Hydrocephalus as a consistent predictor of in-hospital mortality in tuberculous meningitis: no age-specific effect modification in a lifespan cohort.

Background: Tuberculous meningitis (TBM) remains the most lethal manifestation of extrapulmonary tuberculosis, with persistently high in-hospital mortality despite advances in antimicrobial therapy. Hydrocephalus is a frequent and devastating complication of TBM. However, whether its prognostic impact on mortality is consistent across age groups-and whether age modifies this association-remains unresolved due to the lack of lifespan-spanning evidence. Methods: We conducted a single-center retrospective cohort study including non-HIV TBM patients hospitalized between October 2013 and January 2024. To address potential multicollinearity and reduce dimensionality, least absolute shrinkage and selection operator regression was applied for variable selection. This was followed by Firth's penalized logistic regression to estimate the adjusted associations with in-hospital mortality, ensuring robust statistical adjustment for confounding variables. We examined age-related effect modification by incorporating categorical and continuous interaction terms, performing stratified analyses, and applying restricted cubic splines modeling. Model performance was evaluated using discrimination, calibration, and decision curve analysis, with internal validation by bootstrap resampling. Results: Among 1,574 patients, the in-hospital mortality rate was 11.9%. Hydrocephalus was independently associated with increased mortality risk in the overall cohort, and this effect was broadly similar in children and adults. No statistically significant interaction between hydrocephalus and age was detected in categorical, continuous, or spline-based analyses. Predicted mortality increased with older age in both hydrocephalus groups, particularly in older patients, but the relative adverse effect of hydrocephalus remained broadly parallel across age. The prediction model showed excellent discrimination and good calibration, with stable net clinical benefit across a wide range of risk thresholds. Conclusions: Hydrocephalus is a strong predictor of in-hospital mortality in TBM. We found no statistically clear evidence of effect modification by age across the lifespan; however, this finding should be interpreted cautiously because the pediatric subgroup was relatively small and its confidence intervals were wider. These findings support early detection and timely management of hydrocephalus in patients of all ages.