Predicting Fragmented Care: Beneficiary, Physician, Practice, and Market Characteristics

Predicting Fragmented Care: Beneficiary, Physician, Practice, and Market Characteristics

Published: Dec 01, 2022
Publisher: Medical Care, vol. 60, issue 12
Download
Authors

Lisa M. Kern

Carol Urato

Background

Understanding what drives fragmented ambulatory care (care spread across multiple providers without a dominant provider) can inform the design of future interventions to reduce unnecessary fragmentation.

Objectives

To identify the characteristics of beneficiaries, primary care physicians, primary care practice sites, and geographic markets that predict highly fragmented ambulatory care in the United States.

Research Design

Cross-sectional analysis of Medicare claims data for beneficiaries attributed to primary care physicians and practices in 2018. We used hierarchical linear models with random intercepts and an extensive list of explanatory variables to predict the likelihood of high fragmentation.

Subjects

A total of 3,540,310 Medicare fee-for-service beneficiaries met the inclusion criteria, attributed to 26,344 primary care physicians in 9300 practice sites, and 788 geographic markets.

Measures

We defined high care fragmentation as a reversed Bice-Boxerman Index score above 0.85.

Results

Explanatory variables explained only 6% of the variation in highly fragmented care. Unobserved differences between primary care physicians, between practice sites, and between markets together accounted for 4%. Instead, 90% of the variation in high fragmentation was unobserved residual variance. We identified the characteristics of beneficiaries (age, reason for original Medicare entitlement, and dually eligible for Medicaid insurance), physicians (comprehensiveness of care), and practices (size, being part of a system/hospital) that had small associations with high fragmentation.

Conclusions

Variation in fragmentation was not explained by observed beneficiary, primary care provider, practice site, or market characteristics. Instead, the aggregate behavior of diverse health care providers beyond primary care, along with unmeasured patient preferences and behaviors, seem to be important predictors.

See Clearly. Act Quickly.

We can help you shine a spotlight through the dark. Get in touch to learn more about how Mathematica could help advance your mission.

Explore Partnership