Comparing Shared Patient Networks Across Payers.

Area of Science:

• Health Services Research
• Network Science
• Oncology

Background:

• Measuring care coordination using administrative data is crucial for improving healthcare quality.
• Understanding patient-sharing networks among providers is essential for evaluating care coordination.

Purpose of the Study:

• To compare shared patient networks derived from administrative claims data across multiple insurance payers.
• To assess the impact of different data sources on the representation of provider networks.

Main Methods:

• Employed social network analysis on pooled physician data for colorectal cancer patients (2003-2013).
• Utilized North Carolina Central Cancer Registry data linked to Medicare and private insurance claims for oncologists.
• Analyzed provider-level (e.g., patient volume, shared patients) and network-level (e.g., density, clusters) measures.

Main Results:

• Provider patient volume rankings differed between payers for 24.5% of physicians.
• Medicare claims missed 14.6% of shared patient relationships but captured more per provider than private insurance.
• Private networks showed a higher fraction of shared patients and significant variations in network centrality measures compared to Medicare.

Conclusions:

• Shared patient networks constructed from single-payer data exhibit substantial differences compared to multi-payer data.
• Limitations of single-payer claims data should be carefully considered when drawing conclusions about provider behavior and care coordination.