Methods: In a cross-sectional analysis, 27019 men and 26738 women aged 35-74 were identified from 21 European cohorts in the BiomarCaRE consortium. We defined three educational classes (EDs) according to years of schooling and an AL score as the sum of z-scores of 8 selected biomarkers from the cardiovascular, metabolic, and inflammatory systems. We used the Oaxaca-Blinder decomposition to disentangle the educational class gradient in AL score into the differential exposure (DE, attributable to different distribution of smoking and alcohol intake across EDs) and the differential susceptibility (DS, attributable to a different effect of risk factors on AL across EDs) components.
Results: Less educated men (mean AL difference: 0.68, 95%CI: 0.57-0.79) and women (1.52, 1.40-1.64) had higher AL scores. DE accounted for 7% and 6% of the gradient in men and women, respectively. In men, combining smoking and alcohol intake, DS accounted for 42% of the gradient, (smoking DS coefficient=0.177, 26% of the gradient; alcohol DS coefficient=0.109; 16%, not statistically significant). DS contribution increased to 69% in metabolic markers. DS estimates were consistent across age groups; irrespective of comorbidities; and robust to unmeasured confounding. No DS was observed in women.
Conclusions: In men, a differential susceptibility mechanism substantially contributes to the educational class gradient in allostatic load.