Deterioration Index in Critically Injured Patients: A Feasibility Analysis
Abstract
Introduction: Continuous prediction surveillance modeling is an emerging tool giving dynamic insight into conditions with potential mitigation of adverse events (AEs) and failure to rescue. The Epic electronic medical record contains a Deterioration Index (DI) algorithm that generates a prediction score every 15 min using objective data. Previous validation studies show rapid increases in DI score (≥14) predict a worse prognosis. The aim of this study was to demonstrate the utility of DI scores in the trauma intensive care unit (ICU) population. Methods: A prospective, single-center study of trauma ICU patients in a Level 1 trauma center was conducted during a 3-mo period. Charts were reviewed every 24 h for minimum and maximum DI score, largest score change (Δ), and AE. Patients were grouped as low risk (ΔDI <14) or high risk (ΔDI ≥14). Results: A total of 224 patients were evaluated. High-risk patients were more likely to experience AEs (69.0% versus 47.6%, P = 0.002). No patients with DI scores <30 were readmitted to the ICU after being stepped down to the floor. Patients that were readmitted and subsequently died all had DI scores of ≥60 when first stepped down from the ICU. Conclusions: This study demonstrates DI scores predict decompensation risk in the surgical ICU population, which may otherwise go unnoticed in real time. This can identify patients at risk of AE when transferred to the floor. Using the DI model could alert providers to increase surveillance in high-risk patients to mitigate unplanned returns to the ICU and failure to rescue.