Presentation (CTI)

OBJECTIVES: This study evaluates the contributions of healthcare encounter data from Electronic Data Networks (EDN) and from Direct Facility Retrieval (DFR). We explore variations in data availability between methods across patient demographics and investigate how different methods can enhance the completeness of records for analytic use.
METHODS: We collected medical records from 3,842 US-based patients with confirmed multiple sclerosis diagnoses who consented to record retrieval. EDN records came from networked digital sources; DFR records were retrieved from records offices via electronic, fax, and paper transfer and subsequently digitized.
RESULTS:
Record Availability: Across EDN and DFR sources, we retrieved records from approximately 474,000 healthcare visits. All but 1 study patient had DFR records (median: 147 DFR records, IQR 78-280). 3187 (83%) of patients had 50+ EDN records available, 548 (14%) had 1-49 EDN records, and 117 (3%) had 0 EDN records.
Source Complementarity: 63% of visits appeared in EDN and 55% in DFR records; with only 18% of visits having records from both sources. The integration enhanced granularity within existing observation periods, and also expanded the temporal scope of available data. The full dataset with both EDN and DFR data demonstrated a median visit span of 12.72 years, compared to the subset of records from DFR (9.41 years) or EDN (9.34 years) alone.
Demographics: There are no trends in age, sex, race, ethnicity, or geography distinguishing patients with 0 EDN records, <50 EDN records, and 50+ EDN records.
CONCLUSIONS: Neither source alone provides comprehensive coverage: density and follow-up time are improved when both EDN and DFR retrieval were used together. Despite minimal demographic variations in EDN availability, the absence of EDN data for a minority of patients highlights potential systematic gaps in single-source datasets.The integration of EDN and DFR data is crucial for maximizing record capture for observational research.