Frailty burden and symptomatic BPH/LUTS in aging men: evidence from two nationally representative population-based...

The frailty index (FI) is a well-established marker of biological aging and a widely validated predictor of adverse health outcomes in older adults. However, its association with symptomatic benign prostatic hyperplasia/lower urinary tract symptoms (BPH/LUTS) has not been well defined in nationally representative populations. This study aimed to evaluate the association between FI levels and symptomatic BPH/LUTS in men using two large population-based studies.

We analyzed men from the China Health and Retirement Longitudinal Study (CHARLS) and the US National Health and Nutrition Examination Survey (NHANES). Cohort-specific FIs were constructed following the Rockwood-Mitnitski deficit-accumulation approach. Symptomatic BPH/LUTS was operationalized using cohort-specific questionnaire and medication data and included bothersome LUTS with or without physician-diagnosed BPH, depending on data availability in each cohort. Multivariable logistic regression models estimated odds ratios (ORs) and 95% confidence intervals (CIs) for associations of log-transformed FI (continuous) and FI tertiles with BPH/LUTS. Restricted cubic splines (RCS) were used to assess graded associations and potential nonlinearity, and prespecified subgroup analyses with interaction tests evaluated effect modification. As a supplementary analysis, cohort-specific machine-learning models were internally validated for risk stratification, with SHAP used for model interpretation.

The analysis included 8,781 men in CHARLS, of whom 1,043 met criteria for BPH/LUTS (11.9%), and 2,539 men in NHANES, with 559 classified as BPH/LUTS (22.0%). Higher log(FI) was associated with higher odds of BPH/LUTS after full adjustment (CHARLS: OR 1.91, 95% CI 1.71-2.13; NHANES: OR 1.47, 95% CI 1.25-1.73). FI tertiles showed graded associations (Model 3; Q3 vs Q1: CHARLS OR 2.91, 95% CI 2.39-3.56; NHANES OR 1.77, 95% CI 1.32-2.37; both P for trend < 0.001). RCS models indicated significant overall associations without evidence of nonlinearity (CHARLS P for nonlinearity = 0.171; NHANES P for nonlinearity = 0.791). Effect modification was observed for drinking status in CHARLS and for age and smoking status in NHANES (all P for interaction ≤ 0.005). Exploratory machine-learning models showed good internal discrimination (CHARLS CatBoost AUC 0.947; NHANES XGBoost AUC 0.925), but these findings require cautious interpretation because external validation was not performed.

Higher FI levels were associated with a higher prevalence of broadly defined symptomatic BPH/LUTS in two nationally representative populations. Given the cross-sectional design, these findings should be interpreted as associations only, and longitudinal studies are needed to clarify the temporal and causal relationships between frailty burden and BPH/LUTS.