Clinical and psychosocial outcomes by sex among individuals prescribed buprenorphine-naloxone (BUP-NX) or extended-release naltrexone (XR-NTX) for Opioid Use Disorder

Abstract

Background and Objectives:

Limited research has explored sex differences in opioid use disorder medication (MOUD) treatment outcomes. The purpose of this study was to examine MOUD initiation onto buprenorphine-naloxone (BUP-NX) vs. extended-release naltrexone (XR-NTX) by sex, and sex differences in clinical and psychosocial outcomes.

Methods:

Using data from a 24-week open-label comparative effectiveness trial of BUP-NX or XR-NTX, this study examined MOUD initiation (i.e., receiving minimum one XR-NTX injection or first BUP-NX dose) and 24-week self-report outcomes. We used regression models to estimate the probability of MOUD initiation failure among the intent-to-treat sample (N=570), and main and interaction effects of sex on outcomes of interest among the sub-sample of participants who successfully initiated MOUD (n=474).

Results:

In the intent-to-treat sample, odds of treatment initiation failure were not significantly different by sex. In the sub-sample of successful MOUD initiates, the effect of treatment on employment at week 24 was significantly moderated by sex (p=0.003); odds of employment were not significantly different among males by MOUD type; females randomized to XR-NTX vs. BUP-NX had 4.63 times greater odds of employment (p<0.001). Males had significantly lower odds of past 30-day exchanging sex for drugs vs. females (aOR=0.10, p=0.004), controlling for treatment and baseline outcomes.

Discussion and Conclusions:

Further research should explore how to integrate employment support into OUD treatment to improve patient outcomes, particularly among women.

Scientific Significance:

The current study addressed gaps in the literature by examining sex differences in MOUD initiation and diverse treatment outcomes in a large, national sample.

BACKGROUND

Over 1.5 million Americans over the age of 12 had an opioid use disorder (OUD) in 2019.¹ Historically, men have experienced higher rates of opioid misuse compared to women; yet, these disparities have significantly lessened in recent years.^2,3 Research suggests that men and women involved in substance use treatment differ on various clinical and psychosocial characteristics when entering and while in treatment,^4–9 including employment status,⁴ mental health problems,^8,9 risk for experiencing violence, and availability of social support.^4,5 For example, compared to men, women in OUD treatment report greater economic dependence and likelihood of having a partner with substance use disorder (SUD), and less support for engaging in treatment from a partner who is not using drugs.⁵ Men appear to be more likely than women to be involved in the criminal-legal system and to be employed while in OUD treatment.⁷

Less is known about sex differences in OUD medication (MOUD) treatment outcomes. Although most studies to date have not identified sex differences in opioid outcomes,¹⁰ research has been hampered by small sample sizes, poor representation of women, and inconsistencies in measurement of treatment outcomes.^7,11 Studies have focused mostly on opioid use outcomes, with limited consideration of other outcomes, including in domains where clear pre-treatment sex differences have been identified (e.g., psychiatric symptoms and psychosocial functioning).¹² Furthermore, little is known about whether sex interacts with treatment type, such that men and women respond differently to different medication types. Given that women face specific obstacles to treatment entry—such as stigma, caregiving responsibilities, and mental health problems— sex differences in OUD treatment outcomes may also occur.^3,13 Accordingly, research on treatment outcomes in women has been highlighted as an important research opportunity for addressing the opioid overdose crisis.^14,15

Purpose

Given these noted research gaps, the purpose of this study was to examine (1) MOUD initiation onto either buprenorphine-naloxone (BUP-NX) or extended-release naltrexone (XR-NTX) by sex¹, and (2) sex differences across 24-week clinical and psychosocial outcomes among the sub-sample of participants who successfully initiated MOUD. Understanding and addressing sex differences in MOUD initiation and treatment outcomes are important for improving treatment selection and retention and reducing opioid overdose risk among women.^4,11,16 The current study offers a unique opportunity to examine these differences in a large sample of both males and females.

METHOD

Participants

The current study used data from a randomized controlled 24-week open-label comparative effectiveness trial of BUP-NX vs. XR-NTX medication treatment for people with OUD conducted within the National Drug Abuse Treatment Clinical Trials Network (CTN).¹⁷ Recruitment occurred in eight inpatient detoxification units throughout the U.S. Following baseline assessment, participants (N=570) were randomized to receive either BUP-NX or XR-NTX. Detailed site selection and randomization procedures are described elsewhere.¹⁸ To enroll in the study, participants had to express willingness to accept either medication. All participants provided written informed consent, and all sites received local Institutional Review Board approval.¹⁷

Measures

Initiation onto MOUD was defined as receiving at least one XR-NTX injection or receiving the first dose of BUP-NX. Because the response options for sex in the current study were “male”, “female”, “don’t know”, and “refused” (0 participants selected “don’t know” or “refused”), we refer to “males” and “females” when describing participants in the current sample, and “men” and “women” when discussing the extant literature. Outcome data for the current analysis were collected at the week 24 visit via self-report. Measures included: past-month employment (yes/no), criminal activity (yes/no), opioid use relapse (defined as use of non-study opioids for four consecutive weeks or seven consecutive days [yes/no]),¹⁷ opioid, stimulant, cannabis, and heavy alcohol use in the past 7 days (Timeline Follow-Back [TLFB], collapsed to use=yes/no),¹⁹ weekly urine drug screen (abstinent yes/no), weekly opioid craving intensity (0–100 visual analogue scale),¹⁷ and any sex in exchange for drugs or other material goods (yes/no). Mental health status was measured using the Hamilton Rating Scale for Depression (HAM-D; none, mild to moderate, moderate to severe; dichotomized as “no/mild” and “moderate/severe”)²⁰ Physical health status was assessed using pain scores from the EuroQual 5D three-level version (EQ-5D-3L; none, moderate, extreme), as well as a 0–100 visual analogue scale to measure current health state.²¹

Analysis

Sample characteristics

Descriptive statistics (i.e., frequencies and percentages) were used to describe the sample in terms of sex, age, race and ethnicity, education, employment, marital status, mental and physical health status, current health state, opioid craving, sex in exchange for drugs, criminal activity in the past month, and substance use in the past week/month.

MOUD initiation

Logistic regression models were used to estimate the probability of failure to initiate onto treatment medication among the intent-to-treat (N=570) sample. The model included the effects of treatment (XR-NTX vs BUP-NX), sex (males vs. females), and their interaction (site was treated as a random effect).

Sex by treatment interaction

Regression models were run to estimate the effect of sex by MOUD treatment interaction on outcomes of interest among the sub-sample who successfully initiated onto either medication (n=474). Models were run separately for each clinical and psychosocial outcome and included the effect of sex (males vs. females), treatment (XR-NTX vs. BUP-NX), and their interaction, and were adjusted by their respective baseline indicator of the outcome measure. Logistic regression was utilized for binary categorical outcomes, multinomial logistic regression for multilevel categorical outcomes, and linear regression for continuous outcomes using the appropriate link function to match distribution of the outcome/residuals. Categorical outcomes were estimated using a logit link function; model estimates were exponentiated to compute adjusted odds ratios (aOR). Normally distributed outcomes were estimated using an identity link function and model estimates were interpreted in unit change. Models were examined for fit and validity of assumptions, and any violations noted. Corresponding 95% confidence intervals are presented with model estimates.

Main effect of sex

If the sex by treatment interaction was not significant at 5% level of significance, new regression models were run, removing the sex by treatment interaction term from the model to leave only the main effects terms. This model was then used to estimate the main effect of sex (males vs. females) on outcomes of interest among the sub-sample who successfully initiated onto either medication, adjusting for MOUD type and respective baseline indicator of the outcome measure.

RESULTS

Sample characteristics

The intent-to-treat sample was comprised of N=570 participants, n=169 (29.6%) females and n=401 (70.4%) males. The mean age was 33.9 (SD 9.6) years. The race/ethnicity of the participants were as follows: white: n=421, 73.9%; Black: n=57, 10.0%; multiracial/other (see Table 1 notes): n=86, 15.1%; not reported: n=6, 1.1%; and non-Hispanic/Latinx: n=471; 82.6%. Less than half (n=248; 43.5%) had more than a high school diploma and about one-third had completed high school or obtained a GED (n=190; 33.3%). Over 60% had never been married (n=376; 66.0%) and did not have a job (n=360; 63.2%). About a quarter (n=144, 25.3%) met the HAM-D criteria for moderate or severe depression. Nearly 70% of the sample (n=391; 68.6%) responded that their anxiety or depression was moderate or extreme and almost 60% (n=335, 58.8%) responded that their pain was moderate or extreme. The median health state rating was 70.0 (interquartile range [IQR]: 59.0–80.0; 0–100 visual analogue scale [VAS]); the median opioid craving rating was 78.0 (IQR: 47.0–96.0; 0–100 VAS). Most participants (n=301, 90.4% of those who responded) had not engaged in criminal activity or sex exchange (n=319, 97.0% of those who responded to the question) within the past month. In the past month, about 40% of the sample had used alcohol (n=227, 39.8%) or marijuana (n=231, 40.5%), 16.5% (n=95) had used stimulants, and nearly all (n=556, 97.5%) had used opioids. See Table 1 for additional details and for male and female subgroups.

Table 1:

Descriptive demographic and clinical characteristics at baseline for the total sample and by sex (N=570)

	Total (N=570)		Males (N=401)		Females (N=169)
	N	% or M (SD)	N	% or M (SD)	N	% or M (SD)
Age	570	33.9 (9.6)	401	34.5 (9.9)	169	32.4 (8.8)
Race
White	421	73.9%	287	71.6%	134	79.3%
Black	57	10.0%	49	12.2%	8	4.7%
Multiracial/Other*	86	15.1%	59	14.7%	27	16.0%
Not reported	6	1.1%	6	1.5%	0	0.0%
Hispanic/Latinx Ethnicity
Not Hispanic/Latinx	471	82.6%	326	81.3%	145	85.8%
Hispanic/Latinx	99	17.4%	75	18.7%	24	14.2%
Education
<High School	132	23.2%	101	25.2%	31	18.3%
High School Diploma/GED	190	33.3%	136	33.9%	54	32.0%
>High School	248	43.5%	164	40.9%	84	49.7%
Marital Status
Ever married	191	33.5%	129	32.2%	62	36.7%
Never Married	376	66.0%	271	67.6%	105	62.1%
Unknown	3	0.5%	1	0.2%	2	1.2%
Employment in last 30 days
Did not have a job	360	63.2%	248	61.8%	112	66.3%
Had a job	210	36.8%	153	38.2%	57	33.7%
Depression HAM-D
No/mild depression (0–13)	425	74.7%	297	74.2%	128	75.7%
Moderate/severe depression (14+)	144	25.3%	103	25.8%	41	24.3%
EQ-5D-3L: Pain
None	235	41.2%	175	43.6%	60	35.5%
Moderate	309	54.2%	206	51.4%	103	60.9%
Extreme	26	4.6%	20	5.0%	6	3.6%
Health state today - Median (IQR)	570	70.0 (59.0–80.0)	401	70.0 (59.0–80.0)	169	73.0 (60.0–80.0)
VAS opioid craving - Median (IQR)	570	78.0 (47.0–96.0)	401	77.0 (47.0–95.0)	169	80.0 (54.0–97.0)
Sex for drugs in past month (yes)	54	9.7%	11	2.8%	43	26.2%
Any criminal activity in past month (yes)	213	37.4%	153	38.2%	60	35.5%
Past Week Any Alcohol Use
No	409	87.4%	286	84.9%	123	93.9%
Yes	59	12.6%	51	15.1%	8	6.1%
Past Week Any Marijuana Use
No	399	85.3%	279	82.8%	120	91.6%
Yes	69	14.7%	58	17.2%	11	8.4%
Past Week Any Stimulant Use
No	448	95.7%	322	95.5%	126	96.2%
Yes	20	4.3%	15	4.5%	5	3.8%
Past Week Any Opioid Use
No	258	55.1%	173	51.3%	85	64.9%
Yes	210	44.9%	164	48.7%	46	35.1%
Past Month Any Alcohol Use
No	343	60.2%	229	57.1%	114	67.5%
Yes	227	39.8%	172	42.9%	55	32.5%
Past Month Any Marijuana Use
No	339	59.5%	228	56.9%	111	65.7%
Yes	231	40.5%	173	43.1%	58	34.3%
Past Month Any Stimulant Use
No	475	83.3%	339	84.5%	136	80.5%
Yes	95	16.7%	62	15.5%	33	19.5%
Past Month Any Opioid Use
No	14	2.5%	8	2.0%	6	3.6%
Yes	556	97.5%	393	98.0%	163	96.4%

NOTES: EQ-5D-3L= EuroQual 5D three-level version; HAM-D=Hamilton Rating Scale for Depression; IQR=interquartile range; M=mean; SD=standard deviation; VAS=Visual Analogue Scale

^* “Other” racial group includes self-descriptions of Asian, Hawaiian, American Indian, Arabic, Armenian, Brown, Czech, Egyptian, Hispanic, Human, Latinx, Mexican, Salvadoran, Persian, Peruvian, Puerto Rican, Spanish

MOUD initiation

There were no sex differences in rates of successful initiation (intent-to-treat sample, N=570): 93.7% of males and 95.1% of females successfully initiated onto BUP-NX; 70.8% of males and 75.0% of females successfully initiated onto XR-NTX. Among females, those randomized to XR-NTX had 6.98 (95% CI: 2.23, 21.80) times the odds of initiation failure compared to those randomized to BUP-NX (p<0.001). Similarly, among males, those randomized to XR-NTX had 6.55 (95% CI: 3.41, 12.59) times the odds of initiation failure compared to those randomized to BUP-NX (p<0.001; results not reported in tables).

Sex by MOUD treatment interaction

Table 2 shows the descriptive data for the week 24 outcomes. Among the sub-sample of participants who were initiated onto one of the medications (n=474), only one significant sex by MOUD treatment interaction effect emerged in the regression models: the effect of treatment on employment at week 24 was significantly moderated by sex (p=0.003). Among males, the odds of employment were not significantly different by medication type (aOR=0.99, p=0.975); however, among females, those randomized to and successfully initiated onto XR-NTX had 4.63 times greater odds of having a job (69.4%) compared to those randomized and successfully initiated onto BUP-NX (32.7%; aOR=4.63, p<0.001).

Table 2:

Descriptive data for clinical and psychosocial outcomes at week 24 by sex and medication treatment arm (Buprenorphine-Naloxone [BUP-NX] or Extended-Release Naltrexone [XR-NTX]) for the subsample of participants successfully initiated onto either medication (n=474)

			Males				Female
	Total n=474		BUP-NX n=193		XR-NTX n=138		BUP-NX n=77		XR-NTX n=66
	N	% or M (SD)	N	% or M (SD)	N	% or M (SD)	N	% or M (SD)	N	% or M (SD)
Relapse (by 24 weeks) (yes)	256	54.0%	108	56.0%	72	52.2%	42	54.5%	34	51.5%
Employed, past month (Yes)	200	60.1%	88	64.7%	60	64.5%	18	32.7%	34	69.4%
Opioid use, past week (yes)	90	28.6%	39	30.2%	28	31.5%	14	25.9%	9	20.9%
Stimulant use, past week (yes)	73	23.2%	34	26.4%	18	20.2%	11	20.4%	10	23.3%
Cannabis use, past week (yes)	101	32.1%	44	34.1%	27	30.3%	11	20.4%	19	44.2%
Heavy drinking, past week (yes)	32	10.2%	13	10.1%	9	10.1%	5	9.3%	5	11.6%
Depression HAM-D
None/mild	315	96.6%	132	97.1%	87	96.7%	50	96.2%	46	95.8%
Moderate/severe	11	3.4%	4	2.9%	3	3.3%	2	3.8%	2	4.2%
EQ-5D-3L: Pain
None	255	76.8%	100	74.1%	79	84.9%	35	63.6%	41	83.7%
Moderate	64	19.3%	30	22.2%	11	11.8%	16	29.1%	7	14.3%
Extreme	13	3.9%	5	3.7%	3	3.2%	4	7.3%	1	2.0%
Health state today Median (IQR)	323	82.0 (70.0–93.0)	133	81.0 (70.0–92.0)	92	87.5 (75.0–95.5)	53	77.0 (67.0–88.0)	45	85.0 (70.0–95.0)
VAS opioid craving Median (IQR)	325	4.0 (0.0–38.0)	134	5.0 (0.0–38.0)	92	2.0 (0.0–48.0)	54	4.5 (0.0–58.0)	45	3.0 (0.0–27.0)
Sex for drugs, past month (yes)	10	3.0%	2	1.4%	0	0.0%	5	8.9%	3	6.5%
Any criminal activity, past month (yes)	32	9.6%	13	9.6%	10	10.8%	4	7.3%	5	10.2%

NOTES: BUP-NX= Buprenorphine-Naloxone; EQ-5D-3L= EuroQual 5D three-level version; HAM-D=Hamilton Rating Scale for Depression; IQR=interquartile range; M=mean; SD=standard deviation; VAS=Visual Analogue Scale; XR-NTX =extended-release Naltrexone

Main effect of sex

The main effect of sex was explored for all outcome variables for which the sex by MOUD treatment interaction was not significant at the 5% level of significance, among the sample who successfully initiated onto one of the medications (n=474). The main effect of sex was significant on one outcome: males had significantly lower odds of exchanging sex for drugs in the past month in comparison to females (aOR=0.10, p=0.004), when controlling for MOUD treatment and baseline measures corresponding to specific of outcomes.

DISCUSSION

The current study examined a range of clinical and psychosocial outcomes by sex among individuals receiving either BUP-NX or XR-NTX treatment for OUD, as well as sex differences in medication initiation success. This relatively large sample offered a unique opportunity to examine sex differences in opioid treatment and to compare differences across two types of MOUD. Among the intent-to-treat sample, there were no significant sex differences in successful initiation onto either medication. Common differences in initiation success were observed between NR-NTX and BUP-NX; there were significantly greater odds of initiation failure among both males and females in the XR-NTX group, and 25–30% of individuals did not receive the first XR-NTX injection. Initiation onto XR-NTX is challenging due to the requirement that people achieve abstinence from opioids prior to first XR-NTX injection. Strategies such as rapid initiation procedures or greater access to inpatient detoxification with medication initiation services may help to reduce barriers to XR-NTX use.^22,23

Previous studies have reported sex differences in clinical and psychosocial characteristics at treatment entry. A prior analysis of this same dataset found various baseline differences by sex;⁴ however, in the present study most of these sex differences did not persist at 24 weeks following successful initiation onto MOUD treatment. The lack of significant sex differences in MOUD treatment outcomes may suggest that, despite variations in psychiatric and psychosocial severity that are often observed at treatment entry,⁴ differences between males and females in the OUD treatment outcomes explored in this study are likely of limited magnitude, if they exist at all.

Among those who initiated one of the study medications, only one measure significantly differed among males and females: females who successfully initiated onto XR-NTX had significantly greater odds of employment at follow-up compared to females randomized and successfully initiated onto BUP-NX, while there were no significant employment differences across the two medication treatments among males. These findings are consistent with and extend previous studies showing significant sex differences in employment outcomes by type of OUD treatment medication.^4,7

Although findings should be interpreted with caution, it is worth noting several factors that may contribute to differences in employment among females in the current sample. First, females in the XR-NTX group may have experienced less treatment burden than those in the BUP-NX group given that XR-NTX was dispensed by injection approximately every 28 days vs. weekly to biweekly,¹⁷ enhancing employment opportunity. Additionally, given that women are more likely than men to have other responsibilities (e.g., childcare and eldercare),^3,13 MOUD that requires less of a time commitment might be particularly beneficial for women seeking employment. Further, compared to men with OUD, women with OUD may be less economically independent;⁵ thus, treatment that requires a lower time commitment may have greater benefit for women than for men. These findings suggest a need to reduce barriers to employment among women, especially those receiving BUP-NX.

Among participants who successfully initiated onto one of the study medications, the main effect of sex was only significant on one outcome: males had significantly lower odds of exchanging sex for drugs compared to females. Men’s lower likelihood of having caregiving responsibilities^3,13 and economic dependence on someone else compared to women,⁴ as well as less stigma than women who use drugs^3,24,25 may result in fewer barriers to obtaining mainstream employment. These factors may increase women’s likelihood to have sex in exchange for drugs given more employment obstacles. Further research is needed to understand how to integrate employment and OUD treatment services to improve both economic and treatment outcomes for people with OUD,²⁶ with particular emphasis on supporting women seeking and securing employment.

Limitations

The current study is one of the largest, national trials examining effectiveness of BUP-NX and XR-NTX. Despite these strengths, this study has limitations. The findings from this study are specific to males and females who sought OUD treatment and had access to inpatient care, thus limiting generalizability. Additionally, very few studies that focus on sex, including the current study, specifically assess gender identity and examine differences among cisgender and transgender populations; studies also often conflate gender identity with sex assigned at birth.²⁷ Transgender populations unfortunately face higher rates of substance use and SUD,²⁸ as well as barriers to substance use treatment engagement, due to discrimination and stigma.²⁹ Further, the response options in the current study of “male”, “female”, “don’t know”, and “refused” do not reflect emerging best practice recommendations of asking both sex assigned at birth and gender identity,²⁷ and thus may not have accurately captured study participants’ identities. Although all respondents selected “male” or “female” responses, the limited response options may have dissuaded some participants from selecting “don’t know” or “refused”. Future studies should explore OUD treatment outcomes among cisgender and transgender samples. Moreover, it was beyond the scope of this study to examine racial or ethnic differences; future research may investigate whether the patterns that emerged in this study are generalizable to specific racial/ethnic groups. Finally, outcomes were explored among the subsample of participants who successfully initiated medication treatment to understand the impact of medication delivery and there were differences by medication on initiation success; thus, this was not an intent-to-treat analysis.

CONCLUSION

This study explored differences in clinical and psychosocial outcomes between males and females receiving XR-NTX or BUP-NX for OUD. Sex differences were evident in the relationship between treatment type and employment. These findings offer further evidence that treatment options with greater flexibility (e.g., once per month injection of XR-NTX or longer BUP-NX prescriptions) could be associated with improved employment opportunity, especially among females, perhaps related to less burden or the stigma associated with attending treatment. Findings also suggest that females face ongoing financial need and potentially mainstream employment barriers at the end of treatment, further demonstrating a need to support women in obtaining and maintaining employment.