Ventriculoperitoneal shunt infections in Pakistan: a retrospective cohort study of clinical and microbiological outcomes

Ventriculoperitoneal (VP) shunt infection is a serious complication of cerebrospinal fluid diversion, and emerging data from South Asia suggest a shift toward multidrug-resistant Gram-negative pathogens. We aimed to describe the clinical profile, microbiological spectrum, antimicrobial resistance pattern, and clinical outcomes of VP shunt infections at a tertiary care centre in Pakistan, and to identify independent risk factors for infection. This retrospective cohort study included 200 patients with complete records who underwent VP shunt surgery between January 2018 and December 2024. Factors associated with infection were selected by LASSO-penalised Cox regression and analyzed with multivariable Cox proportional hazards regression and bootstrap internal validation; cumulative incidence was estimated by the Kaplan-Meier method. In this predominantly paediatric cohort, VP shunt infection occurred in 42 of 200 patients (21.0%, 95% confidence interval 15.9–27.2%), mostly within 30 days of surgery (cumulative incidence 13.0% at 30 days and 20.1% at 12 months). Non-communicating hydrocephalus (adjusted hazard ratio [aHR] 2.08, 95% CI 1.07–4.07) and previous central nervous system (CNS) infection (aHR 2.39, 95% CI 1.11–5.17) were independently associated with infection, although discrimination was modest (optimism-corrected C-index 0.61). Gram-negative organisms predominated (59.5%), led by Acinetobacter species (23.8%); 20 of 25 Gram-negative isolates were multidrug- or extensively drug-resistant. Susceptibility was highest for colistin (91%) and lowest for ceftazidime (9%); meropenem and piperacillin-tazobactam susceptibility were 68% and 52%. Most staphylococci were methicillin-resistant (9 of 11; 82%). In-hospital mortality was 19.0%; all eight deaths occurred in patients with ventriculitis or brain abscess. VP shunt infection affected approximately one in five patients and was caused predominantly by multidrug-resistant Gram-negative organisms. Non-communicating hydrocephalus and previous CNS infection may identify patients at higher risk but require confirmation in larger prospective cohorts. Empirical antibiotic therapy should be guided by local antibiogram data.