Predicting the need for mechanical ventilation and vasopressor therapy in patients with traumatic brain injury

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Patients with traumatic brain injury (TBI) represent potentially difficult subgroup, as they are prone to chronic critical illness characterized by persistent organ dysfunction. Maintaining vital functions in these patients often requires vasopressor therapy and mechanical ventilation that is associated with adverse outcomes. Identifying the predictors of these interventions is essential for risk stratification and intensive care planning. Objective. To assess the risk factors and develop a model for predicting the need for mechanical ventilation and vasopressor therapy according to admission data in patients with chronic critical illness and traumatic brain injury. Material and methods. We analyzed real-world clinical data using the Russian Intensive Care Database (RICD). Demographic and clinical data on patients within 48 hours after admission (anamnesis, laboratory parameters, and scores) were extracted from the RICD database, and in-hospital outcomes were also taken into account. Results. To predict the need for mechanical ventilation, we included 188 patients. Of these, 25 (13.3%) patients required mechanical ventilation. To predict the need for vasopressor support, we enrolled 244 patients, and 39 (16.0%) ones required vasopressor therapy. We identified significant risk factors for the outcomes using univariate and multivariate analysis. Three parameters (serum albumin, polytrauma and anemia upon admission) were included in nomogram for predicting the need for mechanical ventilation. AUC (predictive accuracy of the model) was 0.753 (95% CI 0.645; 0.862). Regarding the need for vasopressor therapy, univariate and multivariate analyses established only one risk factor (low albumin concentration) with AUC 0.650 (95% CI 0.555; 0.746). The cutoff point was 31.1 g/L. Conclusion. Analysis of patients with traumatic brain injury in chronic critical illness revealed high probability of necessary respiratory and hemodynamic support. This need increased along with prolonged hospital-stay. Low serum albumin on admission was a key predictor of these interventions and highlighted potential prognostic significance of triad of inflammation, hypercatabolism and immunosuppression. Our results demonstrate limitations of left-sided prediction models and the need to develop new right-sided approaches taking into account dynamics of chronic critical illness.