Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2023 Dec 11.
Published in final edited form as: Drug Alcohol Depend. 2023 Aug 9;251:110926. doi: 10.1016/j.drugalcdep.2023.110926

Risks of returning to opioid use at treatment entry and early in opioid use disorder treatment: role of non-opioid substances

Felipe Castillo 1,*, Mei-Chen Hu 1, Ying Liu 1, Raymond R Balise 5, Roger D Weiss 2, John Rotrosen 3, Edward V Nunes 1, Andrew J Saxon 4, Daniel J Feaster 2, Sean X Luo 1
PMCID: PMC10712265  NIHMSID: NIHMS1946543  PMID: 37604012

Abstract

Objective

Patients in treatment with medications for opioid use disorder (MOUD) often report use of other substances in addition to opioids. Few studies exist that examine the relationship between use at treatment entry and early non-opioid use in opioid treatment outcome.

Methodology

We combined and harmonized three randomized, controlled MOUD clinical trials from the National Institutes of Drug Abuse (NIDA) Clinical Trials Network (CTN) (N=2,197) and investigated the association of non-opioid substance use at treatment entry and during early treatment with a return to opioid use. The trials compared MOUD treatment (buprenorphine, methadone, extended-release naltrexone) in populations with opioid use disorder (OUD). Non-opioid substances were identified through harmonizing self-reported use. The primary outcomes were markers of return to opioid use by 12 weeks.

Results

When treatment cohorts were adjusted, no association between self-reported treatment entry use of non-opioid substances and week-12 opioid use was detected. During the first month of treatment, higher use of cocaine (OR 1.41 [1.18–1.69]) and amphetamine (OR 1.70 [1.27–2.26]) was found to be associated with higher likelihood of illicit opioid use by week 12. Exploratory analyses of potential treatment cohort-by-predictor interactions showed that those with heavier cocaine use had a lower rate of returning to opioid use in the extended-release naltrexone group than in the methadone group.

Conclusion

Substance use other than opioids at treatment entry is not associated with relapse. Use of cocaine or amphetamines during the first few weeks of MOUD treatment may signal a worse outcome, suggesting a need for additional interventions.

Introduction

Opioid use disorder (OUD) continues to exert an enormous public health impact in terms of mortality, morbidity and long term disability (Hedegaard, Miniño, and Warner 2020; McDermott and Sun 2017; Peterson et al. 2021). The use of other substances in addition to opioids often carries additional morbidity and mortality and may occur prior to starting and during treatment for OUD (Drake et al. 2019; Mahoney et al. 2021). Whether use patterns of substances other than opioids at and prior to treatment entry relate to the effectiveness of a specific OUD treatment medication has not been examined comprehensively. Typically, studies of baseline predictors of treatment outcome in OUD rely on retrospective reviews of longitudinal observational data (Ford et al. 2021), and different outcome measures including abstinence, adherence, retention, overdose among others are used, though not always in an uniform fashion. Generally, multiple substance use disorder diagnoses in addition to OUD decrease the likelihood of treatment initiation with MOUDs (Lin et al. 2020; Mintz et al. 2021) and worsen outcomes in terms of treatment retention (Rosic et al. 2017; Bunting et al. 2022).

Substance use data during and prior to treatment for OUD are heterogeneous, and capturing the effect of multiple substances at once poses challenges (Connor et al. 2014; Ellis et al. 2023). Interactions between polysubstance use and general health outcomes suggest that those who use multiple substances compared to opioid use only have worse outcomes such as abscesses and overdoses (Betts et al. 2016; Riley et al. 2016). A review of each specific non-opioid substance mostly suggests that there are no obvious, consistent interactions between their use and OUD treatment outcomes. The use of the following common substances in OUD treatment: alcohol (Srivastava, Kahan, and Ross 2008; Mintz et al. 2021), cannabis (McBrien et al. 2019; Olfson et al. 2018; Rosic et al. 2021), and benzodiazepines (Abrahamsson et al. 2017; Brands et al. 2008) all have inconsistent effects on OUD treatment outcomes. Stimulants and cocaine in particular, have been shown to be associated with worse MOUD initiation and retention (Cook et al. 2023), as well as other inconsistent effects (Bovasso and Cacciola 2003; Schottenfeld et al. 1997). This heterogeneity may be due to the diversity of drug use patterns and different mechanisms of actions of each drug. Lastly, while extended-release naltrexone has been shown to be effective for both alcohol use disorder and OUD, to date no study has specifically examined the comparative effectiveness of this agent in individuals with both conditions.

We therefore examined the relationship between treatment entry and early treatment non-opioid substance use and response to MOUD treatment among patients across three large clinical trials, after been combining and harmonizing their data. Between them, the trials included arms for treatment with all three major medications for opioid use disorder—methadone, buprenorphine, and extended-release naltrexone. We hypothesized that the use of non-opioid substances, both at entry into treatment and during the first month of treatment would be associated with higher risk of a return to regular opioid use during treatment with MOUD. We also explored whether associations between non-opioid substance use and a return to use outcome differ by type of medication treatment. Based on existing literature and pharmacodynamic mechanisms, we hypothesized that individuals with heavy use of alcohol may respond better on extended-release naltrexone (XR-NTX) given this medication’s effectiveness as a treatment for alcohol use disorder.

Methods

Data Sources

Data from three large, previously published clinical trials were combined and harmonized. The three trials were conducted under the National Institute of Drug Abuse’s (NIDA) National Drug Abuse Treatment Clinical Trials Network (CTN). Fully de-identified data sources from the three previously published trials were included in the analysis: CTN0027/START (Saxon et al. 2013), CTN0030/POATS (Weiss et al. 2010) and CTN0051/X:BOT (Lee et al. 2018). Briefly, CTN0027 enrolled from May of 2006 to October of 2009 at federally licensed opioid treatment programs; CTN0030 enrolled from June of 2006 to July of 2009 at a diverse group of outpatient opioid treatment facilities, and CTN0051 enrolled from January of 2014 to May of 2016 from a combination of inpatient and outpatient opioid treatment sites. All three trials were pragmatic and broadly included individuals who meet criteria for OUD (DSM-5) or opioid dependence (DSM-IV), with CTN0030 specifically included only individuals who predominantly used prescription opioids. CTN0027 and CTN0030 were outpatient studies. CTN0027 compared patients randomized to open-label buprenorphine or methadone after 12–24 hours of abstaining from opioids and starting the study medications in mild withdrawal. CTN0030 patients were inducted onto buprenorphine only, comparing a psychosocial intervention delivered by non-physician psychotherapist with treatment as usual, a manual-based intervention delivered by the study physician. Lastly, participants in CTN0051 underwent medically supervised withdrawal and were randomized to treatment with buprenorphine or XR-NTX initiated on an inpatient basis, and then completing the trial as outpatients. For the participants maintained on XR-NTX, an injection was delivered every 28 days.

Harmonization

The number of days individuals used a specific non-opioid and opioid substance from all three studies were harmonized using measures of Timeline Follow Back (TLFB, Sobell and Sobell, 1995) at treatment entry data. Use of cannabis (including all forms), cocaine (including all forms and routes of administration), alcohol (any drinks), drinking alcohol to intoxication (defined as 3 or more standard drinks as reported on CTN0051 and CTN0027), amphetamine and derivatives (both amphetamine and methamphetamine), and benzodiazepines (prescribed and non-prescribed) was surveyed. Participants in CTN0030 did not have number of drinks recorded in TLFB, and alcohol use data were instead harmonized from the Addiction Severity Index (ASI-Lite), which contains a survey of “days of drinking to intoxication” (McLellan et al. 1980). We harmonized the data for 28 days, to avoid any effects from weekend-driven use, as some 30-day periods will include five weekends while others will include four.

Similarly, days of non-opioid substance use in the first 28 days after randomization were obtained from the timeline follow back interview (TLFB).

Urine toxicology results were harmonized by study week after randomization, and all non-prescribed use of any opioids was counted as positive opioid use. Weekly urine toxicology and TLFB were collected for all three studies during all treatment phases. We defined relapse, a binary outcome, as the appearance of four consecutive opioid positive or missing urine toxicology tests between treatment weeks 4 and 12, inclusive. In the present analysis, urine toxicology data may not capture a period of abstinence in some patients, and we counted these cases as a “return to use” though they could also be considered “induction failures”.

Statistical Analyses

We converted the number of days of non-opioid substance use into ordinal categories to account for non-normal distribution of this predictor variable: 0 days of use, 1–4 days a month, 5–14 days a month and 15–28 days a month. Use at treatment entry was obtained from the TLFB for the 28 days prior to randomization and early treatment use for the 28 days following randomization. The sample was divided into five groups based on medication treatment assignment as follows: two groups in CTN0027, (methadone and buprenorphine) two groups in CTN0051 (buprenorphine and extended-release naltrexone) and one group in CTN0030 (buprenorphine). Demographic covariates are presented for each of these five groups. Logistic regression was used to test for effects of level of non-opioid substance use at treatment entry and in early treatment on opioid relapse by week 12 based on urine toxicology results. Alpha was set to p = 0.007 for Bonferroni corrected threshold for 7 hypotheses simultaneously tested (for each of the non-opioid substances on the TLFB) and p = 0.05 for exploratory hypotheses for the effects of the non-opioid substance differed by five-group treatment cohorts. Of note, the early treatment effects were estimated using the subsample of participants who did not drop out of the studies in the first 28 days. Interaction effect models were used to estimate treatment specific interactions for non-opioid substances. All analyses were conducted in SAS software Version 9.4 Copyright © (2018) SAS Institute.

Results

Demographics and substance use at treatment entry

Harmonizing three large studies yielded a total of 2,199 participants at treatment entry, and 2 participants were excluded who did not have complete TLFB data. Demographic and treatment group assignment data are tabulated in Table 1. The CTN0051 buprenorphine group had the most participants who used cannabis, whereas the CTN0027 methadone group had the fewest; the CTN0051 groups had more participants with more frequent cocaine use, who used alcohol more than 15 days a month, who were intoxicated more than 15 days a month, and more individuals using benzodiazepines more than 15 days a month.

Table 1.

Socio-demographic characteristics and number of days of use of substances in the month prior to OUD treatment initiation by treatment group cohorts (N=2,197 with complete baseline TFLB data). BUP: buprenorphine; XR-NTX: extended-release naltrexone

CTN27 Methadone (N=528) CTN27BUP (N=739) CTN30BUP (N=360) CTN51 BUP (N=287) CTN51 XR-NTX (N=283) Total (N=2,197) X2 or F-test
Mean (SD) or %
Age 36.80 (10.86) 36.98 (11.24) 32.06 (9.65) 33.68 (9.80) 34.04 (9.46) 35.32 (10.67) F(4, 2192)=18.76, p<.0001
Male 67.80 67.93 58.06 71.78 68.90 66.91 X2(4)=16.86, p=.0021
Non-Hispanic white (%) 69.32 66.17 88.61 65.85 64.66 70.37 X2(9)=79.80, p<.0001
Non-Hispanic African Americans (%) 9.66
10.01		 2.50 11.50 12.01 9.15
Hispanic of any race (%) 15.34 16.91 5.00 18.82 15.90 14.70
Other (%) 5.68 6.90 3.89 3.83 7.42 5.78
Heroin (days/month) 0 5.68 5.41 90.28 13.59 13.43 21.48 X2(12)=1303.53, p<.0001
1–4 3.79 4.47 9.44 4.18 2.83 4.87
5–14 5.68 7.71 0.28 7.67 9.19 6.19
15–28 84.85 82.41 0.00 74.56 74.56 67.46
Prescription Opioids (days/month) 0 79.92 79.16 1.67 58.89 60.78 61.63 X2(12)=1123.60, p<.0001
1–4 7.01 7.58 3.61 14.29 14.49 8.56
5–14 5.49 4.06 4.72 9.06 9.54 5.87
15–28 7.58 9.20 90.00 17.77 15.19 23.9
Cannabis (days/month) 0 72.73 71.31 61.39 54.01 57.60 66.00 X2(12)=77.74, p<.0001
1–4 15.91 15.29 16.67 19.51 19.43 16.75
5–14 5.11 4.74 10.83 8.01 5.65 6.37
15–28 6.25 8.66 11.11 18.47 17.31 10.88
Cocaine (days/month) 0 53.03 60.22 81.94 59.58 63.25 62.36 X2(12)=101.58, p<.0001
1–4 26.70 23.00 15.56 19.86 18.73 21.71
5–14 13.07 8.93 2.50 12.20 10.25 9.47
15–28 7.20 7.85 0.00 8.36 7.77 6.46
Alcohol (days/month) 0 72.92 74.02 60.28 56.45 56.54 66.95 X2(12)=88.83, p<.0001
1–4 18.18 18.67 30.56 25.44 26.50 22.39
5–14 5.49 4.60 6.11 7.67 8.48 5.96
15–28 3.41 2.71 3.06 10.45 8.48 4.69
Alcohol intoxication (days/month) 0 86.93 85.79 76.11 66.55 69.26 79.84 X2(12)=128.74, p<.0001
1–4 9.66 9.74 18.33 16.38 16.61 12.88
5–14 1.89 3.11 3.61 7.32 6.01 3.82
15–28 1.52 1.35 1.94 9.76 8.13 3.46
Methadone (days/month) 0 81.06 79.97 74.44 89.20 92.93 82.20 X2(12)=56.57, p<.0001
1–4 10.23 10.28 12.22 7.67 4.24 9.47
5–14 4.55 5.55 6.11 1.05 1.77 4.32
15–28 4.17 4.19 7.22 2.09 1.06 4.01
Amphetamine (days/month) 0 87.31 87.69 93.06 78.75 83.75 86.80 X2(12)=58.32, p<.0001
1–4 9.85 8.66 5.56 12.20 11.31 9.24
5–14 1.70 3.38 1.39 4.88 2.47 2.73
15–28 1.14 0.27 0.00 4.18 2.47 1.23
Prescription Benzodiazepines (days/month) 0 84.66 83.22 67.22 71.43 76.33 78.52 X2(12)=92.79, p<.0001
1–4 12.88 13.26 23.06 17.07 11.66 15.07
5–14 1.70 2.98 6.11 6.2 7.42 4.23
15–28 0.76 0.54 3.61 4.88 4.59 2.18
Open in a new tab

Associations of use at treatment entry of non-opioid substances on opioid relapse

The odds ratio of a return to opioid use 12 weeks after randomization by specific non-opioid substance use are presented in Table 2. Given that the use of non-opioid substances might be correlated with heavier use of opioids, we adjusted for both heroin and prescription opioid use, and aimed to detect an association independent of baseline opioid use severity. Our analysis showed no effects of use at treatment entry of any non-opioid substances (Table 2A). Estimating these effects without adjusting for the effect of higher severity for individuals who use heroin heavily, and accounting for multiple comparisons, we likewise did not detect any significant effect. We investigated whether there are specific interactions between the use of alcohol, stimulants, benzodiazepines, and marijuana at treatment entry, and the five group treatment cohorts. Analyses showed that there was an interaction effect between cocaine use prior to treatment entry and treatment cohort assignment (F(4, 2156) = 4.17, p = 0.0023, Table 2B). Visualizing this effect (Fig 1) shows that it was driven by the extended-release naltrexone group having lower relapse rate when patients have heavy cocaine use prior to treatment, compared to those who do not use cocaine. Among the CTN0051 extended-release naltrexone group, for each one-unit increase in cocaine use, the odds of returning to use opioids was decreased by 29% (OR 0.71, 95% CI=[0.55–0.93]). Among the CTN0027 methadone group, for each one-unit increase in cocaine use, the odds of returning to use opioids was increased by 38% (OR 1.38, 95% CI=[1.13–1.68]). There were no effects of cocaine use detected on opioid relapse for the remaining three groups (CTN0027 BUP, CTN0030 BUP and CTN0051 BUP).

Table 2:

A. Logistic regression results between each non-opioid substance prior to treatment entry and relapse to opioid use by week 12 (N=2197). B. Exploratory analyses for interaction between treatment group cohort and each non-opioid substance.

A.
Adjusted OR (95%CI) t, p
Substance use 28 days prior to screeninga
Cannabis 0.93 (0.85–1.02) t= −1.46, p=0.1448
Cocaine 1.06 (0.96–1.18) t= 1.14, p=0.2551
Alcohol 0.89 (0.79–1.00) t= −2.04, p=0.0410
Alcohol intoxication 0.89 (0.78–1.01) t= −1.75, p=0.0800
Amphetamine 1.13 (0.94–1.35) t= 1.32, p=0.1855
Benzodiazepines 1.06 (0.92–1.22) t= 0.81, p=0.4201
B.
Adjusted OR (95%CI) t, p F, p
Interaction between treatment group cohort and baseline substance use 28 days prior to treatment entryb
Cannabis CTN27Methadone
CTN27BUP
CTN30BUP
CTN51BUP
CTN51-XR-NTX51		 0.86 (0.70–1.06)
1.00 (0.85–1.19)
0.81 (0.64–1.03)
0.87 (0.70–1.08)
1.09 (0.88–1.35)		 Ref.
t= 1.16, p= 0.25
t= −0.36, p= 0.72
t= 0.06, p= 0.95
t= 1,55, p= 0.12		 F (4, 2156) = 1.28, p = 0.2753
Cocaine CTN27Methadone
CTN27BUP
CTN30BUP
CTN51BUP
CTN51-XR-NTX51		 1.38 (1.13–1.68)
1.10 (0.93–1.30)
1.06 (0.64–1.77)
0.93 (0.72–1.20)
0.71 (0.55–0.93)		 Ref.
t= −1.74, p= 0.08
t= −0.93, p= 0.35
t= −2.38, p= 0.02
t= −3.94, p< .0001		 F (4, 2156) = 4.17, p = 0.0023
Alcohol CTN27Methadone
CTN27BUP
CTN30BUP
CTN51BUP
CTN51-XR-NTX51		 0.94 (0.74–1.19)
0.84 (0.68–1.04)
0.61 (0.42–0.89)
1.06 (0.82–1.36)
0.93 (0.72–1.20)		 Ref.
t= −0.70, p= 0.49
t= −1.92, p= 0.06
t= 0.66, p= 0.51
t= −0.07, p= 0.95		 F (4, 2156) = 1.56, p = 0.1819
Alcohol intoxication CTN27Methadone
CTN27BUP
CTN30BUP
CTN51BUP
CTN51-XR-NTX51		 0.91 (0.65–1.27)
0.80 (0.61–1.05)
0.60 (0.38–0.93)
1.13 (0.87–1.46)
0.89 (0.68–1.16)		 Ref.
t= −0.57, p= 0.57
t= −1.48, p= 0.14
t= 1.01, p= 0.31
t= −0.10, p= 0.92		 F (4, 2156) = 1.75, p = 0.1359
Amphetamine CTN27Methadone
CTN27BUP
CTN30BUP
CTN51BUP
CTN51-XR-NTX51		 1.12 (0.78–1.62)
1.16 (0.82–1.64)
1.06 (0.54–2.06)
1.18 (0.85–1.65)
1.04 (0.70–1.55)		 Ref.
t= 0.15, p= 0.88
t= −0.15, p= 0.88
t= 0.22, p= 0.83
t= −0.27, p= 0.78		 F (4, 2156) = 0.08, p = 0.9882
Benzodiazepines CTN27Methadone
CTN27BUP
CTN30BUP
CTN51BUP
CTN51-XR-NTX51		 0.93 (0.65–1.35)
0.92 (0.68–1.24)
1.16 (0.86–1.55)
1.37 (1.01–1.85)
0.94 (0.69–1.28)		 Ref.
t= −0.07, p= 0.94
t= 0.89, p= 0.37
t= 1.57, p= 0.12
t= 0.04, p= 0.97		 F (4, 2156) = 1.25, p = 0.2876
Open in a new tab
a

Coded using ordinal groups for levels of use (0 – no use, 1 – 1 to 4 days a month, 2 – 5 to 14 days a month and 3 – 15 to 28 days) and controlling for age, gender, race/ethnicity, 5-group treatment cohort (CTN27-methadone, CTN27-buprenorphine, CTN30-buprenorphine, CTN51-burprenorphine, CTN51-extended-release naltrexone), baseline heroin and prescription opioids levels of use in each model. One regression model was created for each one of the non-opioid substances of abuse.

b

Analyses in the section B included interaction effect between treatment group cohort and each non-opioid substance use in each model, besides of the same covariates with main effect of each non-opioid substance use in the section A.

Figure 1.

Figure 1.

Open in a new tab

Relationship between number of days of use at treatment entry for cocainea and the probability of relapse to opioid use by week 12 (N = 2,197). In particular, individuals who use cocaine heavily have a decreased rate of relapse when they were assigned to the XR-NTX group. BUP: buprenorphine; XR-NTX: extended-release naltrexone. P = 0.09 (F-test for interaction).

Associations of first month use of all substances on opioid relapse

We tested a second set of hypotheses that early treatment patterns (first 4 weeks) of non-opioid substance use would be associated with week-12 treatment outcome. When adjusted for opioid use during early treatment, we were able to detect associations between early treatment cocaine use (OR 1.41, 95% CI=[1.18–1.69] and amphetamine use (OR 1.70, 95% CI=[1.27–2.26]) (Table 3A). No interaction effects between treatment cohort assignment and each non-opioid substance use in the first four weeks were detected (Table 3B).

Table 3:

A. Logistic regression results between use of non-opioid substances in the first 4 weeks of treatment after randomization and relapse of opioid use by week 12 (N=1,850 with complete TLFB data). B. Interaction effects between treatment group cohort and non-opioid substance use in the first 4 weeks.

A.
Adjusted OR (95%CI) t, p
Non-opioid substance use in the first 4 weeks of treatment after randomizationa
Cannabis 1.15 (1.00–1.33) t= 1.96, p=0.0507
Cocaine 1.41 (1.18–1.69) t= 3.79, p=0.0002
Alcohol 1.07 (0.88–1.29) t= 0.67, p=0.5006
Alcohol intoxicationc 1.344 (1.01–1.79) t= 2.01, p=0.0442
Amphetamines 1.70 (1.27–2.26) t= 3.60, p=0.0003
Benzodiazepines 1.29 (1.03–1.63) t= 2.17, p=0.0299
B.
Adjusted OR (95%CI) t, p F, p
Interaction between treatment group cohort and Non-opioid substance use in the first four weeks of treatment after randomizationb
Cannabis CTN27Methadone
CTN27BUP
CTN30BUP
CTN51BUP
CTN51-XR-NTX51		 1.12 (0.83–1.51)
1.24 (0.96–1.60)
1.49 (0.97–2.27)
0.91 (0.61–1.35)
1.18 (0.86–1.63)		 Ref.
t= 0.50, p= 0.62
t= 1.07, p= 0.29
t= −0.82, p= 0.41
t= 0.24, p= 0.81		 F (4, 1804) = 0.76, p = 0.5513
Cocaine CTN27Methadone
CTN27BUP
CTN30BUP
CTN51BUP
CTN51-XR-NTX51		 1.23 (0.92–1.65)
1.32 (0.98–1.79)
1.06 (0.53–2.15)
1.57 (0.83–2.97)
2.10 (1.14–3.88)		 Ref.
t= 0.34, p= 0.74
t= −0.38, p= 0.71
t= 0.67, p= 0.50
t= 1.53, p= 0.13		 F (4, 1804) = 0.75, p = 0.5573
Alcohol CTN27Methadone
CTN27BUP
CTN30BUP
CTN51BUP
CTN51-XR-NTX51		 1.12 (0.75–1.67)
0.80 (0.57–1.13)
1.18 (0.71–1.94)
1.58 (0.96–2.58)
1.44 (0.87–2.38)		 Ref.
t= −1.25, p= 0.21
t= −0.15, p= 0.88
t= 1.05, p= 0.29
t= 0.77, p= 0.44		 F (4, 1804) = 1.62, p = 0.1665
Alcohol intoxicationc CTN27Methadone
CTN27BUP
CTN30BUP
CTN51BUP
CTN51-XR-NTX51		 1.69 (0.93–3.09)
1.38 (0.72–2.65)
1.06 (0.65–1.72)
1.61 (0.85–3.02)		 Ref.
t= −0.46, p= 0.65
t= −1.20, p= 0.23
t= −0.13, p= 0.90		 F (3, 1471) = 0.60, p = 0.6153
Amphetamine CTN27Methadone
CTN27BUP
CTN30BUP
CTN51BUP
CTN51-XR-NTX51		 1.61 (0.97–2.67)
1.28 (0.78–2.08)
2.18 (0.58–8.21)
8.14 (2.59–25.55)
1.40 (0.70–2.80)		 Ref.
t= −0.66, p= 0.51
t= 0.42, p= 0.67
t= 2.54, p= 0.01
t= −0.32, p= 0.75		 F (4, 1804) = 2.24, p = 0.0627
Benzodiazepines CTN27Methadone
CTN27BUP
CTN30BUP
CTN51BUP
CTN51-XR-NTX51		 0.82 (0.48–1.38)
1.46 (0.93–2.29)
1.12 (0.70–1.81)
1.99 (0.93–4.23)
1.92 (1.09–3.37)		 Ref.
t= 1.65, p= 0.10
t= 0.88, p= 0.38
t= 1.90, p= 0.06
t= 2.17, p= 0.03		 F (4, 1804) = 1.67, p = 0.1547
Open in a new tab
a

Coded as ordinal groups for levels of use (0 – no use, 1 – 1 to 4 days a month, 2 – 5 to 14 days a month and 3 – 15 to 28 days) and controlling for age, gender, race/ethnicity, 5-group treatment cohort (CTN27-methadone, CTN27-buprenorphine, CTN30-buprenorphine, CTN51-burprenorphine, CTN51-extended-release naltrexone), heroin use, prescription opioids use and each non-opioid substance use on treatment entry in each model.

b

Analyses in section B included two interaction effects between treatment group cohort with each non-opioid substance use in the first 4 weeks, and with the use prior to treatment entry, besides the same covariates with main effects of each non-opioid substance use in the section A.

c

Participants in CTN0030 was excluded due to lack of data on alcohol intoxication after randomization.

Discussion

Co-occurrence of a non-opioid substance use and OUD is a common clinical phenomenon (Jones and McCance-Katz 2019), but how exactly it influences MOUD treatment outcome remains to be seen. Using a harmonized dataset (CTN0094) generated from three large clinical trials, we aimed to identify whether use of non-opioid substances either at the time of treatment entry or in the first month of treatment might influence week-12 treatment outcome. We found no effect for any substances at treatment entry. We found that ongoing use of amphetamine and cocaine in the first 4 weeks of MOUD predicted worse OUD treatment outcome at week 12. Exploratory analyses of treatment by moderator interaction showed that treatment with XR-NTX was associated with reduced negative effects of cocaine use during first 4 weeks compared to treatment with methadone or buprenorphine.

The lack of relationships of heavy use of non-opioid substances prior to treatment with treatment outcome suggests that the clinical lore that “poly-substance” users do worse in MOUD treatment in general may be a small and noisy effect, especially when the underlying sample is large and heterogenous. We were surprised, for example, that we were unable to detect an association of reported cocaine use at treatment entry with relapse at week 12. The inability to detect an association was apparent with or without adjustments for treatment cohort and other demographic variables. This finding also has the hopeful clinical implication that MOUD appears to be just as effective for individuals using non-opioid substances prior to treatment initiation compared to those who do not have a history of other substance use.

Ongoing cocaine and amphetamine use early in treatment, however, was an independent prognostic factors for higher risk of relapse at week 12. This finding suggests that there might have been two groups of opioid users: one group used stimulants and opioids concurrently, described classically as a way to self-medicate opioid use side effects (Khantzian 1987)—this group of patients improved in their stimulant use during treatment. The other group had a stimulant use disorder not sufficiently treated by MOUD. Our results suggest that this second group had an overall worse outcome. Especially intriguing is the finding that cocaine had an interaction with extended-release naltrexone, suggesting a possible strategy to target this group specifically with naltrexone. This strategy is being actively tested in previous (Ling et al. 2016) and future treatment studies (CTN CURB-2 study).

In summary, our analysis did not detect an effect of non-opioid substance use at treatment entry on MOUD relapse outcome. It did show that the use of cocaine and amphetamine in the first 4 weeks predicted a higher relapse risk. The strength of our study was a design that leveraged a unique harmonized clinical trial dataset that allowed us to delineate treatment entry and early substance use patterns quantitatively during the treatment process.

The limitations of this study are related to the data used in harmonization. For example, the total number of days of use of a substance lacking further granular information such as amount of frequency of use within each day is not the most precise way to define pattern of use prior to randomization. The concurrent use of several substances of misuse is also a dynamic phenomenon that may have significant just-in-time predictive effects, which can be modelled with more sophisticated approaches such as Market Basket Analysis and Latent Factor Analysis. These are potential future studies. Additionally, this analysis did not evaluate to which extent patients achieved recovery prior to returning to use opioids, or whether socioeconomic status influenced study enrollment differentially. For example, recruitment may have differed by housing status, employment, and educational attainment across the three studies that was not captured by the present analysis. When compared against a “real-world” data set from publicly-funded substance treatment programs in the US, the present data set underrepresented pregnant patients (as expected), patients 65 and older, and those patients ages 50–64 who identified as other (non-White, non-Black, and non-Hispanic) race/ethnicity or multi-racial (Rudolph et al. 2022). Other “real-world” samples also highlight the role of social determinants of health in the study population (Lister, Greenwald, and Ledgerwood 2017; Ellis et al. 2023). We were only able to harmonize the first 12 weeks—a longer observation period would be more clinically meaningful. However, a follow-up study on one of the harmonized trials shows that treatments converge, and that at 36 weeks, the rate of relapse is not higher than at 24 weeks (Greiner et al. 2021). Harmonizing urine toxicology data from the three trials poses the problem that it could differ from subjective reports. Furthermore, unmeasured and unknown cofounders may be responsible for these results. Finally, future clinical studies ought to identify interventions for those who are using the non-opioid substances in early treatment associated with higher risk of relapse—one might consider testing strategies with extended-release naltrexone or early use of contingency management for those patients who specifically have ongoing cocaine use.

References

  1. Abrahamsson Tove, Berge Jonas, Öjehagen Agneta, and aakansson Anders. 2017. “Benzodiazepine, z-Drug and Pregabalin Prescriptions and Mortality among Patients in Opioid Maintenance Treatment—a Nation-Wide Register-Based Open Cohort Study.” Drug and Alcohol Dependence 174: 58–64. [DOI] [PubMed] [Google Scholar]
  2. Betts Kim S., Chan Gary, McIlwraith Fairlie, Dietze Paul, Whittaker Elizabeth, Burns Lucy, and Alati Rosa. 2016. “Differences in Polysubstance Use Patterns and Drug-Related Outcomes between People Who Inject Drugs Receiving and Not Receiving Opioid Substitution Therapies: Polysubstance Use, Injecting-Related Harms and Treatment.” Addiction 111 (7): 1214–23. 10.1111/add.13339. [DOI] [PubMed] [Google Scholar]
  3. Bovasso Gregory, and Cacciola John. 2003. “The Long-Term Outcomes of Drug Use by Methadone Maintenance Patients.” The Journal of Behavioral Health Services & Research 30 (3): 290–303. [DOI] [PubMed] [Google Scholar]
  4. Brands Bruna, Blake Joan, Marsh David C., Sproule Beth, Jeyapalan Renuka, and Li Selina. 2008. “The Impact of Benzodiazepine Use on Methadone Maintenance Treatment Outcomes.” Journal of Addictive Diseases 27 (3): 37–48. [DOI] [PubMed] [Google Scholar]
  5. Bunting Amanda M., Krawczyk Noa, Choo Tse-Hwei, Pavlicova Martina, McNeely Jennifer, Tofighi Babak, Rotrosen John, Nunes Edward, and Lee Joshua D. 2022. “Polysubstance Use before and during Treatment with Medication for Opioid Use Disorder: Prevalence and Association with Treatment Outcomes.” Journal of Substance Abuse Treatment, June, 108830. 10.1016/j.jsat.2022.108830. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Connor Jason P., Gullo Matthew J., White Angela, and Kelly Adrian B. 2014. “Polysubstance Use: Diagnostic Challenges, Patterns of Use and Health.” Current Opinion in Psychiatry 27 (4): 269–75. 10.1097/YCO.0000000000000069. [DOI] [PubMed] [Google Scholar]
  7. Cook RR, Foot C, Arah OA, Humphreys K, Rudolph KE, Luo SX, Tsui JI, Levander XA, and Korthuis PT 2023. “Estimating the Impact of Stimulant Use on Initiation of Buprenorphine and Extended-Release Naltrexone in Two Clinical Trials and Real-World Populations.” Addiction Science & Clinical Practice 18 (1): 11. 10.1186/s13722-023-00364-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Drake Stacy A., Yang Yijiong, Wolf Dwayne A., Reynolds Thomas, Harper Sherhonda, Hudson Antoinette, and Meininger Janet C. 2019. “Individual and Community Characteristics Associated with Premature Natural and Drug-Related Deaths in 25–59 Year Old Decedents.” PloS One 14 (2): e0212026. 10.1371/journal.pone.0212026. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Ellis Jennifer D., Rabinowitz Jill A., Ware Orrin D., Wells Jonathan, Dunn Kelly E., and Huhn Andrew S. 2023. “Patterns of Polysubstance Use and Clinical Comorbidity among Persons Seeking Substance Use Treatment: An Observational Study.” Journal of Substance Use and Addiction Treatment 146 (March): 208932. 10.1016/j.josat.2022.208932. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Ford Becky R., Bart Gavin, Grahan Brian, Shearer Riley D., and Winkelman Tyler N.A. 2021. “Associations Between Polysubstance Use Patterns and Receipt of Medications for Opioid Use Disorder Among Adults in Treatment for Opioid Use Disorder.” Journal of Addiction Medicine 15 (2): 159–62. 10.1097/ADM.0000000000000726. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Greiner Miranda G., Shulman Matisyahu, Choo Tse-Hwei, Scodes Jennifer, Pavlicova Martina, Campbell Aimee N.C., Novo Patricia, et al. 2021. “Naturalistic Follow-up after a Trial of Medications for Opioid Use Disorder: Medication Status, Opioid Use, and Relapse.” Journal of Substance Abuse Treatment 131 (December): 108447. 10.1016/j.jsat.2021.108447. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Hedegaard Holly, Miniño Arialdi M., and Warner Margaret. 2020. “Drug Overdose Deaths in the United States, 1999–2019.” NCHS Data Brief, no. 394 (December): 1–8. [PubMed] [Google Scholar]
  13. Jones Christopher M., and McCance-Katz Elinore F. 2019. “Co-Occurring Substance Use and Mental Disorders among Adults with Opioid Use Disorder.” Drug and Alcohol Dependence 197: 78–82. [DOI] [PubMed] [Google Scholar]
  14. Khantzian Edward J. 1987. “The Self-Medication Hypothesis of Addictive Disorders: Focus on Heroin and Cocaine Dependence.” In The Cocaine Crisis, 65–74. Springer. [DOI] [PubMed] [Google Scholar]
  15. Lee Joshua D., Nunes Edward V. Jr, Novo Patricia, Bachrach Ken, Bailey Genie L., Bhatt Snehal, Farkas Sarah, Fishman Marc, Gauthier Phoebe, and Hodgkins Candace C. 2018. “Comparative Effectiveness of Extended-Release Naltrexone versus Buprenorphine-Naloxone for Opioid Relapse Prevention (X: BOT): A Multicentre, Open-Label, Randomised Controlled Trial.” The Lancet 391 (10118): 309–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Lin Lewei, Bohnert Amy SB, Blow Frederic C., Gordon Adam J., Ignacio Rosalinda V., Kim H. Myra, and Ilgen Mark A. 2020. “Polysubstance Use and Association with Opioid Use Disorder Treatment in the US Veterans Health Administration.” Addiction. [DOI] [PubMed] [Google Scholar]
  17. Ling Walter, Hillhouse Maureen P., Saxon Andrew J., Mooney Larissa J., Thomas Christie M., Ang Alfonso, Matthews Abigail G., Hasson Albert, Annon Jeffrey, and Sparenborg Steve. 2016. “Buprenorphine+ Naloxone plus Naltrexone for the Treatment of Cocaine Dependence: The Cocaine Use Reduction with Buprenorphine (CURB) Study.” Addiction 111 (8): 1416–27. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lister Jamey J., Greenwald Mark K., and Ledgerwood David M. 2017. “Baseline Risk Factors for Drug Use among African-American Patients during First-Month Induction/Stabilization on Methadone.” Journal of Substance Abuse Treatment 78 (July): 15–21. 10.1016/j.jsat.2017.04.007. [DOI] [PubMed] [Google Scholar]
  19. Mahoney James J., Winstanley Erin L., Lander Laura R., Berry James H., Marshalek Patrick J., Haut Marc W., Marton Jennifer L., et al. 2021. “High Prevalence of Co-Occurring Substance Use in Individuals with Opioid Use Disorder.” Addictive Behaviors 114 (March): 106752. 10.1016/j.addbeh.2020.106752. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. McBrien Heather, Luo Candice, Sanger Nitika, Zielinski Laura, Bhatt Meha, Xi Ming Zhu David C. Marsh, Thabane Lehana, and Samaan Zainab. 2019. “Cannabis Use during Methadone Maintenance Treatment for Opioid Use Disorder: A Systematic Review and Meta-Analysis.” CMAJ Open 7 (4): E665. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. McDermott Kimberly W, and Sun Ruirui. 2017. “Trends in Hospital Inpatient Stays in the United States, 2005–2014,” June, 18. [Google Scholar]
  22. McLellan AT, Luborsky L, Woody GE, and O’Brien CP 1980. “An Improved Diagnostic Evaluation Instrument for Substance Abuse Patients. The Addiction Severity Index.” The Journal of Nervous and Mental Disease 168 (1): 26–33. 10.1097/00005053-198001000-00006. [DOI] [PubMed] [Google Scholar]
  23. Mintz Carrie M., Presnall Ned J., Xu Kevin Y., Hartz Sarah M., Sahrmann John M., Bierut Laura J., and Grucza Richard A. 2021. “An Examination between Treatment Type and Treatment Retention in Persons with Opioid and Co-Occurring Alcohol Use Disorders.” Drug and Alcohol Dependence 226 (September): 108886. 10.1016/j.drugalcdep.2021.108886. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Olfson Mark, Wall Melanie M., Liu Shang-Min, and Blanco Carlos. 2018. “Cannabis Use and Risk of Prescription Opioid Use Disorder in the United States.” American Journal of Psychiatry 175 (1): 47–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Peterson Cora, Li Mengyao, Xu Likang, Mikosz Christina A., and Luo Feijun. 2021. “Assessment of Annual Cost of Substance Use Disorder in US Hospitals.” JAMA Network Open 4 (3): e210242. 10.1001/jamanetworkopen.2021.0242. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Riley Elise D., Evans Jennifer L., Hahn Judith A., Briceno Alya, Davidson Peter J., Lum Paula J., and Page Kimberly. 2016. “A Longitudinal Study of Multiple Drug Use and Overdose Among Young People Who Inject Drugs.” American Journal of Public Health 106 (5): 915–17. 10.2105/AJPH.2016.303084. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Rosic Tea, Kapoor Raveena, Panesar Balpreet, Naji Leen, Chai Darren B., Sanger Nitika, Marsh David C., Worster Andrew, Thabane Lehana, and Samaan Zainab. 2021. “The Association between Cannabis Use and Outcome in Pharmacological Treatment for Opioid Use Disorder.” Harm Reduction Journal 18 (1): 24. 10.1186/s12954-021-00468-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Rosic Tea, Naji Leen, Bawor Monica, Dennis Brittany B., Plater Carolyn, Marsh David C., Thabane Lehana, and Samaan Zainab. 2017. “The Impact of Comorbid Psychiatric Disorders on Methadone Maintenance Treatment in Opioid Use Disorder: A Prospective Cohort Study.” Neuropsychiatric Disease and Treatment 13: 1399. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Rudolph Kara E., Russell Matthew, Luo Sean X., Rotrosen John, and Nunes Edward V. 2022. “Under-Representation of Key Demographic Groups in Opioid Use Disorder Trials.” Drug and Alcohol Dependence Reports 4 (September): 100084. 10.1016/j.dadr.2022.100084. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Saxon Andrew J., Ling Walter, Hillhouse Maureen, Thomas Christie, Hasson Albert, Ang Alfonso, Doraimani Geetha, et al. 2013. “Buprenorphine/Naloxone and Methadone Effects on Laboratory Indices of Liver Health: A Randomized Trial.” Drug and Alcohol Dependence 128 (1–2): 71–76. 10.1016/j.drugalcdep.2012.08.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Schottenfeld Richard S., Pakes Juliana R., Oliveto Alison, Ziedonis Douglas, and Kosten Thomas R. 1997. “Buprenorphine vs Methadone Maintenance Treatment for Concurrent Opioid Dependence and Cocaine Abuse.” Archives of General Psychiatry 54 (8): 713–20. [DOI] [PubMed] [Google Scholar]
  32. Sobell LC, and Sobell MB. 1995. Alcohol Timeline Followback Users’ Manual. Toronto, Canada: Addiction Research Foundation. [Google Scholar]
  33. Srivastava Anita, Kahan Meldon, and Ross Sue. 2008. “The Effect of Methadone Maintenance Treatment on Alcohol Consumption: A Systematic Review.” Journal of Substance Abuse Treatment 34 (2): 215–23. [DOI] [PubMed] [Google Scholar]
  34. Weiss Roger D., Potter Jennifer Sharpe, Provost Scott E., Huang Zhen, Jacobs Petra, Hasson Albert, Lindblad Robert, Connery Hilary Smith, Prather Kristi, and Ling Walter. 2010. “A Multi-Site, Two-Phase, Prescription Opioid Addiction Treatment Study (POATS): Rationale, Design, and Methodology.” Contemporary Clinical Trials 31 (2): 189–99. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES