PERFORMANCE OF CLAIMS-BASED FRAILTY MEASURES IN MEDICARE FFS AND MA - Findings from NHATS

Sep 29, 2026

Frailty in older adults predicts adverse health outcomes, but many studies rely on claims-based measures rather than clinical reference standards such as the Comprehensive Geriatric Assessment–Frailty Index (CGA-FI) and the Fried phenotype (FP). Using National Health and Aging Trends Study data linked to Medicare claims, the study evaluated Kim’s Claims-based Frailty Index (CFI) and Gilbert’s Hospital Frailty Risk Score (HFRS) for detecting frailty among community-dwelling Medicare Advantage (MA) and fee-for-service (FFS) beneficiaries. In 2019, CFI showed high specificity but low sensitivity versus CGA-FI, while HFRS showed high sensitivity but low specificity; discrimination was generally good, with model performance varying by benchmark and settings.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages2

Reading Time5 min

AccessFree to download

Study objective and data source

Frailty models evaluated

Results by sensitivity, specificity, PPV, and discrimination

Sensitivity analyses and implications

1318 Innovation in Aging, 2025, Vol. 9, No. S2 Abstract citation ID: igaf122.4389 PERFORMANCE OF CLAIMS-BASED FRAILTY MEASURES IN MEDICARE FFS AND MA: FINDINGS FROM NHATS Lan Luo1, Xiecheng Chen2, Chan Mi Park3, Sandra Shi1, Kevin Pritchard4, Ellen McCarthy1, and Dae Hyun Kim1, 1. Hebrew SeniorLife, Boston, Massachusetts, United States, 2. Hebrew Rehabilitation Center, Roslindale, Massachusetts, United States, 3. Hebrew SeniorLife, Harvard Medical school, Boston, Massachusetts, United States, 4. Hebrew SeniorLife, Roslindale, Massachusetts, United States Frailty is a state of vulnerability in older adults which predicts adverse health outcomes. Although clinical measures, such as a Comprehensive Geriatric Assessment–Frailty Index (CGA-FI) and the Fried phenotype (FP), are reference standards, claims-based measures are often used for research. Given the rapidly growing Medicare Advantage (MA) population and concerns regarding MA data completeness, we assessed performance of claims-based frailty models in detecting frailty by CGA-FI and FP among community-dwelling MA and fee-for-service (FFS) beneficiaries. We measured the Kim’s Claims-based Frailty Index (CFI) and Gilbert’s Hospital Frailty Risk Score (HFRS) in the National Health and Aging Trends Study linked to Medicare claims. In 2019 (FFS n = 2,153, MA n = 1,672), CFI had high specificity (FFS 0.97, MA 0.99) but low sensitivity (FFS 0.31, MA 0.24) in detecting frailty by CGA-FI. The CFI’s PPV was higher for CGA-FI than FP (FFS: 0.84 vs 0.44, MA: 0.90 vs 0.47). In comparison, HFRS had high sensitivity (FFS 0.91, MA 0.91) but low specificity (FFS 0.38, MA 0.43) for CGA-FI frailty, with Innovation in Aging, 2025, Vol. 9, No. S2 1319 similar results for FP. Model discrimination was good (C-statistics: CFI 0.81, HFRS 0.79), with both models performing better when benchmarked against CGA-FI. 2021 analyses were similar. Insensitivity analysis using only inpatient claims, HFRS sensitivity decreased while specificity increased. These results suggest that claims-based frailty measures may perform comparably in MA and FFS beneficiaries. However, performance characteristics between models vary – highlighting the importance of modelselection for a given population or purpose.

FAQs

What frailty measures were assessed in this study?

The study evaluated Kim’s Claims-based Frailty Index (CFI) and Gilbert’s Hospital Frailty Risk Score (HFRS) using Medicare claims linked to NHATS.

How did CFI perform when detecting frailty against CGA-FI?

In 2019, CFI had high specificity but low sensitivity for frailty detection versus CGA-FI in both FFS and MA groups.

How did HFRS performance compare to CFI across MA and FFS?

HFRS showed high sensitivity but low specificity for CGA-FI frailty detection, and similar patterns were reported when benchmarked against FP.

Sep 29, 2026

Upload by Rainbow Cat