Association between Frailty Assessed Using Two Electronic Medical Record-Based Frailty Assessment Tools and Long-Term Adverse Prognosis in Older Critically Ill Survivors.

Abstract

Frailty has become an independent risk factor for adverse outcomes in critically ill patients. This study aimed to explore the predictive ability of two electronic medical record-based frailty assessment tools, the Hospital Frailty Risk Score (HFRS) and Frailty Index based on physiological and laboratory tests (FI-lab), for long-term adverse prognosis in older critically ill survivors. Retrospective observational study. 9,082 critically ill survivors aged ≥ 65 years. The HFRS and the 33-item FI-lab were constructed based on the published literature. Cox and logistic regression models assessed the association between frailty and 1-year mortality and post-discharge care needs. 2,586 patients died within 1 year of follow-up. In fully adjusted models, frailty assessed using both the HFRS (per point, hazard ratio [HR] 1.06, 95% confidential interval [CI] 1.05-1.06; intermediate frailty risk, HR 2.00, 95% CI 1.78-2.25; high frailty risk, HR 3.06, 95% CI 2.68-3.50) and FI-lab (per 0.01 points, HR 1.03, 95% CI 1.03-1.03; intermediate frailty risk, HR 1.59, 95% CI 1.44-1.76; high frailty risk, HR 2.30, 95% CI 2.06-2.57) was associated with mortality. Addition of frailty indicators improved the predictive validity of the Sequential Organ Failure Assessment score for mortality (HFRS alone ∆ C-index 0.034; FI-lab alone ∆ C-index 0.016; HFRS and FI-lab combined ∆ C-index 0.042). The HFRS but not the FI-lab was associated with higher probability of post-discharge care needs. Both the HFRS and FI-lab could independently predict 1-year mortality in older critically ill survivors. Adding the HFRS to the SOFA score model improved it more than adding the FI-lab. The greatest improvement was achieved when both frailty indicators were used together. These findings suggest that electronic medical record-based frailty assessment methods can be useful tools for predicting long-term outcomes in older critically ill patients.