Predictors of death in pediatric patients requiring cerebrospinal fluid shunts

Despite improved therapeutic strategies and better diagnostic techniques in the management of pediatric hydrocephalus there continues to be a significant mortality rate associated with cerebrospinal fluid (CSF) shunts. The goal of this study was to determine the long-term outcome and predictors of death in these patients.

Data were collected in all patients requiring a CSF shunt presenting to a single tertiary care pediatric institution during a 10-year period. Patients with neoplasms were excluded because their deaths were predominantly related to the tumor. Descriptive statistics were obtained on the patient characteristics, surgical features, and shunt characteristics. The time and cause of death were determined. Kaplan-Meier survival estimates were used to determine overall survival of patients. Univariate analysis was performed using the log-rank test. Multivariate analysis included use of Cox regression model to determine the significance of age (at the time of initial shunt insertion), the number of shunt-related failures and infections, and whether the shunts were complex or multiple in nature in predicting death. Hazard ratios, 95% confidence intervals and probability values were calculated. Of 907 patients, 124 died. The most common causes were myelomeningocele (191 cases), intraventricular hemorrhage (114 cases), and tumor (190 cases) with 7.9, 3.5, and 32.6% dying, respectively, during the study period. Restricting all analyses to cases without neoplasms, the incidence of shunt-related failures was 58.1% in patients who died and 55.3% in those who survived, with an incidence of shunt-related infection of 19.4% in the former and 18.5% in the latter. The overall mortality rates in all patients at 1, 5, and 10 years were 4.5, 8.9, and 12.4%, respectively, from time of initial shunt insertion to death or last follow-up visit. The infection rate per procedure (that is, following the first shunt insertion) was 10.9% (78 of 717 cases). Evaluation of predictors of death revealed a statistically significant effect of infection with a hazard ratio of 1.66 (p = 0.04).

The mortality rate in shunt-treated pediatric patients with hydrocephalus remains high, dependent on the underlying reason for CSF shunt insertion and the subsequent development of infection of the shunt apparatus.