A Systematic Review and Meta-Analysis of Preoperative Frailty Instruments Derived From Electronic Health Data

BACKGROUND: Frailty is a strong predictor of adverse outcomes in the perioperative period. Given the increasing availability of electronic medical data, we performed a systematic review and meta-analysis with primary objectives of describing available frailty instruments applied to electronic data and synthesizing their prognostic value. Our secondary objectives were to assess the construct validity of frailty instruments that have been applied to perioperative electronic data and the feasibility of electronic frailty assessment. METHODS: Following protocol registration, a peer-reviewed search strategy was applied to Medline, Excerpta Medica dataBASE (EMBASE), Cochrane databases, and the Comprehensive Index to Nursing and Allied Health literature from inception to December 31, 2019. All stages of the review were completed in duplicate. The primary outcome was mortality; secondary outcomes included nonhome discharge, health care costs, and length of stay. Effect estimates adjusted for baseline illness, sex, age, procedure, and urgency were of primary interest; unadjusted and adjusted estimates were pooled using random-effects models where appropriate or narratively synthesized. Risk of bias was assessed. RESULTS: Ninety studies were included; 83 contributed to the meta-analysis. Frailty was defined using 22 different instruments. In adjusted data, frailty identified from electronic data using any instrument was associated with a 3.57-fold increase in the odds of mortality (95% confidence interval [CI], 2.68-4.75), increased odds of institutional discharge (odds ratio [OR], 2.40; 95% CI, 1.99-2.89), and increased costs (ratio of means, 1.54; 95% CI, 1.46-1.63). Most instruments were not multidimensional, head-to-head comparisons were lacking, and no feasibility data were reported. CONCLUSIONS: Frailty status derived from electronic data provides prognostic value as it is associated with adverse outcomes, even after adjustment for typical risk factors. However, future research is required to evaluate multidimensional instruments and their head-to-head performance and to assess their feasibility and clinical impact.