What Explains Increasing Diabetes Prevalence across Cohorts? A Causal Mediation Analysis in the 1958 and 1970 British Birth Cohort Studies

Laura Gimeno , UCL
Alice Goisis, UCL
Jennifer Dowd, University of Oxford
George Ploubidis, UCL

The prevalence of diabetes has increased across successive generations, yet few studies have examined the factors explaining these cohort differences. Identifying these factors could inform efforts to reverse the Generational Health Drift in diabetes. We used data from the 1958 and 1970 British birth cohort studies to quantify cohort differences in observer-measured diabetes at age 44-48. A causal mediation approach was used to evaluate the contribution of six hypothesised mediators to this gap: obesity, psychological distress, housing tenure, relative poverty, breastfeeding, and maternal smoking during pregnancy. Interventional indirect and direct effects and the randomised analogue of the proportion mediated (rPM) were estimated using the parametric g formula. Diabetes prevalence was twice as high in the 1970 cohort compared to the 1958 cohort (6.5% vs. 3%). Obesity was the most important mediator of the cohort-diabetes association (rPM = 29.8%), with psychological distress (rPM = 6.4%) and housing tenure (rPM = 6.9%) also contributing. Relative poverty, maternal smoking and breastfeeding were not identified as significant mediators. Given that rates of obesity and mental ill-health have risen and rates of home ownership have declines across cohorts born since 1970, cohort differences may persist or widen as these more recently born cohorts enter midlife. Our results suggest that tackling obesity is key to reversing the Generational Health Drift in diabetes, but that there is also a need to address the impact of broader social determinants of health to fully achieve this goal.

See extended abstract

 Presented in Session 10. Non-Communicable Diseases and Cardio-Metabolic Health