Patient Characteristics Associated with Opioid Abstinence after Participation in a Trial of Buprenorphine versus Injectable Naltrexone

Abstract

Background and Objectives:

Better understanding of predictors of opioid abstinence among patients with opioid use disorder (OUD) may help to inform interventions and personalize treatment plans. This analysis examined patient characteristics associated with opioid abstinence in the X:BOT (Extended-Release Naltrexone versus Buprenorphine for Opioid Treatment) trial.

Methods:

This post-hoc analysis examined factors associated with past-month opioid abstinence at the 36-week follow-up visit among participants in the X:BOT study. 428 participants (75% of original sample) attended the visit at 36 weeks. Logistic regression models were used to estimate the probability of opioid abstinence across various baseline sociodemographics, clinical characteristics, and treatment variables.

Results:

Of the 428 participants, 143 (33%) reported abstinence from non-prescribed opioids at the 36-week follow-up. Participants were more likely to be opioid abstinent if randomized to XR-NTX (compared to BUP-NX), were on XR-NTX at week 36 (compared to those off OUD pharmacotherapy), successfully inducted onto either study medication, had longer time on study medication, reported a greater number of abstinent weeks, or had longer time to relapse during the 24-week treatment trial. Participants were less likely to be abstinent if Hispanic, had a severe baseline Hamilton Depression Rating (HAM-D) score, or had baseline sedative use.

Conclusions:

A substantial proportion of participants were available at follow-up (75%), were on OUD pharmacotherapy (53%), and reported past-month opioid abstinence (33%) at 36 weeks. A minority of patients off medication for OUD reported abstinence and additional research is needed exploring patient characteristics that may be associated with successful treatment outcomes.

1. Introduction

The opioid crisis continues to grip the United States with over 1.6 million people with opioid use disorder (OUD) and over 100,000 drug overdose deaths in the past year (reported in April 2021), the majority related to opioids. This dramatic increase, a nearly 30% increase from the year before, comes amid the rise in highly potent synthetic opioids such as fentanyl and the challenges of the coronavirus-2 (SARS-CoV-2) pandemic (Ahmad et al., 2020; Substance Abuse and Mental Health Services Administration, 2019; Volkow, 2020). FDA-approved medications effective in treating OUD include the full opioid agonist methadone, partial agonist buprenorphine, and the antagonist naltrexone in extended-release formulation (Comer et al., 2006; Krupitsky et al., 2011; Mattick et al., 2003, 2004). These treatments are underutilized and most individuals do not receive medication for opioid use disorder (MOUD) (Substance Abuse and Mental Health Services Administration, 2019; Williams et al., 2019). Even after successful initiation of MOUD, many patients are not retained in treatment (Timko et al., 2016). Patients who stop medication are at increased risk for relapse and overdose, although a minority do succeed in remaining abstinent (Williams et al., 2020; Bailey et al., 2013; Binswanger, 2013).

Most studies examining opioid abstinence have included patients on buprenorphine or methadone, and few have looked at injectable naltrexone. Even fewer studies have examined opioid abstinence in the naturalistic treatment setting. A previous secondary analysis by our group examined opioid use outcomes in a naturalistic follow-up after the X:BOT trial and found similar rates of opioid abstinence for participants on MOUD (sublingual buprenorphine, methadone, injectable naltrexone) and those off medication (33% versus 34%) (Greiner et al., 2021). If medication status (on or off MOUD) in the past 30 days is not significant, then this raises the clinical question of whether duration of MOUD treatment or patient characteristics may be associated with opioid abstinence.

Prior studies that have examined correlates of opioid abstinence among individuals with opioid use disorder (on or off MOUD) suggest that it is positively associated with female gender (Darke et al., 2015), older age (Dreifuss et al., 2013; Naji et al., 2016), more social support (Dennis et al., 2007; Flynn et al., 2003; Zhu et al., 2018), being employed (Dennis et al., 2007; Flynn et al., 2003; McKenna, 2017; Nosyk et al., 2013), less impulsivity (Zhu et al., 2018), older age at first opioid use (Zhu et al., 2018), fewer lifetime treatment episodes (Darke et al., 2015), and longer duration on MOUD (Nosyk et al., 2013; Zhu et al., 2018). Factors negatively associated with abstinence include Hispanic ethnicity (Zhu et al., 2018), substance use among social network (Goehl et al., 1993), criminal and legal involvement (Dennis et al., 2007; Flynn et al., 2003; Nosyk et al., 2013; Scott et al., 2011), IV drug use (Darke et al., 2015; Zhu et al., 2018), more frequent baseline opioid use (Darke et al., 2015), cocaine use (Darke et al., 2007; Joe et al., 1999; Zhu et al., 2018), benzodiazepine use (Dreifuss et al., 2013; Naji et al., 2016; Nosyk et al., 2013), heavy alcohol use (Flynn et al., 2003), comorbid substance use disorders (Marsch et al., 2005), and a greater number of prior treatment episodes (Darke et al., 2015; Dreifuss et al., 2013).

This exploratory secondary analysis provides new data on the association between patient characteristics and non-prescribed opioid abstinence 30 days prior to the 36-week follow-up among participants in the National Drug Abuse Treatment Clinical Trials Network (CTN) Extended-Release Naltrexone (XR-NTX) versus Buprenorphine-Naloxone (BUP-NX) for Opioid Treatment trial (X:BOT) (Lee et al., 2018). We hypothesized that baseline characteristics associated with lesser disease severity, less burden of comorbid psychiatric illness and substance use, absence of pain diagnosis, and psychosocial determinants (employment, stable housing, and absence of substance use among social network) would be associated with opioid abstinence. We also hypothesized that participants who had better outcomes during the 24-week treatment trial (successful induction onto study medication, longer time on medication, a greater number of abstinent weeks, and longer time to relapse) would be more likely to be abstinent from non-prescribed opioids at follow-up.

2. Methods

2.1. Design Overview

This exploratory secondary analysis was based on data from the National Drug Abuse Treatment Clinical Trials Network (CTN) X:BOT trial (clinicaltrials.gov: NCT02032433). The X:BOT trial was a multisite, 24-week, randomized comparative effectiveness trial of buprenorphine-naloxone (BUP-NX) (N=287) versus extended-release naltrexone (XR-NTX) (N=283) for prevention of relapse among individuals with OUD. Study design and rationale have been described previously (Lee et al., 2018; Nunes et al., 2016). Briefly, participants were recruited from eight community-based detoxification programs across the United States. Eligible participants were 18 years or older, English-speaking, had non-prescribed opioid use (heroin, codeine, morphine, oxycodone, methadone, or buprenorphine) within the past month, and met criteria for OUD as per Diagnostic and Statistical Manual of Mental Disorders, 5^th edition (American Psychiatric Association, 2013). Exclusion criteria included serious medical or psychiatric conditions, suicidal ideation, homicidal ideation, other severe substance use disorders, elevated liver transaminases (>5 times the upper limit of normal), chronic pain requiring opioids, methadone maintenance ≥30mg per day, legal status precluding study completion, or reported allergy or sensitivity to BUP-NX or XR-NTX. Females planning pregnancy, pregnant, breastfeeding, or not using a birth control method were excluded. The study was approved by the Institutional Review Board of New York University Medical Center, individual sites obtained local Institutional Review Board approval, and all participants provided written informed consent.

Participants were randomized to BUP-NX or XR-NTX during inpatient admission, stratified by site and by severity of opioid use (high severity operationalized as intravenous heroin use ≥6 bags per day). BUP-NX was initiated after participants had withdrawal symptoms and could tolerate an initial dose. Prior to XR-NTX injection, participants had to complete detoxification and remain off opioids for at least 5 to 7 days, provide a negative opioid urine toxicology, and pass a naloxone challenge.

Participants were followed for up to 24 weeks after randomization and provided with assigned study medication in the context of medical management and outpatient counseling, the latter available (but not required) at community treatment programs. Assessments included weekly self-reported non-prescribed opioid and other substance use, urine toxicology, ratings of opioid cravings, and adverse events. BUP-NX dose was adjusted as clinically indicated (ranging from 8mg to 24mg per day) and XR-NTX (380mg) was administered intramuscularly approximately every 28 days.

Study medications were provided until participant dropout, relapse, or upon completion of the 24-week treatment trial. Participants were encouraged to continue OUD pharmacotherapy at the same community-based treatment program or were referred elsewhere in the community. OUD pharmacotherapy and treatment procedures were naturalistic and may have varied across community settings. Referral for MOUD included methadone, buprenorphine, or injectable naltrexone. Follow-up assessments were at 28 and 36 weeks. Assessments at follow-up visits included the same assessments administered in the 24-week trial. In some cases, follow-up assessments were conducted over the phone and urine toxicology samples were not collected in these instances. This sample included participants who completed the 36-week follow-up (n=428). This timepoint was selected to explore outcomes in the naturalistic treatment setting (i.e., study medications no longer provided and patients receiving treatment in the community). The 28-week follow-up sample was not included in this analysis since there were fewer participants (N=368).

2.2. Baseline Demographic and Clinical Variables

Baseline demographic and clinical characteristics examined for association with the opioid abstinence outcome are listed in Table 1. In addition to demographics, variables were chosen that were associated with severity or treatment outcome of OUD based on prior literature (Zhu et al., 2018; Dreifuss et al., 2013; Darke et al., 2007; Nunes et al., 2011; Sullivan et al., 2006; Goehl et al., 1993). These included severity of opioid use, characteristics of opioid and other substance use, number of family and friends using heroin or other illicit drugs, number of treatment episodes for substance use, self-reported importance of substance use treatment, history of psychiatric disorders, baseline Hamilton Depression Rating Scale (HAM-D) score (Hamilton, 1960), pain, and legal status (on parole or probation). Items were drawn from the medical and psychiatric history, Timeline Followback (TLFB) (Sobell & Sobell, 2000), select items from the Addiction Severity Index (McLellan et al., 1992), and EuroQol (EQ-5D-3L) (Rabin & de Charro, 2001). Opioid use severity was classified as low or high with high use defined as ≥6 bags IV heroin per day in the seven days prior to admission. Other substance use 30 days prior to inpatient admission was assessed using TLFB and included stimulants, sedatives, heavy alcohol (≥5 drinks per day for males and ≥4 drinks per day for females), and cannabis.

Table 1.

Demographics and baseline clinical characteristics of participants at week 36.

	Total Sample (n=428)
Measure	n	% or Mean (SD)
Demographics
Age (years)	428	34.0 (10.1)
Gender
Male	297	69.4%
Female	131	30.6%
Race
White Only	308	72.0%
Black Only	46	10.7%
Other	74	17.3%
Hispanic Ethnicity	80	18.7%
Education
<HS	94	22.0%
HS/GED	144	33.6%
>HS	190	44.4%
Unemployed	266	62.1%
Legal Status
On parole or probation	57	13.3%
Living Situation
Currently Homeless	114	26.6%
Number of family/friends using heroin or other illicit drugs
0	163	38.4%
1–2	95	22.4%
3+	167	39.3%
Opioid Use
Age of onset opioid use	428	21.2 (7.1)
Duration of opioid use (years)	428	12.9 (9.5)
Primary opioid use
Heroin	356	83.4%
Opioid analgesics	71	16.6%
Route of use
IV use	300	70.1%
No IV use	128	29.9%
Severity of opioid use
Low	253	59.1%
High (≥ 6 bags IV heroin daily)	175	40.9%
Other Substance Use
Tobacco smoker	388	90.7%
Stimulant use (30 days prior to admission)	231	54.1%
Sedative use (30 days prior to admission)	119	27.9%
Heavy alcohol use (30 days prior to admission)	111	26.0%
Cannabis use (30 days prior to admission)	197	46.1%
Treatment History
Current treatment is first opioid treatment in lifetime	154	36.0%
Number of times treated for any substance use
0	33	7.7%
1	87	20.3%
2–5	197	46.0%
6+	111	25.9%
Self-reported Importance of Current Treatment
Not at all	9	2.1%
Slightly	5	1.2%
Moderately	8	1.9%
Considerably	37	8.6%
Extremely	369	86.2%
Psychiatric Symptoms and History
History of psychiatric disorder	289	67.5%
HAM-D score
Normal (< 8)	212	49.6%
Mild (8–13)	104	24.4%
Moderate (14–18)	70	16.4%
Severe (> 18)	41	9.6%
Pain
Moderate pain or discomfort	236	55.1%
Extreme pain or discomfort	17	4.0%

Abbreviations: SD=Standard Deviation; HS=High School; GED=General Educational Development; IV=intravenous; HAM-D=Hamilton Rating Depression Scale

2.3. Treatment Variables

Treatment-related variables included randomization assignment (BUP-NX or XR-NTX), successful induction onto study medication (binary: inducted/not inducted), time to relapse (in weeks), time on study medication (in weeks), number of abstinent weeks during the 24-week trial, site (1–8), and medication status at week 36 (BUP-NX, XR-NTX, methadone, or off MOUD).

Randomization assignment

Randomization assignment was defined as randomization to either BUP-NX or XR-NTX in the 24-week treatment phase.

Induction status

Induction status was dichotomous (inducted versus not inducted) onto study medication (BUP-NX or XR-NTX) in the 24-week treatment phase.

Time to relapse

Time to relapse was defined as the use of non-study opioids any time after 20 days post-randomization: at the start of 4 consecutive opioid use weeks or at the start of 7 consecutive days of self-reported opioid use days. A use week was defined as any week during which a participant reported at least one day of nonprescribed opioid use or provided a urine sample positive for non-study opioids. If a urine sample was missing, then relapse was based on the self-report from TLFB since some assessments were done remotely over the phone at week 36. Of the 428 participants who completed week 36 assessments, 49 (11%) had missing urine samples.

Time on study medication

Time on study medication was defined as number of weeks during the 24-week trial that participants reported being on medication (BUP-NX or XR-NTX). Time was computed between induction and the date of the last dose for BUP-NX or 4 weeks after the last XR-NTX injection.

Number of abstinent weeks

Number of abstinent weeks during the 24-week trial was defined using results from weekly urine samples. Abstinence was operationalized as no non-prescribed opioid use in the past week indicated by a negative urine sample for non-prescribed opioids [heroin, codeine, morphine, oxycodone, methadone and buprenorphine (if not on maintenance treatment)].

Medication status at week 36

Medication status at week 36 was determined using the Medical Management Log and self-report by participants on whether or not they had continued to receive treatment in the community as planned (on MOUD versus off MOUD) and which medication (BUP-NX, XR-NTX, or methadone) in the 30 days prior to the 36-week follow-up.

2.4. Outcome Measure

Past-month opioid abstinence

Past-month opioid abstinence reported at week 36 was dichotomous (abstinent versus not abstinent) and operationalized as no non-prescribed opioid use in the past 30 days using self-report with TLFB and a negative (or missing) urine sample for non-prescribed opioids [heroin, codeine, morphine, oxycodone, methadone and buprenorphine (if not on maintenance treatment)]. Unlike in the main study, missing urine samples were not considered to be positive since some follow-up assessments were done remotely over the phone.

2.5. Data Analyses

Descriptive statistics were reported for sample characteristics and by opioid abstinence status; with means and standard deviations (for continuous variables) or frequency and percentages (for categorical variables). To explore potential differences in baseline characteristics between participants available at the 36-week follow-up (N=428) and those missing (N=142), separate logistic regression models were fit to assess whether baseline patient and clinical characteristics were associated with week 36 missingness.

To assess baseline correlates of past-month opioid abstinence at week 36, separate logistic regression models were fit to estimate the probability of opioid abstinence for each baseline and study measure. Model-estimated odds ratios (OR) with 95% confidence intervals (CI) were presented from all logistic regression models.

The interest of this secondary analysis was exploratory, with the goal of generating hypotheses for future work. As such, all tests were two-tailed with alpha level held at 0.05 for statistical significance, and were not corrected for multiple comparisons, in order to identify and investigate associations that may be potentially important. Pairwise comparisons were interpreted only when the omnibus tests were significant to maintain limits on the possibility of Type I error. SAS^® version 9.4 was used for all analyses.

3. Results

3.1. Sample Description

At week 36, 428 (75%) of 570 participants completed assessments and had TLFB data. This follow-up sample included 225 (78%) of the 287 participants initially randomized to BUP-NX and 203 (72%) of 283 participants randomized to XR-NTX). The majority of participants at 36 weeks (375 (88%)) had successfully inducted onto study medication, and of those about half reported being on medication at week 36. Of the 225 participants on MOUD, 120 (53%) were on BUP-NX, 88 (39%) on XR-NTX (of which 87 were on XR-NTX and 1 participant reported switching to oral naltrexone), and 17 (8%) on methadone. Of the 53 participants at week 36 that failed induction onto study medication, 27 (51%) reported being on MOUD at follow-up.

Baseline demographic and participant characteristics are shown in Table 1 for the total sample available at week 36 (N=428). Most participants were white, in their early 30s, unemployed, and over half had a high school education or less. Thirteen percent of participants were on parole or probation, 27% were homeless, and 62% reported having family or friends using heroin or other illicit drugs. The majority had IV use with primary opioid of heroin (as opposed to opioid analgesics), had low severity of opioid use (<6 bags IV heroin use per day), and first began using opioids at an average age of 21 with an average of 13 years of opioid use. Thirty-six percent of participants reported this being their first opioid treatment episode and 86% reported substance use treatment being extremely important to them. Ninety-one percent were current smokers and many reported other substance use (54% reported stimulants, 28% sedatives, 26% heavy alcohol, and 46% cannabis). Sixty-eight percent had a history of a psychiatric disorder. Half of the participants had normal baseline HAM-D scores, about a quarter had mild, and a quarter had moderate or severe. Over half of participants reported moderate to extreme physical pain on the EQ-5D-3L.

Participants available at week 36 (N=428) significantly differed on three baseline measures compared to those missing (N=142, Supplemental Table 1). Participants whose primary opioid was heroin (77%) (compared to opioid analgesics) (OR [95% CI] = 1.59 [1.00, 2.53]) and with IV use (78%) were more likely to be available at week 36 compared to those with no IV use (69%) (OR [95% CI] = 1.57 [1.06, 2.33]). Participants on probation or parole (62%) were less likely to be available at follow-up compared to those who were not (78%) (OR [95% CI] = 0.47 [0.29, 0.76]). Otherwise, participants evaluated at week 36 did not differ significantly from those who were not available, across demographic and other baseline clinical variables.

3.2. Opioid Abstinence and Baseline Correlates

At the 36-week follow-up, 143 (33%) of 428 participants reported past-month abstinence from non-prescribed opioids, and 285 (67%) were not abstinent (Table 2). Among the demographic and baseline clinical characteristics examined, only three were found to be significantly associated with past-month opioid abstinence (Figure 1). Hispanic participants were less likely to be abstinent from non-prescribed opioids (15%) compared to non-Hispanics (38%) (OR [95% CI] = 0.29 [0.15, 0.56]). Participants with severe baseline HAM-D scores (OR [95% CI] = 0.23 [0.09, 0.62]) were less likely to be abstinent from non-prescribed opioids compared to those with normal HAM-D scores. Those with sedative use were less likely to be abstinent at week 36 compared to those with no sedative use (24% versus 37%) (OR [95% CI] = 0.55 [0.34, 0.89]).

Table 2.

Demographics and baseline clinical characteristics by week 36 opioid abstinence groups and logistic regression model results.

	Opioid Abstinent (n=143)		Not Opioid Abstinent (n=285)		Model Results
Measure	n	Row % or Mean (SD)	n	Row % or Mean (SD)	OR	95% CI	p-value
Demographics
Age (years)	143	33.4 (9.7)	285	34.3 (10.3)	0.99	(0.97, 1.01)	0.3810
Gender
Male	95	32.0%	202	68.0%	ref
Female	48	36.6%	83	63.4%	1.23	(0.80, 1.89)	0.3477
Race							0.0729
White	113	36.7%	195	63.3%	ref
Black	11	23.9%	35	76.1%	0.54	(0.26, 1.11)	0.0947
Other	19	25.7%	55	74.3%	0.60	(0.34, 1.06)	0.0764
Hispanic ethnicity	12	15.0%	68	85.0%	0.29	(0.15, 0.56)	0.0002
Non-Hispanic	131	37.6%	217	62.4%
Education							0.6446
<HS	30	31.9%	64	68.1%	ref
HS/GED	45	31.3%	99	68.8%	0.97	(0.55, 1.70)	0.9141
>HS	68	35.8%	122	64.2%	1.19	(0.70, 2.01)	0.5186
Not employed	94	35.3%	172	64.7%	1.26	(0.83, 1.92)	0.2798
Legal Status
On parole or probation	23	40.4%	34	59.6%	1.41	(0.79, 2.50)	0.2405
Living Situation
Currently homeless	36	31.6%	78	68.4%	0.89	(0.56, 1.41)	0.6286
Number of family/friends using heroin or other illicit drugs							0.0814
0	45	27.6%	118	72.4%	ref
1–2	39	41.1%	56	58.9%	1.83	(1.07, 3.12)	0.0276*
3+	58	34.7%	109	65.3%	1.40	(0.87, 2.23)	0.1641
Opioid Use
Age of onset opioid use	143	21.4 (7.4)	285	21.1 (7.0)	1.01	(0.98, 1.03)	0.6707
Duration of opioid use (years)	143	12.0 (8.8)	285	13.3 (9.9)	0.99	(0.96, 1.01)	0.2134
Primary opioid use
Heroin	114	32.0%	242	68.0%	0.72	(0.43, 1.23)	0.2279
Opioid analgesics	28	39.4%	43	60.6%	ref
Route of use
IV use	102	34.0%	198	66.0%	1.09	(0.70, 1.70)	0.6928
No IV use	41	32.0%	87	68%
Severity of opioid use
Low	78	30.8%	175	69.2%	ref
High (≥ 6 bags IV heroin daily)	65	37.1%	110	62.9%	1.33	(0.88, 1.99)	0.1746
Other Substance Use
Tobacco smoker	130	33.5%	258	66.5%	1.05	(0.52, 2.10)	0.8980
Stimulant use (30 days prior to admission)	86	37.2%	145	62.8%	1.45	(0.96, 2.18)	0.0767
Sedative use (30 days prior to admission)	29	24.4%	90	75.6%	0.55	(0.34, 0.89)	0.0142
Heavy alcohol use (30 days prior to admission)	41	36.9%	70	63.1%	1.23	(0.78, 1.93)	0.3719
Cannabis use (30 days prior to admission)	61	31.0%	136	69.0%	0.81	(0.54, 1.22)	0.3071
Treatment History
Current treatment is first opioid treatment in lifetime	53	34.4%	101	65.6%	1.07	(0.71, 1.63)	0.7414
Number of times treated for any substance use							0.2917
0	8	24.2%	25	75.8%	ref
1	34	39.1%	53	60.9%	2.00	(0.81, 4.97)	0.1328
2–5	69	35.0%	128	65.0%	1.68	(0.72, 3.94)	0.2289
6+	32	28.8%	79	71.2%	1.27	(0.52, 3.11)	0.6063
Self-reported Importance of Current Treatment							0.7214
Not at all	3	33.3%	6	66.7%	ref
Slightly	3	60.0%	2	40.0%	3.00	(0.31, 29.03)	0.3419
Moderately	2	25.0%	6	75.0%	0.67	(0.08, 5.57)	0.7076
Considerably	14	37.8%	23	62.2%	1.22	(0.26, 5.69)	0.8020
Extremely	121	32.8%	248	67.2%	0.98	(0.24, 3.98)	0.9727
Psychiatric Symptoms and History
History of psychiatric disorder	92	31.8%	197	68.2%	0.81	(0.53, 1.23)	0.3194
HAM-D score							0.0130
Normal (< 8)	79	37.3%	133	62.7%	ref
Mild (8–13)	40	38.5%	64	61.5%	1.05	(0.65, 1.71)	0.8366
Moderate (14–18)	19	27.1%	51	72.9%	0.63	(0.35, 1.14)	0.1257
Severe (> 18)	5	12.2%	36	87.8%	0.23	(0.09, 0.62)	0.0037
Pain							0.0798
Moderate pain or discomfort	68	28.8%	168	71.2%	0.62	(0.41, 0.94)	0.0249*
Extreme pain or discomfort	6	35.3%	11	64.7%	0.84	(0.30, 2.38)	0.7391

^*Omnibus p-value is not significant.

Abbreviations: SD=Standard Deviation; HS=High School; GED=General Educational Development; IV=intravenous; HAM-D=Hamilton Rating Depression Scale

Figure 1.

Factors significantly associated with past-month opioid abstinence at week 36.*

*Odds ratios are presented on the x-axis with a log scale. 95% confidence intervals are shown for odds ratios.

Abbreviations: XR-NTX=extended-release injectable naltrexone; BUP-NX=buprenorphine-naloxone; MOUD=medication for opioid use disorder; HAM-D=Hamilton Rating Depression Scale

3.3. Treatment Factors Associated with Opioid Abstinence

Several treatment variables were found to be significantly associated with opioid abstinence at week 36 (Table 3; Figure 1). Participants randomized to XR-NTX were more likely to be abstinent from non-prescribed opioids compared to those randomized to BUP-NX (44% versus 24%) (OR [95% CI] = 2.47 [1.63, 3.74]). Participants who successfully inducted onto either study medication were more likely to be abstinent from non-prescribed opioids compared to those not inducted (36% versus 15%) (OR [95% CI] = 3.16 [1.45, 6.92]). Other significant positive correlates of opioid abstinence included: longer time on study medication with an average of 16 weeks (SD 9.0) for those opioid abstinent compared to 14 weeks (SD 9.7) for those not abstinent (OR [95% CI] = 1.03 [1.01, 1.05]), longer time to relapse with an average of 23 weeks (IQR 7.4–23.4) for those opioid abstinent compared to 13 weeks (IQR 3.1–23.4) for those not abstinent (OR [95% CI] = 1.04 [1.02, 1.07]), more abstinent weeks with an average of 15 weeks (SD 8.4) for those opioid abstinent compared to 12 weeks (SD 9.1) for those not opioid abstinent at week 36 (OR [95% CI] = 1.04 [1.02, 1.07]), and those that reported being on XR-NTX at week 36 (53%) compared to those that reported being off medication (34%) (OR [95% CI] = 2.33 [1.40, 3.90]).

Table 3.

Treatment variables by week 36 opioid abstinence groups and logistic regression model results.

	Opioid Abstinent (n=143)		Not Opioid Abstinent (n=285)		Model Results
Measure	n	Row % or Mean (SD)	n	Row % or Mean (SD)	OR	95% CI	p-value
Sites							0.1757
Site 1	21	33.9%	41	66.1%	ref
Site 2	23	39.7%	35	60.3%	1.28	(0.61, 2.71)	0.5118
Site 3	11	26.2%	31	73.8%	0.69	(0.29, 1.65)	0.4066
Site 4	25	36.8%	43	63.2%	1.14	(0.55, 2.34)	0.7306
Site 5	18	24.0%	57	76.0%	0.62	(0.29, 1.30)	0.2049
Site 6	16	32.7%	33	67.3%	0.95	(0.43, 2.10)	0.8926
Site 7	22	47.8%	24	52.2%	1.79	(0.82, 3.92)	0.1453
Site 8	7	25.0%	21	75.0%	0.65	(0.24, 1.78)	0.4023
Treatment Assignment
BUP-NX	54	24.0%	171	76.0%	ref
XR-NTX	89	43.8%	114	56.2%	2.47	(1.63, 3.74)	<.0001
Induction Status
Successful induction onto BUP-NX or XR-NTX	135	36.0%	240	64.0%	3.16	(1.45, 6.92)	0.0040
Relapse History
Relapse	57	25.3%	168	74.7%	0.46	(0.31, 0.70)	0.0002
Time to relapse in weeks	143	23.4 (7.4–23.4)	285	13.3 (3.1–23.4)	1.04	(1.02, 1.07)	0.0003
Opioid Abstinence History
Number of abstinent weeks	142	15.2 (8.4)	277	12.0 (9.1)	1.04	(1.02, 1.07)	0.0006
MOUD History
Time on study medication in weeks	143	16.3 (9.0)	276	13.7 (9.7)	1.03	(1.01, 1.05)	0.0092
MOUD status at week 36							0.0002
Off MOUD	68	33.5%	135	66.5%	ref
BUP-NX	28	23.3%	92	76.7%	0.60	(0.36, 1.01)	0.0546
XR-NTX*	47	53.4%	41	46.6%	2.28	(1.37, 3.79)	0.0016
Methadone	0	0.0%	17	100.0%	0.00	--	0.8224

Abbreviations: SD=Standard Deviation; BUP-NX=buprenorphine-naloxone; XR-NTX=extended-release injectable naltrexone; MOUD=medication for opioid use disorder

^*1 participant in XR-NTX group on oral naltrexone.

4. Discussion

We evaluated patient characteristics and treatment variables as correlates of non-prescribed opioid abstinence among those available at the 36-week follow-up in the X:BOT study. Most baseline patient characteristics generally considered to be related to greater severity of OUD were not significantly associated with opioid abstinence at follow-up, while success in treatment, in particular earlier treatment during the 24-week trial, was associated with later abstinence. Psychosocial determinants such as employment, stable housing, and absence of substance use among participant’s social network were also found to be nonsignificant. These findings are similar to a follow-up study by Weiss et al (Weiss et al., 2015) that looked at opioid use outcomes at 42 months among a subset of individuals who had participated in a trial of sublingual buprenorphine treatment for different durations, and were randomized to receive or not receive additional drug counseling in conjunction with standard medical management for prescription opioid dependence (Weiss et al., 2011). Engagement in buprenorphine therapy was significantly associated with abstinence at 42-month follow-up, whereas most baseline patient characteristics were not.

Baseline characteristics may be less associated with the outcome in these analyses due to the nature of treatment in the initial randomized-controlled trials (RCTs). Baseline characteristics and psychosocial determinants may influence treatment engagement and outcomes more so in the naturalistic setting, whereas in an RCT, patients are engaging in treatment in a controlled setting. Patients are less likely to encounter barriers to entering and continuing in treatment, accessing medication (when study-provided), or encountering other community-level barriers.

In this report, individuals who had better outcomes during the 24-week treatment trial (longer time on study medication, a greater number of abstinent weeks, and longer time to relapse) were more likely to report opioid abstinence at 36 weeks. Our measure of opioid abstinence was limited to the past month. A few studies have explored longer term opioid use outcomes (Darke et al., 2015; Y. I. Hser et al., 2001; Soyka et al., 2017; Weiss et al., 2015; Williams et al., 2020). Hser and colleagues found that maintaining opioid abstinence for at least 5 years substantially increases the likelihood of future stable abstinence (Y. I. Hser et al., 2001; Y.-I. Hser, 2007).

Longer time on study medication (BUP-NX or XR-NTX) during the 24-week treatment trial was also positively associated with non-prescribed opioid abstinence at follow up. A retrospective longitudinal cohort analysis of Medicaid claims found that patients were at high risk for relapse and overdose after buprenorphine discontinuation, irrespective of treatment duration (up to 18 months) (Williams et al., 2020). The optimal length of treatment on MOUD continues to be an important clinical question, and more research is needed exploring this in naturalistic treatment settings.

The majority of patients who discontinue medication are at increased risk for relapse and overdose, although a minority do succeed in remaining abstinent (Williams et al., 2020; Bailey et al., 2013; Binswanger, 2013). As seen in a previous secondary analysis by our group, we found similar rates of opioid abstinence in the past month between participants on MOUD and those off medication at week 36, although relapse rates were higher among those off MOUD (Greiner et al., 2021). In one trial performed in Russia (Krupitsky et al., 2011), 30% of patients that received placebo and counseling alone after inpatient detoxification were abstinent throughout the 24-week study compared to the 50% randomized to XR-NTX. Still, more evidence supports better outcomes and abstinence for the majority of patients on MOUD compared to those off medication (Krupitsky et al., 2011; Weiss et al., 2019, 2015; Y.-I. Hser et al., 2016; Y.-I. Hser, 2007).

However, in this analysis, only XR-NTX (compared to off MOUD) was found to be significantly associated with opioid abstinence at week 36. In addition, those randomized to XR-NTX (versus BUP-NX) in the main trial were more likely to be abstinent at follow-up. A separate analysis of the X:BOT per-protocol and completers sample also found that compared to XR-NTX, BUP-NX had significantly greater proportions of opioid-positive urine tests (Mitchell et al., 2021). These findings may be related to the different pharmacological profiles of naltrexone and buprenorphine. While patients are within 28 days of their last XR-NTX injection, opioid effects are fully blocked for most patients and there is typically less opioid use (Nunes et al., 2020). Risk of opioid use and relapse increases when patients miss the next injection. In contrast, sublingual BUP-NX depends on daily adherence and patients may be more likely to intermittently miss doses and are more vulnerable to opioid use at those times. Another consideration when comparing these medications is that there was a substantial induction hurdle for injectable naltrexone in the parent study with 28% dropping out of treatment before induction (versus 6% in the buprenorphine group); and nearly all induction failures had early relapse (Lee et al., 2018).

Different measurements provide different views of treatment outcomes. The outcome of opioid abstinence may not be the best indicator of successful treatment. OUD is a chronic condition and improvement in use patterns may facilitate important recovery goals without complete abstinence. Buprenorphine is a partial opioid agonist with daily dosing and individuals may be able to use opioids intermittently but not necessarily meet criteria for relapse, while still demonstrating improved clinical and psychosocial outcomes. A recent X:BOT analysis by our group found that more individuals on XR-NTX at 36-week follow-up were abstinent from non-prescribed opioids compared to BUP-NX (53% versus 23%), but did not differ significantly across other opioid and substance use outcomes (Greiner et al., 2021). This study’s finding of greater rates of recent abstinence at 36 weeks in those treated with XR-NTX complements, but does not contradict or reverse the main findings of the parent study that rates of study-defined relapse were greater at 24 weeks in the intention-to-treat (ITT) analysis and no different in the per protocol analysis.

Severe baseline HAM-D scores were negatively associated with opioid abstinence; consistent with many other studies that suggest baseline depressive symptoms and/or a prior diagnosis of depression are linked to more severe opioid use and other substance use disorder outcomes (Hasin et al., 2002; Najt et al., 2011; Nunes et al., 2004; Rounsaville et al., 1986). However, a secondary analysis of a large, multi-site randomized trial of buprenorphine for prescription opioid dependence found that a diagnosis of lifetime major depressive disorder had the strongest association with better outcomes for OUD. This counterintuitive finding may be related to the potential antidepressant properties of buprenorphine (Peciña et al., 2019; Saxena & Bodkin, 2019) or could also be related to a greater motivation to engage in treatment among those with a depression history. A subsequent analysis of these data (Peckham et al., 2020) did not find evidence for buprenorphine treatment leading to a reduction in depressive symptoms. Regardless of medication choice, treatment of OUD and other substance use disorders leads to improvement in depressive symptoms, (Dean et al., 2004; Krupitsky et al., 2016; Latif et al., 2019; Mysels et al., 2011; Romero‐Gonzalez et al., 2017). More research is needed examining the complex interplay between psychiatric and substance use disorders and treatment outcomes.

With regard to other substance use, baseline sedative use was found to be negatively associated with opioid abstinence. Other studies have shown that in addition to sedative use (Dreifuss et al., 2013; Naji et al., 2016; Nosyk et al., 2013), cocaine (Darke et al., 2007; Joe et al., 1999; Zhu et al., 2018) and heavy alcohol (Flynn et al., 2003) use were associated with non-abstinence and worse opioid use outcomes. There is a need for comprehensive and integrated treatment to address comorbid psychiatric and other substance use disorders among individuals with OUD.

Hispanic participants were less likely to be opioid abstinent (compared to non-Hispanic participants), similar to other trials (Alegría et al., 2006; Zhu et al., 2018). A previous study found that Hispanic individuals were less likely to be in active treatment (22% versus 38%), more likely to report not receiving the necessary treatment services (27% of Hispanics compared to 13% of whites), more likely to experience delays in care (23% versus 11%), and tended to be less satisfied with substance use disorder treatment compared to white respondents (Wells et al., 2001). In recent years, the opioid epidemic has significantly impacted minorities with a dramatic increase in opioid use and overdose deaths among Hispanic, African American, and American Indian/Alaska Native populations (The Opioid Crisis and the Hispanic/Latino Population: An Urgent Issue., 2020). There is an urgent need to better identify needs, understand cultural contexts of OUD treatments, identify potential barriers to care, and adapt models of care to improve outcomes among minoritized populations.

Strengths of the current analysis include the relatively high follow-up rate at week 36 (75% of the original sample), and the availability of diverse sociodemographic and treatment characteristics. Limitations of this report include that this was a secondary analysis and was not randomized or part of an a priori design. As statistical tests were not corrected for multiple comparisons, it is crucial that all findings are considered preliminary, and need replication and validation in future research. Non-prescribed opioid abstinence was self-reported and may include recall error or other biases; however prior studies have demonstrated adequate reliability and validity of such data in research settings (Carey, 1997; Fals-Stewart et al., 2000). Opioid abstinence was operationalized differently than in the parent study and missing urine samples at follow-up assessments were not considered to be positive since some assessments were conducted by phone at week 36, albeit a small proportion (49 (11%) of participants). Medication status was based on patient report and did not include confirmatory urine screening for opioid agonists (buprenorphine or methadone) or documented administration of XR-NTX. Associations between treatment participation and the outcome of opioid abstinence may not be causal, but rather reflect motivation and thus inherently better outcome at follow-up. Lastly, these results may not be generalizable to individuals with OUD who initiate medication in an outpatient setting. XR-NTX often requires an inpatient treatment admission whereas buprenorphine can be started immediately during an inpatient or outpatient admission; the main study recruited and enrolled participants while they were in an inpatient setting.

5. Conclusions

In summary, a substantial proportion of participants were available at follow-up (75% of the original sample), were on MOUD (53%), and reported past-month abstinence from non-prescribed opioids (33%) at 36 weeks (three months or longer after study-provided treatment ended). A minority of patients off medication reported abstinence from non-prescribed opioids and future research should explore characteristics of patients that may be more successful off medication or after a certain amount of time on medication. More research is needed exploring key sociocultural and contextual considerations such as access to care and cultural responsiveness in the delivery of treatment interventions for Hispanic and other minoritized populations with OUD.