Enhancing the accuracy of a multivariable prediction model to identify medical patients suitable for Same Day Emergency Care services

ObjectivesTo test the performance of the Glasgow Admission Prediction Score (GAPS) and Ambulatory Score (Amb score), and derive and validate a novel score for the identification of Emergency Department (ED) attendances suitable for treatment by Same Day Emergency Care (SDEC) services. DesignRetrospective diagnostic study using routinely collected data from electronic healthcare records. SettingThree hospitals in the diverse urban setting of Birmingham, UK, between April 2023-March 2024. ParticipantsAdult patients with an unplanned hospital attendance requiring internal medicine assessment. Main Outcome MeasuresSuitability for treatment by SDEC services, defined as being discharged alive with a length of stay of <12 hours ("LOS<12"). ResultsData were included for 152,877 attendances, with a median age of 58 years (interquartile range: 38 to 76), and of which 54.3% were by female patients and 68.4% of White ethnicity; the outcome of LOS<12 was achieved in 45.0% (N=68,752). The GAPS and Amb score had moderate predictive accuracy, with areas under the receiver operating characteristic curve (AUROCs) of 0.741 (95% CI: 0.738 to 0.744) and 0.733 (95% CI: 0.730 to 0.736), respectively. A novel score was produced, comprising the factors from the GAPS and Amb score, as well as the National Early Warning Score 2 (NEWS2) and primary presenting complaint. When applied to an internal validation set (N=27,078), the resulting SDEC Triage Tool (SDEC-T) achieved an AUROC of 0.850 (95% CI: 0.845 to 0.854), with performance being similar across the three hospitals (AUROC range: 0.845 to 0.858). ConclusionsThe novel score derived within this diverse cohort has superior accuracy to the existing Amb score and GAPS for the identification of patients suitable for treatment in SDEC.