The Base Deficit, International Normalized Ratio, and Glasgow Coma Scale (BIG) Score, and Functional Outcome at Hospital Discharge in Children With Traumatic Brain Injury

The prognostic value of an age-adjusted BIG score in adult patients with traumatic brain injury

Background

The base deficit, international normalized ratio, and Glasgow Coma Scale (BIG) score was previously developed to predict the outcomes of pediatric trauma patients. We designed this study to explore and improve the prognostic value of the BIG score in adult patients with traumatic brain injury (TBI).

Methods

Adult patients diagnosed with TBI in a public critical care database were included in this observational study. The BIG score was calculated based on the Glasgow Coma Scale (GCS), the international normalized ratio (INR), and the base deficit. Logistic regression analysis was performed to confirm the association between the BIG score and the outcome of included patients. Receiver operating characteristic (ROC) curves were drawn to evaluate the prognostic value of the BIG score and novel constructed models.

Results

In total, 1,034 TBI patients were included in this study with a mortality of 22.8%. Non-survivors had higher BIG scores than survivors (p < 0.001). The results of multivariable logistic regression analysis showed that age (p < 0.001), pulse oxygen saturation (SpO2) (p = 0.032), glucose (p = 0.015), hemoglobin (p = 0.047), BIG score (p < 0.001), subarachnoid hemorrhage (p = 0.013), and intracerebral hematoma (p = 0.001) were associated with in-hospital mortality of included patients. The AUC (area under the ROC curves) of the BIG score was 0.669, which was not as high as in previous pediatric trauma cohorts. However, combining the BIG score with age increased the AUC to 0.764. The prognostic model composed of significant factors including BIG had the highest AUC of 0.786.

Conclusion

The age-adjusted BIG score is superior to the original BIG score in predicting mortality of adult TBI patients. The prognostic model incorporating the BIG score is beneficial for clinicians, aiding them in making early triage and treatment decisions in adult TBI patients.

Pediatric severe traumatic brain injury mortality prediction determined with machine learning-based modeling

INTRODUCTION

Severe traumatic brain injury (sTBI) is a leading cause of mortality in children. As clinical prognostication is important in guiding optimal care and decision making, our goal was to create a highly discriminative sTBI outcome prediction model for mortality.

METHODS

Machine learning and advanced analytics were applied to the patient admission variables obtained from a comprehensive pediatric sTBI database. Demographic and clinical data, head CT imaging abnormalities and blood biochemical data from 196 children and adolescents admitted to a tertiary pediatric intensive care unit (PICU) with sTBI were integrated using feature ranking by way of a forest of randomized decision trees, and a model was generated from a reduced number of admission variables with maximal ability to discriminate outcome.

RESULTS

In total, 36 admission variables were analyzed using feature ranking with variable weighting to determine their predictive importance for mortality following sTBI. Reduction analysis utilizing Borata feature selection resulted in a parsimonious six-variable model with a mortality classification accuracy of 82%. The final admission variables that predicted mortality were: partial thromboplastin time (22%); motor Glasgow Coma Scale (21%); serum glucose (16%); fixed pupil(s) (16%); platelet count (13%) and creatinine (12%). Using only these six admission variables, a t-distributed stochastic nearest neighbor embedding algorithm plot demonstrated visual separation of sTBI patients that lived or died, with high mortality predictive ability of this model on the validation dataset (AUC = 0.90) which was confirmed with a conventional area-under-the-curve statistical approach on the total dataset (AUC = 0.91; P < 0.001).

CONCLUSIONS

Machine learning-based modeling identified the most clinically important prognostic factors resulting in a pragmatic, high performing prognostic tool for pediatric sTBI with excellent discriminative ability to predict mortality risk with 82% classification accuracy (AUC = 0.90). After external multicenter validation, our prognostic model might help to guide treatment decisions, aggressiveness of therapy and prepare family members and caregivers for timely end-of-life discussions and decision making.

LEVEL OF EVIDENCE

III; Prognostic.

Validation of the Japan Coma Scale for the prediction of mortality in children: analysis of a nationwide trauma database

Objective

The Japan Coma Scale (JCS) is widely used in clinical practice to evaluate levels of consciousness in Japan. There have been several studies on the usefulness of JCS in adults. However, its usefulness in evaluating children has not been reported. Therefore, this study aimed to assess the usefulness of the JCS for the prediction of mortality in children.

Methods

This is a multicenter cohort study which used data from a national trauma registry (Japan Trauma Data Bank). This study included patients under 16 years of age who were treated between 2004 and 2015.

The primary outcome measure was in-hospital mortality. Two models were used to examine each item of the Glasgow Coma Scale (GCS) and the JCS. Model A included the discrete levels of each index. In model B, data regarding age, sex, vital signs on arrival to hospital, the Injury Severity Score, and blunt trauma were added to each index. The effectivity of the JCS score was then evaluated using the area under the curve (AUC) for discrimination, a calibration plot, and the Hosmer-Lemeshow test for calibration.

Results

A total of 9045 patients were identified. The AUCs of the GCS and JCS were 0.929 (95% confidence interval (CI) 0.904 to 0.954) and 0.930 (95% CI 0.906 to 0.954) in model A and 0.975 (95% CI 0.963 to 0.987) and 0.974 (95% CI 0.963 to 0.985) in model B, respectively. The results of the Hosmer-Lemeshow test were 0.00 (p=1.00) and 0.00 (p=1.00) in model A and 4.14 (p=0.84) and 8.55 (p=0.38) in model B for the GCS and JCS, respectively.

Conclusions

We demonstrated that the JCS is as valid as the GCS for predicting mortality. The findings of this study indicate that the JCS is a useful and relevant tool for pediatric trauma care and future research.

Development of a Mortality Prediction Tool in Pediatric Severe Traumatic Brain Injury

Severe traumatic brain injury (sTBI) is a leading cause of pediatric death, yet outcomes remain difficult to predict. The goal of this study was to develop a predictive mortality tool in pediatric sTBI. We retrospectively analyzed 196 patients with sTBI (pre-sedation Glasgow Coma Scale [GCS] score <8 and head Maximum Abbreviated Injury Scale (MAIS) score >4) admitted to a pediatric intensive care unit (PICU). Overall, 56 patients with sTBI (29%) died during PICU stay. Of the survivors, 88 (63%) were discharged home, and 52 (37%) went to an acute care or rehabilitation facility. Receiver operating characteristic (ROC) curve analyses of admission variables showed that pre-sedation GCS score, Rotterdam computed tomography (CT) score, and partial thromboplastin time (PTT) were fair predictors of PICU mortality (area under the curve [AUC] = 0.79, 0.76, and 0.75, respectively; p < 0.001). Cutoff values best associated with PICU mortality were pre-sedation GCS score <5 (sensitivity = 0.91, specificity = 0.54), Rotterdam CT score >3 (sensitivity = 0.84, specificity = 0.53), and PTT >34.5 sec (sensitivity = 0.69 specificity = 0.67). Combining pre-sedation GCS score, Rotterdam CT score, and PTT in ROC curve analysis yielded an excellent predictor of PICU mortality (AUC = 0.91). In summary, pre-sedation GCS score (<5), Rotterdam CT score (>3), and PTT (>34.5 sec) obtained on hospital admission were fair predictors of PICU mortality, ranked highest to lowest. Combining these three admission variables resulted in an excellent pediatric sTBI mortality prediction tool for further prospective validation.