Approaches to Identify Nursing Home Specialists Using Medicare Claims Data

• Published in: Medical care Vol. 63; no. 7; p. 520
• Main Authors: Riester, Melissa R, Ryskina, Kira L, White, Elizabeth M, Hayes, Kaleen N, Harris, Daniel A, Zullo, Andrew R
• Format: Journal Article
• Language: English
• Published: United States 01.07.2025
• Subjects: Aged; Cross-Sectional Studies; Female; Humans; Insurance Claim Review - statistics & numerical data; Male; Medicare - statistics & numerical data; Medicare Part B - statistics & numerical data; Medicare Part D - statistics & numerical data; Nursing Homes - statistics & numerical data; Physicians - statistics & numerical data; Specialization - statistics & numerical data; United States; United States; Medicare Part B; older adults; Medicare Part D; nursing homes
• ISSN: 1537-1948
• DOI: 10.1097/MLR.0000000000002161

Physicians and advanced practice clinicians who practice in nursing homes (NHs) are becoming increasingly specialized. Studies have identified clinicians as NH specialists using multiple data sources; yet, researchers' access to several sources may be limited due to required data purchases. Examine the concordance of 2 approaches to measure NH specialization versus a standard approach using clinician-level Medicare Data on Provider Practice and Specialty (MD-PPAS). These alternative approaches leveraged: (1) publicly available clinician-level Medicare Part B data; and (2) patient-level Medicare Part D Event claims linked to publicly available clinician-level Medicare Part D prescribers data. Yearly cross-sections from 2016 to 2020. Physicians and advanced practice clinicians with at least one Medicare-paid service to NH residents and at least 100 total services in a given year. Nursing home specialists were classified as clinicians with ≥90% of annual services provided to NH residents. Between 2016 and 2020, NH specialists comprised 49,542 of 321,267 eligible clinician-years (15.4%) in MD-PPAS data; 35,983 of 189,992 eligible clinician-years (18.9%) in Part B data; and 31,148 of 1,101,484 eligible clinician-years (2.8%) in Part D data. Compared with the MD-PPAS approach, the concordance was greater for the Part B approach (sensitivity 71.8%, specificity 99.7%) than the Part D approach (39.4%, 97.6%). There were large differences in the numbers of eligible clinicians and NH specialists identified by 3 approaches. The Part B approach was reasonably concordant with the MD-PPAS approach and could be considered by researchers without the financial resources required to purchase MD-PPAS data.