A longitudinal analysis of metabolic biomarkers and selected predictors of overall, cancer and CVD mortality in US adults

Objective: To assess the role of glucose metabolism and inflammatory biomarkers, as well as that of sociodemographic characteristics, as predictors of overall, CVD and cancer mortality in US adults. Methods: We analyzed a sample of 6760 subjects with 12-18 years or mortality follow-up available data; subjects were included in the NHANES III cohort study linked to National Death Index records through December 31, 2006. To examine cause-specific mortality, deaths were coded using ICD-9 and ICD-10 codes. Serum cholesterol, fasting glucose, triglycerides, insulin, C-reactive protein (CRP) and leptin were assessed as predictors of overall, CVD and cancer mortality using multivariate Cox and Weibull regression models adjusted for sociodemographic and comorbidity variables. Standard errors of the mean were estimated by Taylor Series Linearization to incorporate sample weights and account for the NHANES complex sample design. Statistical analyses were done using STATA V11. Results: Multivariate analyses showed that glycated hemoglobin and CRP are significant independent predictors of overall, CVD and cancer mortality, after adjusting for smoking and sociodemographic factors. Overweight and obesity did not predict premature mortality but may be in the causal pathway by increasing insulin resistance and augmenting glycated hemoglobin and inflammation, thus supporting the obesity mortality paradox. Conclusions: Glycated hemoglobin and CRP may serve as proxies of a common metabolic pathway leading to overall and disease-specific mortality. In addition, psychosocial factors are also independent predictors of mortality in models including the aforementioned biomarkers; thus, interventions targeting biomarkers may not be sufficient to curb mortality or extend survival if social and environmental factors are not addressed.