Application of Claims-based Frailty Index to a Structured Electronic Health Record Data.

Area of Science:

• Gerontology and Health Services Research
• Clinical Informatics
• Health Outcomes Research

Background:

• The Claims-based Frailty Index (CFI) is validated for Medicare claims data.
• Its applicability to structured electronic health record (EHR) data remains unstudied.

Purpose of the Study:

• To assess the performance of the CFI when applied to structured EHR data.
• To compare frailty prevalence and predictive accuracy for adverse outcomes between EHR and Medicare claims datasets.

Main Methods:

• Applied the CFI to a large structured EHR dataset (Explorys) and Medicare fee-for-service (FFS) data.
• Compared frailty prevalence (CFI ≥0.25) in older adults across datasets.
• Evaluated the predictive performance (Odds Ratio and AUC) of frailty for adverse clinical outcomes within one year.

Main Results:

• Included 526,681 individuals from Explorys and 346,070 from Medicare.
• Frailty prevalence was lower in Explorys (7.4-7.1%) compared to Medicare (14.2%) among heart failure patients.
• Frailty showed similar predictive performance for hospital/ER visits (AUCs 0.656-0.697) and ADE-related encounters (AUCs 0.654-0.676) across datasets.

Conclusions:

• The CFI can be applied to structured EHR data, though it identifies fewer frailty cases than Medicare claims.
• The CFI demonstrates comparable performance in predicting adverse clinical outcomes using both EHR and Medicare claims data.
• Structured EHR data is a viable source for frailty assessment using the CFI.