Network Analysis of Medical Claims Data Suggests Network-Based, Regional Targeting and Intervention Delivery Strategies to Increase Access to Office Based Opioid Treatment (OBOT) for Opioid Use Disorder (OUD)

Abstract

Opioid overdose and Opioid Use Disorder (OUD) statistics underscore an urgent need to significantly expand access to evidence-based OUD treatment. Office Based Opioid Treatment (OBOT) has proven effective for treating OUD. However, limited access to these treatments persists. Recognizing the need for significant investment in clinical, behavioral, and translational research, the Indiana State Department of Health and Indiana University embarked on a research initiative supported by the “Responding to the Addictions Crisis” Grand Challenge Program. This brief presents recommendations based on existing research and our own analyses of medical claims data in Indiana, where opioid misuse is high and treatment access is limited. The recommendations cover target providers, intervention focus, priority regions, and delivery methods.

• What do we already know about this topic?

• Although prescription buprenorphine for OUD treatment is very effective, OBOT remains underutilized.

• How does this research contribute to the field?

• This brief presents recommendations from previous research and our own network analysis of medical claims data designed to identify strategies that can effectively increase OBOT prescribing behavior among providers.

• What are this research’s implications toward theory, practice, or policy?

• To expand access to evidence-based OUD treatment, it is essential to target providers in high-need communities based on the patients they treat, activate peer connections with OBOT prescribers and leverage those connections for intervention delivery. Using up-to-date empirical data, implementing network-based/peer strategies, and concentrating efforts on areas of greatest need, we can optimize the provision of OBOT and improve outcomes for individuals with OUD.

Introduction

Opioid Use Disorder (OUD) statistics underscore an urgent need to significantly expand access to evidence-based OUD treatment in the United States.^1,2 Office Based Opioid Treatment (OBOT), involving the prescription of buprenorphine or buprenorphine/naloxone, has proven effective. However, limited access to these treatments persists, as many providers choose not to offer them despite clinical evidence of effectiveness. A recent study found that 86.6% of individuals diagnosed with OUD did not receive such medications. ³

Background

Although U.S. physicians were authorized to prescribe buprenorphine for OUD treatment in 2000, and this authority was expanded to other qualifying prescribers in 2018, OBOT remains underutilized. ⁴ Several factors contribute to this underutilization, including perceptions of low treatment need, lack of provider familiarity with prescribing requirements and treatment regimes, professional attitudes toward medications for substance dependence, and social and cultural stigmas associated with addiction. ⁵ Consequently, there is a limited number of OBOT providers in the United States, with most concentrated in urban areas on the east and west coasts.^{6 -9}

Objective

Recognizing the need for significant investment in clinical, behavioral, and translational research, the Indiana State Department of Health and Indiana University embarked on a research initiative supported by the “Responding to the Addictions Crisis” Grand Challenge Program. This brief presents recommended strategies that can effectively increase OBOT prescribing behavior among providers based on previously published research as well as our own analyses of medical claims data in Indiana, where opioid misuse is high and treatment access is limited.^{10 -12} The recommendations we propose are supported by our own and previously published research and cover target providers, intervention focus, priority regions, and delivery methods.

In these and other related studies, we have used claims data to generate provider networks based on shared patients, extract provider characteristics including summary information about their patients, integrate data regarding regional characteristics related to public health and public opinion to assess factors related to provider behavior at multiple levels with a primary focus on provider networks and provider characteristics.^{13 -16} Network connections inferred from shared patients replicate provider communication patterns and are predictive of diffusion and uptake of innovative medical treatments.^{17 -19}

In multi-level logistic regressions across 4 quarters of medical claims data in 2019 to 2020 (n = 8022 Indiana prescribers), we found that having at least one connection to another OBOT prescriber (OR = 3.69, P < .001), having more patients (OR = 1.62, P < .001), having more high-risk patients (defined by average daily morphine milligram equivalents, diagnosis of OUD, or non-fatal overdose; OR = 1.82, P < .001), and practicing in a county with more federally identified buprenorphine treatment programs (OR = 1.15, P < .001) increased the odds a provider prescribed OBOT; having more cancer patients (OR = 0.12, p =< .001) and practicing in a county with a methadone clinic (OR = 0.22, P < .05) decreased the odds a provider prescribed OBOT. In rare event logistic regressions across the same 4 quarters of medical claims data (n = 6974 Indiana providers), we found that having at least one connection to another OBOT prescriber (OR = 2.89, P < .001), having more patients (OR = 1.85, P < .001), and having more high risk patients (OR = 1.57, P < .001) increased the odds a provider would begin prescribing OBOT; having more shared-patient connections to other providers decreased the odds a provider would begin prescribing OBOT (OR = 0.50, P < .05).

Recommendations

1. Targeting Providers. Interventions should primarily target providers who have the largest number of high-risk patients identified based on factors such as high average daily opioid intake, non-fatal overdoses, or OUD diagnoses.^13,14 Additionally, interventions should prioritize providers who have at least one colleague already providing OBOT. Qualitative research on medications delivered within similar stigmatized contexts^15,16,20,21 and our own analyses of claims data ¹³ indicate that having even one colleague familiar with the processes and policies can influence a provider’s decision to initiate or maintain OBOT provision.

2. Intervention Focus. Interventions should address not only treatment recommendations and logistics but also community and professional stigma related to individuals with OUD.^{22 -24} It is crucial to communicate up-to-date information on local OUD prevalence and OBOT utilization, while providing clear and unbiased advice for diagnosing OUD ²⁵ and discussing OBOT with patients. ²⁶

3. Priority Regions. While all providers should be familiar with OBOT, resource constraints necessitate targeted intervention. Priority should be given to geographic regions with the highest need, based on indicators such as opioid use, OUD prevalence, and non-fatal overdose rates. Instead of relying solely on existing administrative boundaries, our research suggests a data-driven network-based approach could be employed to identify communities of providers prescribing to the same set of patients. ¹⁴ Analyzing provider networks in this way can determine high-risk areas more effectively and thus better allocate resources. ²⁷

4. Delivery Methods. Providers in this and other contexts where highly effective and innovative pharmaceutical treatments have been introduced have expressed a preference for content delivered by knowledgeable and respected peers rather than pharmaceutical industry representatives or government sources.^{15,16,20,21,28,29} Therefore, we recommend providing Continuing Medical Education sessions facilitated by peers or implementing a peer mentoring process to offer feedback and support. This approach can create or strengthen connections to existing OBOT providers, fostering peer change agents within communities.

Conclusion

To expand access to evidence-based OUD treatment, it is essential to target providers based on patient profiles, activate connections with OBOT mentors, engage well-connected OBOT prescribers as change agents, and leverage peers for intervention delivery. Using up-to-date empirical data, implementing network-based strategies for targeting providers, and concentrating efforts on areas of greatest need bounded using network approaches, we can optimize the provision of OBOT and improve outcomes for individuals with OUD. ³⁰ Future studies will employ simulation approaches that model these recommendations^11,31,32 to facilitate rapid and effective implementation of interventions with a high probability of increasing OBOT prescribing and reducing opioid related deaths.