(A-219) Characterization of Healthcare Providers of a Cohort of Patients from Claims Data Based on Propagated Information from Linkage to Patient Encounters

Background: The healthcare providers (HCPs) characteristics in claims data studies of patient cohorts are typically limited to only specialty, clinic size, and broad geographic area. Furthermore, some specialties or sub-specialties are not accurately captured on the claims.
Data on the HCPs’ encounters with their patients outside of the original study cohort can provide insights on the HCPs’ patient composition and practice patterns.

Objectives: To infer likely sleep specialty and describe the practice of providers of patients with rare sleep conditions using linkage to their other patients and encounters.

Methods: The study identified a cohort of individuals with a diagnosis of narcolepsy (NT) or IH, rare sleep conditions, from Optum’s Market Clarity database from January 2015 to December 2021 (N=68,328), and the HCPs who managed them (N=67,706). Stratified sampled 1,000 HCPs, including sleep and other specialties, were further characterized by linked data from all their other patients’ encounters over a year (N= 91,022 patients). Diagnoses, medications, and tests provided insight on the patient composition and practice of the HCPs; counts and proportions of types of encounters and diagnoses of patients were summarized.
Under-recognition of sleep specialists by the billing taxonomy was addressed by a heuristic of identifying providers seeing majority sleep patients (>50% of patients with a sleep condition as a primary diagnosis), “likely-sleep specialists”.

Results: The HCP cohort included sleep specialists (N=270), neurologists (N=178), and other specialties (N=541 of internal medicine, pulmonology, psychiatry, and cardiology). A total of 319 HCPs were likely-sleep specialists.
Commonly diagnosed conditions included depression, anxiety, diabetes, metabolic disorders, and cardiovascular disease, which were more frequently diagnosed by non-sleep specialists. Obstructive sleep apnea, obesity, hypertension, hypersomnia and insomnia were common and more frequently diagnosed by sleep specialists.
The “likely-sleep specialists” group was very similar to the sleep specialists group.
NT/IH patients accounted for a median of 2% of all patients with medical claims among sleep specialists. A pattern of sub-specialty within specialty emerged from the distribution of the number of NT/IH patients seen by sleep specialists.

Conclusions: Information on the HCPs of patients can be derived from linked patient encounters in claims data and enable inference on the HCPs’ specialty and practice. The study demonstrated how data linkage could be leveraged for additional characterization, identification, and categorization, helping to address common limitations of claims-based studies.