Association between admission serum potassium and quantitative external lumbar drainage responsiveness in patients with hydrocephalus: a retrospective study

Objective To evaluate the association between untreated admission serum potassium levels and quantitative responsiveness to external lumbar drainage (ELD) in patients with hydrocephalus. Methods This single-center retrospective study included 349 unique patients with hydrocephalus after reconciliation and deduplication of clinical records collected between 2020 and 2026. Serum potassium was defined as the first measurement obtained at hospital admission before treatment. ELD responsiveness was defined as an improvement of ≥10% in Timed Up and Go or 10-m walking time, ≥3 points in the Mini-Mental State Examination (MMSE), or ≥2 points in the Montreal Cognitive Assessment (MoCA). Multivariable logistic regression was used to assess the association between potassium and ELD responsiveness after adjustment for study period and eight clinical covariates. Sensitivity analyses examined medication exposure, potassium range, and alternative outcome definitions. Model performance was evaluated using bootstrap internal validation. Results Among 349 patients, 228 (65.3%) met the quantitative criteria for ELD responsiveness. Higher admission potassium was independently associated with greater odds of ELD responsiveness (adjusted OR per 0.1 mmol/L increase, 1.117; 95% CI, 1.038–1.201; p = 0.003). The association remained significant after exclusion of diuretic users (OR, 1.099; 95% CI, 1.017–1.188; p = 0.017). However, no significant association was observed among patients with potassium ≥3.5 mmol/L (OR, 1.043; 95% CI, 0.914–1.190; p = 0.536) or when a stricter definition of nonresponse was applied (OR, 1.076; 95% CI, 0.962–1.204; p = 0.200). The combined clinical-plus-potassium model achieved an apparent AUC of 0.757 and an optimism-corrected AUC of 0.725 (approximate 95% CI, 0.685–0.797). The corrected improvement in discrimination after adding potassium was small and uncertain. Conclusion Higher untreated admission serum potassium was independently associated with quantitative ELD responsiveness in this heterogeneous hydrocephalus cohort. However, the association was not demonstrated within the non-hypokalemic range, and the incremental predictive contribution of potassium beyond clinical variables remained uncertain. These findings support serum potassium as a potential clinical correlate of ELD responsiveness that warrants further validation in well-characterized prospective cohorts.