A. Mukhi1, J. Vosswinkel1, R.S. Jawa1 1Stony Brook University Medical Center, Division Of Trauma, Department Of Surgery, Stony Brook, NY, USA
Introduction: The quick Sequential Organ Failure Assessment score (qSOFA), while conventionally applied for identification of sepsis, has been evaluated in pediatric and adult blunt trauma admissions, but not specifically in the older adult population. We assessed the validity of qSOFA as an Emergency Department (ED) tool to predict adverse outcomes in the admitted older adult blunt trauma patient population using conventional and machine learning (ML) algorithms.
Methods: Retrospective data was gathered from the trauma registry at an American College of Surgeons (ACS) verified Level 1 Trauma center for older adult blunt trauma patients (age 65+ y) hospitalized between years 2017 and 2023. qSOFA scores are the sum of binary scores for 3 variables (Respiratory Rate ≥ 22, Systolic Blood Pressure ≤ 100 mmHg, and Glasgow Coma Score ≤ 14). Univariate and multivariable analyses were performed. As very few patients had qSOFA =3, qSOFA 2 and 3 categories were combined in multivariable ML algorithms with an 80-20 train-test data split.
Results: 5,660 admissions met inclusion criteria (Table 1). Higher qSOFA scores were associated with higher in-hospital mortality rates, higher ICU admission rates and with higher National Trauma Data Standard (NTDS) assigned complications. The performance of several ML algorithms utilizing the following predictors (qSOFA, age, gender, comorbid count, injury mechanism and Injury Severity Score (ISS)) was assessed for the prediction of in-hospital mortality, ICU admission and complications. The Area under the ROC curve (AUCROC) scores for Logistic Regression (LR), Random Forest Classifier (RFC) and Decision Tree Classifier (DTC) for mortality were 0.82, 0.73, 0.55, for ICU Admissions were 0.73, 0.69, 0.62, and for complications were 0.70, 0.63, 0.55. LR had the highest AUCROC for each of these outcomes. Specifically, for mortality, qSOFA was found to be a significant (p<0.001) independent predictor with qSOFA =1 (OR =3.3, 95% CI [2.3,4.8]) and qSOFA =2/3 (OR =8.4 [4.7,15.1]). For ICU admissions, qSOFA =1 (OR =1.7, 95% CI [1.5, 2.0]) and qSOFA =2/3 (OR =2.6 [1.7, 4.0]) were significant (p<0.001) predictors. For complications, qSOFA =1 (OR =1.6, 95% CI [1.3, 2.0]) and qSOFA =2/3 (OR =2.7 [1.7, 4.1]) were significant (p<0.001) predictors.
Conclusion: Admitted older adult blunt trauma patients with higher ED qSOFA scores have higher mortality, higher ICU admissions, and higher complications rates in univariate and multivariable analyses. The qSOFA can serve as a quick and easy scoring system to allocate resources and determine adverse outcomes.