naltrexone and 11.6% for buprenorphine-naloxone). Despite these differences in magnitude and overdose event classification, the between-group risk reduction for overdose observed in the X:BOT reanalysis was similar to that observed in our target trial emulation.
While considered rare events in the context of RCTs, the all-cause mortality (1.4%) and nonfatal opioid overdose (11.6% and 13.9%) rates observed within 24 weeks of follow-up in our study are very high. We note these data are from 2014 to 2018 in alignment with the X:BOT target trial, a period during which overdose rates were substantially increasing in Massachusetts.27 Despite the proven efficacy of MOUD, one substantive reason for poor outcomes in those initiating MOUD in nontrial settings is poor retention. In a prior study using the same target trial emulation framework also using the Massachusetts PHD,21 several authors from the present study found that the 24-week risk of treatment interruption (a composite outcome of medication discontinuation, incarceration, MMOW readmission, or death) to be 82% among those initiating buprenorphine-naloxone and 93% among those initiating XR naltrexone, with a risk difference favoring buprenorphine-naloxone by 11 pp (95% CI, −13 to −10 pp). In that study, we also formally reanalyzed X:BOT data with a harmonized protocol and found that X:BOT participants exhibited substantially lower treatment interruption rates of 68% for buprenorphine-naloxone and 72% for XR naltrexone.21 Worse outcomes in observational vs RCT settings may be due to differences in populations willing to participate in trials or other aspects of participating in a trial such as financial incentives to attend study visits. Regardless of reason, these differences highlight the importance of studying MOUD treatment effectiveness in less idealized settings.
The results from our study are largely consistent with other observational studies that have compared buprenorphine-naloxone and XR naltrexone, albeit with varied methodologies and settings. Two studies of commercial insurance claims examined fatal or nonfatal overdose within medical encounters.4,14 The first found that active treatment with buprenorphine-naloxone but not XR naltrexone was associated with reduced risk of opioid overdose compared with no treatment.14 The second found that after a new diagnosis of OUD, compared with no treatment, individuals initiating buprenorphine-naloxone or methadone, but not naltrexone (oral or injectable) were found to have reduced risk of opioid overdose at as long as 12 months of follow-up.4 Retrospective cohort studies of individuals with OUD initiating MOUD in Australia found no differences between XR naltrexone and buprenorphine-naloxone in both mortality and opioid overdose outcomes.28,29 Similarly, a study of Medicaid claims from 2 states also found no difference between buprenorphine-naloxone and XR naltrexone for the composite end point of overdose or death.30
Limitations
This study has several limitations. First, as with any study using administrative data, exposures and outcomes are subject to misclassification. In particular, many nonfatal opioid overdoses are managed in the community without contact with an ambulance or hospital and therefore would not be captured in our study’s data sources.31 Additionally, incomplete or inaccurate linkages between PHD datasets may contribute to misclassification, although the PHD’s match rates are considered excellent for administrative data.32 We do not have any reason to believe that any misclassification in this study differed systematically by treatment group. Second, results are subject to unmeasured confounding, including confounding by indication, that may not be eliminated through use of a target trial framework. However, our comparison of treatment options that are in clinical equipoise at the time of MMOW discharge helps mitigate confounding by indication by restricting the study sample to individuals with similar treatment indications.33 In addition, the PHD allowed us to adjust for a large selection of demographic and clinical characteristics that are known to be associated with treatment choice and outcomes. Third, our results may not be generalizable outside Massachusetts, a state with higher-than-average opioid overdose mortality, high rates of insurance coverage, and high availability of harm reduction and treatment services.
Conclusions
In this comparative effectiveness study of buprenorphine-naloxone vs XR naltrexone, initiation of buprenorphine-naloxone after MMOW discharge was associated with lower incidence of nonfatal opioid overdose compared with XR naltrexone, but no difference in all-cause mortality. Rates of opioid overdose and mortality are high following MMOW, even among individuals who successfully initiate treatment with either buprenorphine-naloxone or XR naltrexone. However, adverse outcomes may be associated with poor treatment retention, which should remain a focus of ongoing and new interventions. These data provide further evidence of improved outcomes for individuals treated with buprenorphine-naloxone compared with XR naltrexone.
eTable 1. Description of the Datasets and Sources Included in the Massachusetts Department of Public Health Data Warehouse
eTable 2. Variable Specification and Data Sources
eTable 3. Key Components of the Protocol for a Hypothetical Target Trial and the Observational Emulation Based on Data From the Massachusetts Public Health Database
eFigure 1. Diagram of Study Design, Including Time Periods for Identifying Exclusion Criteria, Baseline and Time-Varying Covariates, Treatment Assignment, and Follow-Up
eFigure 2. Percentage of Medically Managed Opioid Withdrawal (MMOW) Discharge Episodes That Initiated Assigned Treatment With Possession of the Assigned Treatment or Any MOUD Over Time
Data Sharing Statement
References
- 1.Mattick RP, Breen C, Kimber J, Davoli M. Buprenorphine maintenance versus placebo or methadone maintenance for opioid dependence. Cochrane Database Syst Rev. 2014;2014(2):CD002207. doi: 10.1002/14651858.CD002207.pub4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Nielsen S, Larance B, Degenhardt L, Gowing L, Kehler C, Lintzeris N. Opioid agonist treatment for pharmaceutical opioid dependent people. Cochrane Database Syst Rev. 2016;(5):CD011117. doi: 10.1002/14651858.CD011117.pub2 [DOI] [PubMed] [Google Scholar]
- 3.Sees KL, Delucchi KL, Masson C, et al. Methadone maintenance vs 180-day psychosocially enriched detoxification for treatment of opioid dependence: a randomized controlled trial. JAMA. 2000;283(10):1303-1310. doi: 10.1001/jama.283.10.1303 [DOI] [PubMed] [Google Scholar]
- 4.Wakeman SE, Larochelle MR, Ameli O, et al. Comparative effectiveness of different treatment pathways for opioid use disorder. JAMA Netw Open. 2020;3(2):e1920622. doi: 10.1001/jamanetworkopen.2019.20622 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Substance Abuse and Mental Health Services Administration . Medications for opioid use disorder for healthcare and addiction professionals, policymakers, patients, and families. Treatment Improvement Protocol (TIP) 63. 2020. Accessed June 8, 2026. https://library.samhsa.gov/sites/default/files/pep21-02-01-002.pdf [PubMed]
- 6.Caldiero RM, Parran TV Jr, Adelman CL, Piche B. Inpatient initiation of buprenorphine maintenance vs. detoxification: can retention of opioid-dependent patients in outpatient counseling be improved? Am J Addict. 2006;15(1):1-7. doi: 10.1080/10550490500418989 [DOI] [PubMed] [Google Scholar]
- 7.Liebschutz JM, Crooks D, Herman D, et al. Buprenorphine treatment for hospitalized, opioid-dependent patients: a randomized clinical trial. JAMA Intern Med. 2014;174(8):1369-1376. doi: 10.1001/jamainternmed.2014.2556 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Comer SD, Sullivan MA, Yu E, et al. Injectable, sustained-release naltrexone for the treatment of opioid dependence: a randomized, placebo-controlled trial. Arch Gen Psychiatry. 2006;63(2):210-218. doi: 10.1001/archpsyc.63.2.210 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Krupitsky E, Nunes EV, Ling W, Gastfriend DR, Memisoglu A, Silverman BL. Injectable extended-release naltrexone (XR-NTX) for opioid dependence: long-term safety and effectiveness. Addiction. 2013;108(9):1628-1637. doi: 10.1111/add.12208 [DOI] [PubMed] [Google Scholar]
- 10.Lee JD, Friedmann PD, Kinlock TW, et al. Extended-release naltrexone to prevent opioid relapse in criminal justice offenders. N Engl J Med. 2016;374(13):1232-1242. doi: 10.1056/NEJMoa1505409 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Sordo L, Barrio G, Bravo MJ, et al. Mortality risk during and after opioid substitution treatment: systematic review and meta-analysis of cohort studies. BMJ. 2017;357:j1550. doi: 10.1136/bmj.j1550 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Larochelle MR, Bernson D, Land T, et al. Medication for opioid use disorder after nonfatal opioid overdose and association with mortality: a cohort study. Ann Intern Med. 2018;169(3):137-145. doi: 10.7326/M17-3107 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Santo T Jr, Clark B, Hickman M, et al. Association of opioid agonist treatment with all-cause mortality and specific causes of death among people with opioid dependence: a systematic review and meta-analysis. JAMA Psychiatry. 2021;78(9):979-993. doi: 10.1001/jamapsychiatry.2021.0976 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Morgan JR, Schackman BR, Weinstein ZM, Walley AY, Linas BP. Overdose following initiation of naltrexone and buprenorphine medication treatment for opioid use disorder in a United States commercially insured cohort. Drug Alcohol Depend. 2019;200:34-39. doi: 10.1016/j.drugalcdep.2019.02.031 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Kornør H, Lobmaier PPK, Kunøe N. Sustained-release naltrexone for opioid dependence. Cochrane Database Syst Rev. 2025;5(5):CD006140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Lee JD, Nunes EV Jr, Novo P, et al. Comparative effectiveness of extended-release naltrexone versus buprenorphine-naloxone for opioid relapse prevention (X:BOT): a multicentre, open-label, randomised controlled trial. Lancet. 2018;391(10118):309-318. doi: 10.1016/S0140-6736(17)32812-X [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Lee JD, Nunes EV, Novo P, et al. Commentary on Ajazi et al (2021) Re-analysis of the X:BOT Trial. J Addict Med. 2022;16(4):382-385. doi: 10.1097/ADM.0000000000000930 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Ajazi EM, Dasgupta N, Marshall SW, et al. Revisiting the X:BOT naltrexone clinical trial using a comprehensive survival analysis. J Addict Med. 2022;16(4):440-446. doi: 10.1097/ADM.0000000000000931 [DOI] [PubMed] [Google Scholar]
- 19.Dasgupta N, Ajazi EM, Schwartz TA, Marshall SW. Misclassification of overdose events in the X:BOT study. Lancet. 2023;402(10401):526-527. doi: 10.1016/S0140-6736(23)00113-7 [DOI] [PubMed] [Google Scholar]
- 20.Christine PJ, Lodi S, Hsu HE, et al. Target trial emulation for comparative effectiveness research with observational data: promise and challenges for studying medications for opioid use disorder. Addiction. 2024;119(7):1313-1321. doi: 10.1111/add.16473 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Lodi S, Yan S, Bovell-Ammon B, et al. Comparative effectiveness of extended-release naltrexone versus buprenorphine-naloxone on treatment interruption: comparing findings from a reanalysis of the X:BOT RCT and harmonized target trial emulation using population-based observational data. Addiction. 2025;120(8):1634-1645. doi: 10.1111/add.70040 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Larochelle M. COMPEET MOUD Aim 2 Protocol. Accessed December 23, 2025. https://osf.io/rkbxt/overview
- 23.Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46(3):399-424. doi: 10.1080/00273171.2011.568786 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Lodi S, Günthard HF, Gill J, et al. Effectiveness of transmitted drug resistance testing before initiation of antiretroviral therapy in HIV-positive individuals. J Acquir Immune Defic Syndr. 2019;82(3):314-320. doi: 10.1097/QAI.0000000000002135 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Hernán MA, Robins JM. Causal Inference: What If. Taylor & Francis; 2020. [Google Scholar]
- 26.Tanum L, Solli KK, Latif ZE, et al. Effectiveness of injectable extended-release naltrexone vs daily buprenorphine-naloxone for opioid dependence: a randomized clinical noninferiority trial. JAMA Psychiatry. 2017;74(12):1197-1205. doi: 10.1001/jamapsychiatry.2017.3206 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Data brief: opioid-related overdose deaths among Massachusetts residents. Massachusetts Department of Public Health . June 2024. Accessed January 21, 2026. https://www.mass.gov/doc/opioid-related-overdose-deaths-among-ma-residents-june-2024-0/download
- 28.Kelty E, Joyce D, Hulse G. A retrospective cohort study of mortality rates in patients with an opioid use disorder treated with implant naltrexone, oral methadone or sublingual buprenorphine. Am J Drug Alcohol Abuse. 2019;45(3):285-291. doi: 10.1080/00952990.2018.1545131 [DOI] [PubMed] [Google Scholar]
- 29.Kelty E, Hulse G. Fatal and non-fatal opioid overdose in opioid dependent patients treated with methadone, buprenorphine or implant naltrexone. Int J Drug Policy. 2017;46:54-60. doi: 10.1016/j.drugpo.2017.05.039 [DOI] [PubMed] [Google Scholar]
- 30.Ross RK, Nunes EV, Olfson M, et al. Comparative effectiveness of extended-release naltrexone and sublingual buprenorphine for treatment of opioid use disorder among Medicaid patients. Addiction. 2024;119(11):1975-1986. doi: 10.1111/add.16630 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Smith RJ, McMahan VM, McCann NC, et al. Emergency medical system response, emergency department visits, and hospital admissions in response to non-fatal opioid overdoses reported by a cohort of overdose survivors in San Francisco and Boston, 2019-2022. Drug Alcohol Depend Rep. 2025;16:100351. doi: 10.1016/j.dadr.2025.100351 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.An assessment of fatal and nonfatal opioid overdoses in Massachusetts (2011-2015). Massachusetts Department of Public Health. August 16, 2017. Accessed April 10, 2025. https://www.mass.gov/doc/legislative-report-chapter-55-opioid-overdose-study-august-2017/download
- 33.Sendor R, Stürmer T. Core concepts in pharmacoepidemiology: confounding by indication and the role of active comparators. Pharmacoepidemiol Drug Saf. 2022;31(3):261-269. doi: 10.1002/pds.5407 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Description of the Datasets and Sources Included in the Massachusetts Department of Public Health Data Warehouse
eTable 2. Variable Specification and Data Sources
eTable 3. Key Components of the Protocol for a Hypothetical Target Trial and the Observational Emulation Based on Data From the Massachusetts Public Health Database
eFigure 1. Diagram of Study Design, Including Time Periods for Identifying Exclusion Criteria, Baseline and Time-Varying Covariates, Treatment Assignment, and Follow-Up
eFigure 2. Percentage of Medically Managed Opioid Withdrawal (MMOW) Discharge Episodes That Initiated Assigned Treatment With Possession of the Assigned Treatment or Any MOUD Over Time
Data Sharing Statement
