Is cerebrospinal fluid sampling necessary at the time of first ventriculo-peritoneal shunt insertion in paediatric patients?

OBJECTIVE

We aim to evaluate whether intraoperative cerebrospinal fluid (CSF) sampling during ventriculo-peritoneal (VP) shunt insertion can predict future VP shunt infection or guide its management.

METHODS

83 paediatric patients undergoing VP shunt insertion between February 2013 and July 2019 were retrospectively identified. Patient demographics, presence of pre-operative extra ventricular drain (EVD), pre-operative CSF results, and intra-operative CSF results were identified from patient case notes and electronic clinical databases. All included patients were followed up for a minimum of 6 months for identification of shunt infection.

RESULTS

90 VP shunt insertions were performed in 83 patients. Age at time of shunt insertion ranged from 5 days to 15.8 years (mean 44.2 months). Tumours were the most common aetiology for hydrocephalus (n = 24). 67 cases (74.4%) had intra-operative CSF samples, of which 2 revealed the presence of bacteria. Only 1 patient with intra-operative CSF sampling positive for growth developed shunt infection during follow up. Two cases developed a shunt infection despite normal intra-operative CSF results. Three cases did not have intra-operative CSF sampling but developed a shunt infection during follow up. Intra-operative CSF culture achieved 33.3% sensitivity and 98.4% specificity for predicting future shunt infection (p = 0.154). The Receiver Operator Characteristic (ROC) curve of intra-operative white cell count (WCC) and shunt infection at 6 months follow up yielded an Area Under the Curve (AUC) of 50.3%.

CONCLUSION

Our results show that intraoperative CSF sampling as a method to predict future risk of shunt infection and to help inform future antibiotic prescribing is unreliable. Given an AUC of 50.3%, it is no better than chance as a diagnostic tool. Further larger studies are needed to substantiate this.