Diagnosing Cerebrospinal Fluid Shunt Failure: Systematic Review and Meta-Analysis of Noninvasive and Minimally Invasive Tests in Children and Adults

Neurosurgery 98:1206–1220, 2026

This systematic review and meta-analysis evaluates diagnostic performance of noninvasive and minimally invasive tests (X-ray shunt series, CT, MRI, nuclear scintigraphy, thermal flow) for cerebrospinal fluid shunt failure in pediatric and adult patients, synthesizing 33 studies and 8,077 patients. Pooled accuracy was 82%, sensitivity 67%, and specificity 88%, with X-ray series showing lower sensitivity but higher specificity compared with CT.

The analysis highlights substantial study bias—especially in patient selection and reference standards—and concludes no existing test is adequate as a standalone diagnostic for shunt failure, underscoring the need to optimize diagnostic pathways, reduce harm, and develop improved point-of-care functional tests.

Objective Assess diagnostic performance of noninvasive/minimally invasive tests (X-ray shunt series, CT, MRI, nuclear scintigraphy, thermal flow evaluation) for suspected CSF shunt failure in pediatric and adult patients.

Methods Systematic review/meta-analysis (PROSPERO CRD42024596142) including patients with permanent CSF shunts who underwent the above index tests; performance (accuracy, sensitivity, specificity) modeled via meta-regression incorporating test type, age group, follow-up timeframe, and shunt type.

Evidence base PubMed search (Oct 22, 2024) screened 465 articles; 33 studies included, totaling 8077 patients.

Pooled performance Overall pooled accuracy 82%, sensitivity 67%, specificity 88% across included tests.

Key comparative finding (sensitivity) X-ray shunt series had significantly lower sensitivity than CT (OR 0.106, 95% CI 0.056–0.201; overall effect of test type on sensitivity P = .013).

Key comparative finding (specificity) No overall effect of test type on specificity (P = .192), but X-ray shunt series showed higher specificity than CT in subgroup analysis (OR 4.04, 95% CI 1.938–8.424).

Covariates No significant effects of age group, follow-up timeframe, or shunt type on accuracy, sensitivity, or specificity.

Conclusion/limitations No test performed well enough to serve as a standalone diagnostic for shunt failure; evidence limited by high risk of bias, especially in the reference standard domain for most studies.