2059-P: Provider Data Quality in Medicare Advantage vs. Fee-for-Service among Adults Newly Diagnosed with Type 2 Diabetes

Introduction and Objective: Incomplete provider data in Medicare Advantage (MA) claims can reduce provider attribution accuracy, which is important for network analyses and population health efforts. This study examines the extent and patterns of missing provider information in MA claims compared with Medicare Fee-for-Service (FFS) among beneficiaries with diabetes. Methods: Using the Arkansas All-Payer Claims Database (APCD), we identified beneficiaries newly diagnosed with type 2 diabetes and extracted all medical claims from two months before through one year after diagnosis for MA and FFS populations (2015-2019). Claims were classified using Restructured BETOS classification system, which group HCPCS codes into clinically meaningful categories. Results: For MA beneficiaries (n=8,334), we identified 116,256 outpatient E/M claims, of which 3,394 (2.9%) had missing NPIs. Using provider name and ZIP code available in the APCD provider file, 634 (19%) could be matched to identify an NPI; most others lacked provider and organization names. After NPI assignment, 915 claims lacked specialty codes, and 88% were resolved using taxonomy and ZIP linkage. For the Medicare FFS cohort (n=17,425), we identified 66,011 outpatient E/M claims and 272,068 carrier E/M claims. Attending NPIs and performing physician NPIs were available for 100% of outpatient claims and carrier claims, respectively. Unlike MA, 100% of FFS claims had individual-level NPIs, while MA claims were 83% individual and 17% organization NPIs. Conclusion: Though the vast majority of MA claims have NPIs or provider information to identify NPIs, these claims are not as complete as FFS claims. These data quality issues affect provider attribution accuracy for diabetes care and highlight the need for improved provider data reporting standards within MA plans. Disclosure M. Rezaeiahari: None. J. Henske: None. B.C. Martin: Other - I receive royalties from Trestle Tree LLC for the commercialization of an opioid risk prediction too which is unrelated to this investigation.; Current; Trestle Tree LLC. K. Fujimoto: None. M. Bhai: None. M.L. Williams: None. Funding National Institute of Diabetes and Digestive and Kidney Diseases of the National Institute of Health [K25DK136966], Access to the Arkansas All-Payer Claims Database for this study was provided by the support from the Arkansas Biosciences Institute/Arkansas Insurance Department/Arkansas Healthcare Transparency Initiative Collaboration.