OBJECTIVE: Prediction of death and functional outcome is essential for determining treatment strategies and allocation of resources for patients with severe traumatic brain injury (TBI). The aim of this study was to evaluate, by using pupillary status and Glasgow Coma Scale (GCS) score, if patients with severe TBI who are ≤ 15 years old have a lower mortality rate and better outcome than adults with severe TBI. METHODS: A retrospective cohort analysis of patients suffering from severe TBI registered in the Trauma Registry of the German Society for Trauma Surgery between 2002 and 2013 was undertaken. Severe TBI was defined as an Abbreviated Injury Scale of the head (AIShead) score of ≥ 3 and an AIS score for any other part of the body that does not exceed the AIShead score. Only patients with complete data (GCS score, age, and pupil parameters) were included. To assess the impact of GCS score and pupil parameters, the authors also used the recently introduced Eppendorf-Cologne Scale and divided the study population into 2 groups: children (0–15 years old) and adults (16–55 years old). Each patient's outcome was measured at discharge from the trauma center by using the Glasgow Outcome Scale. RESULTS: A total of 9959 patients fulfilled the study inclusion criteria; 888 (8.9%) patients were ≤ 15 years old (median 10 years). The overall mortality rate and the mortality rate for patients with a GCS of 3 and bilaterally fixed and dilated pupils (19.9% and 16.3%, respectively) were higher for the adults than for the pediatric patients (85% vs 80.9%, respectively), although cardiopulmonary resuscitation rates were significantly higher in the pediatric patients (5.6% vs 8.8%, respectively). In the multivariate logistic regression analysis, no motor response (OR 3.490, 95% CI 2.240–5.435) and fixed pupils (OR 4.197, 95% CI 3.271–5.386) and bilateral dilated pupils (OR 2.848, 95% CI 2.282–3.556) were associated with a higher mortality rate. Patients ≤ 15 years old had a statistically lower mortality rate (OR 0.536, 95% CI 0.421–0.814; p = 0.001). The rate of good functional outcomes (Glasgow Outcome Scale Score 4 or 5) was higher in pediatric patients than in the adults (72.2% vs 63.1%, respectively). CONCLUSIONS: This study found that severe TBI in children aged ≤ 15 years is associated with a lower mortality rate and superior functional outcome than in adults. Also, children admitted with a missing motor response or fixed and bilaterally dilated pupils also have a lower mortality rate and higher functional outcome than adults with the same initial presentation. Therefore, patients suffering from severe TBI, especially pediatric patients, could benefit from early and aggressive treatment.
The evolution of intracranial pressure (ICP) of critically ill patients admitted to a neurointensive care unit (ICU) is difficult to predict. Besides the underlying disease and compromised intracranial space, ICP is affected by a multitude of factors, many of which are monitored on the ICU, but the complexity of the resulting patterns limits their clinical use. This paves the way for new machine learning techniques to assist clinical management of patients undergoing invasive ICP monitoring independent of the underlying disease. An institutional cohort (ICP-ICU) of patients with invasive ICP monitoring (n = 1346) was used to train recurrent machine learning models to predict the occurrence of ICP increases of ≥22 mmHg over a long (>2 h) time period in the upcoming hours. External validation was performed on patients undergoing invasive ICP measurement in two publicly available datasets [Medical Information Mart for Intensive Care (MIMIC, n = 998) and eICU Collaborative Research Database (n = 1634)]. Different distances (1–24 h) between prediction time point and upcoming critical phase were evaluated, demonstrating a decrease in performance but still robust AUC-ROC with larger distances (24 h AUC-ROC: ICP-ICU 0.826 ± 0.0071, MIMIC 0.836 ± 0.0063, eICU 0.779 ± 0.0046, 1 h AUC-ROC: ICP-ICU 0.982 ± 0.0008, MIMIC 0.965 ± 0.0010, eICU 0.941 ± 0.0025). The model operates on sparse hourly data and is stable in handling variable input lengths and missingness through its nature of recurrence and internal memory. Calculation of gradient-based feature importance revealed individual underlying decisions for our long short time memory-based model and thereby provided improved clinical interpretability. Recurrent machine learning models have the potential to be an effective tool for the prediction of ICP increases with high translational potential.