Aims: To develop an electronic frailty index (eFI) in Queensland Health hospitals using routinely collected patient data.
Methods: We used data from 196,239 hospital inpatients aged ≥18 years admitted to 14 digital hospitals in Queensland between January 2023–June 2024. Patient data sourced from the statewide integrated electronic medical record included nurse-administered risk/functional assessments and routine pathology results. Mortality was determined by linkage to Queensland death register. Associations between eFI scores and adverse outcomes were expressed as mean differences (MDs), odds ratios (ORs) and hazard ratios (HRs), quantified following adjustment for patient demographic and clinical characteristics.
Results: The analytical sample included 103,707 patients (53% of baseline cohort) who had data available for the eFI (mean age 63.0 years, 52.7% male). After screening 64 potential deficits, the final eFI included 34. Scores were positively skewed, ranged 0.00–0.78 (median 0.13) and had a submaximal limit (99th percentile) of 0.49. Mean eFI scores increased by 0.03 per decade of age and were 0.01 points higher in females than males. Each 0.1 increase in eFI was associated with longer hospital stay (MD 2.6 days, 95% CI 2.5–2.7), higher odds of in-hospital mortality (OR 2.25, 95% CI 2.16–2.34) and higher mortality risk across an average of 8.4 months of follow-up (HR 1.92, 95% CI 1.88–1.95).
Conclusions: An eFI derived entirely from routinely collected hospital data demonstrated strong, graded associations with adverse outcomes. This scalable eFI enables system-wide frailty measurement and provides a foundation for prospective patient risk stratification and service planning.