"Patient and Provider Medication Preferences Affect Treatment Outcomes among Adolescents and Young Adults with Opioid Use Disorder"

Abstract

Background:

The opioid epidemic in the United States has not spared youth or young adults, as evidenced by a six-fold increase in opioid use disorder (OUD) diagnoses in the last two decades. Given this dramatic rise, a call for greater uptake and accessibility of medications for opioid use disorder (MOUDs) among youth and young adults has ensued, resulting in an increasing number of MOUD treatment pathways for this vulnerable population.

Methods:

This secondary data analysis seeks to characterize patient and provider preferences for MOUD treatment pathways, and test for associations between baseline MOUD treatment preferences and opioid use and treatment adherence outcomes. Participants included 288 youth and young adults (age 15-21 years), recruited from a residential treatment program in Maryland. The study assessed patient preferences at baseline (n = 253) and provider preferences at patient treatment discharge (n = 224). Mixed-effects negative binomial regression models were conducted for opioid use outcomes, and logistic regressions were conducted for treatment adherence outcomes.

Results:

Results indicate that congruence of treatment with patients’ (Incidence Rate Ratio [IRR] = 0.65) and providers’ (IRR = 0.66) preferences was significantly associated with reduced self-reported days of opioid use in the past 90 days, but only for patients receiving extended-release naltrexone (XR-NTX). Results also indicated that patients were less likely to switch medication treatment pathways (e.g., from XR-NTX to buprenorphine, or vice versa) during follow-up if they received their preferred treatment at baseline, a finding which held true for both XR-NTX (Odds Ratio [OR] = 0.32) and buprenorphine (OR = 0.22).

Conclusions:

Receipt of MOUD congruent with patient and provider preferences was associated with reduced opioid use and greater treatment adherence in this sample of youth and young adults with OUD.

1.0. Introduction

The US opioid crisis has substantial impact on adolescents and young adults. The turn into the 21^st century has been accompanied by a dramatic rise in the rates of opioid misuse and accidental overdose in the US (National Center for Injury Prevention and Control, 2021); in 1999 there were fewer than 10,000 deaths due to opioids each year in the US, in 2021 there were more than 75,000 (National Center for Health Statistics, 2021; National Institute on Drug Abuse, 2021). Over ten million people a year in the US now misuse opioids, and two out of every three overdose deaths in the US are due to opioid misuse (Centers for Disease Control and Prevention, 2022).

A dramatically increased prevalence of both heroin and synthetic opioid misuse are now being observed among adolescents and young adults (hereafter “youth”) in the US (Banta-Green & Cooley, 2018; Carney et al., 2018; Chang et al., 2018). In less than twenty-five years, opioid use disorder (OUD) diagnosis rates among US youth have increased nearly six-fold (Hadland et al., 2017) and unintentional opioid poisonings have increased three-fold (Carney et al., 2018), such that over 3,300 US youth now die each year from opioid overdose (CDC Wonder, 2020), making it a leading cause of death. A recent study found that these impacts were even greater during the COVID-19 pandemic, when the number of adolescent years of life lost to unintentional drug overdose more than doubled from 2019 (39,579 years) to 2020 (84,179 years) – a 113% increase. Synthetic opioids, including fentanyl and its derivatives, accounted for 81% of overdose deaths in 2020, compared to 67% in 2019 (Hermans et al., 2022). This dramatic rise in opioid misuse among US youth has prompted growing interest in the provision of developmentally appropriate treatment for OUD, including through use of medications for opioid use disorder (MOUDs) (Hadland et al., 2017; Mitchell et al., 2021; Pecoraro et al., 2013; Robinson & Wilson, 2020; Wenzel & Fishman, 2021a; Yule et al., 2018).

There are over 1.25 million people currently receiving MOUDs in the US, a population which increased in size 142% from 2016-2018 alone (Centers for Disease Control and Prevention, 2022). Two of the most commonly prescribed MOUDs among US youth specifically are buprenorphine, a partial opioid agonist, and extended-release naltrexone (XR-NTX), an opioid antagonist. Unfortunately, MOUD for youth in any form lags behind rates observed among adult populations (Alinsky et al., 2020a; CommitteeonSubstanceUseandPrevention, 2016). Only about one-quarter of all US youth misusing opioids receive medication for their opioid use disorder (Banta-Green & Cooley, 2018; Hadland et al., 2018; Hadland et al., 2017), with particularly low rates (e.g., less than 10%) observed among adolescents (Carney et al., 2018; Feder et al., 2017). Only about one-quarter of the addiction treatment facilities in the US even offer treatment for persons under the age of 18, and only 23% of those offer MOUDs (Alinsky et al., 2020a).

Perhaps relatedly, most US youth who experience an opioid-related overdose receive no addiction treatment services whatsoever within the first 30 days after the incident, and evidence suggests fewer than 2% are offered pharmacotherapeutic support (Alinsky et al., 2020b). As such, there is relatively little research on the efficacy of MOUDs among US youth compared to adults (Camenga et al., 2019; Carney et al., 2018; Vo et al., 2018). What evidence does exist suggests that MOUDs are well-tolerated and efficacious in promoting treatment retention and abstinence from opioid use or relapse (Borodovsky et al., 2018; Hadland et al., 2018; Mitchell et al., 2021; Wenzel et al., 2021b; Woody et al., 2008). It may also be the case, however, that retention and abstinence outcomes of MOUDs among US youth are related not only to the pharmaceutical efficacy of the treatments themselves, but also to the preconceptions, prejudices, and preferences held by both the patient, the parent/caregiver, and their provider prior to treatment initiation.

Though patient preferences for treatment have, in general, been shown to affect retention and efficacy in the treatment of certain other diseases and disorders (e.g., Azhar et al., 2021; Chung et al., 2021; Losi et al., 2021; Wade et al., 2019), little work has been done to clarify how patient preferences for different OUD treatment influence outcomes. A recent open-label clinical trial in Norway compared buprenorphine and XR-NTX patients and found that patient preferences had no effect on treatment outcomes in the short term, but that over the longer term (i.e., through 104 weeks) increased treatment adherence and avoidance of relapse were both associated with pre-treatment preference, but only for XR-NTX (Gaulen et al., 2022). Very little is known about how providers’ preferences for certain OUD treatments might affect that treatment’s efficacy, or how providers’ and patients’ initial preferences may influence patients’ decisions to discontinue one treatment for another over time.

What little evidence does exist suggests some providers have an overall negative, or stigmatizing view of patients with OUD (Bapat et al., 2020), and perhaps in particular the selection of agonist treatments (Molfenter et al., 2015). Though some evidence suggests that patients will work to form and maintain beliefs which can steer them toward their preferred MOUD preference (Uebelacker et al., 2016), this may be more difficult to achieve for youth seeking treatment for OUD, as their stated preferences may carry less weight to providers or families or hold less sway than preferences expressed by their adult counterparts, especially if they are under the age of 18. Opinions and perspectives provided by parents, providers, legal guardians, family physicians, housing units, school policies, and more may all interfere with the direct association between a young patient’s preferences and the treatment they receive. As such, what evidence does exist linking adult patient preferences to treatment outcomes may differ from what is observed among US youth.

This secondary analysis of data from a recent randomized clinical trial (Mitchell et al., 2021) seeks to characterize both patient and provider preferences for certain MOUD treatments over others, to test for associations between these preferences and OUD treatment outcomes, and to determine whether initial preferences influence patients’ decisions to “switch” treatment options over time (e.g., changing from a buprenorphine regimen to an XR-NTX regimen, or vice versa).

2. Methods

For a full discussion of parent study methods and outcomes, please see Mitchell et al., 2021. The WIRB-Copernicus Group (WCG IRB) approved this study.. A federal Certificate of Confidentiality was obtained and the study was monitored by a Data Safety and Monitoring Board. Since XR-NTX is not approved for pediatric use, the research team obtained an Investigational New Drug (IND) permit from the US FDA to study this medication in patients who were 15 to 17 years of age. The study was registered at ClinicalTrials.gov (NCT 02599818).

2.1. Participants

In the larger trial, 288 youth were enrolled from a residential treatment program in Maryland between October 2013 and January 2018. Inclusion criteria were: (1) between 15 and 21 years of age, inclusive; (2) current moderate or severe OUD (diagnosed at admission); (3) receiving treatment for opioid withdrawal; (4) able and willing to provide informed consent; and, (5) for participants under 18 years of age, parental or guardian consent and participant assent. Exclusion criteria included: (1) liver function test levels 4x greater than normal; (2) unstable medical or psychiatric illness which could impede participation; (3) suicide attempt in the past 6 months; (4) allergic to naloxone and/or naltrexone; (5) current chronic pain condition for which opioids were deemed necessary for ongoing care; (6) blood coagulation disorder; (7) Body Mass Index > 40; (8) if female: pregnant, lactating, unwilling or unable (due to parental objection) to use FDA-approved contraceptive methods; and (9) needing detoxification from benzodiazepines during inpatient treatment.

2.2. Treatment Conditions

The parent study design was a randomized trial in which participants were assigned on a 1:1 basis to XR-NTX or not-XR-NTX. The not-XR-NTX group, otherwise known as TAU, was a heterogeneous category of individuals receiving buprenorphine or counseling only. However, as described in Mitchell et al. (2021) there was considerable non-adherence to the random assignment due to a number of factors spanning patient preference/choice and external constraints. Thus, the current analysis focuses systematic attention to preferences from both the youth and the youth’s provider during residential care, and the subsequent impact on the treatment received, rather than the allocated treatment.

2.2.1. XR-NTX (n = 82 youth).

Participants receiving XR-NTX were required to have undergone a medically assisted withdrawal (with buprenorphine over approximately five-seven days) prior to initiation of XR-NTX. Following approximately seven days without buprenorphine or other opioids, participants received an oral naltrexone lead-in over two-to-three days, followed by an intramuscular injection prior to discharge. Individuals receiving XR-NTX were able to receive another XR-NTX injection once every four weeks for up to an additional six months.

2.2.2. Not-XR-NTX (n = 206).

Participants who did not receive XR-NTX, generally received: 1) treatment without buprenorphine (n = 112), following a buprenorphine dose taper similar to those in the XR-NTX condition, or 2) buprenorphine maintenance treatment during their residential treatment episode (n = 94). Decisions regarding which treatment was most appropriate were collaborative with input from providers, participants, and (especially in the case of adolescents under the age of 18) family members. Participants receiving buprenorphine were provided (on average) 12–20 mg e per day, with treatment plan adjustments during the follow-up period as clinically appropriate.

2.3. Assessments

2.3.1. Baseline Demographics.

The study collected participant demographic information including gender, age, and both racial and ethnic identity during the baseline interview.

2.3.2. Baseline Patient Preference.

During the baseline interview and before participants were randomized to a study condition, the study asked patients (n = 253 responses) to report if they had a treatment preference and could indicate buprenorphine, XR-NTX, oral naltrexone, counseling only, other, or no preference. Patients indicated why this was their treatment preference including, “had good experiences,” “heard good things,” “heard bad things,” “had bad experiences,” “avoids medications I’ve diverted in the past,” and “helps me truly ‘be clean’/don’t want dependence on a medication.” Participants graded each of these explanations on a Likert scale (ranging 1-5) from “not at all” to “completely.”

2.3.2. Discharge Provider Preference.

At discharge, the provider completed an assessment indicating their preference for each of their participant’s treatment following the residential treatment episode (n = 224 responses) using the same options presented to participants.

Providers were also asked which treatment the participant received at discharge, as well as what factors influenced the participant in receiving that treatment. Factors included: random allocation, provider recommendation, patient preference, patient family preference, transitional housing requirements, availability from a local provider, covered by insurance, and other.

2.3.3. Qualitative Data Source.

As mentioned above, part of the discharge provider preference assessment included asking providers which treatment they preferred for a particular patient. Subsequently, providers were asked “the patient seemed to be a good candidate for this treatment because…” Only a subsample of providers elected to elaborate on their treatment preferences (n=115), and these responses provided the basis of the qualitative analyses.

2.4. Statistical Analysis

The study described each variable using methods best suited for its distributional properties, including counts and percentages for nominal variables, or means and standard deviations for counted or continuous variables. Missing data was not imputed. Inferential analyses took the form of either mixed effects negative binomial regressions (outcome variable: days of opioid use in the past 90 days at 3- and 6-month follow-ups; control variable: opioid use in the past 90 days at baseline) or logistic regressions (outcome variable: 1 = patient “switched” from one treatment to another over time [e.g., a patient began treatment using XR-NTX and crossed over to buprenorphine during the study observation period, or vice versa). The study constructed variables indicating congruence between (1) the treatment received and (2) patient or provider preferences by comparing participants’/providers’ stated preferences against what treatment was actually provided; incongruence was coded as “0,” while congruence was coded as “1.” If a participant indicated no treatment preference at baseline, they were coded as “0.”

2.5. Qualitative Analysis

The study collected and aggregated open-ended provider responses from the discharge provider preference assessment for analysis. Two experienced qualitative researchers on the research team analyzed this data using a content analysis to group responses thematically. Provider responses related to each OUD treatment medication pathway were analyzed within each pathway category, and then examined across each pathway category for similarities and differences. Each of the qualitative researchers reviewed the content of the open-ended questions separately, then generated themes independently. The two researchers compared thematic areas and only utilized themes that generated consensus. Themes related to provider preferences for patients in each treatment medication pathway are presented as aggregate themes, and not related to any individual treatment consideration.

3.0. RESULTS

Youth participant demographics are provided in Table 1; most participants identified as male (54%), median age was 20 years (range: 15 thru 21), and the vast majority identified as non-Hispanic white (81%).

Table 1:

Patient Sociodemographics (N = 288)

	Mean or N	[SD] or %
Gender
Female	133	46%
Age
Years	19.4	[1.25]
Racial/Ethnic Identity
White	233	81%
Black	27	9%
Latinx	13	5%
Multiple/Other	15	5%

3.1. Quantitative Results

As shown in Table 2, patients and providers reported strongly divergent preferences for treating opioid use. Whereas 42% of participants reported a preference for buprenorphine, only 13% of providers shared this preference, predominantly preferring XR-NTX instead (56% of providers vs. 38% of patients). Providers were also significantly more likely to express no preference relative to patients (25% vs. 5%), while patients were more likely to prefer a non-MOUD “counseling only” approach (13% or patients vs. 1% of providers). As Table 3 shows, when choosing from a list of reasons why they might prefer one treatment option over another participants who reported a preference for buprenorphine were observed to be more likely to endorse having had a good experience with buprenorphine in the past, extended release naltrexone was disproportionately preferred because of a participant having heard bad things about other medications, and participants who preferred counseling over medication-based treatments disproportionately did so because they were avoiding medications they had already tried and did not like. In contrast, providers could only endorse three answers regarding their preferences: (1) it worked well in the past; (2) other treatments had not worked in the past; or, (3) the patient was a “good candidate” for the medication. Providers who preferred XR-NTX said it worked well in the past (5 [3%]), that other treatments had not worked well in the past (59 [40%]), or that they simply felt the patient was a good candidate for XR-NTX (82 [56%]). Providers who preferred buprenorphine said it worked well in the past (9 [30%]), that other treatments had not worked well in the past (2 [7%]), or that they simply felt the patient was a good candidate for buprenorphine (19 [63%]; results not tabulated).

Table 2:

Patient and Provider Treatment Preferences

	Patients (N = 253)		Providers (N = 224)
	n	%	n	%
XR-NTX	95	38%	125	56%
Buprenorphine	105	42%	28	13%
Counseling Only	33	13%	2	1%
No Preference	12	5%	56	25%
Other Preference	0	0%	12	5%
Oral Naltrexone	8	3%	1	1%
	n = 35 missing		n = 64 missing

Table 3.

Participant-Endorsed Reasons for Treatment Preference at Baseline

	Had Good Experiences		Heard Good Things		Heard Bad Things About Other Meds		Had Bad Experience with Other Meds		To Avoid Meds I Don't Like		Helps Me Stay "Clean"
	N	%	N	%	N	%	N	%	N	%	N	%
Not at all	150	62%	44	18%	109	45%	157	65%	172	71%	64	27%
A little	12	5%	22	9%	35	15%	13	5%	7	3%	8	3%
Moderately	14	6%	52	22%	42	17%	12	5%	16	7%	29	12%
Mostly	35	15%	76	32%	33	14%	27	11%	11	5%	56	23%
Completely	30	12%	47	20%	22	9%	32	13%	35	15%	84	35%
Overall Median:	"Not At All"		"Mostly"		"A little"		"Not At All"		"Not At All"		"Mostly"
Buprenorphine Median:	"Moderately"		"Mostly"		"Not at All"		"Not at All"		"Not at All"		"A little"
XR-NTX Median:	"Not at All"		"Mostly"		"Moderately"		"Not at All"		"Not at All"		"Completely"
Oral NTX Median:	"Not at All"		"Not at All"		"A little"		"Not at All"		"Not at All"		"Moderately"
Counseling Only Median:	"A little"		"A little"		"A little"		"Not at All"		"A little"		"Completely"

Table 4 provides the results of four mixed effects negative binomial regressions, each including 3- and 6-month follow-up data and estimating the association between having either received (patients’ perspective; n = 217) or prescribed (providers’ perspective; n = 189) one’s preferred medication (either XR-NTX or buprenorphine) and self-reported opioid use. Results indicate that congruence between either patients’ (Incidence Rate Ratio [IRR] = 0.39; 400 person-time observations included in analysis) or providers’ (IRR = 0.39; 347 observations) preferences, compared to non-congruence, significantly reduced self-reported days of opioid use in the past 90 days, but only for patients receiving XR-NTX. For patients/providers who received/prescribed buprenorphine, there were no associations between having had one’s preferences met and rates of self-reported opioid use.

Table 4:

Negative Binomial Regressions of Having Received/Prescribed One's Preferred Treatment and Patient's Opioid Use

	Patients' Perspective (N = 217; 400 follow-up observations)				Providers' Perspective (N = 189; 347 follow-up observations)
	Received XR-NTX		Received Buprenorphine		Prescribed XR-NTX		Prescribed Buprenorphine
	IRR a	95% CI	IR R a	95% CI	IRR a	95% CI	IR R a	95% CI
Received/Prescribed Preferred Treatment	0.39***	0.23; 0.65	1.18	0.77; 1.81	0.39***	0.23; 0.67	0.94	0.38; 2.37

Outcome: Days of Opioid Use at 3- & 6-month follow-ups [count; past 90 days]

Control: Days of Opioid Use at Baseline

^aIRR: Incidence Rate Ratio

^***p ≤ 0.001

Table 5 provides the results of four logistic regressions, each estimating the association between having received (n = 253)/prescribed (n = 224) one’s preferred medication and patients’ subsequent tendency to “crossover” or “switch” to a different treatment during treatment. Evidence suggests patients were significantly less likely to crossover to a different treatment pathway if they originally received the treatment they preferred, for both XR-NTX (Odds Ratio [OR] = 0.32) as well as buprenorphine (OR = 0.22). Evidence suggests a similar finding for when providers’ original preferences were met, though only for XR-NTX (OR = 0.36).

Table 5:

Logistic Regressions of Having Received One’s/Provider's Preferred Treatment and Subsequent Treatment Crossover by Patient and Subsequent Treatment Crossover by Patient

	Patients' Perspective (N = 253)				Providers' Perspective (N = 224)
	Originally Received XR-NTX		Originally Received Buprenorphine		Originally Prescribed XR-NTX		Originally Prescribed Buprenorphine
	OR a	95% CI	OR a	95% CI	OR a	95% CI	OR a	95% CI
Received/Prescribed Preferred Treatment	0.32*	0.11; 0.94	0.22**	0.08; 0.64	0.36*	0.13; 0.96	0.29	0.04; 2.29

Outcome: Treatment Crossover [binary; 1 = Patient "switched" from one treatment pathway to another during treatment]

^aOR: Odds Ratio

^*p ≤ 0.05

^**p ≤ 0.01

3.2. Qualitative Results

Overall, the research team collected 115 responses from providers indicating why a patient would be a good candidate for the treatment pathway the provider believed they were most appropriate to receive. Although some providers declined to elaborate on their treatment preference recommendation, the responses that were provided allowed for sufficient variation in narrative to be analyzed using a content analysis approach. These results are presented by medication (XR-NTX=72 responses and Buprenorphine=18 responses) and treatment pathway (No Preference/Counseling/Oral Naltrexone/Other=25 responses).

3.2.1. XR-NTX.

Providers indicated that the principal reason they believed a particular patient was a good candidate to receive XR-NTX was consistency with patient choice. Providers indicated, “he wanted it,” “his choice,” “she wanted vivitrol,” and “very motivated and patient choice.” Other primary reasons included a history of impulsive and non-treatment compliant behavior (“non-compliant and impulsive,” “not committed to abstinence,” “poor compliance weekly (with oral) naltrexone,” and “history of poor compliance”) and a history of illicit buprenorphine use or co-occurring other substance use disorders, primarily alcohol (“history of benzodiazepine and alcohol use disorders,” “she has a history of abusing suboxone off the street,” “opiates and ETOH,” and “history of benzodiazepine use disorder and of cooccurring alcohol use disorder”). Other secondary reasons include previously failed methods or MOUDs, treatment naïve or a short period of OUD, and family considerations (parent prescribed pain medication, significant other with OUD, or a significant risk for relapse given complicated family environment).

3.2.2. Buprenorphine.

The two most significant reasons providers believed a patient was a good candidate for buprenorphine were patient choice (“he wanted it,” “preferred by patient,” and “it was patient’s preference”) and because this was the patient’s first treatment episode (“first substance (opioid) use disorder treatment,” “patient hasn’t received MAT for opioid use disorder in past,” and “no history of suboxone treatment in past”). Other secondary reasons include the patient seems reliable and motivated, some history of success with buprenorphine in the past, intense cravings, and non-compliant or unwilling to try XR-NTX.

3.2.3. No Preference/Counseling/Oral Naltrexone/Other.

There were very few notes from providers when they indicated any of the remaining treatment pathways, but generally these comments considered the patient’s preference (“her choice,” “patient preference,” and “she wanted”) or treatment/use history (“recent onset of opioid use (less than two months) with tendency to abuse suboxone,” “use of other substances (non-opioids),” and “relapsed with no MAT”) to support the recommendation.

4.0. DISCUSSION

This secondary analysis of data from a study of MOUD among youth and young adults characterized both patient and provider preferences for MOUD and other treatment pathways, tested for associations between these preferences and OUD treatment outcomes, and determined whether initial preferences influence patients’ decision to change treatment pathways over time. Overall, the quantitative results indicate that youth and young adult participants had a slightly stronger preference for buprenorphine treatment over XR-NTX, but providers had a noticeably stronger preference for XR-NTX over any other treatment pathway, with no preference as secondary. Results indicated improved reductions in past 90-day opioid use outcomes if a patient and/or provider preferred XR-NTX and that patient received XR-NTX; fewer self-reported days of opioid use were associated with such congruence between preferences and actual medication received. This association did not hold for participants receiving buprenorphine or no-medication treatment. Patient preference congruence was also associated with less crossover to other treatment for buprenorphine and XR-NTX, but provider congruence was only associated with less crossover for patients receiving XR-NTX.

As the number of FDA-approved medication options and formulations for the treatment of OUD increases steadily over time, the saliency of patient treatment preference and integrating patient involvement in decision-making will also likely continue to grow. As the availability of these medications expand, patients will be in a more frequent position of being presented with various options for treatment, opening the door for patients and providers to engage in a model of shared decision-making (Friedrichs et al., 2016). Previous studies have demonstrated that patients want to be informed about their options when more than one treatment is available for a health condition (Frosch & Kaplan, 1999; Guadagnoli & Ward, 1998), and patients who are matched with preferred treatments have improved treatment adherence and symptom-related outcomes (Swift & Callahan, 2009), and longer retention rates (Graff et al., 2009). Therefore, a collaborative process that leads to a joint treatment decision-making between the patient and the provider could be particularly useful, especially in situations where the patient is presented with two or more equivalent treatment options (Elwyn et al., 2000; Légaré & Witteman, 2013). In these situations, patients’ decisions are largely driven by preferences related to individual risks and benefits specific to their history and current life circumstances. This navigation of treatment options between the patient and the provider in a shared decision-making process has the potential to improve patient outcomes, symptomatology, and treatment retention rates in the future given the direction of our early findings.

Shared decision-making models for treatment planning occur at the individual level and consider contextual factors related to the specific patient. Interestingly, patients and providers in our study had differing treatment preferences overall when considered in the aggregate. Youth and young adult patients largely preferred buprenorphine (42%) or XR-NTX (38%) following their residential treatment episode, but providers largely preferred XR-NTX (56%) with a small minority preferring buprenorphine for their patients in comparison (13%). Early MOUD preference studies conducted with adults (within the last decade when different MOUDs have been more commonly available) have found patient preferences for buprenorphine to largely be attributable to negative perceptions or barriers to accessing methadone treatment (Gryczynski et al., 2013; Uebelacker et al., 2016), and many adults at the time were willing to try XR-NTX despite lack of prior experience with the medication (Uebelacker et al., 2016). Due to the gradual uptake and availability of MOUDs to youth and young adults (Bagley et al., 2021), patient preference studies are lagging behind those of adults and future research in this area is needed.

Among the providers in this study, however, the qualitative results indicate that preferences for patients receiving XR-NTX largely surrounded the patient preferring XR-NTX, having a history of impulsive or non-treatment compliant behavior, as well as a history of illicit buprenorphine use or co-occurring substance use disorder, such as alcohol. In the open-ended responses, providers clearly articulated a preference for XR-NTX when the patient wanted to receive XR-NTX, or the patient would be unlikely to be adherent to a buprenorphine maintenance treatment plan in the community. Providers only seemed to prefer buprenorphine for their patients if the patient themselves strongly preferred buprenorphine, had not attempted buprenorphine maintenance in the past, or achieved some success with buprenorphine maintenance previously. In the case of both medications, provider preference was primarily influenced by patient preference first, followed by clinical indicators the provider felt would be predictive of treatment adherence and ultimately fewer days of opioid use following the residential treatment episode.

Ultimately providers weigh patient preference so heavily because there is evidence in the broader medical literature that treatment congruence with patient preference improves outcomes (Greenfield et al., 1985), but the evidence in the field of substance use disorders is mixed (Friedrichs et al., 2016). One Norwegian clinical trial examining the relationship between preference for XR-NTX and treatment/opioid use outcomes, measured baseline preference for antagonist treatment (XR-NTX) (characterizing preferences as high, medium, and low) among all participants and then randomized to either monthly injections of XR-NTX or daily sublingual buprenorphine (Gaulen et al., 2022). Individuals in the buprenorphine group had a significantly higher risk of first relapse to opioids than those in the XR-NTX group, across all preference levels, as well as significantly greater number of days of opioid use during the follow-up period. In the XR-NTX group, risk of relapse to opioids was significantly higher among participants continuing with XR-NTX with the lowest and medium preference levels for XR-NTX, compared with the highest preference levels for XR-NTX. Throughout the follow-up period, treatment adherence rates to XR-NTX were twice as high for participants with a high preference for XR-NTX compared to those with the lowest preference, both for those participants who continued XR-NTX from baseline and those who switched to XR-NTX mid-study. In that study, congruence between preferred treatment and received treatment resulted in fewer days of illicit opioid use during the shorter-term follow-up period, but greater baseline preference for XR-NTX resulted in longer-term reduced opioid use and improved treatment adherence.

In our study of youth and young adults, congruence between patient preference for XR-NTX and reduced opioid use during the follow-up period held consistent with the Gaulen et al. (2022) study, as well as the mixed outcomes for those preferring and receiving buprenorphine. Similarly, our results indicate that individuals receiving the treatment congruent with their baseline preference were also less likely to switch to a different medication during the study. Taking these results together, it appears that early evidence suggests that patients experience improved opioid use and treatment outcomes if they receive XR-NTX as the treatment medication of their choice. For both XR-NTX and buprenorphine, if the patient does not receive the treatment medication pathway of their choosing at treatment onset, patients are more likely to switch to their preferred pathway later in their recovery efforts, possibly increasing the risk of treatment disengagement and medication non-adherence, and ultimately the potential for overdose.

Providers in our study seem to have acted with this sense in mind, given how much they prioritized the patient’s preference in their own descriptions. Future research should consider semi-structured qualitative interviews with providers to further explore these nuanced attitudes and break down the construction of these preferences by patient characteristics, treatment and opioid use history, and other clinically relevant determinates. While patient preference appears central to predicting treatment outcomes, these other meaningful areas are likely still meaningful to consider, and may strengthen the argument for greater integration of shared decision-making models in settings where MOUDs are prescribed to patients of all ages, to ensure the patient is aware of all considerations that should weigh into preferring one medication treatment pathway other another.

4.1. Limitations

The limitations to this study include, first, recruitment of youth and young adults from a residential specialty substance use disorder treatment program in the mid-Atlantic area, which may limit generalizability to other youth and young adults in other settings. Similarly, because the providers in this sample had extensive experience with youth and young adults in this setting, their sophisticated understanding of this specialty population may be dissimilar to other providers in other settings. Additionally, results may be biased by any non-random missing data given that a significant number of participants and/or physicians did not complete the treatment preference measures. Despite these limitations, the study also has strengths, including a focus on an understudied group with OUD that is at high risk of poor outcomes.

4.2. Conclusions

Patient and provider preferences for a medication treatment did ultimately result in reduced opioid use (in the case of XR-NTX) and less medication switching (for both XR-NTX and buprenorphine) if the patient received the medication congruent with their preference. However, while patient preference was the most salient factor in shaping provider preferences, other clinically relevant factors were also important to providers to consider. Given that these additional factors may also contribute to opioid use and treatment outcomes, a greater emphasis on shared decision-making processes between patients and providers may allow for more comprehensive guidance to patients who are increasingly confronted with multiple, effective medication treatments for OUD. Promoting greater shared decision-making between patients and providers reduces the risk of medication non-adherence in the future, thus increasing the likelihood of improved outcomes.

Highlights.

• Patients were less likely to switch treatments if they received their preference.

• This finding held for patients receiving either XR-NTX or buprenorphine.

• Preference-receipt congruence reduced opioid use and improved adherence.