Long-term outcomes after shunt surgery for normal pressure hydrocephalus: a competing-risks analysis of revision and mortality

Abstract Background Shunt surgery improves symptoms in normal pressure hydrocephalus (NPH), but its long-term morbidity is incompletely characterised because most series report short follow-up and analyse revision risk with methods that ignore the competing risk of death in an elderly, comorbid population. We evaluated the long-term cumulative incidence of revision and mortality using a competing-risks framework. Methods This retrospective single-centre cohort included 114 consecutive adults who underwent ventriculoperitoneal (VP, n = 80) or lumboperitoneal (LP, n = 34) shunting for NPH between January 2015 and December 2024, with follow-up administratively closed on 31 January 2026 (median follow-up 36 months; maximum 116 months). The cumulative incidence of revision was estimated with the Aalen–Johansen estimator, treating death without revision as a competing event; associations with revision were assessed by cause-specific Cox regression, with aetiology-adjusted and idiopathic-only sensitivity analyses. Shunt-related mortality was adjudicated by a multidisciplinary panel of neurosurgery and neurology specialists against pre-specified criteria. Results At least one shunt-related complication occurred in 45 patients (39.5%); 42 (36.8%) underwent revision. The cumulative incidence of revision was 19.4% at 12 months, 30.8% at 36 months and 39.4% at 60 months; the conventional Kaplan–Meier method overestimated the cumulative incidence of revision by up to 4% points at five years (43.4% vs. 39.4%). Older age (cause-specific hazard ratio [HR] 1.08 per year, 95% CI 1.03–1.14) and higher Evans index (HR 2.61 per 0.1-unit, 95% CI 1.57–4.34) were independently associated with revision, whereas shunt type was not (VP vs. LP HR 1.01, 95% CI 0.52–1.95; p = 0.98); the shunt-type estimate remained non-significant after adjustment for aetiology and in an idiopathic-only analysis. All-cause mortality was 33.3% (38/114); 7 deaths (6.1%; 18.4% of deaths) were adjudicated as shunt-related, and the remaining 31 were not. Among patients with available paired assessments, iNPHGS scores in the idiopathic subgroup and mRS scores in the overall cohort improved significantly from baseline at all postoperative assessments through two years. Conclusions The long-term cumulative incidence of revision after shunting for NPH is substantial but, when the competing risk of death is modelled, lower than conventional estimates suggest. Mortality in this elderly cohort is driven predominantly by comorbidity rather than shunt failure, and older age and greater ventricular enlargement—not shunt type—are associated with revision. These findings support the use of competing-risk approaches when estimating long-term revision incidence in NPH cohorts with substantial competing mortality.