Cumulative BMI and incident prediabetes over 30 years of follow-up: The CARDIA study

Objective: This study examined how cumulative BMI (cBMI) is associated with incident prediabetes in a biracial observational cohort study followed from young adulthood to middle age. Methods: Black and White men and women (n = 4190) from the Coronary Artery Risk Development in Young Adults (CARDIA) study, ages 18 to 30 years in 1985 to 1986 and free of prediabetes or diabetes at baseline, were followed for 30 years. Cox regression was used to determine how cBMI was associated with incident prediabetes after controlling for traditional cardiovascular risk factors. Results: Over 30 years of follow-up, 46.2% of the sample developed prediabetes. Mean cBMI was 801.4 BMI-years for those with prediabetes and 658.3 BMI-years for those without (p < 0.0001). After multivariable adjustment, the hazard rate ratio for the highest cBMI quartile was 2.064 (95% CI: 1.793-2.377) relative to the lowest quartile. The second and third quartiles did not differ from the first quartile, consistent with a nonlinear trend. Conclusions: The cumulative burden of higher weight and longer duration was associated with incident prediabetes, but this association was statistically significant only after a higher threshold was reached. Strategies for prevention of prediabetes in middle age may focus on avoiding overweight in young adulthood to limit duration.