External validation of the hospital frailty risk score among hospitalised home care clients in Canada: a retrospective cohort study
Journal Articles
Overview
Research
Identity
Additional Document Info
View All
Overview
abstract
AbstractBackgroundThe Hospital Frailty Risk Score (HFRS) is scored using ICD-10 diagnostic codes in administrative hospital records. Home care clients in Canada are routinely assessed with Resident Assessment Instrument-Home Care (RAI-HC) which can calculate the Clinical Frailty Scale (CFS) and the Frailty Index (FI).ObjectiveMeasure the correlation between the HFRS, CFS and FI and compare prognostic utility for frailty-related outcomes.DesignRetrospective cohort study.SettingAlberta, British Columbia and Ontario, Canada.SubjectsHome care clients aged 65+ admitted to hospital within 180 days (median 65 days) of a RAI-HC assessment (n = 167,316).MethodsCorrelation between the HFRS, CFS and FI was measured using the Spearman correlation coefficient. Prognostic utility of each measure was assessed by comparing measures of association, discrimination and calibration for mortality (30 days), prolonged hospital stay (10+ days), unplanned hospital readmission (30 days) and long-term care admission (1 year).ResultsThe HFRS was weakly correlated with the FI (ρ 0.21) and CFS (ρ 0.28). Unlike the FI and CFS, the HFRS was unable to discriminate for 30-day mortality (area under the receiver operator characteristic curve (AUC) 0.506; confidence interval (CI) 0.502–0.511). It was the only measure that could discriminate for prolonged hospital stay (AUC 0.666; CI 0.661–0.673). The HFRS operated like the FI and CFI when predicting unplanned readmission (AUC 0.530 CI 0.526–0.536) and long-term care admission (AUC 0.600; CI 0.593–0.606).ConclusionsThe HFRS identifies a different subset of older adult home care clients as frail than the CFS and FI. It has prognostic utility for several frailty-related outcomes in this population, except short-term mortality.
