Sex‐ and age‐specific determinants of diabetes: Insights from BKMR and cox modelling of metabolic and lifestyle risk factors in a Korean cohort

Abstract Aim Priorities for public health interventions may vary depending on demographic characteristics. This study investigated sex‐ and age‐specific associations of diabetes‐related metabolic and lifestyle risk factors in middle‐aged and older adults. Methods Using long‐term data from the community‐based cohort of the Korean Genome and Epidemiology Study, sex‐ and age‐specific sub‐cohorts were constructed at ages 45, 55 and 65 years. As diabetes‐related risk factors, five metabolic factors (waist circumference [WC], mean arterial pressure, triglycerides [TG], high‐density lipoprotein cholesterol and alanine aminotransferase) and five lifestyle factors (sleep duration, alcohol consumption, cumulative smoking, physical activity and dietary diversity score) were considered. The associations between these risk factors and incident diabetes were estimated for each sex‐ and age‐specific sub‐cohort. Additionally, hierarchical Bayesian Kernel Machine Regression was used to determine the relative importance of individual risk factors in the development of diabetes and to quantify their joint association. Results The incidence of diabetes ranged from 8.8% to 13.4% across the sub‐cohorts. Overall, metabolic risk factors—particularly WC and TG—appeared to play a more prominent role in the development of incident diabetes than lifestyle‐related factors. Among lifestyle factors, alcohol consumption and cumulative smoking showed an association with the development of diabetes only in women. The estimated collective association of the 10 risk factors with incident diabetes ranged from 0.39 to 0.46 in men and from 0.26 to 0.45 in women, with relatively weaker association observed in older women. The ranking of relative importance of individual risk factors varied depending on sex and age sub‐cohorts. Conclusions These findings suggest that sex‐ and age‐specific intervention prioritization may be helpful in improving the effectiveness of public health strategies for diabetes prevention.