The utility of routine intraoperative CSF during ventricular shunt insertion: a 10-year retrospective cohort study.

Objective: Ventricular shunt insertion is a common procedure in pediatric neurosurgical practice. In many areas of medicine there is a push toward rationalization of healthcare resources and a reduction in low-value tests or procedures. The intraoperative sampling of CSF at the time of shunt insertion is one traditional aspect of care that has not been rigorously evaluated. Additionally, the role of CSF cell counts and chemistry in predicting shunt dysfunction is often discussed but poorly studied. A recent meta-analysis found a correlation between elevated CSF protein levels and shunt failure in patients with tuberculous meningitis, but not other pathologies, which limits the generalizability of those findings to Western populations. The aims of this study were to assess the clinical utility of intraoperative CSF sampling during insertion of ventricular shunts and to describe any association of routine intraoperative CSF sample parameters with shunt failure or infection.

Methods: A retrospective review of a prospectively maintained surgical database covering 10 years of consecutive cases from a quaternary Australian pediatric neurosurgical center serving a population of 5.3 million was conducted. Statewide electronic medical records were reviewed to collate data on demographics, postoperative imaging, CSF biochemistry, instances of shunt failure, shunt revisions, and mortality. Patients undergoing insertion of a new ventricular shunt were included, while all cases of shunt revision were excluded.

Results: During the study period, 1485 shunt procedures were performed, of which 427 involved the placement of a new ventriculoperitoneal shunt system. The mean patient age was 5.2 years (range premature-18 years). Of the 427 cases, 377 (88%) underwent routine CSF sampling. The most common indications for shunt revision were proximal catheter obstruction (51/173, 29%), followed by infection (29/173, 17%) and valve blockage (23/173, 13%). During the study period, 3 patients with existing intracranial hardware had overt ventriculitis identified at the time of intraoperative sampling, resulting in shunt removal. One patient with an existing ventricular reservoir had a delayed clinically significant infection identified on intraoperative cultures. Elevated CSF protein levels were associated with shunt failure during follow-up (area under the curve 0.625). The identified cutoff of 300 mg/L was significantly associated with a reduced time to shunt failure in both univariate and multivariate analyses.

Conclusions: It may be reasonable to consider omission of routine intraoperative CSF sampling at the time of shunt insertion in patients without existing intracranial hardware. Elevated CSF protein levels are associated with a reduced time to shunt failure in a dose-dependent manner.