Using an electronic frailty index and patient reported outcomes to predict sarcopenia risk

Abstract

The study goal was to identify sarcopenia using information from electronic health records (EHR) or prospectively collected patient reported outcomes tools among male and female adults participating in our Musculoskeletal Functioning, Imaging, and Testing (FIT Core) protocol. Participants (n = 1304) were grouped according to number of sarcopenia test thresholds (adapted from the European Working Group on Sarcopenia in Older People) into those meeting ≥ 1 (labeled Sarcopenia-1), or ≥ 2 (Sarcopenia-2) test thresholds for sarcopenia or no sarcopenia for meeting none. The electronic frailty index (eFI) was calculated from participant EHR using a published cumulative health deficits model. The Patient-Reported Outcomes Measurement Information System's (PROMIS) Physical Function (PF) and Mobility domains were assessed prospectively. A logistic regression model based on their EHR data was applied to describe each individual's proportionate risk for having Sarcopenia-1 and Sarcopenia-2. eFI correlated positively with Sarcopenia-1 (r = 0.77, p < 0.01) and Sarcopenia-2 (r = 0.66, p < 0.01). After multivariable adjustment, Sarcopenia-1 was best predicted by eFI (AUC 0.76, p = 0.029) and PROMIS Mobility T-score (AUC 0.76, p < 0.01). Sarcopenia-2 was best predicted by eFI (AUC 0.91, p = 0.01) and the PROMIS PF T-score (AUC 0.91, p = 0.03). The eFI and PROMIS PF and Mobility scores may be useful to predict sarcopenia risk when direct muscle quantifying measurements are unavailable.