Background Given substantial heterogeneity in health status between older adults, there have been many efforts to develop metrics of biological and physiological ageing that explain variability in health outcomes better than chronological age (CA) alone. Physiological age (PA) based on clinical indicators may explain variation in health outcomes better than epigenetic or telomere-based biological clocks because clinical indicators capture downstream physiological changes more closely related to health outcomes. Though gender and education are key characteristics that contribute to heterogeneity in ageing-related outcomes, how PA differs by gender and education level is not yet clear. Furthermore, previous studies examining PA are primarily cross-sectional; understanding how PA changes with increasing CA is a necessary next step. In the present study, we examine how gender and education interact to inform longitudinal PA trajectories.

Methods We used three waves of data (2004/05-2012/13) from 8,891 participants aged 50-100 years from the English Longitudinal Study of Ageing. PA was derived in a healthy subsample of the analytic sample using principal component analysis of clinical biomarkers. We then validated PA in the analytic sample by examining associations with incident ageing-related chronic conditions, memory impairment, and functional limitations using Cox proportional hazards models adjusted for chronological age and gender. We used joint models adjusted for birth cohort to produce eight-year trajectories of PA from ages 50, 60, and 70 years in men and women at three levels of education. These models involve simultaneous estimation of linear mixed effects and survival models to account for data missing-not-at-random in the longitudinal process.

Results After adjusting for CA and gender, an increase in PA was associated with increased incidence of functional limitations, memory impairment, diabetes, lung disease, cardiovascular disease, stroke, high blood pressure, high cholesterol, arthritis, dementia (p<0.001 for all), and cancer (p=0.04). Gender differences in PA were minor at baseline, but PA increased faster in women than in men (p=0.004); as a result, at the end of eight years of follow-up, women had PA up to 2.6 years (95% confidence interval=1.8-3.4) older than men. Stratification by education level showed that these gender differences occurred among less educated women only (p interaction=0.007); women educated above secondary level maintained PA younger than men for the duration of follow-up.

Conclusion Higher education may be important to reduce gender disparities in physiological ageing. Policies to promote gender equity in higher education may contribute to reducing gender inequalities in a wide range of health outcomes in old age.