Association of nine composite inflammatory indices with cardiovascular diseases in US adults: national health and nutrition examination survey (NHANES, 2005-2018).

backgroundComposite inflammation-immune indices derived from routine blood counts and albumin are thought to capture innate-adaptive imbalance, thrombo-inflammation, and nutritional status. Whether a panel of such indices is consistently associated with cardiovascular diseases (CVD) and improves discrimination in key subgroups has not been evaluated head‑to‑head in a nationally representative sample. METHODS AND

resultsWe conducted a cross-sectional prevalence analysis of seven National Health and Nutrition Examination Survey (NHANES) cycles (2005-2018), including 31,536 adults after excluding those < 18 years, pregnant, with cancer, or missing CVD or inflammatory indices. CVD was defined from standardized questionnaire items, and nine prespecified inflammatory indices (SII, SIRI, PIV, NLR, PLR, NPR, LMR, PAR, HALP) were computed from routine laboratory data. Associations were examined using survey-weighted logistic regression (quartiles and per-standard-deviation [SD] increments) in the complete-case analytic sample, with false discovery rate (FDR) correction, restricted cubic splines, and stabilized inverse probability of treatment weighting (IPTW); body mass index (BMI)-stratified analyses tested interactions, and incremental discrimination was evaluated in normal-weight adults (BMI 18.5-24.9 kg/m²) using the change in area under the receiver operating characteristic curve (ΔAUC), integrated discrimination improvement (IDI), continuous net reclassification index (cfNRI), and calibration metrics. Multiple imputation of missing covariates and exclusion of extreme index values were conducted as sensitivity analyses. Participants with CVD were older, had higher BMI, more cardiometabolic comorbidities, less favorable socioeconomic and lifestyle profiles, higher neutrophil-based indices (NPR, NLR, SIRI, PIV), and lower LMR. In fully adjusted models, each 1-SD increase in log-transformed NPR, NLR, and SIRI was associated with higher odds of CVD (ORs 1.25, 1.17, and 1.23, respectively), whereas higher log-transformed LMR was on average associated with lower odds (per-SD OR 0.82). Restricted cubic splines showed approximately linear dose-response relations for NPR, NLR, and SIRI, pronounced U-shaped associations for PLR and PAR, and a non-linear pattern for LMR with the lowest risk at intermediate values. IPTW analyses confirmed excess risk in the top versus bottom quartiles of neutrophil-dominant indices (for example, NPR IPTW-adjusted OR 1.41, with a 2.2% absolute increase in CVD prevalence), whereas other markers contributed little. Associations were strongest among normal-weight adults and generally attenuated in overweight and obese strata. In this subgroup, SIRI and LMR provided the largest discrimination gains (ΔAUC ≈ 0.006, with the highest IDI and cfNRI), while maintaining good calibration relative to the baseline clinical model.

conclusionsAcross nine indices evaluated on a single platform using complex survey methods, NPR and SIRI consistently tracked higher CVD burden, LMR and PAR tracked lower burden, and index-specific nonlinearity was evident. Across markers, incremental gains in discrimination were modest (ΔAUC approximately 0-0.006), with prespecified analyses indicating that these small improvements were most apparent in normal-weight adults. These findings support pragmatic use of complete blood count (CBC)-based composites for inflammation-centric risk dialogues and motivate prospective studies to determine whether incorporating these indices into cardiovascular risk stratification improves clinical decision-making and outcomes.