Integrating machine learning and traditional statistics to evaluate the Uric acid-to-HDL cholesterol ratio for identification of prevalent COPD in older adults

Abstract Background The uric acid-to-high-density lipoprotein cholesterol ratio (UHR) is a novel inflammatory and metabolic biomarker, but its association with chronic obstructive pulmonary disease (COPD) in older adults remains unclear. This study investigated the relationship between UHR and COPD prevalence among adults aged ≥ 60 years. Methods This cross-sectional study analyzed 5,277 participants aged ≥ 60 years from NHANES 2013–2018. Weighted multivariable logistic regression estimated the UHR–COPD association. Restricted cubic spline (RCS) analysis characterized the dose–response relationship, and receiver operating characteristic (ROC) curves evaluated model performance. Subgroup, sensitivity, and machine learning analyses were conducted. Variables with missingness of 20% or higher were not imputed and were instead examined in complete-case analyses. Results Of 5,277 participants, 404 (7.7%) had COPD. Elevated UHR was independently associated with higher COPD prevalence. In the fully adjusted model, each standard deviation (SD) increase in UHR corresponded to 23% higher odds (OR 1.23; 95% CI 1.05–1.44; P = 0.015). The highest versus lowest UHR quartile showed significantly higher risk (OR 1.71; 95% CI 1.09–2.68; P for trend = 0.019). RCS confirmed a linear dose–response relationship (P for non-linearity = 0.449). The model had good discrimination (AUC = 0.785). Adding UHR to 15 conventional predictors gave a small but statistically detectable increment in discrimination (AUC 0.784 to 0.788; DeLong P = 0.049), although the reclassification gains were small. After further adjustment for alcohol, the top-quartile association (OR 2.201, 95% CI 1.202 to 4.031; P = 0.018) and the quartile trend ( P = 0.012) were retained, whereas the per-standard-deviation estimate was attenuated (OR 1.241, 95% CI 0.980 to 1.571; P = 0.085). Conclusions Elevated UHR is associated with a higher prevalence of COPD in older adults. UHR may serve as a low-cost, routinely measurable marker to aid in the identification of prevalent, physician-diagnosed COPD in geriatric care. Because COPD status was ascertained at a single survey time point in a cross-sectional analysis, no prospective-screening inference is intended.