Modelling success after perinatal post-haemorrhagic hydrocephalus: a single-centre study

Introduction Post-haemorrhagic hydrocephalus is common amongst premature infants and one of the leading indications for paediatric cerebrospinal fluid (CSF) diversion. Permanent CSF diversion is often delayed until the infant is older but there is no clear consensus on the timing for this. The outcomes for permanent shunting in this patient group are poor, with higher rates of failure and infection compared to other aetiologies of hydrocephalus. Methods We conduct a single-centre retrospective review of infants with post-haemorrhagic hydrocephalus requiring a permanent shunt insertion over a 5-year period. Demographic and clinical data from time of shunt insertion were collected and used to generate generalised linear models (GLMs) to predict shunt success at 12 months after insertion. Results Twenty-six infants underwent permanent shunting in this period for post-haemorrhagic hydrocephalus, with 10 suffering shunt failure within the first 12 months. The best-performing GLM was able to predict shunt success with a sensitivity of 1 and specificity of 0.90, with head circumference, weight, and corrected age at the time of shunt insertion being the most significantly associated variables for shunt success in this model. Conclusion Our proof-of-principle study suggests that highly accurate prediction of shunt success for infants with post-haemorrhagic hydrocephalus is possible using routinely available clinical variables. Further work is required to test this model in larger cohorts and validate whether pre-operative use can improve outcomes for this patient group.