TARGETED PREVENTION OF PEDIATRIC SHUNT INFECTION: DEVELOPMENT OF A USABLE RISK SCORE FOR LMIC SETTINGS

Abstract

Background: Hydrocephalus is a major pediatric neurosurgical condition, particularly in low- and middle-income countries (LMICs), where shunt infection remains a major cause of morbidity and resource strain. Despite advances in shunt technology, infection rates remain 5–15%. This study aimed to identify modifiable risk factors for early postoperative infection and develop a pragmatic, point-based risk score for use in LMICs Methods: We retrospectively analyzed 1,185 pediatric cerebrospinal fluid (CSF) shunt procedures performed at Lady Reading Hospital, Peshawar (2019–2024). Early infection (≤90 days) was defined using CDC criteria. A matched case–control model (127 infections, 381 controls) and time-to-event analysis identified independent predictors. Logistic regression and bootstrap validation (500 resamples) assessed model performance. Regression coefficients were translated into an integer-based risk score. Results: Early shunt infection occurred in 10.7% of cases, with a median onset of 21 days. Multivariable analysis identified five independent predictors: absence of an antibiotic-impregnated catheter (aOR 2.52, 95% CI 1.48–4.29), operative time >90 min (aOR 2.10, 95% CI 1.22–3.62), prior external ventricular drain (aOR 1.89, 95% CI 1.18–3.02), low weight-for-age (aOR 1.85, 95% CI 1.12–3.06), and cloudy/purulent CSF at surgery (aOR 3.45, 95% CI 1.62–7.33). The internally validated model showed excellent discrimination (AUC = 0.83, Brier = 0.072). The derived score stratified infection risk as low (4.5%), intermediate (18.8%), and high (41.6%). Conclusions: This study confirms established predictors and identifies new, practical markers nutritional status and intraoperative CSF clarity for early pediatric shunt infection. The derived, easy-to-use score enables targeted preventive strategies such as selective use of antibiotic-impregnated catheters or deferred surgery in high-risk children, optimizing outcomes and costs in resource-limited LMIC settings.