Modeling the impact of socioeconomic disparity, biological markers and environmental exposures on phenotypic age using mediation analysis and structural equation models

Abstract

Introduction

Average age is increasing worldwide, raising the public health burden of age-related diseases, as more resources will be required to manage treatments. Phenotypic Age is a score that can be useful to provide an estimate of the probability of developing aging-related conditions, and prevention of such conditions could be performed efficiently studying the mechanisms leading to an increased phenotypic age. The objective of this study is to characterize the mechanisms that lead to aging acceleration from the interactions among socio-demographic factors, health predispositions and biological phenotypes.

Methods

We present an approach based on the combination of mediation analysis and structural equation models (SEM) to better characterize these mechanisms, quantifying the interactions between biological and external factors and the effects of preexisting health conditions and socioeconomic disparities. We use two independent cohorts of the NHANES dataset: we use the largest (n = 13,186) to select the variables that enlarge the gap between phenotypic and chronological ages, we then create a SEM based on nested linear regressions to quantify the influence of all sociodemographic variables expressed in three latent variables indicating ethnicity, socioeconomic status and preexisting health status. We then replicate the model and apply it to the second cohort (n = 4,425) to compare the results.

Results

Results show that phenotypic age increases with poor glucose control or obesity-related biomarkers, especially if combined with a low socioeconomic status or the presence of chronic or vascular diseases, and provide a framework to quantify these relationships. Black ethnicity, low income/education and a history of chronic diseases are also associated with a higher phenotypic age. Although these findings are already known in literature, the proposed SEM-based framework provides an useful tool to assess the combinations of these heterogeneous factors from a quantitative point of view.

Conclusion

In an aging society, phenotypic age is an important metric that can be used to estimate the individual health risk, however its value is influenced by a myriad of external factors, both biological and sociodemographic. The framework proposed in this paper can help quantifying the combined effects of these factors and be a starting point to the creation of personalized prevention and intervention strategies.