188 Predicting Posttraumatic Hydrocephalus

Abstract

Abstract INTRODUCTION Posttraumatic hydrocephalus (PTH) is a common complication of traumatic brain injury (TBI) and often has a high risk of clinical deterioration and worse outcomes. The incidence and risk factors for the development of PTH after decompressive craniectomy (DC) has been assessed in previous studies, but rare studies identify patients with higher risk for PTH among all TBI patients. This study aimed to develop and validate a risk scoring system to predict PTH after TBI. METHODS Demographics, injury severity, duration of coma, radiologic findings, and DC were evaluated to determine the independent predictors of PTH during hospitalization until 6 months following TBI through logistic regression analysis. A risk stratification system was created by assigning a number of points for each predictor and validated both internally and externally. The model accuracy was assessed by the area under the receiver operating characteristic curve (AUC). RESULTS >Of 526 patients in the derivation cohort, 57 (10.84%) developed PTH during 6 months follow up. Age >50 (Odd ratio [OR] = 1.91, 95% confidence interval [CI] 1.09 3.75, 4 points), duration of coma = 1 w (OR = 5.68, 95% CI 2.57 13.47, 9 points), Fisher grade III (OR = 2.19, 95% CI 1.24 4.36, 5 points) or IV (OR = 3.87, 95% CI 1.93 8.43, 7 points), bilateral DC (OR = 6.13, 95% CI 2.82 18.14, 9 points), and extra herniation after DC (OR = 2.36, 95% CI 1.46 4.92, 5 points) were independently associated with PTH. Rates of PTH for the low- (0-12 points), intermediate- (13-22 points) and high-risk (23-34 points) groups were 1.16%, 35.19% and 78.57% (P < 0.0001). The corresponding rates in the validation cohort, where 17/175 (9.71%) developed PTH, were 1.35%, 37.50% and 81.82% (P < 0.0001). The risk score model exhibited good-excellent discrimination in both cohorts, with AUC of 0.839 versus 0.894 (derivation versus validation) and good calibration (Hosmer-Lemshow P = 0.56 versus 0.68). CONCLUSION A risk scoring system based on clinical characteristics accurately predicted PTH. This model will be useful to identify patients at high risk for PTH who may be candidates for preventive interventions, and to improve their outcomes.