A Literature Review Examining Primary Outcomes of Medication Treatment Studies for Opioid use Disorder: What outcome should be used to measure opioid treatment success?

Abstract

Background and Objectives

Medications for opioid use disorder (MOUD) reduce opioid use and overdose; however, MOUD clinical trials have used varying primary outcomes to document treatment success. We conducted a literature review to assess and critically examine the methodologies used in MOUD treatment studies.

Methods

Published studies in English that examined MOUD (buprenorphine, methadone, or extended-release naltrexone (XR-NTX)) were included. The methods and frequencies of measuring primary opioid outcomes, including urine drug tests (UDTs) and self-report of opioid use were compared among studies.

Results

A total of 20 studies fit the inclusion criteria. Each study assessed opioid use; only 12 had opioid use as a primary outcome. Other primary outcomes included retention in treatment (N=6), and 2 had other primary outcomes (death and opioid withdrawal symptoms). Opioid use was assessed through both self-report and UDTs in 15 studies. Two studies did not use UDTs. Differences were found in the methods used for how opioid use, retention in treatment, self-report of opioid use, and UDTs were measured.

Discussion and Conclusions

The different primary outcomes used and operational definitions in each study make comparisons between studies difficult. The use of both self-report and UDTs for opioid use has several advantages, and if possible, researchers should use both measures.

Scientific Significance

This is the first review critically examining outcome measures from MOUD treatment studies. Creating a standard for opioid treatment outcomes in MOUD studies will allow for generalizable results that can inform both researchers and clinicians to better care for those with OUD.

Introduction

In recent years, there has been a steep rise in opioid use in the United States that has led to dramatic increases in overdoses and deaths. As a result, there has been an urgent need to curb the opioid epidemic. Since the prescription opioid epidemic in the late 1990’s, heroin use has increased since 2002 among both males and females from all socioeconomic statuses¹, and from 1999 to 2017, more than 700,000 people have died from a drug overdose, over half of which were due to opioids². The increase in persons developing opioid use disorder (OUD) along with increases in opioid overdose deaths (that are often related to fentanyl use) has led the US Government to declare the opioid epidemic a public health emergency in late 2017³, and in 2018 the NIH introduced the HEAL (Helping to End Addiction Long-term^SM) Initiative to fund opioid-related research⁴.

The most effective form of treatment for OUD are U.S. Food and Drug Administration (FDA) medications to treat OUD (MOUD), which will help curtail the opioid epidemic, especially by increasing access to and retention to these medications. Currently there are three effective approved forms of MOUD: methadone, buprenorphine (comes in different formulations⁵), and XR-NTX (daily, oral naltrexone is not effective for treating OUD)⁶. These medications have been shown to reduce opioid use, opioid overdose deaths⁷, and the transmission of infectious diseases such as HIV and Hepatitis C, as well as improve viral suppression among persons with HIV^8,9. Clinical research trials of MOUD often use opioid abstinence as a measure of treatment efficacy and effectiveness. This is predominantly because the FDA requires decreased drug use, usually defined as abstinence as a successful endpoint for approval of forms of MOUD (although the FDA is considering other outcome measures that have not yet been implemented)^5,10,11. However, there are substantial concerns with relying on opioid abstinence (and other opioid use outcomes), because there are inconsistencies in how opioid use is measured from both self-report and biological measures (urine toxicology screens, hair, and saliva testing). Abstinence is often used because of the assumption that it leads to positive clinical and socio-behavioral outcomes and because other measures, such as quality of life require a longer and more indefinite period of time to evaluate¹². However, studies have not examined if there is a direct causal relationship between abstinence and these other long-term outcomes. Some studies use other outcomes, such as opioid treatment success, time to first opioid use, or Diagnostic and Statistical Manual of Mental Disorders (DSM) diagnosis of OUD after continued abstinence. Standardized methods to evaluate the effectiveness of MOUD treatments is necessary to provide results that can be generalized, and to inform those who are providing care to people with OUD. We conducted a literature review to assess and critically examine various opioid treatment outcomes and methods used in recent MOUD studies to provide information and guidance to researchers and healthcare providers.

Methods

PubMed was searched for studies examining MOUDs for this literature review. The search criteria used were (opioid dependence or opioid use disorder) AND (naltrexone OR buprenorphine OR vivitrol OR methadone OR suboxone). Among the articles found, we chose 20 recently published studies that were most relevant and met the following inclusion criteria: English language, examined buprenorphine, methadone, and/or XR-NTX (studies that used oral naltrexone as a comparison group for these medications were included), and were recent (cut-off year used was 2002, when the FDA approved primary care physicians’ to prescribe buprenorphine³). Studies had to examine MOUD for DSM-IV opioid dependence or DSM-5 OUD¹³.

The following variables were compiled and summarized by 2 authors (BEB/XZ): intervention; study design; population; primary and opioid related outcomes and methodologies; and results of each study. The operational definition of the primary outcome of interest or the measure of treatment success used to analyze the effectiveness of the type/s of MOUD in the study were abstracted and compared among studies. The primary outcome measures were organized into 3 categories: 1) opioid use (or lack of opioid use, typically defined as abstinence), 2) retention in treatment or study, and 3) other outcomes (including death and withdrawal symptoms). We noted if studies had more than one primary outcome, and all outcomes (both primary and secondary) related to opioid use are noted in the results and corresponding tables. The methods used to measure primary outcomes including biological outcomes (such as UDTs and self-report of opioid use), and the frequencies of method use, were reviewed and compared. Other opioid outcomes that were common among studies were also abstracted.

Results

Twenty studies (Table 1) were found that fit the inclusion criteria for this literature review. All studies included participants with DSM-5 OUD or DSM-IV opioid dependence. Variations in the studies were found on the following characteristics: targeted participant population, medication interventions and comparison groups, study design, length of follow-up periods, primary outcomes, opioid outcome measures (biological vs. self-report vs. a combination of both biological and self-report), opioid outcome definition (including abstinence, time to opioid use, etc.), and other outcomes (retention in treatment and study, and overdose).

Table 1.

Studies

	Authors	Intervention and Study Design	Population	Duration of Intervention	Primary and Opioid-Related Outcomes	N	Results of Primary Outcome/s
1	Lee et al.1 (2016)	XR-NTX vs. TAU RCT 1:1, open-label	OUD and previous criminal justice system involvement 5 sites, U.S.	6 months (24 weeks)	-Time to an opioid relapse event: defined as 10 or more days of opioid use in a 28-day period as assessed by self-report (via TLFB) or by testing of urine samples obtained every 2 weeks; a positive or missing sample was computed as 5 days of opioid use. -opioid relapse event (binary) -percentage of 2-week intervals with no opioid use (self-report or positive urine sample) -percentage of days with self-reported opioid use -post-treatment rates of opioid use (self-report and urine samples)	N=153	-XR-NTX longer median time to relapse compared to TAU (10.5 vs. 5.0 weeks, p<0.001, HR=0.49 (95% CI 0.36–0.68)).
2	Ling et al. 2	Probufine (4 BUP implants) vs. Placebo RCT 2:1, placebo Controlled, double-blind	OUD U.S., multi-site (18) outpatient clinics	6 months (24 weeks)	- Proportion of 48 urine samples (obtained 3x/week) that were negative for illicit opioids during the first through 16th week of the trial -percentages of the 24 urine samples that were negative from weeks 17–24 -proportion of treatment failures -proportion who completed study -patient and clinician rated withdrawal scales -craving scale	N=163	-The Probufine group had more urine negative samples from weeks 1–16 (40.4%) compared to the placebo group (28.3%, p=.04).
3	Krupitsky et al.3	XR-NTX vs. Placebo Double-blind, placebo controlled RCT, 1:1	IDU OUD 13 clinical sites, Russia	6 months (24 weeks)	- # of confirmed opioid abstinence weeks during weeks 5–24, assessed by urine drug tests (obtained weekly) and self-report of non-use (via TLFB), and # of patients with confirmed abstinence -self-reported opioid free days -opioid craving scores -number of days of retention -relapse to physiological opioid dependence	N=250	-The median proportion of weeks of confirmed abstinence was 90.0% (95% CI 69.9–92.4) in the XR-NTX group compared with 35.0% (11.4–63.8) in the placebo group (p=0.0002). -XR-NTX group 36% had total confirmed abstinence, placebo group had 23% (p=0.0224).
4	Rosenthal et al. 4	Probufine (6-month BUP implant) vs. daily SL BUP RCT, 1:1, active –controlled, double-blind, double-dummy	OUD U.S., multi-site (21) outpatient substance use disorder clinics	6 months (26 weeks)	- Proportion of participants without evidence of opioid use (based on urine test results (urine samples obtained 1x/month + at 4 random times for a total of 10 samples) and self-report (via TLFB) -treatment retention -time to first illicit opioid use -% of illicit opioid use per month -cumulative % of negative illicit opioid urine results at 6 months	N=177	More participants in the BUP implant group (96.4%) had no evidence of opioid use compared to the SL BUP group (87.6%,p< .001).
5	Mattick et al. 5	SL BUP vs. MMT, Randomized, controlled, double-blind, double-placebo parallel group design.	OUD seeking treatment, 3 public clinics, Australia	13 weeks	-days of retention in treatment: 1) # who completed study; 2) # of days from first dose of study medication to completion (91 days) or last dose of study medication -the absence of morphine in urine samples (obtained randomly every 2 weeks). -self-report of drug use -withdrawal symptoms in previous 24 hours	N=405	-Patients in the buprenorphine group were retain in treatment for a mean of 59.2 days (SD=35.9), patients in methadone group were retained for a mean of 66.8 days (SD=33.1) -59% of methadone patients completed the trail compared to 50% of buprenorphine patients (p=0.061, in survival analysis p=0.037) -No significant between group differences in morphine-positive urines (data shown in figure).
6	Fiellin et al. 6	Counseling + SL BUP vs. RCT of 3 conditions: 1. standard medical management and 1x/week BPN dispensing 2. Standard medical management and 3x/week BPN dispensing 3. Enhanced medical management and 3x/week medication dispensing	OUD, Office-based primary care centers, U.S.	6 months (24 weeks)	-self-reported frequency of illicit opioid use -the percentage of opioid negative urine tests (urine samples obtained every week) -self-reported maximum number of consecutive weeks of abstinence from illicit opioids (verified from urinalysis). -proportion of patients remaining in study (did not meet criteria for protective transfer, did not miss medication for >7 days, did not miss >3 counseling sessions).	N=166	-All 3 groups reduced frequency of illicit opioid use from 5.3 days/week to 0.4 days/week during maintainence phase (p=NS for differences between 3 groups), -No difference between the 3 groups in mean percentage of opioid urine negatives (44% vs. 40% vs 40%, p=0.82) -No difference between the 3 groups in the maximum duration of continuous abstinence from illicit opioids (6.7 vs. 5.7 vs. 5.5 weeks, p=0.54)
7	Lee et al. 7 (2017)	XR-NTX vs. SL BUP Randomized, open-label, comparative effectiveness trial	OUD U.S., multi-site (8) community-based inpatient services but followed as outpatients	6 months (24 weeks)	-Time to relapse, defined as: any non-study opioid use 20 days post-randomization, either 1) at the start of 4 consecutive opioid weeks (an opioid use week was defined as any week a participant reported (via TLFB) at least 1 day of non-study opioid use, or had urine sample positive for non-study opioids), OR 2) at the start of 7 consecutive opioid use days. -proportion of participants successfully inducted onto an initial dose of study medication, frequency of non-study opioid use, assessment of weekly urine toxicology samples, opioid craving	N=570	Intention to treat analyses: -relapse events greater for XR-NTX (65%) compared to SL BUP (57%), OR=1.44 (95% CI 1.02–2.01, p=0.036) -XR-NTX group had a median of 8.4 weeks of relapse free time compared to the SL BUP who had a median of 14.4 weeks (HR=1.36 (95% CI 1.10–1.68, p=0.004), however, these differences were largely due to the high proportion of induction failures seen in the XR-NTX group (25% of participants), compared to the SL BUP group (3%).
8	Tanum et al. 8	XR-NTX vs. daily SL BUP RCT, multicenter open-label	OUD Norway, multi-site (5) urban outpatient addiction clinics	12 weeks	-Retention in study/treatment, defined as the number of days until dropout from study medication and by the number of participants completing the study at week 12. -Proportion of opioid-negative urine drug tests (weekly) during study - Number of days of use of heroin and other illicit opioids (self-report via TLFB) -heroin craving, thoughts about heroin	N=232	-Groups had similar retention in study (XR-NTX had mean of 69.3 days (SD=25.9) in study vs. SL BUP mean of 63.7 days (SD=29.9); 24 participants dropped out of the study from the XR-NTX group, and 29 from the SL BUP group. -XR-NTX had greater proportion of opioid negative UDTs mean=0.9 (SD=0.3), vs. SL BUP 0.8 (0.4), (p<.001). At week 12, lower heroin and other illicit opioid use in the XR-NTX (mean days of heroin use=1.1, other illicit opioids=2.0) group compared to daily SL BUP (mean days of heroin use=4.1, other illicit opioids=4.4, p=0.003, 0.060. respectively).
9	Haight et al.9	BUP-XR, double-blind, placebo-controlled; Randomized 4:4:1:1 – 300 mg/300 mg, 300 mg/100 mg, or volume matched placebo	OUD 36 treatment centers, U.S.	24 weeks	- Percent abstinent from opioid use as measured by percentage of each participant’s negative weekly urine samples and self-reports of illicit opioid use (via TLFB) from week 5 to week 24 -Treatment success (≥80% opioid abstinence during weeks 5–24) -treatment retention, opioid withdrawal symptoms, opioid craving	N=489	Percent abstinence was 41.3% (SD=39.7) in the BUP-XR 300 mg/300 mg group, 42.7% (38.5) in the BUP-XR 300 mg/100 mg group, and 5.0% (17.0) in the placebo group (comparison of both BUP-XR groups vs. placebo p<0.0001).
10	Lofwall et al. 10	Subcutaneous Buprenorphine (monthly injectable extended-release, CAM2038) vs. flexible dosing of SL BUP up to 32 mg daily Randomized, Double-Blind, Double Dummy, active control	Moderate-severe OUD outpatient setting 35 sites, U.S.	24 weeks	- Proportion of urine toxicology results negative for illicit opioids -responder rate (no evidence of illicit opioid use via urine test result and self-report- TLFB used) -mean % of opioid negative urine samples (urine samples obtained at each study visit (weeks 1–12, 16, 20 and 24) plus 3 random visits during weeks 12–24) examined by cumulative distribution function -study retention	N=428	-Subcutaneous buprenorphine group had a greater proportion of opioid urine negatives (35%) compared to SL BUP (28%) (p<0.001) - Subcutaneous buprenorphine had greater percent of responders (37% vs. 31%, p<0.001)
11	Solli et al.11	Prospective cohort study (9 month follow-up) of a 3-month RCT8 to assess the effectiveness, safety, feasibility of long-term treatment with XR-NTX on patients continuing XR-NTX and participants inducted on XR-NTX (switched from BP-NLX or dropped out of RCT)	Norwegian multi-center OUD	9 months	-Retention in treatment (# of weeks in treatment and # of participants who completed study) -opioid abstinence -use of heroin/other illicit opioids (via TLFB) -Money spent on drugs/alcohol within 4 weeks preceding each study attendance -Craving for heroin measured by visual analogue scale	N = 117	-Continuing XR-NTX group spent a mean of 25.6 (22.3–29.0) weeks in treatment vs. 25.4 (22.4–28.4) weeks among those inducted on XR-NTX (p=NS). -52% of the continuing XR-NTX group completed treatment vs. 48% of those inducted on XR-NTX -54% of the continuing XR-NTX group reported abstinence from all opioids vs. 44% of those inducted on XR-NTX (p=NS).
12	Strang et al. 12	XR-NTX vs. standard oral NTX vs. placebo; three arm, parallel group, placebo controlled, double blind, double dummy, randomized controlled trial	OUD, completion of detoxification, 3 centers, England	12 weeks,	-Proportion of heroin-negative urine drug screen results (obtained 3x/week) at the end of the 12-week post-randomization time point -treatment retention -Adherence to oral study medication -Heroin craving score -Self-reported heroin use	N=6	Sample was too small to draw meaningful conclusions, individual level data presented in report
13	Mokri et al.13	Double-blind, double-dummy, single site parallel randomized clinical trial to oral NTX vs. SL BUP	Patients with OUD in an out-patient clinical research program, Tehran, Iran	12 weeks	-Duration of verified initial opioid abstinence (self-report via TLFB, days to first opioid-positive or missed urine test, urine samples obtained 2x/week) -total duration of verified abstinence (# of opioid negative urine tests) -treatment retention -proportions with sustained (12-week) verified opioid abstinence -Number of opioid negative UDTs	N=102	BNX group had mean of 28.8 (20.0–37.5) days of opioid abstinence vs. 21.6 (14.4–28.7) days in the NTX group (p=0.205)
14	Pinto et al.14	Prospective patient-preference study (patients chose methadone or buprenorphine)	OUD, 3 sites within one community drug service covering a large rural area and two urban centers in Norfolk, UK	6 months	-Positive outcome: retention in treatment at 6 months (if subjects did not take maintenance drug for 7 consecutive days they were considered not retained) OR successful detoxification (patients who were recorded to be opiate-free at the point of discharge from the service during the trial) -No illicit drug use via urine toxicology (study used first urine toxicology report each month from clinical notes)	N=361	At 6 months, 50% of BMT patients had a positive outcome (retention or detoxification) vs. 70% of MMT patients, OR=0.43 (95% CI 0.20–0.67, p<0.001).
15	Hser et al. (2014) 15	START study: participants in a 24-week, Open-label, 2:1 randomization of SL BUP vs methadone	OUD, entering opioid treatment program, U.S.	24 weeks	-Treatment completion (continuing in the assigned medication group for 24 weeks without being withdrawn) - Treatment retention (days in treatment since randomization until the last day of medication during the 24 weeks of treatment) -Weekly urine drug screens	N=126	-46% of Bup participants completed treatment vs. 74% of MET participants, p<0.01 -mean (SD) number of days in treatment was 103.8 (66.9) for BUP and 141.3 (50.8) for MET, p<0.01
16	Hser et al.16 (2016)	Long-term follow up of START study: participants in a 24-week, Open-label, 2:1 randomization of SL BUP vs methadone	OUD, entering opioid treatment program, U.S.	2–8 years post-randomization (mean=4.5 years)	-Occurrence and date of death (NDI searched through Nov 2014) -Current opioid use (self-reported days of opioid use in past 30 days, via TLFB, at the follow-up interview, or a positive urine test) -opioid use during follow-up period, defined as self-reported days of opioid use per month -Treatment status from enrollment to follow-up	N=795	-23 (3.6%) deaths in the BUP group vs. 26 (5.8%) in the MET group, p=0.10 -No difference in time to death between 2 groups (HR not shown, p=0.10) -Current opioid use higher in BUP group (51%) than MET group (41%), p<0.01
17	Kunoe et al.17	Randomized, open-label trickle-inclusive design; naltrexone implant vs. patient’s usual aftercare	OUD, receiving abstinence-oriented in-patient treatments, Norway	6 months	-Self-report of opioid use via TLFB, verified against hair samples at 6-months (number of days of drug use in the previous 30 days) - Addiction Severity Index (ASI): 4 point frequency-of-use scale for the whole 6-month period (0, no use; 1, maximum two or three times per month; 2, two or three times per week; 3, daily or almost daily use) -number of overdoses at 6-month follow-up	N=56	-For 180-day TLFB, the implant group reported an average of 37 days of opioid use (SD=63.8), compared to 97.1 days (SD=80.9) in controls, p<0.01 -ASI frequency scale: implant group scored an average of 0.8 (SD=0.98) vs. 1.5 (SD=1.3) in the control group, p<0.05 -There was no significant difference between groups on self-reported overdoses (3 in the implant group vs. 4 in the control group),
18	White et al. 18	Open-label, maintenance on 8 mg SL BUP then switched to 2 Probuphine implants OR maintainance on 16 mg SL BUP and switched to 4 Probuphine implants	OUD, 3 treatment centers in Australia who were on maintenance treatment with SL BUP	6 months (24 weeks)	-Opioid withdrawal symptoms by self-report (via subjective opioid withdrawal scale, SOWS) or clinician-administered rating scales (objective opioid withdrawal scale, OOWS) -craving for heroin, assessed with a 100mm visual analog scale -heroin use (self-report since last visit, calculated as days of use per month) and urine toxicology screens	N=12	-Over 6 months, mean SOWS score for the 2-implant group was 0.97±2.18 and for 4-implant group was 0.49±1.19; the mean OOWS mean scores were 0.49±0.90 in the 2-implant group and 0.49±0.78 in the 4-implant group. -Average craving value during 6 months of Probuphine treatment was 1.77±8.17 for the 2-implant group and 2.25±6.44 for the 4-implant group -66% reported no heroin/opioid use; 2 implant group reported using heroin 0.62±0.85 and the 4-implant group reported using heroin 4.42±3.21 days per month
19	Woody et al.19	RCT, 12 weeks of treatment with SL BUP or a 14-day outpatient detox	OUD, aged 15–21 in 6 community programs	12 weeks	-Opioid-positive urine test results at weeks 4, 8, 12 -dropout from assigned condition -self-reported opioid use -injecting drugs -enrollment in addiction treatment outside the assigned condition	N=152	-At week 4, 61% of detox patients had positive UDT results 26% vs. of 12-week, buprenorphine-naloxone patients, OR (from GEE model)= 7.05, p<0.001) -At week 8, 54% of detox patients had positive UDT results vs. 23% of 12-week buprenorphine-naloxone patients, OR=5.07, p=0.001) -At week 12, 51% of detox patients had positive UDT results vs. 43% of 12-week buprenorphine-naloxone patients, OR=1.84, p=0.18
20	Wang et al.20	Double-blind placebo-controlled study comparing SL BUP vs. placebo	OUD, 5 clinics in China	6 weeks	-Retention in treatment, defined as time from randomization to treatment completion or treatment failure (3 consecutive opioid-positive urine tests, dropout, discontinuation or discharge from study, or use of ancillary medication treatment for severe withdrawal over 3 days following randomization) -maximum consecutive days of abstinence from opioids, (longest period of consecutive opioid-negative UDTs with no missing urines) -COWS -VAS of heroin craving -percentage of opioid-negative UDTs throughout the treatment period (absence of UDTs treated as positive).	N=260	-Fewer treatment failures in the buprenorphine/naloxone group (63%) compared to placebo (92%, p<0.001) -median time to treatment failure: buprenorphine/naloxone, 32 days (95% CI, 26–38); placebo, 6 days (95% CI, 5–8), HR=0.28 (95% CI, 0.21–0.38; P < 0.001).

Abbreviations: BUP=buprenorphine

BUP-XR= extended release buprenorphine

COWS=clinical opioid withdrawal scale

IDU= injection drug use

MMT= methadone maintenance treatment

RCT= randomized controlled trial

NTX= naltrexone

OOWS= objective opioid withdrawal scale

OUD= opioid use disorder

SL BUP= sublingual buprenorphine

SOWS= subjective opioid withdrawal scale

TAU=treatment as usual

TLFB= Timeline Followback

UDT= urine drug test

VAS=visual analog scale

XR-NTX= extended-release naltrexone

^1.Lee JD, Friedmann PD, Kinlock TW, et al. Extended-Release Naltrexone to Prevent Opioid Relapse in Criminal Justice Offenders. The New England journal of medicine. Mar 31 2016;374(13):1232–1242.

^2.Ling W, Casadonte P, Bigelow G, et al. Buprenorphine implants for treatment of opioid dependence: a randomized controlled trial. JAMA. Oct 13 2010;304(14):1576–1583.

^3.Krupitsky E, Nunes EV, Ling W, Illeperuma A, Gastfriend DR, Silverman BL. Injectable extended-release naltrexone for opioid dependence: a double-blind, placebo-controlled, multicentre randomised trial. Lancet. Apr 30 2011;377(9776):1506–1513.

^4.Rosenthal RN, Lofwall MR, Kim S, et al. Effect of Buprenorphine Implants on Illicit Opioid Use Among Abstinent Adults With Opioid Dependence Treated With Sublingual Buprenorphine: A Randomized Clinical Trial. JAMA. Jul 19 2016;316(3):282–290.

^5.Mattick RP, Ali R, White JM, O’Brien S, Wolk S, Danz C. Buprenorphine versus methadone maintenance therapy: a randomized double-blind trial with 405 opioid-dependent patients. Addiction. 2003;98(4):441–452.

^6.Fiellin DA, Pantalon MV, Chawarski MC, et al. Counseling plus buprenorphine-naloxone maintenance therapy for opioid dependence. The New England journal of medicine. Jul 27 2006;355(4):365–374.

^7.Lee J, Nunes E, 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. The Lancet. November 14, 2017 2017.

^8.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. Dec 1 2017;74(12):1197–1205.

^9.Haight BR, Learned SM, Laffont CM, et al. Efficacy and safety of a monthly buprenorphine depot injection for opioid use disorder: a multicentre, randomised, double-blind, placebo-controlled, phase 3 trial. The Lancet. 2019;393(10173):778–790.

^10.Lofwall MR, Walsh SL, Nunes EV, et al. Weekly and Monthly Subcutaneous Buprenorphine Depot Formulations vs Daily Sublingual Buprenorphine With Naloxone for Treatment of Opioid Use Disorder: A Randomized Clinical Trial. JAMA Intern Med. Jun 1 2018;178(6):764–773.

^11.Solli KK, Latif ZE, Opheim A, et al. Effectiveness, safety and feasibility of extended-release naltrexone for opioid dependence: a 9-month follow-up to a 3-month randomized trial. Addiction. Oct 2018;113(10):1840–1849.

^12.Strang J, Kelleher M, Mayet S, et al. Extended-release naltrexone versus standard oral naltrexone versus placebo for opioid use disorder: the NEAT three-arm RCT. Health Technol Assess. Jan 2019;23(3):1–72.

^13.Mokri A, Chawarski MC, Taherinakhost H, Schottenfeld RS. Medical treatments for opioid use disorder in Iran: a randomized, double-blind placebo-controlled comparison of buprenorphine/naloxone and naltrexone maintenance treatment. Addiction. May 2016;111(5):874–882.

^14.Pinto H, Maskrey V, Swift L, Rumball D, Wagle A, Holland R. The SUMMIT trial: a field comparison of buprenorphine versus methadone maintenance treatment. J Subst Abuse Treat. Dec 2010;39(4):340–352.

^15.Hser YI, Saxon AJ, Huang D, et al. Treatment retention among patients randomized to buprenorphine/naloxone compared to methadone in a multi-site trial. Addiction. Jan 2014;109(1):79–87.

^16.Hser YI, Evans E, Huang D, et al. Long-term outcomes after randomization to buprenorphine/naloxone versus methadone in a multi-site trial. Addiction. Apr 2016;111(4):695–705.

^17.Kunoe N, Lobmaier P, Vederhus JK, et al. Naltrexone implants after in-patient treatment for opioid dependence: randomised controlled trial. Br J Psychiatry. Jun 2009;194(6):541–546.

^18.White J, Bell J, Saunders JB, et al. Open-label dose-finding trial of buprenorphine implants (Probuphine) for treatment of heroin dependence. Drug Alcohol Depend. Jul 1 2009;103(1–2):37–43.

^19.Woody GE, Poole SA, Subramaniam G, et al. Extended vs short-term buprenorphine-naloxone for treatment of opioid-addicted youth: a randomized trial. JAMA. Nov 5 2008;300(17):2003–2011.

^20.Wang X, Jiang H, Zhao M, et al. Treatment of opioid dependence with buprenorphine/naloxone sublingual tablets: A phase 3 randomized, double-blind, placebo-controlled trial. Asia Pac Psychiatry. Mar 2019;11(1):e12344.

1. Differences in Participant populations

While all of the studies included persons with OUD or opioid dependence, some of the studies recruited specific subpopulations of those with OUD, including: one study recruited participants with previous criminal justice involvement¹⁴, while the majority of studies recruited persons from the community who were treatment seeking. Some studies that assessed XR-NTX recruited participants from inpatient detoxification centers^15,16, or participants who were enrolled in abstinence based treatment¹⁷. All studies but one recruited adults age 18 and older; one study recruited solely youth aged 15–21¹⁸ (Table 1).

2. Medication Interventions and Comparison Groups

All the studies included in this review included MOUD as the main intervention. The breakdown of the MOUD interventions included: 1) methadone (4 studies)^19–22; 2) naltrexone (extended-release used in 6 studies^{14–16,23–25}, oral naltrexone in 2 studies^25,26 and naltrexone implant in 1 study¹⁷); 3) buprenorphine (sublingual buprenorphine used in 12 studies^{14,16,18–22,26–30}, buprenorphine implant used in 3 studies^30–32, buprenorphine extended-release used in 2 studies^28,33, some studies used >1 MOUD, see Table 1).

Comparison groups for the main MOUD intervention varied for these studies, and included treatment as usual(TAU, N=2^14,17), placebo(N=5^{23,25,29,31,33}), or another form of MOUD(N=11, START Study counted only once^{15,16,19,20,22,24–26,28,30,32}). The comparison group of one study assessed different forms of medication management and how often sublingual buprenorphine was dispensed²⁷, and in one study the comparison group was those who were detoxed from opioids¹⁸ (Table 1).

3. Design of Studies

The majority of studies were randomized controlled trials (N=17). One study was a prospective cohort study²⁴, one was a preference clinical trial²⁰, and one was a non-randomized open label trial³².

4. Differences in follow–up periods

There was substantial variation in the study intervention follow-up periods. One study was a long-term follow-up study which followed subjects for an average of 4.5 years²¹. The remaining 19 studies ranged from 6 to 36 weeks in duration. Among these studies, the mean time participants were followed (study duration) was 20.8 weeks (SD=7.4, median=24 weeks).

5. Primary outcomes

Most studies had more than one primary outcome, however, opioid use was the primary outcome or one of the primary outcomes in 16 studies (80%)^{14–19,21,23–28,30,31,33}, 6 studies had a primary outcome of retention in treatment or study^{16,19,20,22,24,29}, and 2 studies had other primary outcomes (death²¹, opioid withdrawal symptoms¹⁸, and the number of overdoses¹⁷).

6. Opioid Outcome Measures

Opioid outcome measures were divided into biological and self-report measures. All of the 20 studies included in this review assessed opioid use even if it was not included as the primary outcome; 15 studies assessed opioid use through both self-report and UDTs, 2 studies through self-report alone, and another 3 studies used UDTs alone^20,29,31.

a). Biological opioid outcome measures

A total of 18 studies assessed opioid use with UDTs. Of the 2 studies that did not use UDTs, one used hair samples to assess opioid use¹⁷, and the other acknowledged the lack of UDTs as a study limitation²⁴.

Frequency of the UDT assessments also varied across studies (Table 3). Most studies assessed UDTs once a week (N=6)^{15,16,22,23,27,33}, 2 studies assessed UDTs two times a week, 2 assessed them three times a week^26,29, 2 assessed them every two weeks^14,19, and 2 assessed them one time a month^18,20. One study assessed UDTs every month plus at 4 random time-points³⁰; another study assess UDTs at every study visit (approximately once a month) and 3 additional random visits²⁸; and one study assessed UDTs at different time intervals throughout the study with UDTs occurring less frequently during the time period of the study³². In one long-term follow-up study, a UDT was done at the single in-person follow-up interview²¹.The range of frequency of UDT assessments for studies that used daily compared to non-daily MOUD did not differ (among studies that used daily MOUD^{18–22,26,27,29}, the range of UDTs measurements were 2 times week to once a month; for non-daily MOUD studies^{14,17,23,24,31–33}, the range was 3 times a week to every 2 weeks; and for studies that used both daily and non-daily MOUD^{15,16,25,28,30}, the range was 3 times a week to once a month).

Table 3.

Frequency of Urine Toxicology Screens

Frequency of Urine Toxicology Screens	Number of Studies	Studies
3 times a week	2	Ling et al.2, Strang et al. 12
2 times a week	2	Mokri et al.13, Wang et al., 20
1 time a week	6	Krupitsky et al.3, Fiellin et al. 6, Lee et al. 7, Tanum et al. 8, Haight et al.9, Hser et al. (2014) 15
Every 2 weeks	2	Lee et al.1, Mattick et al. 5
Once a month	2	Pinto et al.14, Woody et al.19
Once a month + 4 random time points	1	Rosenthal et al. 4
At every study visit (approximately once a month) + 3 additional random visits	1	Lofwall et al. 10
At different time intervals throughout the study with UDTs* occurring less frequently during the time period of the study	1	White et al. 18
Once at a single in-person Follow-up interview	1	Hser et al.16

UDT= Urine Drug Testing

Another issue that showed variance in UDTs measurement of opioid use included how missing data points were handled. In most studies (N=14) missing urine samples were considered as positive for opioid use. One study allowed one missing sample out of 6 if the remaining 5 were negative²⁰, and in 3 studies it was not specified how missing UDTs were handled^21,22,32. In addition to treating missing UDTs as positive, some studies imputed missing UDT data in different ways to conduct sensitivity analyses. This varied throughout studies; 2 studies imputed the previous UDT result when a UDT was missing,^18,27 1 study imputed missing UDTs as positive only when patients were still receiving treatment²⁷. Another study imputed missing UDTs by randomly generated binary outcome (positive or negative for opioid use) using a 20% relative penalty against the buprenorphine implant group based on UDT results from each group³⁰.

b). Self-Report Measurement of Opioid Use

A total of 17 studies measured opioid use with self-report. The timeline follow-back (TLFB) survey³⁴ was used most often (11 studies)^{14–17,21,23,24,26,28,30,33}. Studies that did not use the TLFB used the Opiate Treatment Index³³ and a 7-day drug and alcohol use self-report form²⁵. Additional measures of self-reported opioid use included the Addiction Severity Index (ASI)³⁵, which was used in 4 studies^16,17,24,26. Three studies however, did not describe the measurement tool used for collection of self-reported opioid use^18,22,27.

The frequency in which self-report of opioid use was collected also varied (Table 4). Six studies asked participants once a week about their opioid use^{15,19,23,26,27,33} and 5 studies asked once a month^{16,17,22,24,30}. Other studies asked at each study visit, with self-report more frequent at the beginning of the study^{14,18,25,28,32}. The range of how often self-report was asked for studies that used daily compared to non-daily MOUD did not differ (among studies that used daily MOUD^{18–22,26,27,29}, the range of when self-report was asked were once a week to once a month (then every 3 months); for non-daily MOUD^{14,17,23,24,31–33}, the range of assessment was 2 times a week (1 study³²) to once a month; and for studies that used both daily and non-daily MOUD^{15,16,25,28,30}, the range was once a week to once a month).

Table 4.

Frequency of self-report of opioid use

How often Self-report data was ascertained	Number of Studies	Studies
Once a week	6	Krupitsky et al.3, Mattick et al. 5, Fiellin et al. 6, Lee et al. 7, Haight et al.9, Mokri et al.13
Once a month	5	Rosenthal et al. 4, Tanum et al. 8, Solli et al.11, Hser et al. (2014) 15, Kunoe et al.17
At each study visit, with self-report more frequent at the beginning of the study	5	Lee et al.1, Lofwall et al. 10, Strang et al. 12, White et al. 18, Woody et al.19

Missing self-report data was handled differently among these studies. Most studies counted missing self-report data as a participant having relapsed or used opioids^14,23,28,33. Several studies used the Last Observation Carried Forward (LOCF) approach, a way to impute missing data^16,17,24. Other studies used different imputation methods^25,30 or used mixed effects models, which handle missing data¹⁵.Three studies did not count missing data in analyses (they only used the data provided)^18,19,27. One study did not report self-report opioid use because the majority of participants (66%) under-reported opioid use when compared with UDT results²⁶.

c). Other Self-Report Measures utilized

Other self-report methods used included heroin/opioid craving (13 studies)^{15–17,19,23–25,28–33}; most used a visual analog scale (VAS) to assess craving. One study adapted the Minnesota Cocaine Craving Scale³⁶ to assess opioid craving²⁵, no other studies used standardized craving measures. The Subjective Opiate Withdrawal Scale (SOWS) was used in 6 studies^{19,28,30–33} and the Clinical Opiate Withdrawal Scale was used in 5 studies^28–31,33. One study used the Objective Opioid Withdrawal Scale³².

7. Opioid Use Outcome Definitions

Opioid use via self-report and UDTs was defined in various ways: the percent of opioid negative urine toxicology samples (N=16), the number or percent of days study participants used opioids in a given time period (N=13), proportion of study participants who used opioids (N=11), the number or percent of week intervals with no opioid use (N=4), and time to opioid use (N=4, Table 2).

Table 2.

Opioid Use Outcome Definitions

Authors	Time to drug use	Proportion used drugs (binary variable)	# or % of days used opioids	% or # of week intervals with no opioid use	% or # of opioid neg samples
Lee et al.1	X	X	X	X	X
Ling et al. 2			X		X
Krupitsky et al.3		X	X	X
Rosenthal et al. 4	X	X	X
Mattick et al. 5					X
Fiellin et al. 6			X	X	X
Lee et al. 7	X	X	X	X	X
Tanum et al. 8			X		X
Haight et al.9			X		X
Lofwall et al. 10					X
Solli et al.11		X	X
Strang et al. 12					X
Mokri et al.13	X	X	X		X
Pinto et al.14		X			X
Hser et al. (2014) 15					X
Hser et al.16		X	X		X
Kunoe et al.17		X	X
White et al. 18		X	X		X
Woody et al.19		X			X
Wang et al.20					X
TOTAL	4	11	13	4	16

Other opioid outcome definitions included the number of times participants used heroin or opioids per day¹⁹, greater than 80% opioid abstinence during specified trial period³³, and maximum consecutive days of opioid abstinence²⁹.

Of the 11 studies that measured an opioid use event (or lack thereof, i.e., abstinence), 8 defined opioid use/relapse as any opioid use^{18,21,22,24,26,27,30,32}. Other studies were more flexible with the amount of opioid use in their relapse definition: 1 study defined relapse as ≥10 days of opioid use in a 28-day period OR through positive or missing UDTs, where a positive or missing UDT was calculated as 5 days of opioid use¹⁴; One study defined relapse as a return to physiological opioid dependence as defined by a positive naloxone challenge²³. Another study defined opioid use in two different ways, either at the start of 7 consecutive days of self-reported opioid use, or 4 consecutive opioid use weeks, where at least one day of non-prescribed opioid use contributed to an opioid use week¹⁵; and one study defined opioid abstinence as 6 negative UDTs or 5 negative UDT and 1 missing²⁰.

8. Other Outcomes

a). Retention in MOUD Treatment Definition

A total of 16 studies measured retention on MOUD treatment. For studies using injectable forms of MOUD, the number of injections participants received was a common measure of retention^14,37. Three studies measured treatment retention as time to last dose of study medication^16,19,28 or time to treatment failure. In the study that examined Probuphine^®(4 buprenorphine implant rods that deliver treatment for 6 months), treatment failure included receiving a fifth implant and requiring sublingual buprenorphine (SL BUP) over certain time periods³¹. One study considered MOUD treatment failure as 3 consecutive opioid-positive UDTs, among other criteria²⁹, while another study defined failure of MOUD treatment retention if MOUD was missed for 7 consecutive days²⁰.

MOUD treatment failure was analyzed either as the proportion of participants who reached treatment failure in each group^29,31 and/or time to event analyses²⁹. Other studies measured the number of days, weeks, or months participants were in MOUD treatment (mean or median)²¹, or as a time to event analysis^{16,19,20,22,23,25,26}) and/or the percent who completed treatment²⁰.

b). Retention in Study Outcome

Most studies defined retention in the study as the number or percent of participants who completed the study trial (N=15)^{15–17,19,20,22–28,30,31,33}. One study measured the number of participants retained at different study time points, including the last study visit²⁹, and another measured the percent who dropped out during the first 30 days of the study²¹. Studies also measured the time in the trial (mean days, percent of study visits attended, or time to study discontinuation)^17,21,33.

c). Overdose outcome

One study assessed drug overdose as an outcome¹⁷, and 13 studies provided information drug overdose as part of the study’s adverse events^{14–16,18–20,23–26,28,33}. In some studies it was not specified or unclear how overdose events or deaths were obtained^{15,19,20,26,33}. Many studies used self-report for overdose events^{15–17,23–25,28}. One study noted that registry databases were used to obtain death information¹⁶, and another found out of a participant death through a newspaper obituary and subsequent medical examiner report¹⁸. Reports of overdose included overdose event (N=10 studies)^{14–17,19,23–25,28,33} or death from overdose (N=8 studies)^{14,15,17,18,20,26,28,33}. Three studies reported only drug overdose deaths^18,20,26, and one study assessed the number of participants with more than one overdose as well as the number of overdoses in each treatment group³⁷. Some studies specified if the overdose they were measuring was an opioid overdose^16,19,33, while others did not specify type of drug overdose^{14,15,17,23,26}. One study examined both opioid and non-opioid overdoses²⁴, and five studies listed the drugs that were involved in each participant’s overdose^17–20,28.

Discussion

Studies of MOUD are necessary in order to establish treatment efficacy and effectiveness, however, there is a lack of standardization regarding what outcomes are measured and how they are measured. MOUD (buprenorphine, methadone, XR-NTX) has been proven to be evidence-based, safe, and highly effective for the treatment of OUD based on different outcome measures³⁸, but there has been no standardized way to assess outcomes from MOUD studies, and researchers and clinicians are unable to compare or generalize findings. This is the first review to address this issue and critically examine the details of such measures from MOUD treatment studies.

Most of the studies were randomized controlled trials, and while they are considered the ‘gold standard’ for assessing treatment effectiveness, they do have some limitations. First, they exclude participants who do not wish to be randomized and thus disregard the importance of patient preference. Patient preference for MOUD is linked to beneficial health outcomes, including reduced substance use³⁹. A well-designed randomized control trial, however, provides the strongest evidence to show a treatment has an effect, while assuring that both groups are comparable. Pinto et al. also noted that randomization and blinding, “divorce the results from patients’ expectations about the effectiveness… of the two alternative treatments,” (pg.341) however, many of the studies were unable to blind participants to their treatment assignment due to how these medications are administered (e.g. implant versus oral versus injectable).

Buprenorphine was the most common MOUD studied. This is likely due to reasons including: 1) this review had an inclusion criterion of 2002 or later of published studies, given that 2002 was when buprenorphine was FDA approved in the U.S. for clinical use; and 2) several formulations of buprenorphine have been developed that include implants, injectable and different forms of sublingual tablets and trans-mucosal films. Nine studies analyzed naltrexone-based treatments, and most analyzed XR-NTX, which was FDA approved in 2010 to treat OUD. Oral naltrexone was used in two studies; however, it is not an effective treatment for OUD due to adherence issues⁶. Methadone was assessed in four studies. It has been available since the 1960s in the U.S., and thus there is a considerable amount of literature showing its effectiveness.

Comparison groups for the type of MOUD studied varied; most studies used participants inducted onto another form of MOUD as the comparison group. Treatment with MOUD is associated with fewer overdose deaths compared to placebo, and better study and treatment retention, which is likely why placebo was often not used as a comparison group³⁸. Placebo controlled trials are the standard for assessing the efficacy of new drugs, but concern about the risk of serious harm among those assigned to placebo⁴⁰ (i.e., overdose deaths), likely played a role in the choice of another form of MOUD as the comparison group (5 studies used placebo as their comparison group).

The studies included in this review all addressed opioid use as an outcome through either self-report and/or UDTs, however, the methods used to assess opioid use varied. The most used self-report method was the Timeline Follow-back (TLFB) method, that asks participants to record opioid use each day since the last study visit using a calendar-based method. The TLFB has been shown to be a reliable and valid measure of alcohol and drug use, including opioids^34,41, and can be used to collect data retrospectively to compensate for missed study visits. Limitations of the TLFB t include that participants may choose not to report opioid use and decreased accuracy of recall when the interval between follow-up visits are long. This is particularly important given the range of times in which self-report of drug use was collected, with some studies asking participants as often as one week, while others asked participants to recall their drug use as far back as the past year²¹. Other measures used to assess self-report of drug use included the Addiction Severity Index (ASI)³⁵. All of the studies that used the ASI also used the TLFB, since calendar-methods (where daily substance use is documented) have proved to be more precise compared to asking participants about the quantity and frequency of drug use over an extended time period³⁴.

Urine drug testing was the most common biological measure of opioid use used among the reviewed studies. UDTs are effective for measuring recent opioid use and are simple to conduct⁴². In the studies examined, however, urine samples were taken at different time intervals, varying from 3 times a week to once a month. Due to these intervals between tests, all opioid use could not have been measured biologically, leaving self-report as the only source of data. Nine studies mentioned that they assessed for urine specimen tampering (most through temperature checks), and 8 did not mention if this was assessed, although it is common practice. Participants can tamper with samples before the sample is tested, however, to discourage patients from tampering, healthcare providers can monitor patients closely, as well as remind them that results are only for research purposes and cannot be shared with anybody outside of the study; only one study mentioned that this was done¹⁹. Another aspect of UDTs to consider is how missing data was treated in statistical analyses. In most of these studies, missing urine samples were imputed as positive for opioids, as this is the most conservative measure of opioid use and treatment effectiveness. Some studies did not mention how they handled missing samples, and one study allowed one missing sample²⁰. Some studies conducted sophisticated imputation methods for sensitivity analyses. Imputing missing samples as positive disregards reasons for missing appointments, which may underestimate the progress a participant is making throughout the treatment period and thus not correctly measure the intended outcome. It was beyond the scope of this review, but it is important to note that there is variability in urine drug screens⁴³.

A combination of self-report and UDTs can measure opioid use with greater accuracy and reliability⁴², thus many studies used both measures. If a urine sample is missing, self-report results can be used to obtain the missing data, or UDTs can corroborate patients’ self-report of opioid use. When only self-reports are used, social desirability or concern about legal consequences may lead some participants to not be honest about their opioid use. Additionally, the reliance on the subject’s memory regarding opioid use makes self-reports unreliable. Using both methods contributes to the collection of the most accurate record of opioid use. The National Institute of Drug Abuse (NIDA)-funded Clinical Trials Networks (CTN) investigators have recommended using both measures as a standard in drug use trials⁴⁴.

Opioid use and abstinence were measured differently among these studies. Most studies defined opioid use or relapse as any opioid use, even if participants only used opioids once, while other studies allowed minimal or intermittent opioid use before defining a relapse event. Opioid use that may be considered a relapse event in one study could fail to meet the criteria in another study. Although opioid abstinence and time to opioid relapse both inherently measure time without opioid use, the operational definitions used in each study were different. Further, the methods in handling opioid use data varied. Some studies measured time to opioid use, while others measured the proportion of participants abstinent or the number of days of opioid use. Due to these inconsistencies in both definition and analysis, opioid use and abstinence cannot be effectively compared, and the ability to generalize results is hindered. Further, these different measures could encourage researchers to only report those measures that are statistically significant, and exclude non-significant results, commonly known as inflation bias or p-hacking⁴⁵. To avoid this, studies could report multiple measures of opioid use and should delineate their methods for outcome measures in detail, before data collection begins.

Retention in MOUD treatment was also used as a primary outcome measure. Because of our cut-off time point of 2002 or later for this review, many studies assessed injectable forms of MOUD, and thus a common measure of retention was the number of injections a participant received. However, there were other measures of treatment retention across studies that were often based on their study design and study visits. For example, some studies created a definition of MOUD treatment failure, and measured this as an outcome. For studies assessing retention in treatment, this is often used as a proxy for relapse because it is assumed that those who stop treatment return to drug use. This definition, however, may be limited by neglecting participants who chose another treatment if they did not like the study intervention form of MOUD, or who were unable to continue treatment due to circumstances such as incarceration.

Given that one benefit of MOUD treatment is a reduction in opioid overdoses, it was surprising that only one study measured overdose as a primary outcome. The remaining studies assessed overdose as an adverse event, and some only reported overdose deaths. Further, in some studies it was unclear how overdose events were ascertained, and undercounts are likely if they only relied on patient report. In these studies, opioid overdose was low, which could explain why this is not often used as a primary outcome. Given the importance of overdose as an outcome, it should be ascertained in MOUD treatment studies, along with relevant details about the overdose, including type/s of drugs used before the overdose event. Ascertaining overdose should be done comprehensively, through both self-report and review of medical records and death reports. Participants may not disclose overdose events due to stigma, or participants may not go to the emergency department for an overdose if they have naloxone available, therefore, ascertaining overdose in multiple ways is critical.

Although all these studies showed benefit of treatment with MOUD, the results from each cannot be effectively compared. In order to increase the comparability of MOUD studies, operational definitions of treatment success should be standardized and used by all researchers. NIDA has created some data harmonization projects, such as projects that use seek, test, treat, and retain model of care (STTR), which aims to increase the comparability of clinical research on HIV and drug abuse⁴⁶, and the Justice Community Opioid Innovation Network (JCOIN) is planning to harmonize data for upcoming studies⁴.

Most studies used opioid abstinence as an endpoint because the US FDA previously required clinical trials to have opioid abstinence as a treatment outcome¹⁰. However, this target is difficult to achieve given that many participants are unable to achieve complete abstinence from opioids. Opioid use is expected to occur during MOUD treatment, yet it most often occurs as isolated incidents and does not necessarily represent a treatment “failure.” Thus, reduction in opioid use, or number of opioid free days may be a more suitable outcome measure.

In these studies, opioid use (often defined as relapse) are most common endpoints, likely because they are the easiest and timeliest to measure. These short-term outcomes are associated with clinically meaningful endpoints, such as decreased overdoses and death but further research is needed to correlate these with more distal outcomes, such as quality of life improvement and improvement in psychosocial functioning that includes housing and employment status. Given the well-known link between OUD and infectious diseases like HIV, Hepatitis C Virus (HCV) and bacterial and fungal infections (e.g. endocarditis), studies evaluating MOUD effects on infectious disease outcomes are also needed (e.g. reduction in HIV and HCV seroconversion; HIV viral load suppression; and HCV cure)^9,47,48. Measuring distal outcomes require long follow-up periods, which are more costly, risk lower retention on study, and are more likely to have confounders associated with them which would need to account for in analyses.

Direct comparisons between studies were not possible because of the different primary outcomes used, as well as the lack of standardized definitions for each outcome. Further, given the differences in outcomes and how they are measured, meta-analysis will be difficult to conduct for systematic reviews of this topic. The most common primary outcome measures in the studies reviewed included opioid use (as a binary outcome such as abstinence, or as a continuous outcome) and retention on treatment. Some studies allowed minimal opioid use among those they considered abstinent, thus there was not even a standardized definition of abstinence, further limiting the comparability of results. The choices of primary outcomes are sometimes dictated by government agencies, as the FDA requires opioid abstinence as an endpoint, and some studies may ignore other meaningful outcomes, such as psychosocial outcomes⁴⁹. Overdose is an important clinical outcome related to opioid and drug use, yet only one study out of the 20 reviewed considered overdoses as a primary outcome.

Our literature review has some limitations, including that it was not a systematic review, and thus we could have missed studies that fit our inclusion criteria. We only chose 20 studies, so our review is not comprehensive, however these studies show the breadth of different primary opioid treatment outcomes in recent, major studies and how these outcomes are measured and utilized as ‘treatment success’. It also illuminates the need for the field to come to an agreement on definitions and methodologies used for abstinence, treatment success/failure, and relapse.

The goal of opioid treatment is to improve physical and psychosocial functioning, thus studies that evaluate MOUD treatment should consider measuring outcomes that reflect this. Researchers and clinicians should understand that other outcomes can occur without complete abstinence from opioids. Measures such as increases in opioid free days, and reduction in use can reflect this, however, research is needed to show that there is a direct causal relationship between decreased use and improved physical and psychosocial functioning. Other outcomes should also be explored, including social support, job readiness, employment status and housing, and infectious disease outcomes.

The lack of comparable standards for outcomes in these studies create barriers for analyzing the effects of MOUD across studies, and standardization should be applied to future studies, especially given the increase in funding for OUD research. The use of both self-report and urine toxicology screens for opioid use has several advantages, and if possible, researchers should use both measures. However, consensus is needed on how to best combine self-report and biological measures, or if one measure should be used to validate the other. Addressing these issues will allow more generalizable results which will help inform a greater understanding of MOUD, which can enhance the efforts of both researchers and clinicians to provide more effective care for those with OUD.