Validation of the Glycemic Stress Index in Pediatric Neurosurgical Intensive Care

Background: Studies have suggested that both the degree and the duration of hyperglycemia are independent risk factors for adverse outcome both in pediatric anesthesia and in critically ill children. In a recent paper, we combined intraoperative glycemic variations and length of surgery creating a metabolic glucose-related stress index called "Glycemic Stress Index" (GSI). Aim: To validate GSI for predicting PICU stay in a population of children undergoing different major neurosurgical procedures. Methods: A total of 352 patients with craniotomy were enrolled. Basic clinical data and PICU length of stay were recorded real time. Intraoperative blood loss has been determined considering the estimated red cell volume loss ratio. GSI was calculated and subjected to ROC analysis having as targets PICU length of stay >100 or >200 h. Results: The overall mean PICU stay was 35 h. Correlation analysis confirmed a low but highly significant direct correlation between GSI and PICU length of stay. ROC analysis showed an area under the ROC curve (AUC) of 0.74 (p = 0.03) for GSI to predict PICU stay >200 h and an AUC of 0.67 (p = 0.01) to predict PICU stay >100 h. Best predictive cutoff values were 4.5 and 3.9, for PICU stay >200 and >100 h, respectively. Overall accuracy for the test is higher in predicting PICU stay >200 h. Conclusions: GSI significantly predicts prolonged PICU stay after major neurosurgery in a mixed population of children affected by different neurosurgical conditions