Opioid Use Disorder Medications Among Youth in Primary Care: Subgroup Analysis of the PROUD Trial Open Access

In the United States, annual drug overdose deaths have been increasing for decades, although they decreased in 2023 for the first time since 2018 (105 007 deaths in 2023 compared with 107 941 deaths in 2022).¹ Among youth younger than age 25 years, drug overdose contributes to most unintentional deaths caused by poisoning, which is the second leading cause of death after motor vehicle accidents.² The rates of drug overdose deaths in the United States are higher among adults ages 35–44 years (60.8 per 100 000) than youth younger than age 25 years (13.5 per 100 000).¹ However, premature deaths owing to drug overdose among youth lead to many years of life lost, with drug overdose among adolescents contributing to more than 200 000 years of life lost between 2015 and 2019.³ The importance of timely treatment of young patients with opioid use disorder (OUD) is critical given the heightened risk for overdose because of the presence of fentanyl in the illicit drug supply.

Medications for treating OUD, including buprenorphine and injectable extended-release naltrexone (XR-NTX), are recommended by practice guidelines for patients with OUD and can be offered in primary care in the United States.^4–7 Yet, numerous previous studies have demonstrated low use of these evidence-based medications among US youth with OUD,^8,9 and few prescribers report comfort with offering them.¹⁰ Both receipt of medications for OUD and retention in care are notably lower among youth compared with older age groups.^8,11–15 Increasing access to medications for OUD in primary care settings is greatly needed and could be achieved by addressing barriers to prescribing these life-saving medications.⁵ 

The Primary care OUD (PROUD) trial was a pragmatic trial testing a model for increasing delivery of medications for OUD in primary care clinics.¹⁶ The PROUD intervention implemented the Massachusetts Model,^17,18 which uses trained nurse care managers to support prescribing medications for OUD (ie, buprenorphine or XR-NTX). The PROUD intervention aimed to address some prescribing barriers through the patient and provider supports offered by nurse care managers. Lack of psychosocial support for patients and burdensome documentation for providers have been described as barriers previousely.^19–21 The PROUD trial was associated with a significant increase in the number of medication treatment days in intervention clinics relative to usual care clinics.^22,23 Yet, it is unclear whether the PROUD trial benefited youth, who often have worse treatment outcomes than adults.^24,25 

The present study conducted a secondary analysis of the PROUD trial data and reports on the patient-level estimates of outcomes before and after the implementation of the PROUD intervention for patients aged 16–25 years. The PROUD trial offered nurse care management to patients as young as age 16 years, because buprenorphine is approved by the US Food and Drug Administration (FDA) for OUD in patients aged at least 16 years. In the present subgroup analysis, we included youth through age 25 years given prior literature highlighting challenges young adults, or “emerging adults,” experience in accessing OUD medication treatment.²⁶ The objective was to compare treatment among youth in clinics randomized to PROUD and usual care as well as receipt of naloxone. Offering youth access to medications for OUD through a nurse care manager may partially address concerns regarding stigma, which could be a particularly salient barrier to treatment among youth.²⁷ The results could inform health systems about whether implementing a nurse care management model, which effectively increased access to medications for OUD in the broader primary care population, can achieve this aim among youth specifically.

The data for the present study include a subgroup of youth seen at primary care clinics participating in the PROUD pragmatic cluster-randomized trial, which included 6 health systems from 5 states.¹⁶ Within each health system, 2 primary care clinics agreed to participate, with 1 randomized to the intervention and 1 to usual care. The nurse care management model was implemented at intervention clinics to support medication treatment of OUD to primary care patients aged 16–90 years. Baseline and follow-up data were collected from electronic health records (EHRs), and in 2 health care systems health care claims were collected. The youth sample comprised adolescents and young adults aged 16–25 years at the time of randomization. Randomization occurred on February 28, 2018, except for 1 health system that randomized on August 31, 2018. Patients were eligible if they had a primary care visit during the 5-year study period, which included 3 years before randomization (except for 1 health system with only 2.8 years of data available before randomization) and 2 years after randomization (except for the health system with a delayed randomization). The study was reviewed and approved by the Advarra institutional review board, and the study received waivers of informed consent and US Health Insurance Portability and Accountability Act authorization. This study was registered at ClinicalTrials.gov (identifier NCT03407638).

The primary outcome of interest was receipt of any medication treatment of OUD (buprenorphine or XR-NTX) during the 2-year postrandomization period. Other secondary outcomes evaluated included the specific type of medication received and any receipt of naloxone. Newly initiated treatment was also evaluated and defined as receipt of medication treatment after randomization without a prior year of EHR-documented OUD. Treatment outcomes were based on medication orders and procedure codes. We also report on the receipt of at least 1 medication order for buprenorphine with a dose at least 16 mg, a recognized effective dose.

Mental health conditions, of particular interest to youth, including depression, anxiety, attention-deficit/hyperactivity disorder, posttraumatic stress disorder, conduct disorder, deliberate self-harm, and opioid overdose, were examined. Comorbidities were defined as present if there were International Classification of Diseases diagnosis codes documented during either the pre- or postrandomization periods, respectively. Patient demographic characteristics included age, sex, race or ethnicity, insurance type, and zip code–level neighborhood characteristics (median household income, percent unemployed, and percent below the federal poverty level). Patient race and ethnicity are considered a social, rather than a biological, construct and were defined based on documentation in the EHR. If a patient was documented as Hispanic (or Latino/Latinx) in the EHR, they were categorized as Hispanic and not assigned a racial group, because not all health systems collected race for these patients. “Other race” was recorded in the EHR when a person did not identify as belonging to any of the listed racial groups (and was not Hispanic) (Table 1).

Descriptive statistics were used to characterize the patient population and are presented overall and by clinic type (PROUD intervention or usual care) as well as by time frame (pre- vs postrandomization). To compare the odds of the binary outcome of any medication for OUD between PROUD intervention and usual care clinics, marginal generalized estimating equations with a binomial distribution and a logit link was used, with the 12 clinics as clusters, and adjusted for sex (female vs male), age group (16–17 vs 18–25 years), and race or ethnicity (white vs all other). We included race and ethnicity in descriptive analyses and as an adjustment variable both to describe our sample and to account for known racial and ethnic disparities in OUD medication treatment.^15,28 The Fay and Graubard sandwich estimator was applied to correct for the small cluster bias.²⁹ Analyses were repeated among a subgroup of patients with documented OUD during the study period, defined as any International Classification of Diseases diagnosis of OUD or buprenorphine prescription. Patients who were prescribed buprenorphine but did not have a recent documented OUD were included in the denominator, given that OUD may be underdiagnosed, but the formulations of buprenorphine we included were for an OUD indication. For the continuous measure of days of OUD treatment, we present results as both mean and median (with IQR) owing to the skewed distribution of days of OUD treatment.

Among 47 085 primary care patients who were aged 16–25 years at the time of clinic randomization for the PROUD trial, 26 562 had a visit to a usual care clinic and 20 523 had a visit to a PROUD intervention clinic over the 5-year study eligibility period. Demographic characteristics of patients between the 2 clinics were similar: both had a mean age of 21 years and were majority female (Table 1). However, a slightly higher proportion of usual care clinic patients were of Hispanic ethnicity (34.8% vs 28.2%) and had Medicaid health insurance coverage (43.6% vs 34.2%) compared with those in the PROUD intervention clinic.

The most common documented mental health diagnoses in the population were anxiety and depression (15.0% and 11.5%, respectively); the prevalence of these conditions did not differ markedly between patients in the usual care or intervention clinics (Table 2). The overall prevalence of documented OUD diagnosis among youth was low both before (0.2%, n = 113) and after (0.3%, n = 153) clinic randomization (Table 2; Supplemental Table 1) and was similar for both the usual care and PROUD intervention clinic patients. The most common nonopioid substance use disorders were tobacco use disorder (3.8%), cannabis use disorder (1.8%), and alcohol use disorder (0.9%), and their prevalence was similar during both the baseline and follow-up periods (Table 2; Supplemental Table 1).

A higher proportion of patients in the PROUD intervention clinics (0.2%, n = 42) received any OUD treatment during the 2-year postrandomization period relative to usual care clinic patients (0.1%, n = 25) (Table 3). However, this difference in treatment did not reach statistical significance in adjusted analyses (odds ratio = 1.75; 95% CI, 0.63–4.89; P = .25), which accounted for the patient’s age group, sex, and race or ethnicity. The most-prescribed medication treatment of OUD during the 2-year follow-up period was buprenorphine (n = 64), with XR-NTX prescribed among fewer patients (n = 7). The number of patients per clinic who received any medication treatment ranged from 0 to 21; 2 clinics at 1 site had no youth treated with medications. The prescription of medications for OUD seemed to be predominantly among patients who were newly initiating treatment in both usual care and intervention clinics. A higher proportion of patients were prescribed naloxone (0.3%, n = 119) than were prescribed any medications to treat OUD (0.1%, n = 67) during the follow-up period (Tables 3 and 4). Among the total primary care sample, naloxone prescribing appeared to increase among patients in the intervention clinics from baseline to follow-up but not in usual care (Table 4). Among patients with OUD, prescribing appeared to increase from baseline to follow-up similarly in usual care and intervention clinic patients, such that approximately one-third of all patients with OUD were prescribed naloxone during follow-up (33.3%).

Among patients with OUD, descriptive analyses suggested that a slightly higher proportion of patients in the PROUD intervention clinics received any medication treatment of OUD (32.1%) relative to patients in usual care clinics (26.6%) (Table 5). Buprenorphine was again more commonly prescribed than XR-NTX overall and for patients in either clinic setting. Among patients who received any medication treatment of OUD, the median days of medication treatment were higher in the usual care clinic (102 days) than in the intervention clinic (30 days) at baseline, although this represented only 7 patients in usual care clinics and 16 in intervention clinics (Figure 1). During the 2-year follow-up period, the median days of medication treatment was 64 (IQR, 24–206) days for 25 usual care clinic patients compared with 81.5 (IQR, 30–177) days in 42 intervention clinic patients.

Among a large sample of youth with primary care visits at 1 of 12 clinics in 6 large health systems, we did not detect a significant difference in the proportion of patients receiving medication treatment in intervention clinics relative to usual care clinics during the follow-up period of the PROUD trial. Overall, only a small number of all youth received any medication treatment of OUD after implementation of the PROUD intervention (<1%), and a minority of those with OUD (29.8%). Among patients who received any OUD medication, the number of treatment days ranged widely, but the median medication treatment length was less than 3 months in both intervention and usual care clinics. Although the PROUD trial, which tested the Massachusetts Model of nurse care management for office-based addiction treatment in primary care, demonstrated significant increases in overall clinic-level population-based measures of OUD treatment among patients aged 16–90 years,²² we did not observe a significant patient-level difference in medication treatment among a youth sample of those aged 16–25 years.

The low proportion of youth receiving OUD medication treatment is consistent with other prior studies suggesting that youth who could benefit these medications often do not receive them¹¹ and are less likely to receive them than other older-aged adults.^12,13,24 The present study was a post-hoc subgroup analysis of the PROUD trial, and the small number of youth receiving OUD treatment likely contributed to a lack of statistically significant findings, as evidenced by the wide CI that indicated low precision. However, it is also possible that offering these medications through primary care with nurse care management support was not enough to increase youth engagement. Youth may experience continued barriers, such as stigma associated with medication treatment of OUD, ambivalence, or co-occurring mental health conditions.²⁷ Medical providers’ hesitancy to use medications for OUD among youth may continue to act as another barrier to such care.³⁰ 

Although prescribing medications for OUD was infrequent, buprenorphine was the most-prescribed medication in our sample, which is consistent with prior studies.^31–33 Among youth with OUD, the proportion receiving any medication treatment of OUD was higher, as would be expected; yet most did not receive medication treatment. Unlike prior studies, we did not assess the use of oral naltrexone. Oral naltrexone is not approved by the FDA or recommended for OUD treatment. However, prior studies have found that naltrexone may be more frequently prescribed among adolescents than young adults, and the inclusion of oral naltrexone could contribute to this finding,³¹ particularly if providers or families are hesitant to use injectable medications.³⁴ A lower rate of XR-NTX is also not surprising given it is more difficult to initiate, requires a period of abstinence before injection, and is not FDA-approved for patients younger than 18 years, which could discourage prescribing.

It was encouraging that naloxone prescribing appeared to be higher in the intervention sites and among youth with OUD at follow-up. Yet, it is important to note that not all youth who could be prescribed naloxone may have been eligible for medications for OUD, given that youth can experience an overdose without OUD. Naloxone could be prescribed as an overdose prevention strategy for all youth, given that they (or their peers) may be at risk for overdose if they ever misuse illicit pills or other substances, which increasingly contain fentanyl.³⁵ Yet, there may have been missed opportunities to systematically prescribe naloxone to all youth with OUD, who are at particularly high risk for overdose. Although the PROUD intervention was intended to address barriers to prescribing in primary care settings by offering integrated nurse care management, it did not address youth-specific barriers to care. In addition, youth may have inaccurate beliefs about, or experience stigma related to, medication treatment of OUD.

There are several reasons the intervention may not have had the intended effect among youth. First, the intervention was not specifically designed for youth and may not have been as engaging to a youth audience as one tailored to this patient group. Second, youth (as compared with older patients) are not as reliant on primary care,³⁶ and fewer primary care appointments limit opportunities for engagement. Third, it may be necessary to be more flexible in care delivery once patients are engaged, whether that be through primary care or other service settings. For example, a recent pilot study suggested that home delivery of medications and family involvement may be promising approaches for adolescents and young adults to receive medications for OUD and potentially retain them in treatment.³⁷ Finally, we observed a similar proportion of youth diagnosed with OUD receiving medication in intervention clinics relative to what was observed in the total intervention clinic sample reported previously (32.1% vs 31.1%).²² Therefore, our lack of statistically significant findings may reflect the lower absolute numbers of youth referred for OUD medication treatment, which could be because of a lower prevalence of OUD among youth or underdiagnosis among youth. Although the prevalence of documented OUD diagnosis was low in our sample, it was comparable to that observed in other primary care studies (0.7%–2.8%)^38–40 and could reflect broad underidentification of OUD in primary care.

The PROUD trial was a pragmatic study that required health systems to agree to participate. We do not know whether the findings for these health systems would generalize to other settings. The trial concluded before the removal of the federal x-waiver requirement (December 29, 2022), which required prescribers to register with the program, and imposed limits on the total number of patients they could treat.⁴¹ We do not know whether findings would differ in this new landscape. However, recent studies suggest prescribing barriers remain.³⁰ The PROUD trial intervention was not designed to be youth-centered, and therefore efforts were not taken to ensure adequate numbers of youth were included to conduct outcome analyses in this subgroup. In addition, owing to challenges with the implementation of the trial intervention, some intervention sites experienced a delay in the availability of a nurse care manager. Given that our follow-up window was only 2 years, it is possible that with these delays the follow-up period for the present study was not long enough to detect an effect. In analyses using follow-up data through year 3, continued increases in treatment were observed in intervention clinics.²³ Finally, data on health care use external to the health system was not available for 4 of the 6 health care systems, and we may have missed some externally provided medication treatment. We also do not have information on how many youths were approached by nurse care managers and how many refused services.

This post-hoc analysis of the PROUD intervention reveals that even after implementation of a nurse care management support intervention for OUD medication prescribing in primary care, we did not find evidence that medication treatment meaningfully increased among youth, specifically adolescents and young adults. Greater efforts are likely needed to encourage prescribing for adolescents and young adults that are youth-centered and further reduce barriers to care.

EHR

electronic health record

FDA

Food and Drug Administration

OUD

opioid use disorder

PROUD

Primary care Opioid Use Disorder

XR-NTX

injectable extended-release naltrexone.

We would like to acknowledge the significant contribution of Dr Katharine Bradley to this work.