Association between weight-adjusted waist index (WWI) and multimorbidity: Evidence from NHANES and prospective validation in CHARLS

BACKGROUND: The burden of chronic diseases is known to be significantly increased by obesity, yet conventional measures like waist circumference (WC) and body mass index (BMI) do not accurately capture central adiposity. The Weight-Adjusted Waist Index (WWI), calculated as waist circumference (cm)/√weight (kg), has been proposed to assess abdominal fat distribution independent of overall body size and may theoretically outperform BMI and WC by reducing muscle-mass-related misclassification. Evidence regarding its association with multimorbidity remains limited. This study sought to elucidate these associations using nationally representative NHANES data and to verify the results in a prospective Chinese cohort (CHARLS). METHOD: Adults ≥20 years in NHANES 2017-2023 were included, applying sampling weights. Multimorbidity was defined as the coexistence of ≥2 chronic conditions, identified based on self-reported physician-diagnosed diseases in both NHANES and CHARLS. In NHANES, multivariable logistic regression, RCS, GAM, interaction analysis and Sensitivity analyses were performed, adjusting for demographics, socioeconomic status, lifestyle behaviors, and dietary factors. ROC curve analysis was then performed to evaluate and compare the predictive performance of WWI, BMI, and WC. WWI, BMI, and WC were analyzed in separate regression models to avoid multicollinearity interference when comparing predictive performance. External validation was conducted using Cox regression and RCS in CHARLS 2011-2028, adjusting for demographic, socioeconomic, and lifestyle variables. RESULT: In NHANES, WWI was significantly positively associated with multimorbidity (OR = 1.74; 95% CI: 1.59-1.90; P < 0.001). Participants in the highest quartile of WWI had over threefold higher odds of multimorbidity compared with those in the lowest quartile (OR = 3.04; 95% CI: 2.94-3.73; P < 0.001). RCS indicated a predominantly linear association, while GAM suggested a modest J-shaped pattern. Sex significantly modified the association (P for interaction <0.05). Sensitivity analysis in adults ≥45 years yielded consistent results. In the CHARLS cohort (n = 2985), higher WWI significantly predicted incident multimorbidity (HR = 1.14; 95% CI: 1.06-1.22). WWI showed the highest discriminative ability for multimorbidity (AUC = 0.709) compared with BMI and WC. CONCLUSIONS: WWI remained a significant and consistent predictor of multimorbidity, demonstrating greater discriminatory ability than BMI or WC. While RCS suggested a predominantly linear association, GAM revealed potential non-linearity at extreme WWI levels, indicating that dose-response patterns warrant further confirmation. Validation in the CHARLS cohort supports temporal association, offering greater causal plausibility. Collectively, these results highlight WWI as a practical marker for identifying high-risk individuals and for guiding early prevention strategies.