Development of a Predictive Model for Identifying High-Risk Older Adults for Geriatric Emergency Department Screening.

Objective: Geriatric emergency department (GED) programs face challenges in risk-stratifying older adults. Existing tools, like the identification of seniors at risk (ISAR) score model, have modest predictive accuracy and are difficult to implement sustainably. We undertook a quality improvement initiative to develop, evaluate, and validate a screening score (SS) within a large, integrated health care system. This score, which can be deployed and automated at emergency department (ED) triage, identifies older adults at risk of subsequent ED or hospital use and short-term mortality. Methods: The GED-SS score model was developed from a multicenter cohort of ED patients aged ≥70 years with an Emergency Severity Index > 1 from January 1, 2018, until December 31, 2019. The composite outcome was ≥3 days of acute care (ED, observation, or inpatient) or death within 90 days. The GED-SS score model was prospectively validated in 1313 ED patients and then compared with the nurse-performed ISAR score screenings. Results: The GED-SS score model showed better discrimination with an area under the curve (AUC) value of 0.73 vs 0.66 for the ISAR score model identification. At a 43% sensitivity threshold (based on an ISAR score model cutoff value of ≥ 3), the GED-SS score model had higher specificity (86% vs 78%) and positive predictive value (59% vs 47%), while also flagging fewer patients (23.3% vs 29.1%). Conclusion: We developed and validated a GED-SS score model to identify older adult ED patients at increased risk of short-term mortality and acute care utilization. The model, which uses structured data facilitating automatic calculation, was prospectively validated and performed comparably to or better than the ISAR score model.