The Last Mile to First Treatment: Search for Opioid Use Disorder Medication

• In Event: Policies related to Opioid Use and Health Outcomes

Abstract

The opioid crisis remains a significant public health challenge. Buprenorphine is a medication that reduces cravings and withdrawal symptoms for opioid use disorder (OUD). Despite its effectiveness, around 87% of patients who could benefit from buprenorphine remain untreated (Kyrawczyk et al., 2022). A commonly overlooked reason for this treatment gap is limited pharmacy-level access to buprenorphine. Even patients who overcome barriers to treatment—seeking care, obtaining prescriptions—often struggle to fill their prescriptions due to pharmacy availability. Using novel administrative insurance data from Washington State, I identify patients diagnosed with opioid use disorder who received buprenorphine prescriptions and track whether these prescriptions were subsequently filled. By linking this insurance data to detailed pharmacy shipment records, I can determine if buprenorphine was available at each patient's pharmacy following their doctor's visit. When buprenorphine is available at a patient's "default" pharmacy (the one most frequently visited in a given year), only 30% fail to fill their prescriptions—a rate comparable to other medications. However, if buprenorphine is unavailable at this pharmacy, the rate of unfilled prescriptions jumps to 80%.

If patients were willing to actively search for pharmacies with buprenorphine availability, limited availability would pose only a minor issue. To understand why patients rarely search beyond their default pharmacy,  I develop a structural model of patient sequential search behavior, building on Choi et al. (2018) and Moraga-Gonzalez et al. (2023). I leverage variation arising from situations where patients deviate from their default pharmacy. If a patient obtains buprenorphine elsewhere, it indicates at least one additional search, allowing me to identify preferences and search costs separately. Importantly, I allow search costs to vary based on patients' beliefs about pharmacy availability. The estimated model demonstrates that uncertainty about pharmacy availability strongly influences whether patients decide to search, indicating that uncertainty, rather than pharmacy availability alone, creates significant barriers.

Using this model, I evaluate two policy scenarios. First, reducing uncertainty about buprenorphine availability at pharmacies—without ensuring universal stocking—would increase treatment rates by approximately 45%, achieving roughly 72% of the benefits of universal stocking at significantly lower cost. Second, the current shift toward electronic prescribing (e-prescriptions) for controlled substances increases the search cost—patients must revisit prescribers to modify prescriptions if their initial pharmacy lacks buprenorphine—but simultaneously reduces uncertainty about availability. The model finds that the benefits of reduced uncertainty substantially outweigh the increased search costs. For search costs to offset these benefits, they would need to increase by threefold, an improbable scenario. Thus, expanding e-prescribing is likely to substantially improve treatment uptake.

Author

• Lance Gui, University of Arizona
  Presenting Author