Mapping physician networks with self-reported and administrative data.

Area of Science:

• Health Services Research
• Network Analysis
• Medical Informatics

Background:

• Understanding physician professional relationships is crucial for healthcare delivery.
• Administrative data offers a potential source for mapping these connections.
• Previous methods for mapping physician networks have limitations.

Purpose of the Study:

• To determine if shared patients in administrative data correlate with physician professional relationships.
• To validate patient sharing as a method for identifying physician connections.

Main Methods:

• A web-based survey was administered to 616 physicians.
• Physician relationships were assessed via survey and compared to patient sharing data from 2006 Medicare.
• Physicians reviewed individualized rosters of colleagues based on shared patient data.

Main Results:

• Increased patient sharing significantly correlated with recognized professional relationships.
• Up to 82% of relationships were identified with nine shared patients.
• Patient sharing demonstrated a diagnostic accuracy (AUC) of 0.73 (95% CI: 0.70-0.75).
• Primary care physicians were more likely to recognize relationships than specialists.

Conclusions:

• Patient sharing in administrative data serves as an effective "diagnostic test" for physician relationships.
• This validated method enables future research into mapping physician networks.
• The findings support the use of administrative data for understanding physician collaboration.