(283) Daily Prediction of Next-Day Disease Severity in Critically Ill Patients

Introduction: Predictive models for intensive care unit (ICU) patients mainly focus on mortality, but short-term disease severity is more relevant for day-to-day decision-making. This study aimed to develop and validate a daily prediction model for next-day disease severity in ICU patients.

Methods: Data from the Simple Intensive Care Studies II prospective cohort study of acutely admitted critically ill adults, including data collected during the first 7 days of admission, was used to fit a mixed-effects model for next-day deterioration. Deterioration was defined as a decline in total and organ-specific (respiration, coagulation, liver, cardiovascular, central nervous system, renal) Sequential Organ Failure Assessment (SOFA) scores. Acute physiology variables from the APACHE IV model, along with other laboratory variables and a time-related indicator, were used for predictor selection. Performance was evaluated using the area under the receiver operating characteristic curve (AUC), calibration plots, and decision curve analysis.

Results: 1,009 patients were included in the analysis. The final predictive model for overall next-day deterioration included 6 predictors: total SOFA score on the current day, the minimum value of arterial pH, the Glasgow Coma Scale, mean arterial blood pressure, mechanical ventilation, and the interaction term between mechanical ventilation and an indicator variable denoting the first day of admission. The AUC of the model was 0.74 (95%CI: 0.70−0.78). For a probability threshold for identifying a high risk of deterioration between 0.2 and 0.5, the model exhibited more benefit than treating all or none. Among the six organ-specific prediction models next-day deterioration of respiration, cardiovascular, and renal function showed fair performance (AUCs 0.79 (95%CI: 0.76−0.82), 0.79 (95%CI: 0.74−0.85), and 0.81 (95%CI: 0.77−0.85) respectively).

Conclusions: Our model for daily prediction in ICU patients can predict the deterioration of next-day disease severity in overall, respiration, cardiovascular, and renal function. Within a certain range of probability thresholds, the model provides clinical benefit and may support ICU doctors in making intervention decisions.