Electronic health record embedded frailty index and outcomes among older adults receiving cancer chemotherapy
Heidi D Klepin, Scott Isom, Kathryn E Callahan, Nicholas M Pajewski, Umit Topaloglu, Lynne I Wagner, Jennifer Gabbard, Armida Parala-Metz, Janet A ToozeAbstract
Purpose
We evaluated whether an automated electronic health record (EHR)–based frailty index (eFI) is associated with hospitalization and overall survival (OS) in older adults receiving chemotherapy.
Patients and Methods
We identified 509 consecutive patients aged ≥65 years with newly diagnosed lung, colorectal, or breast cancer treated with chemotherapy (2017–2020). An automated eFI score was calculated using data from ≥2 visits in the 2 years before chemotherapy, including vital signs, smoking status, diagnoses, selected laboratory values, and functional measures. Frailty was categorized using a cumulative deficit approach as fit (eFI ≤0.10), pre-frail (0.10–0.21), or frail (>0.21). Hospitalizations at 3 and 6 months were compared across groups. Adjusted associations between eFI and hospitalization and mortality were assessed using Cox regression models. OS was estimated by Kaplan–Meier methods and compared using log-rank tests.
Results
Among 509 patients (median age, 72.2 years), 27.7% were fit, 42.4% pre-frail, 14.9% frail, and 14.9% had non-calculable eFI. Six-month hospitalization increased with frailty: 24% (fit), 42% (pre-frail), 62% (frail), and 50% (non-calculable) (P<.0001). Median OS was 38 months (fit), 23 months (pre-frail), 17 months (frail), and 10 months (not calculable) (p < 0.0001). In adjusted models, mortality risk was higher in pre-frail (HR 1.7, 95% CI 1.2–2.3), frail (HR 2.0, 95% CI 1.3–3.1), and not calculable groups (HR 2.6, 95% CI 1.7–4.0) compared with fit patients.
Conclusion
EHR-derived frailty was associated with hospitalization and survival, supporting the eFI as a scalable approach to risk stratification in older adults undergoing chemotherapy.