Demographic, behavioral, dietary, and clinical predictors of high-sensitivity C-reactive protein: The National Health and Nutrition Examination Surveys (NHANES)

Abstract

Abstract Funding Acknowledgements Type of funding sources: None. Introduction High-sensitivity C-reactive protein (hs-CRP), a marker of inflammation, is associated with atherosclerosis, and recent studies indicate that therapies targeting inflammation are associated with reductions in cardiovascular (CV) risk. However, factors predictive of elevated hs-CRP in the general population have not been elucidated. Purpose To determine independent predictors of hs-CRP levels in an ambulatory adult population in the US. Methods This study included 5412 adults from the NHANES 2015-2016 cohort and 5856 adults from the NHANES 2017-2018 cohort. The sample was scientifically selected to ensure representativeness of the larger US population. Multivariable logistic regression analysis was used to identify independent predictors of elevated hs-CRP (≥3 mg/L) utilizing the NHANES 2015-2016 cycle (derivation set). The model was verified utilizing the independent NHANES 2017-2018 cycle (validation set). Candidate variables comprised demographic, behavioral, and clinical factors, including standard CV risk factors. Models based on diet and nutrition were developed separately on a subsample of subjects who answered dietary questions. Model discrimination was assessed using area under the receiver-operator characteristic curve (c-statistic). Results Significant independent predictors of high hs-CRP included: increased age (OR 1.09; 95% CI 1.03-1.14 per decade; P=0.024), increased BMI (OR 1.12; 95% CI 1.10-1.14; P<0.001), elevated white blood cell count (OR 1.21; 95% CI 1.15-1.28 per 1000 white blood cells/uL; P=0.002), Black vs White race (OR 1.31; 95% CI 1.10-1.56; P=0.037), female sex (OR 1.57; 95% CI 1.36-1.80; P=0.003), and self-reported poor vs excellent health (OR 1.73; 95% CI 1.04-2.22; P=0.012). The model had excellent discrimination with a c-statistic of 0.77 in the 2015-2016 derivation cycle and 0.76 in the 2017-2018 validation cycle. None of the standard CV risk factors contributed to the model. The only significant nutritional predictor of lower hs-CRP was fiber intake (OR 0.99; 95% CI 0.98-0.99; P=0.035) and the only significant dietary predictor was fruit intake (OR 0.67; 95% CI 0.4-0.934; P=0.045). The dietary and nutrition models had poor discrimination with c-statistics of 0.60 in the 2015-2016 cohort and 0.61 in the 2017-2018 cohort for both models. In a sensitivity analysis, adding dietary factors to the main model did not improve discrimination. Conclusion Older age, female sex, Black race, increased BMI, higher white blood cell count, and self-reported poor health were independent predictors of elevated hs-CRP levels. Higher fruit and fiber consumption were associated with lower hs-CRP levels in univariate but not multivariable models. Additional studies are needed to determine if behavioral modifications (weight loss, increased fiber and fruit intake) can lower hs-CRP and whether this translates to reduced risk for cardiovascular disease.