Community-level factors influencing the duration of buprenorphine treatment in individuals with opioid use disorder: a cohort study using US longitudinal claims data

Abstract

Introduction

Buprenorphine is effective in reducing opioid-related morbidity and mortality; however, many patients discontinue treatment prematurely. While previous research has focused on individual-level predictors of retention, the influence of community context remains underexplored. This study aims to examine how community-level factors, such as social vulnerability, availability of buprenorphine-waivered providers and access to behavioural health services, affect the duration of buprenorphine treatment episodes for individuals with opioid use disorder (OUD).

Methods

We conducted a retrospective cohort study using longitudinal claims data from 2006 to 2022. Adults aged 18 and older who initiated buprenorphine treatment for OUD and maintained continuous enrolment were included. Treatment episodes were defined by refill continuity, monitored until a gap of greater than 14 days occurred. Patient ZIP codes were linked to Social Vulnerability Index scores, provider density and behavioural health facility availability. The primary outcome was treatment duration, classified into short-term (0–3 months), medium-term (3–6 months), extended medium-term (6–12 months) and long-term (>12 months). Associations were identified using stepwise multinomial logistic regression.

Results

From the 131 169 individuals (303 528 buprenorphine treatment episodes), most of them were males (60.8%), aged 18–34 years (47.7%), and commercially insured (76.0%). For medium-term duration, individuals living in low-vulnerable areas (OR: 1.14 (1.36–1.47)), with high provider density (1.07 (1.02–1.12)) and high mental health service availability (1.20 (1.15–1.25)) were associated with longer retention. Effect size increased for extended medium-term durations, including low vulnerability (1.73 (1.67–1.80)) and high mental health service availability (1.31 (1.26–1.37)). In long-term treatment episodes, those living in low-vulnerable areas (1.95 (1.89–2.02)), high provider density (1.1 (1.06–1.15)) and high mental health services access (1.58 (1.52–1.64)) presented an increased effect size in the odds.

Conclusions

Social vulnerability, provider availability and mental health service access are significantly associated with buprenorphine treatment duration. Addressing these access inequities could improve retention for individuals with OUD.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Buprenorphine is an evidence-based medication for opioid use disorder (OUD) that reduces the overdose risk and improves long-term outcomes. However, early treatment discontinuation is common, and most studies focus on individual-level predictors of retention. The influence of community-level factors, such as social vulnerability, provider availability and access to behavioural health services, has been limitedly studied.

WHAT THIS STUDY ADDS

• Using national prescription claims data, including 303 528 buprenorphine episodes, this study showed that individuals living in communities with lower social vulnerability, higher buprenorphine-waivered provider density and greater mental health service availability presented significantly longer treatment durations. These associations remain after adjustment for demographics and clinical factors.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• The findings highlight the role of community-level context in the retention for buprenorphine treatment for individuals with OUD. Improving access to buprenorphine prescribers and expanding behavioural health services in socially vulnerable areas may enhance treatment retention and reduce inequities in OUD care.

Introduction

Opioid use disorder (OUD) continues to be a critical public health emergency in the USA, with significant health, economic and societal consequences. In 2023, more than 81 000 opioid-involved overdose deaths were reported, accounting for approximately 75% of all drug overdose deaths.¹ Despite data from 2024 showing a decrease in overdose deaths involving opioids,² opioid-related harms remain a problem. Emergency department visits for non-fatal opioid overdoses have continued to rise, with an increase of around 8% in February 2025.³ The total economic impact of the opioid crisis, including healthcare costs, lost productivity, criminal justice expenditures and premature mortality, is estimated to exceed $1.5 trillion annually.⁴ OUD is a chronic and relapsing condition characterised by continued opioid use, including both legal and illegal opioids, despite experiencing harmful consequences.⁵ In 2022, an estimated 6.1 million people aged 12 or older reported having an OUD.⁶ The burden of OUD is not equally distributed across communities and has disproportionately affected socioeconomically at-risk populations, contributing to widening health differences.^{7 8}

The US Food and Drug Administration has approved medications to treat OUD (MOUD), including buprenorphine, methadone and extended-release naltrexone. These treatments reduce the risk of overdose and improve long-term outcomes.9,11 Buprenorphine, a partial opioid agonist approved in 2002, is available in office-based settings and has become more routinely used in community-based treatment.¹² Randomised clinical trials and observational data demonstrate that buprenorphine is effective in reducing cravings, suppressing illicit opioid use and lowering mortality, particularly when delivered as part of a comprehensive treatment approach that includes behavioural health support.^{13 14}

Despite the clinical efficacy of buprenorphine, treatment retention remains a major challenge.¹⁵ Many patients discontinue therapy prematurely, often within the first 90 days of treatment, placing them at elevated risk for relapse, overdose and other adverse outcomes.^{16 17} Even in well-resourced clinical settings, 6-month treatment duration averages between 40% and 50%, highlighting the need to better understand and address barriers to sustained treatment engagement.18,20

Although predictors of treatment duration at the individual patient level, such as demographic characteristics, psychiatric comorbidity and co-occurring substance use, have been extensively studied in the association with MOUD duration, growing evidence suggests that structural and contextual factors also play an important role.21,25 Differences in treatment duration persist by race, ethnicity, sex and insurance status.^{26 27} Non-white patients and Medicaid enrolees are less likely to initiate and remain on MOUD, and women face unique barriers, including stigma, caregiving responsibilities and lack of gender-responsive care models.28,31

Beyond individual-level factors, community context, including availability of treatment providers, transportation infrastructure and the broader socioeconomic environment, may influence treatment adherence and outcomes.32,34 Yet, few studies have systematically examined how community-level vulnerability affects buprenorphine retention.^{35 36} The Social Vulnerability Index (SVI), developed by the Centers for Disease Control and Prevention (CDC), is a composite measure of community-level risk based on factors such as income, education, housing, transportation access and minority status.³⁷ When paired with data on provider density and service availability, the SVI may offer a more comprehensive understanding of barriers to MOUD continuity.

This study aims to examine the association between community-level contextual factors and buprenorphine treatment duration in individuals with OUD. Specifically, we assess how social vulnerability, buprenorphine-waivered provider availability and access to behavioural health services are associated with the duration of buprenorphine treatment episodes using a large national prescription claims database, primarily comprising individuals with private insurance in the USA. We hypothesise that patients residing in more socially vulnerable communities or areas with fewer treatment resources will have shorter treatment durations, independent of individual-level characteristics. Our study can inform targeted policies and interventions that improve the duration of buprenorphine treatment and reduce variations in OUD care.

Materials

Patient and public involvement statement

Patients and members of the public were not involved in the design, conduct, reporting or dissemination of this research. The study used fully deidentified, aggregated data that had been previously collected across the USA, with no direct contact with individuals or access to identifiable personal information.

Study design and data sources

We conducted a retrospective cohort study utilising IQVIA PharMetrics Plus for Academics Closed, a national database of longitudinal prescription and medical claims between 2006 and 2022. This database includes prescription claims (including market product names, quantities dispensed and days of supply) as well as medical claims associated with diagnosis and procedure codes. These data enable the identification of clinical encounters such as outpatient visits, psychiatric visits and behavioural health services. Additional information includes enrolment details collected from contributing health plans and demographics (age, sex, insurance type and ZIP code at the three-digit level)

IQVIA PharMetrics Plus for Academics Closed primarily consists of data from commercially insured patients. These patients are drawn from national, regional and self-insured employer health plans, which together make up the majority of the database. A small subset of Medicaid and Medicare beneficiaries is also included. However, these populations may be under-represented relative to their share in the US healthcare system overall, as the database was designed around commercial plan enrolment.

This study was reviewed and determined to be exempt by the Institutional Review Board of the University of Texas Health Science Center at Houston (HSC-SPH-23–1114) because it used deidentified data and involved no direct contact with human participants; therefore, informed consent was not required. No animals were involved in this study. The study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology reporting guidelines.

We incorporated data on buprenorphine-waivered providers and licensed mental health treatment facilities from the Substance Abuse and Mental Health Services Administration (SAMHSA).^{38 39} These data include the locations of providers offering medication-assisted treatment services as well as the facilities that provide therapy or psychotropic medications for mental health issues, across the USA. We obtained 2022 SVI from the US CDC.³⁷ The SVI is a validated composite measure of community vulnerability to social and environmental stressors, constructed from 15 census tract-level variables across four domains: (1) socioeconomic status, (2) household composition and disability, (3) minority status and language and (4) housing type and transportation. Each tract is assigned a percentile score ranging from 0 to 1, with higher scores indicating greater vulnerability.

Geographic data processing

Buprenorphine-waivered providers and mental health services

Because patient location data in IQVIA PharMetrics Plus for Academics Closed is available only at the three-digit ZIP code level, we aggregated provider-level data to match this geographic resolution. We extracted the first three digits from each provider’s five-digit ZIP code, grouped the providers accordingly and calculated the number of unique buprenorphine-waivered providers and mental health services in each three-digit ZIP region. We normalised these counts using population estimates from the American Community Survey⁴⁰ to obtain per-capita provider densities (number of providers per 10 000 population). Finally, we classified them into low, moderate-low, moderate-high and high mental health/buprenorphine waivered providers density according to the quantiles calculated for each group, respectively.

This approach follows recommendations from the US Department of Health and Human Services and SAMHSA to use ZIP code-based geolocation for assessing the mental health workforce. Aggregation to three-digit ZIP codes provides a regionally granular but stable unit of analysis, preserving the spatial resolution.

Social Vulnerability Index

To align SVI data with patient ZIP code geography, we aggregated census tract-level SVI values to three-digit ZIP codes using a population-weighted method. First, we linked census tracts to ZIP Code Tabulation Areas using crosswalks from the US Department of Housing and Urban Development.⁴¹ We then weighted each tract’s SVI based on the number of residential addresses in its corresponding ZIP code. Weighted SVI values were aggregated to five-digit ZIP codes and subsequently grouped by three-digit ZIP region. We calculated the population-weighted average SVI for each three-digit ZIP code, rescaled to be on the range of 0 to 1 and classified SVI into four levels of vulnerability according to the quantiles: low, low–medium, medium–high and high.

Study population and cohort identification

We included individuals aged 18 years and older who initiated buprenorphine treatment between 2006 and 2022. Buprenorphine prescriptions were identified using (1) market product names (see online supplemental Figure S1) and (2) the Uniform System of Classification code 78340.²³ This included both sublingual formulations (99.65% of prescriptions) and a small proportion of long-acting injectable formulations (0.35%). To ensure complete treatment trajectories, we required a minimum of 12 months of continuous enrolment, as determined by IQVIA PharMetrics Plus for Academics enrolment data. In terms of prescription information, we replace missing days of supply for the median days of supply for the other prescriptions and cap the maximum days of supply to 90 days.²³ We also excluded individuals with missing demographic or residential information (age, sex, insurance type or ZIP code).

Measurements

Buprenorphine treatment episode

We defined the start of a new treatment episode as the date when a buprenorphine prescription was first filled, and considered the episode to be discontinued if there was a gap of 14 days or more between the end of the days’ supply of one prescription and the fill of a subsequent prescription.²³ This refill-based definition follows previous literature examining treatment continuity using prescription claims data.

Buprenorphine treatment duration

The primary outcome was buprenorphine treatment duration, defined as the length of each continuous treatment episode. We classified episodes into four categories: (1) short-term (0 to 3 months); (2) medium-term (3 to 6 months); (3) extended medium-term (6 to 12 months) and (4) long-term (more than 12 months). These categories reflect common retention thresholds used in OUD and substance use disorder (SUD) treatment research and are aligned with clinical benchmarks for sustained treatment engagement.⁴²

Explanatory variables

Explanatory variables included both individual-level demographic characteristics and community-level contextual factors. Patient-level demographic information included sex (male or female) and insurance type (private insurance, Medicaid, Medicare or self-pay/uninsured). Insurance type was determined based on the primary payer listed for the first buprenorphine prescription during the treatment episode. For each patient, we also recorded their prior history of mental health conditions and SUDs during the 12 months before initiating buprenorphine treatment. This was based on the International Classification of Diseases, 9th and 10th Revision (ICD-9 and ICD-10) codes, which include diagnoses for alcohol use disorder, non-OUD, post-traumatic stress disorder (PTSD), depression, anxiety, schizophrenia and bipolar disorder.²² ICD-10 codes were mapped to ICD-9 codes using the Center for Medicare and Medicaid Services general equivalence mappings.⁴³ Additionally, we used Current Procedural Terminology (CPT) and Healthcare Common Procedure Coding System (HCPCS) codes to identify health service utilisation, such as outpatient medical services for general evaluation and patient management, outpatient psychiatric services addressing mental health and buprenorphine counselling, telehealth services and buprenorphine treatment services from 12 months before initiating buprenorphine treatment through 21 months after initiation of buprenorphine treatment.⁴⁴ A complete list of codes is found in online supplemental Table S1 and S2.

Community-level variables included SVI, buprenorphine-waivered provider density and mental health service availability. Buprenorphine-waivered provider density was calculated as the number of unique waivered prescribers per 10 000 residents in each three-digit ZIP code. Mental health service density was calculated as the number of licensed mental health and substance use treatment providers per 10 000 population.

Statistical analysis

We used a stepwise multinomial logistic regression model to estimate the association between community-level and individual-level factors and buprenorphine treatment duration. This was done to evaluate different combinations of covariates and choose a model based on the lowest Akaike information criterion (AIC) value, reflecting the best trade-off between model fit and complexity (see online supplemental Table S3).⁴⁵ Covariates included SVI category, buprenorphine-waivered provider density, mental health service availability, sex, age at buprenorphine treatment episode, insurance type, health service utilisation indicators and the presence of mental health diagnoses.

To assess model fit and reduce multicollinearity, we conducted a pairwise correlation analysis to evaluate associations between categorical predictors.⁴⁶ Variables with high correlation were reviewed for potential exclusion or consolidation. We used likelihood ratio tests to evaluate the contribution of each predictor to model performance, with a significance threshold of p<0.05 (see online supplemental Table S4). All statistical analyses were conducted using R software V.4.3.1.

Results

Descriptive statistics

In our cohort, 131 169 individuals (303 528 treatment episodes) had at least one buprenorphine episode related to OUD (see table 1). Among these individuals, 60.8% were male and 39.2% were female. The majority were aged between 18 and 34 years (47.7%) and resided in the Southern region of the US (34.2%) during their first buprenorphine episode. Regarding insurance coverage, most patients were enrolled in commercial insurance (76.0%), while 14.7% were covered by Medicaid. Individuals living in areas with low-medium social vulnerability represented the largest group in the cohort (32.2%). Similar proportions were observed for the individuals residing in areas with a high density of buprenorphine-waivered providers (33.2%). On the contrary, those living in areas with a moderate-low density of mental health services comprised the highest proportion in the cohort (32.5%). Regarding co-occurring diagnoses, a significant number of individuals had a history of depression (22.3%) and anxiety (22.6%) within 12 months prior to their first buprenorphine prescription. Furthermore, most individuals reported having outpatient visits (90.6%) and psychiatric consultations (19.4%). As shown in online supplemental Figure S2, the heatmap of correlations among covariates included in the multinomial logistic regression model indicates low levels of correlation. All pairwise correlation coefficients were below 0.5, with the majority under 0.1, suggesting no evidence of multicollinearity.

Table 1. Characteristics at the individual and episode level of the individuals with at least one buprenorphine episode^*.

Characteristics	Number of individuals/episodes (%)
	Individual level (%)	Episode level (%)
Total	131 169	303 528
Sex
Male	79 743 (60.8%)	188 601 (62.1%)
Female	51 426 (39.2%)	114 927 (37.9%)
Region of residence
East (E)	22 388 (17.1%)	55 188 (18.2%)
Midwest (MW)	35 610 (27.1%)	71 766 (23.6%)
South (S)	44 882 (34.2%)	102 673 (33.8%)
West (W)	28 289 (21.6%)	73 901 (24.3%)
Age at episode
18–34	62 583 (47.7%)	134 272 (44.2%)
35–54	52 755 (40.2%)	128 411 (42.3%)
55–85	15 831 (12.1%)	40 845 (13.5%)
Insurance type
Commercial	99 735 (76.0%)	244 828 (80.7%)
Medicaid	19 277 (14.7%)	34 051 (11.2%)
Medicare	11 973 (9.1%)	24 289 (8.0%)
Self-insured	184 (0.1%)	360 (0.1%)
SVI category
Low	17 506 (13.3%)	36 762 (12.1%)
Low–medium	42 281 (32.2%)	93 440 (30.8%)
Medium–high	38 118 (29.1%)	86 864 (28.6%)
High	33 264 (25.4%)	86 462 (28.5%)
Buprenorphine treatment density group
Low provider density	16 333 (12.5%)	39 002 (12.8%)
High provider density	43 483 (33.2%)	96 350 (31.7%)
Moderate–high provider density	38 569 (29.4%)	89 721 (29.6%)
Moderate–low provider density	32 784 (25.0%)	78 455 (25.8%)
Mental health and substance use services density group
Low mental health services density	27 600 (21.0%)	70 601 (23.3%)
High mental health services density	29 673 (22.6%)	62 011 (20.4%)
Moderate–high mental health services density	31 286 (23.9%)	69 087 (22.8%)
Moderate–low mental health services density	42 610 (32.5%)	101 829 (33.5%)
Schizophrenia diagnosis	1039 (0.8%)	1885 (0.6%)
Bipolar disorder diagnosis	7424 (5.7%)	15 577 (5.1%)
Anxiety disorder diagnosis	29 588 (22.6%)	63 281 (20.8%)
PTSD diagnosis	4090 (3.1%)	7807 (2.6%)
HIV diagnosis	324 (0.2%)	719 (0.2%)
Depression diagnosis	29 306 (22.3%)	64 339 (21.2%)
Outpatient visits	118 782 (90.6%)	276 052 (90.9%)
Psychiatric visits	25 472 (19.4%)	69 276 (22.8%)
Telehealth services	3061 (2.3%)	5430 (1.8%)
Induction and maintenance treatment	2143 (1.6%)	5767 (1.9%)
Buprenorphine treatment services	1188 (0.9%)	2037 (0.7%)

^*All variables were derived from adjudicated medical and pharmacy claims using diagnosis (ICD) and procedure (CPT/HCPCS) codes. Buprenorphine treatment episodes were constructed from prescription claims identified by market product names and the Uniform System of Classification (USC) code 78340. Additional details on the construction of explanatory variables are provided in the Methods section, with the complete list of codes available in the online supplemental material

PTSD, post-traumatic stress disorder; SVI, Social Vulnerability Index.

Multinomial regression model results

We used a stepwise multinomial regression and selected the best model based on the AIC. The results from the models are in online supplemental Figure S3–S14. ORs and 95% CIs were estimated for all three groups: medium-term, extended medium-term and long-term treatment episodes, with short-term as the reference group. In the group of individuals with medium-term buprenorphine episode duration (see figure 1), community-level factors and healthcare access factors showed significant associations with treatment durations. Individuals residing in areas of low (OR (95% CI) 1.41 (1.36 to 1.47)), low–medium (1.35 (1.31 to 1.39)) and medium–high vulnerability (1.18 (1.14 to 1.21)) were significantly associated with higher odds of treatment retention compared with those residing in areas with high social vulnerability. High buprenorphine-waivered provider density (1.07 (1.02 to 1.12)), high and moderate–high density of mental health services (1.2 (1.15 to 1.25); 1.11 (1.07 to 1.16)) were associated with an increase in the odds of being in the group of medium-term duration compared with the shorter term. Compared with commercially insured individuals, those with Medicaid had higher odds of medium-term retention (1.21 (1.17 to 1.26)), and Medicare recipients had slightly higher odds (1.06 (1.01 to 1.11)). Self-insured patients had higher odds (1.71 (1.29 to 2.27)), suggesting longer retention among this group. Substance use, such as alcohol and non-opioid drugs, (0.90 (0.86 to 0.95); 0.93 (0.90 to 0.95)), as well as co-occurring schizophrenia (0.81 (0.70 to 0.93)), was significantly associated with lower odds of having longer treatment duration. Healthcare utilisation was significantly associated with treatment duration, with individuals having outpatient visits (1.08 (1.04 to 1.12)) and individuals having induction and maintenance treatment (1.13 (1.05 to 1.22)) presenting higher odds of having medium-term durations.

Figure 1. Forest plot of adjusted ORs and 95% CI for medium-term versus short-term buprenorphine treatment duration. PTSD, post-traumatic stress disorder.

Compared with medium-term duration, extended medium-term shows similar trends with increased effect sizes for most of the significant predictors for medium-term duration (see figure 2). Social vulnerability was significantly associated with longer treatment durations, with individuals in low (1.73 (1.67 to 1.80)), low–medium (1.56 (1.51 to 1.61)) and medium–high (1.32 (1.28 to 1.36)) social vulnerability areas had significantly higher odds of extended medium-term retention compared with high social vulnerability counterparts. There was an increase in the effect size for high buprenorphine-waivered providers when the treatment duration increased to extended-medium term, significantly increasing the odds of having longer treatment durations (1.13 (1.08 to 1.18)). Similar patterns were observed for the mental health service density, with both high and moderate–high mental health services being associated with increased odds of having longer treatment durations (1.31 (1.26 to 1.37); 1.15 (1.10 to 1.20)).

Figure 2. Forest plot of adjusted ORs and 95% CI for extended medium-term versus short-term buprenorphine treatment duration. PTSD, post-traumatic stress disorder.

Additionaly, in the extended medium-term duration, new significant covariates are identified, with individuals diagnosed with depression (0.93 (0.90 to 0.96)), anxiety (0.95 (0.92 to 0.98)) and bipolar diagnosis (0.91 (0.86 to 0.96)) were associated with lower odds of having longer treatment durations, while receiving telehealth services was associated with higher odds (1.19 (1.09 to 1.29)). The association with induction and maintenance treatment reversed, now linked to lower odds of longer treatment (0.90 (0.82 to 0.98)). Compared with shorter treatment durations, individuals enroled in Medicare and Medicaid and those self-insured had higher odds of remaining in treatment for longer periods. These associations remained significant and became stronger with increasing treatment duration, suggesting a cumulative influence of structural and individual-level factors on extended retention. With long-term durations, the ORs for social vulnerability, buprenorphine-waivered provider density and high mental health service density all increased progressively (see figure 3). In the same way, healthcare utilisation exhibited stronger associations, with individuals using telehealth services having greater odds of longer treatment duration (1.31 (1.22 to 1.40)). Other significant covariates are identified, with individuals diagnosed with PTSD (0.89 (0.83 to 0.96)) and those who had psychiatric visits (0.89 (0.87 to 0.92)) were associated with lower odds of having longer treatment durations. Additionally, as treatment duration increases, worse retention is observed in those individuals enroled in Medicaid compared with individuals enroled in commercial insurance, with Medicaid enrollees having lower odds of longer treatment durations (0.88 (0.85 to 0.91)). These findings suggest that as treatment duration increases, differences associated with insurance type, sex, co-occurring psychiatric condition diagnosis and healthcare utilisation involvement become more pronounced, while the influence of contextual structural factors intensifies.

Figure 3. Forest plot of adjusted ORs and 95% CI for long-term versus short-term buprenorphine treatment duration. PTSD, post-traumatic stress disorder.

Discussion

This retrospective cohort study utilised US national claims data, primarily composed of commercially insured patients, to investigate how community-level contextual factors and individual characteristics affect the duration of buprenorphine treatment among individuals with OUD. Our research found that social determinants of health, represented by the SVI, along with aspects of the healthcare system such as the number of providers and availability of mental health services, play a significant role in treatment duration. Specifically, individuals in communities with lower social vulnerability (ie, better socioeconomic status, greater transportation, etc) and better access to healthcare services (ie, greater availability of healthcare services such as mental health, buprenorphine-waivered providers, etc) tended to remain on buprenorphine longer. This pattern indicates an association where challenges such as economic instability, limited-service availability and geographic obstacles may contribute to hinder long-term treatment for those most in need, although these mechanisms were not directly examined in this study.

Individuals living in areas with lower social vulnerability, marked by greater economic stability and better access to services, were more likely to remain in buprenorphine treatment for longer periods. When comparing medium-term to long-term treatment, the effect size increased by approximately 38%. These findings suggest that structural barriers, such as transportation challenges, provider shortages and concentrated poverty, may disrupt long-term engagement with buprenorphine therapy, even when clinical need is present.³⁶

We also found that having more buprenorphine-waivered providers nearby was associated with longer treatment durations. Patients in regions with a higher number of prescribers tended to stay engaged in their care, as the increase in the effect size by approximately 2.80% from medium-term to long-term shows. One possible explanation is that they faced fewer barriers in booking appointments, experienced reduced travel burdens and benefited from stronger continuity of care. These findings support previous research indicating that the geographic accessibility of medications for MOUD affects retention.^{47 48} Our results reinforce prior evidence that expanding the number of buprenorphine prescribers, especially in underserved regions, may be an important strategy to help close gaps in treatment access.^{49 50}

Furthermore, having robust mental health and substance use services in a community was associated with patients remaining in treatment. Individuals in areas with a strong mental health framework, such as access to psychiatrists, licensed counsellors and diverse treatment options, tended to maintain their buprenorphine use for more than 6 months, as shown by the 32% increase in effect size from medium-term to long-term duration. This supports earlier evidence suggesting that integrated care for mental health and SUDs leads to better outcomes.⁵¹ In addition to eliminating structural barriers, increasing the availability of healthcare providers may help foster stronger therapeutic alliances, which are supportive and collaborative relationships between patients and clinicians. Previous studies have shown that patients who establish a strong therapeutic alliance early in treatment experience greater reductions in psychological distress and improved adherence to treatment.^{52 53} A comprehensive approach to managing OUD may depend not only on access to services but also on the opportunity to build and sustain these alliances, which can alleviate the burden on patients and enhance the likelihood of long-term engagement.

We also noted differences based on insurance type and gender. Within this predominantly commercially focused database, individuals with Medicaid had shorter treatment durations compared with those with private insurance, indicated by a 38% reduction in effect size from medium to long-term treatment duration. While this aligns with prior research showing challenges faced by Medicaid recipients, including fewer providers willing to accept their insurance or stricter authorisation processes,54,56 these results should be interpreted cautiously, given the limited representation of Medicaid enrollees in our cohort. Additionally, female individuals were less likely to remain in treatment beyond 3 or 6 months, maintaining a consistent effect size from medium-term to long-term treatment duration. This is consistent with the existing literature, indicating that women face distinct challenges in treatment access and adherence, such as caregiving responsibilities, stigma and lack of gender-responsive care environments.57,59

Together, these findings emphasise the importance of community context in shaping OUD treatment outcomes. Our study extends prior research by demonstrating that social vulnerability, mental health capacity and insurance type are significant predictors of MOUD retention across a national population using claims data.

These insights have meaningful implications for clinical practice and health policy. First, increasing the number of buprenorphine-waivered providers in high-SVI areas could be crucial for improving individual retention. Policymakers should consider initiatives to attract providers to socially vulnerable communities. Second, investing in integrated mental health services could foster long-term engagement, particularly for those with co-occurring conditions. Third, there is a need to further explore the insurance-related challenges faced by Medicaid patients, such as coverage limits and reimbursement rates, which can disrupt continuity of care. Finally, our findings suggest that addiction care needs to be more responsive to gender differences. Strategies such as flexible appointment times, on-site childcare and programmes focused on women could help mitigate the challenges women face in maintaining buprenorphine treatment.

While the majority of prescriptions in our dataset were for sublingual formulations (99.65%), we also captured a small proportion of long-acting injectable and implantable buprenorphine products, such as Sublocade and Bunavail. Although their contribution to our analysis was minimal due to limited uptake in the US during the study period, these formulations represent an important and growing development internationally. Long-acting depot formulations and implantable options can reduce the burden of daily dosing and may improve adherence, patient autonomy and quality of life.60,63 Evidence from Europe and Australia suggests that these formulations can expand therapeutic choices and may help mitigate inequities in treatment access, particularly in healthcare systems constrained by budgetary pressures.64,66 In the US, the adoption of long-acting injectables has been slower, but early research indicates that they may offer unique advantages for patients in rural or resource-limited areas, where continuity of care is challenging.⁶⁷ These developments align with our findings that community-level vulnerability and provider availability strongly influence treatment retention. Future research should examine how long-acting buprenorphine interacts with social and healthcare system factors to improve treatment outcomes.

This study should be interpreted in light of several limitations. First, the analysis relies on administrative claims data, which lack detailed clinical context, including provider reasoning for treatment discontinuation, patient preferences or non-billable services such as counselling or peer support. Second, although we adjusted for psychiatric comorbidities and insurance type, the dataset lacked reliable data on race, ethnicity, housing status and other individual-level social determinants of health that may influence treatment retention. These unmeasured factors could contribute to the observed variation in treatment duration. Third, the IQVIA PharMetrics Plus for Academics Closed database primarily consists of data from commercially insured individuals, with only limited representation of Medicaid and Medicare enrollees. Therefore, our findings are most representative of treatment patterns in commercially insured populations and may not fully apply to groups that rely more significantly on public insurance programmes.

In conclusion, in this study, we found that social vulnerability, provider availability and mental health service access are significantly associated with buprenorphine treatment duration. Patients in more vulnerable communities or areas with limited healthcare resources were less likely to remain in treatment. Addressing geographic inequities in the distribution of providers and behavioural health services may enhance treatment retention and reduce inequities in outcomes for individuals with OUD.

Data availability statement

No data are available.