Telomere length and socioeconomic status at neighborhood and individual levels among 80,000 adults in the Genetic Epidemiology Research on Adult Health and Aging cohort

Abstract

Background: Telomere length (TL) may serve as a biologic marker of aging. We examined neighborhood and individual-level socioeconomic status (SES) in relation to TL.

Methods: The study included 84,996 non-Hispanic white subjects from the Genetic Epidemiology Research on Adult Health and Aging (GERA) cohort, part of the Research Program on Genes, Environment and Health. Relative TL (T/S) was log₂ transformed to improve normality and standardized to have mean 0 and variance 1. Neighborhood SES was measured using the Neighborhood Deprivation Index (NDI), and individual SES was measured by self-reported education level. We fit linear regression models of TL on age, sex, smoking, body mass index, comorbidities, NDI, and education level. We tested for differences in the associations by sex and nonlinearity in the association of NDI with TL.

Results: Each SD increase in NDI was associated with a decrease of 0.0192 in standardized TL, 95% confidence interval (CI) = -0.0306, -0.0078. There was no evidence of nonlinearity in the association of NDI with TL. We further found that less than high school education was associated with a decrease of 0.1371 in standardized TL, 95% CI = -0.1919, -0.0823 as compared to a college education. There were no differences in the associations by sex.

Conclusions: We found evidence that both lower neighborhood SES and lower individual-level SES are associated with shorter TL among non-Hispanic whites. Our findings suggest that socioeconomic factors may influence aging by contributing to shorter TL.

Keywords: aging; neighborhood factors; socioeconomic status; telomere length.

Figure 1.

Association between standardized TL and education level in the fully adjusted model, showing the estimated effect size and 95% CI for each education level as compared to the reference level of those with a college education. Fully adjusted model includes age, sex, smoking, BMI, comorbidities, diet, NDI, and education level.

Figure 2.

Association between TL and NDI overall and by sex in the fully adjusted model, showing the estimated effect size and 95% CI associated with an increase of one SD in the NDI. Plot (A) shows the main analysis of NDI and TL and plot (B) shows the sensitivity analysis restricting to subjects who lived at their current address for at least 5 years. Fully adjusted model includes age, sex, smoking, BMI, comorbidities, diet, NDI, and education level. There were no statistically significant differences in the association between NDI and TL by sex (P-value for interaction 0.8973).

Figure 3.

Relationship between the NDI and TL when modeled as a spline effect. The association between NDI and TL was statistically significant when modeled as a spline effect (P-value = 0.0007) and when modeled as a linear effect (P-value = 0.0008). There was no statistically significant evidence of nonlinearity in the association between NDI and TL (P-value = 0.1848).