Epigenetic aging markers in the association between frailty and mortality among U.S. adults

Abstract Background Frailty reflects diminished physiological reserve and increased vulnerability to adverse health outcomes. It has been linked to biological aging, including epigenetic age acceleration (EAA), a DNA methylation–based marker of aging, but the extent to which EAA accounts for the frailty–mortality association remains unclear. Methods We analyzed three U.S. cohorts—NHANES (1999–2002), HRS (2016), and HANDLS (2004–2009)—with mortality follow-up through 2019–2022. Frailty was defined using harmonized adaptations of the Fried phenotype and FRAIL scale. EAA was derived from five epigenetic clocks (Horvath, Hannum, PhenoAge, GrimAge, DunedinPoAm). Additive Bayesian networks, Cox proportional hazards models, and counterfactual four-way decomposition were used to assess potential mediation and moderation of the frailty–mortality association by EAA, adjusting for age, sex, race/ethnicity, and socioeconomic status. Results Frailty was strongly associated with higher all-cause mortality in NHANES and HRS. GrimAge and DunedinPoAm showed the strongest mediation. In NHANES, GrimAge accounted for 33% (p < 0.001) and DunedinPoAm mediated 17% (p = 0.006) of the association. In HRS, DunedinPoAm mediated 9% (p = 0.040) and GrimAge 16% (p = 0.020). Other clocks showed limited mediation. HANDLS findings were consistent. Higher socioeconomic status was associated with slower aging and lower frailty risk. Female sex was inversely associated with multiple epigenetic clocks but positively associated with frailty. Conclusions Epigenetic aging, particularly GrimAge and DunedinPoAm, may explain part of the frailty–mortality association, supporting a role for biological aging pathways linking frailty to mortality.