Comparing buprenorphine treatment outcomes among Kentucky Medicaid patients by prescriber specialty and type

Abstract

Background:

Buprenorphine is vital to improving health outcomes for patients with opioid use disorder (OUD). Little is known about variation in buprenorphine retention and related outcomes across different outpatient prescribers and settings.

Methods:

Our 2020–2021 sample includes 10,709 Kentucky Medicaid-enrolled patients with OUD who received most of their buprenorphine prescriptions from primary care physicians, specialty care physicians, or advanced practice providers (APP). Adjusted multivariable linear and logistic regression, respectively, estimated the percentage of days with buprenorphine coverage and the proportion of patients who experienced non-fatal overdose after the first prescription.

Results:

Compared to specialty physicians and APPs, those with primary care prescribers had a higher percentage of days with buprenorphine (primary: 69.6, 95% confidence interval (CI): [66.7, 72.4]; specialty: 60.1, 95% CI: [54.7, 65.4]; APP: 64.3, 95% CI: [62.1, 66.5]; p<0.01). A lower proportion of patients with APP prescribers experienced non-fatal overdose (APP: 4.0, 95% CI: [3.4, 4.7]; primary: 5.5, 95% CI: [4.1, 6.9]; specialty: 5.7, 95% CI: [4.2, 7.2]; p<0.01).

Conclusions:

Kentucky Medicaid-enrolled patients with primary care physicians had the highest treatment retention, while patients with APPs were less likely to experience a non-fatal overdose. Our findings suggest that although multiple provider types can effectively prescribe buprenorphine, the potentially meaningful differences observed in retention and non-fatal overdose warrant further longitudinal investigation to assess whether these differences by provider type and setting persist over time.

Introduction

Buprenorphine is one of two U.S. Food and Drug Administration-approved, gold-standard medications to treat opioid use disorder (OUD) that is prescribed in office-based medical settings (National Academies of Sciences et al., 2019). The 2016 Comprehensive Addiction and Recovery Act (CARA) extended buprenorphine prescribing authority to advanced practice providers (APPs), including nurse practitioners and physician assistants.

Recent observational studies have been mixed regarding buprenorphine retention and other quality outcomes across different settings or provider types (Gertner et al., 2020; Levin et al., 2022; Olfson et al., 2020; Xu et al., 2024; Young et al., 2025; Zhu et al., 2024). While some have found that buprenorphine retention was higher in certain specialty settings (Levin et al., 2022; Xu et al., 2024; Young et al., 2025), others found that providers in primary care provided comparable care to providers in specialty settings (Gertner et al., 2020; Zhu et al., 2024). As noted, these studies examine buprenorphine retention and treatment outcomes either by provider type or outpatient care setting. We could find no studies that explicitly examine buprenorphine-related outcomes for OUD patients who receive buprenorphine from APP prescribers relative to specialty and primary care physicians, combining both setting and provider type in this comparison. Addressing this gap can facilitate further understanding of OUD treatment context and its impact on treatment outcomes more broadly.

In 2021, Kentucky had the third-highest age-adjusted drug overdose mortality rate in the United States (Centers for Disease Control and Prevention, 2022). Kentucky’s Medicaid program insures approximately 30 percent of its population (Kaiser Family Foundation, 2024) and in 2021, approximately 67 percent of opioid overdose deaths in Kentucky occurred among those insured by Medicaid (Hardcastle et al., 2023). Federal and state policies dictate certain aspects of clinical practice that may impact buprenorphine treatment. For example, while the X waiver was eliminated in 2023 at the federal level, Kentucky regulations still require all buprenorphine prescribers to maintain said waiver (Murphy, 2024). These requirements include additional training and caps on the number of patients who can prescribe buprenorphine and are particularly stringent for APPs. APPs in Kentucky are also subject to stringent scope of practice requirements for prescribing buprenorphine, including physician supervision (Andraka-Christou et al., 2022; Spetz et al., 2021), making Kentucky a uniquely rigid policy environment for prescribing buprenorphine (Silwal et al., 2025).

Considering the research gap in examining provider type and setting together for buprenorphine-related outcomes—as well as the policy context of Kentucky, specifically—our study examined associations between buprenorphine services from different provider types and treatment-related outcomes among Medicaid-enrolled adults with OUD in Kentucky in 2020–2021.

Methods

Data and study sample

This study used de-identified data from HealthVerity, a proprietary commercial administrative claims database, from January 2020 to December 2021. The full HealthVerity sample included 37,164 Medicaid beneficiaries aged 18–64 with a primary or secondary OUD diagnosis. We further restricted the sample to patients with complete data for all covariates of interest and with at least one buprenorphine pharmacy claim with a corresponding national provider identifier (NPI) for the prescriber during the two-year study period. We linked the publicly available National Plan and Provider Enumeration System (NPPES) data, updated as of July 2023, to HealthVerity data by NPI to obtain detailed information on provider specialty. We then determined individual receipt of buprenorphine from a primary care physician, specialty care physician, or advanced practice provider (APP) based on the primary taxonomy code (i.e., the provider specialty classification code) included in the NPPES data. After the restrictions, the final sample included 10,709 patients with complete covariates (see supplementary figure 1 for flow chart).

Measures

Exposures.

The primary exposure for this study was whether patients received the majority (>50 percent) of their buprenorphine prescriptions from a primary care physician, specialty care physician, or APP. Buprenorphine prescriptions were identified using national drug codes (see supplementary appendix 1). We first aggregated an individual’s buprenorphine prescriptions by provider type (i.e., primary care physician, specialty care (non-emergency) physician, emergency department physician, APP, or missing). We then calculated the sum of prescriptions for each of the five designated categories, and patients were assigned to the provider type with the highest sum of prescriptions for the respective category. We checked patients in which the sums of prescriptions could be equal across categories (e.g., 11 prescriptions from a primary care physician and 11 prescriptions from an APP) but did not find any patients who met this criterion.

Primary care physicians were defined as a prescriber’s primary taxonomy code corresponding to an internal or family medicine physician. APPs were identified using prescriber’s primary taxonomy code corresponding to nurse practitioner/registered nurse or physician assistant. While specialty information was available through NPPES for nurse practitioners and registered nurses, this information was not available for physician assistants. As a result, we intentionally did not account for specialty types for APPs. Patients for whom the majority of their buprenorphine claims were from an unknown provider type (i.e., the NPI or the primary taxonomy code were missing), the majority of their buprenorphine claims were from an emergency department physician (i.e., primary taxonomy code corresponded to emergency department), or the majority of their claims were from facilities in which the provider type was not clear (i.e., ambulatory care facilities) were excluded from the analysis. Patients for whom the majority of buprenorphine claims were from an emergency department physician were dropped because 1) we were interested in comparing provider type and setting together, and 2) experiences with emergency department visits for emergency and bridge buprenorphine would not be directly comparable to outpatient care visits. All remaining patients were classified as receiving buprenorphine from specialty care physicians (e.g., psychiatry, obstetrics/gynecology) (see supplementary figure 2).

Outcomes.

This study examined five buprenorphine treatment-related outcomes. All outcomes were measured at the patient level. Primary outcomes included 1) the average mean percent of days in the study period with buprenorphine coverage (calculated by dividing the average buprenorphine days’ supply by days of Medicaid coverage that was calculated based on first and last days’ of Medicaid enrollment, starting from the first buprenorphine prescription) and 2) the predicted probability of patients who experienced non-fatal overdose after their first buprenorphine prescription (including overdose that occurred on the same day). Non-fatal overdoses were identified using ICD-10 codes (T40.0–T40.4, T40.6). We identified all non-fatal overdoses that occurred on or after the first day of a patient’s buprenorphine prescription and then created a binary measure of whether the patient had any overdose on/after the first day of the first prescription (1) or not (0). Secondary outcomes included 3) the average number of buprenorphine prescriptions (range: 1–104 prescriptions), 4) the average buprenorphine days’ supply (range: 1–730 days), and 5) the predicted probability of patients who had 80 percent or more days with buprenorphine, consistent with a measure used in other studies of buprenorphine coverage (Wartko et al., 2023).

Covariates.

Demographic covariates included patient age (18–25, 26–40, 41–55, 56–64 years), sex (female, male), and race/ethnicity (non-Hispanic White, all other race/ethnicity). Additional covariates included serious mental illness (SMI), defined as a mental, emotional, or behavioral disorder that causes severe functional impairment or substantially limits daily life activities. We identified SMI by an individual having at least two unique claims with an ICD-10 code corresponding to a diagnosis of schizophrenia, bipolar disorder, major depressive disorder, anxiety or post-traumatic stress disorder, or another serious psychiatric condition, such as borderline personality disorder (see supplementary appendix 2). We also included co-occurring other substance use disorder (SUD), identified by an individual having at least two unique claims with an ICD-10 code corresponding to alcohol use disorder or disorders corresponding to sedative, stimulant, hallucinogen, or inhalant use (see supplementary appendix 2).

Analytic Approach

Analyses were conducted at the patient level. First, we descriptively examined all covariates by primary buprenorphine provider and compared differences using chi-squared tests. Next, we used multivariable linear regression to estimate the association of provider type with the 1) mean days of buprenorphine coverage, 2) number of buprenorphine prescriptions, and 3) buprenorphine days’ supply, adjusting for age, sex, race/ethnicity, SMI, and other SUD. Multivariable logistic regression estimated the association of provider type with 1) experience of non-fatal overdose after their first buprenorphine prescription and 2) having 80 percent or more days covered with buprenorphine, adjusting for categorized age, sex, race/ethnicity, SMI, and other SUD. To enhance interpretability, we estimated the regression-adjusted predicted mean outcomes for all multivariable linear regression models, and the regression-adjusted predicted proportions for all multivariable logistic regression models. 95% confidence intervals were calculated for all regression analyses. Wald tests were used to test for differences by provider type. Two-sided p-values of <0.05 were considered statistically significant. Standard errors were clustered at the provider level.

All analyses were conducted using SAS, version 9.4, and STATA, version 17.1. The Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines were used to guide analyses and reporting of findings.(von Elm et al., 2007) This study met criteria for non-human subjects research from the Johns Hopkins Bloomberg School of Public Health Institutional Review Board review.

Results

Our sample included 10,709 Kentucky patients with OUD enrolled in Medicaid who received the majority of their buprenorphine prescriptions from a primary care physician, specialty physician, or APP during the two-year study period (table 1). Approximately 37 percent of patients in the sample were aged 26–35 years (n=3,948), and over half were female (56.4 percent, n=6,044). Most were non-Hispanic White (96.4 percent, n=10,324). While about 56 percent (n=5,958) of the sample received the majority of their buprenorphine prescriptions from APPs, about 25 percent received the majority of prescriptions from primary care physicians (n=2,693). Distributions of sex, co-occurring SMI, and co-occurring SUD significantly differed across primary care physicians, specialty care physicians, and APPs (table 1). Of note, patients in the sample who received the majority of their buprenorphine prescriptions from APPs also had higher co-occurring other substance use disorder (32% APP vs 25.3% specialty care physician and 19.6% primary care physician; p<0.001) and co-occurring serious mental illness (36.4% APP vs 33.5 specialty care physician and 35.7% primary care physician; p=0.05).

Table 1.

Sociodemographic characteristics among 10,709 Medicaid-enrolled adults with OUD in Kentucky, 2020–2021

	Total N=10,709			Adults Receiving Majority of Buprenorphine Prescriptions from Primary Care Physicians n=2,693 %=25.2			Adults Receiving Majority of Buprenorphine Prescriptions from Specialty Care Physicians n=2,058 %=19.2			Adults Receiving Majority of Buprenorphine Prescriptions from Advanced Practice Providers n=5,958 %=55.6
	N	%		n	%		n	%		n	%		p-value
Age
18–25 years	683	6.4		155	5.6		132	6.4		396	6.7		0.49
26–35 years	3,948	36.9		965	35.8		771	37.5		2,212	37.1
36–45 years	3,667	34.2		950	35.3		704	34.2		2,013	33.8
46–64 years	2,411	22.5		623	23.1		451	21.9		1,337	22.4
Sex
Female	6,044	56.4		1,569	58.3		1,175	57.1		3,300	55.4		0.03
Male	4,665	43.6		1,124	41.7		883	42.9		2,658	44.6
Race/Ethnicity
Non-White	385	3.6		86	3.2		86	4.2		213	3.6		0.19
White, non-Hispanic	10,324	96.4		2,607	96.8		1,972	95.8		5,745	96.4
Co-occurring SMI
No	6,887	64.3		1,731	64.3		1,369	66.5		3,787	63.6		0.05
Yes	3,822	35.7		962	35.7		689	33.5		2,171	36.4
Co-occurring SUD
No	7,751	72.4		2,164	80.4		1,537	74.7		4,050	68.0		<0.001
Yes	2,958	27.6		529	19.6		521	25.3		1,908	32.0

Percentages may not add due to rounding. SMI= serious mental illness; SUD = substance use disorder

Receiving buprenorphine treatment is defined as having at least one buprenorphine claim during the study period. Majority of prescriptions was determined based on sums of buprenorphine prescriptions an individual received by care provider during the study period.

Non-White race group includes Black, non-Hispanic, Hispanic, and mixed race.

SMI diagnoses were defined based on two or more claims with ICD-10 codes corresponding to schizophrenia, major depressive disorder, anxiety or post-traumatic stress disorder, or another serious psychiatric illness (e.g., borderline personality disorder).

SUD diagnoses were defined based on two or more claims with ICD-10 codes corresponding to alcohol use, stimulant use, hallucinogen use, or inhalant use disorders.

P-values less than 0.05 are bolded.

Compared to specialty care physicians and APPs, those who received the majority of their prescriptions from primary care physicians had a higher adjusted average mean percent of days with buprenorphine coverage (primary: 69.6, 95% CI: [66.7, 72.4]; specialty: 60.1, 95% CI: [54.7, 65.4]; APP: 64.2, 95% CI: [62.1, 66.5]; p<0.01) and mean average buprenorphine days’ supply (primary: 453.8, 95% CI: [431.9, 475.8], specialty: 380.0, 95% CI: [343.6, 416.3]; APP: 390.3, 95% CI: [374.5, 406.0]; p<0.001). There was no significant difference between the adjusted average number of buprenorphine prescriptions prescribed across the three provider types (primary: 34.6, 95% confidence interval (CI): [32.1, 37.1]; specialty: 34.2, 95% CI: [29.9, 38.56]; APP: 34.9, 95% CI: [33.1, 36.6]) (table 2A). The predicted probability of patients having a non-fatal overdose following their first buprenorphine prescription was significantly lower among those receiving the majority of their prescriptions from APPs compared to primary care and specialty care physicians (APP: 4.0, 95% CI: [3.4, 4.7]; primary: 5.5, 95% CI: [4.1, 6.9]; specialty: 5.7, 95% CI: [4.2, 7.2]) (table 2B). Finally, the adjusted predicted probability of patients with more than 80 percent of days with buprenorphine coverage was also higher among patients with primary care physicians relative to specialty care and APP prescribers (primary: 57.0, 95% CI: [53.4, 60.6]; specialty: 47.0, 95% CI: [41.7, 52.3]; APP: 50.5, 95% CI: [47.8, 53.1]; p<0.001).

Table 2A.

Buprenorphine treatment continuous outcomes by primary provider type among 10,709 Medicaid-enrolled adults with OUD in Kentucky, 2020–2021

	Adults Receiving Majority of Buprenorphine Prescriptions from Primary Care Physicians n=2,693 %=25.2			Adults Receiving Majority of Buprenorphine Prescriptions from Specialty Care Physicians n=2,058 %=19.2			Adults Receiving Majority of Buprenorphine Prescriptions from Advanced Practice Providers n=5,958 %=55.6
Buprenorphine Treatment-Related Continuous Outcomes	Adjusted mean	95% confidence interval (CI)		Adjusted mean	95% CI		Adjusted mean	95% CI		p-value
Average mean percent of days with buprenorphine coverage	69.6	66.7, 72.4		60.1	54.7, 65.4		64.3	62.1, 66.5		0.001
Secondary Outcomes
Average number of buprenorphine prescriptions	34.6	32.1, 37.1		34.2	29.9, 38.6		34.9	33.1, 36.6		0.95
Average buprenorphine days’ supply	453.8	431.9, 475.8		380.0	343.6, 416.3		390.3	374.5, 406.0		<0.001

All models adjust for age, sex, race/ethnicity, co-occurring SMI, and co-occurring SUD.

Majority of prescriptions was determined based on sums of buprenorphine prescriptions an individual received by care setting during the study period.

Average mean percent of days with buprenorphine coverage is a percentage, defined as the number of days an individual had a buprenorphine prescription by the total number of days enrolled in Medicaid during the study period, starting when they received their first buprenorphine prescription.

P-values less than 0.05 are bolded.

Table 2B.

Buprenorphine treatment binary outcomes by primary provider type among 10,709 Medicaid-enrolled adults with OUD in Kentucky, 2020–2021

	Adults Receiving Majority of Buprenorphine Prescriptions from Primary Care Physicians n=2,693 %=25.2			Adults Receiving Majority of Buprenorphine Prescriptions from Specialty Care Physicians n=2,058 %=19.2			Adults Receiving Majority of Buprenorphine Prescriptions from Advanced Practice Providers n=5,958 %=55.6
Buprenorphine Treatment-Related Binary Outcomes	n	Adjusted predicted probability (%) 95% CI		n	Adjusted predicted probability (%) 95% CI		n	Adjusted predicted probability (%) 95% CI		p-value
Non-fatal overdose following first buprenorphine prescription	139	5.5 4.1, 6.9		120	5.7 4.2, 7.2		254	4.0 3.4, 4.7		0.014
Secondary outcome
≥80% of days with buprenorphine coverage	1,568	57.0 53.4, 60.6		963	47.0 41.7, 52.3		2,964	50.5 47.8, 53.1		0.002

All models adjust for age, sex, race/ethnicity, co-occurring SMI, and co-occurring SUD.

Majority of prescriptions was determined based on sums of buprenorphine prescriptions an individual received by care setting during the study period.

≥80% of days with buprenorphine coverage is defined as the percentage of study participants having 80% or more days covered with buprenorphine treatment during the study period, starting when they received their first buprenorphine prescription.

P-values less than 0.05 are bolded.

Discussion

In this study of Kentucky's 2020–2021 Medicaid program, patients with OUD received nearly 80 percent of buprenorphine prescriptions from primary care physicians and APPs. Patients with primary care physicians had the highest proportion of days with buprenorphine relative to specialty care physician or APP patients. Compared to primary care and specialty care physician patients, a lower probability of APP patients experienced non-fatal overdose after their first buprenorphine prescription.

Prior studies have found that buprenorphine retention outcomes are equivalent across provider types with primary care versus specialty training (Gertner et al., 2020; Levin et al., 2022; Olfson et al., 2020; Zhu et al., 2024), with some exceptions (Young et al., 2025). All these studies examine buprenorphine retention over one year or less, most often examining episodes of at least 180 days, and largely focused on setting rather than setting with provider type. We examined buprenorphine retention—defined by percentage of days covered—for up to two years, while also combining setting and provider type. As few studies have examined a retention period for this time period and to our knowledge, have not conducted this type of comparison before, our finding that patients of primary care physicians had longer buprenorphine treatment retention is novel. Additional research is needed to further investigate these comparisons with a longer buprenorphine treatment retention outcome. However, primary care training and practice allowing for improving chronic disease management over time may be beneficial for OUD treatment (McLellan et al., 2014). For example, qualitative work suggests that patients with OUD who received buprenorphine in primary care appreciated the associated flexibility and team-based care component (Incze et al., 2023). This is evidenced by studies that have shown buprenorphine treatment retention to be improved in settings that incorporate multidisciplinary care (i.e., integration of addiction services into primary care, nurse care management) (Lagisetty et al., 2017; Lapham et al., 2024; Wartko et al., 2023).

A common justification for providing buprenorphine in specialty care is that providers in these settings have the necessary expertise to treat co-occurring, complex psychiatric comorbidities (Gertner et al., 2020; Xu et al., 2024). We observed differences in treatment retention even after adjusting for co-occurring SMI and SUD, and in our sample, these comorbidities were highest among APP patients. However, we were unable to adjust for severity of OUD, and patients with more severe OUD may have been overrepresented among specialty physicians. This could have impacted buprenorphine retention and overdose outcomes observed. Additional study is needed to understand the full context of care services patients with OUD receive from different provider types in outpatient settings. Future longitudinal research can assess the quality of care provided for co-occurring chronic diseases—including mental illness and other substance use disorders—among different provider types and settings (Heidari et al., 2022).

To our knowledge, this is one of the first studies that has examined differences in non-fatal overdose among Medicaid-enrolled individuals receiving buprenorphine across different care providers and settings. The predicted probability of patients having a non-fatal overdose after the first buprenorphine prescription was lower among those who received the majority of prescriptions from APPs compared to primary care and specialty care physicians. APPs in Kentucky have additional licensing requirements for practice supervision (e.g., collaborative practice agreements) and may be situated in settings with strong care coordination with primary or specialty care physicians, potentially reducing the likelihood of non-fatal overdose (Chapman et al., 2024; Lagisetty et al., 2017; Lapham et al., 2024; Wartko et al., 2023). It is also possible that physicians were transitioning less clinically complex patients with lower risk of overdose to APPs. However, the observed significant association may be driven by the examination of a rare outcome in the larger sample of APP patients relative to the smaller primary and specialty care physician samples. Nonetheless, across all groups, fewer than six percent of patients were predicted to have a non-fatal overdose after the first buprenorphine prescription, speaking to the effectiveness of buprenorphine.

Our study reiterates the need for additional resources for buprenorphine prescribers. Barriers to prescribing among primary care providers—comprising both physicians and many APPs—include lack of training, concerns about medication diversion, and stigma about buprenorphine and substance use (Harder et al., 2021; Kelley et al., 2023; Stone et al., 2021). Reductions in workload, protected time, and collaborative care models may address some of these aforementioned barriers (Austin et al., 2023; Kelley et al., 2023; Lapham et al., 2024; Lin et al., 2019; Wartko et al., 2023). In addition, resource offerings combining continuing education with addiction medicine consults and practice facilitation can increase buprenorphine prescribing capacity in primary care settings (Hall et al., 2025). This is particularly impactful in rural areas with decreased access to addiction medicine specialty care (Harder et al., 2021; Zittleman et al., 2022). Finally, Kentucky’s buprenorphine prescribing regulations—promulgated in 2015—stipulate rigid requirements that exceed requirements of federal law (Murphy, 2024), and place additional barriers on APPs prescribing buprenorphine. Federal policies including the 2018 SUPPORT Act (which included provisions that were implemented during this study period) and elimination of the X waiver in 2023 decreased prescribing barriers for all outpatient provider types and provided technical assistance and training. Future policy efforts should be consistent at the state and federal levels to ensure access to buprenorphine.

This study was subject to several limitations. First, our study focused on Kentucky, a state with a population that is predominantly White, non-Hispanic, has a high prevalence of OUD, and has a uniquely restrictive regulatory environment for buprenorphine. As such, our findings may not be generalizable to the general U.S. population with OUD or some states’ adult Medicaid populations with OUD. Second, as noted, our sample was nearly all White, non-Hispanic, limiting our ability to examine racial and ethnic differences in buprenorphine-related outcomes in primary versus specialty care. However, the demographics of our sample were reflective of the state of Kentucky, as its population was over 80 percent White, non-Hispanic in 2020 (Census Bureau, 2020). Third, our descriptive findings should not be interpreted as causal due to the potential for confounding. For example, we could not control for factors such as patient selection into different provider types, which includes factors like proximity to certain providers and other unobserved patient preferences. Much of our sample appeared to have been continued on buprenorphine from a prior year, suggesting a willingness to engage in treatment. Outcomes from the sample that continued buprenorphine from the prior year in comparison to those who initiated buprenorphine later in the study period may look different. This sample also had higher buprenorphine treatment retention across provider types compared to previous literature (Cooper et al., 2023; Xu et al., 2024), although there is significant variation in the literature regarding definitions of treatment retention. Fourth, we were precluded from including patients who had buprenorphine prescriptions that did not have an associated NPI, and NPPES data do not consistently list primary taxonomy codes for providers (n=5,250 dropped from the analysis). Fifth, we did not specifically account for certification in addiction medicine for provider types. As such, primary care providers in this study may have similar training to specialty care providers. Similarly, we could not distinguish specialties of physician assistants, and our analysis could have been strengthened if we were able to examine outcomes by primary care-practicing APPs from specialty care-practicing APPs. Sixth, we did not require continuous enrollment in Medicaid to maximize sample size and given the federal continuous Medicaid enrollment requirement from March 2020-March 2023 due to the COVID-19 pandemic. We also could not account for patients who were dually eligible for Medicare and Medicaid. Seventh, we did not have access to data on fatal overdose, and results using fatal overdose as an outcome may differ from our non-fatal overdose findings. Finally, data analyzed in this study took place during the COVID-19 pandemic and, therefore, may not be generalizable in the post-pandemic landscape, as some studies suggest prescribing environments differed during the pandemic (Lei et al., 2023).

Despite these limitations, our study is one of the first to descriptively examine buprenorphine prescribing and outcome differences across different provider types and settings of Medicaid enrollees in Kentucky, a state disproportionately impacted by the opioid epidemic. Extensions of this work should explore racial/ethnic and geographic differences in buprenorphine prescribing, as differences in buprenorphine access and prescribing among these two domains have been documented in Kentucky more broadly (Cooper et al., 2020; Myers & Borders, 2016). Longitudinal analyses of other physical and mental health-related outcomes among adults with OUD receiving buprenorphine from different provider types and in different settings are also needed. Finally, patient and provider experiences, therapeutic alliance, and other OUD-related outcomes that may vary by provider types in different care settings can be qualitatively explored.

Our study findings suggest that positive OUD treatment outcomes, such as increased retention and reduction in non-fatal overdose, are associated with receipt of buprenorphine from multiple provider types in different settings. Additional training and clinical resources, as well as reductions in workload and protected time, may improve buprenorphine treatment retention and related outcomes across providers. Finally, state and federal policies should be consistent and align with evidence-based clinical practice.