Background: Randomized trials have examined short-term effects of lifestyle interventions for diabetes prevention only among high-risk individuals. Prospective studies have examined the associations between lifestyle factors and diabetes in healthy populations but have not characterized the intervention. We estimated the long-term effects of hypothetical lifestyle interventions on diabetes in a prospective study of healthy women, using the parametric g-formula.
Methods: Using data from the Nurses' Health Study, we followed 76,402 women from 1984 to 2008. We estimated the risk of type 2 diabetes under eight hypothetical interventions: quitting smoking, losing weight by 5% every 2 years if overweight/obese, exercising at least 30 minutes a day, eating less than three servings a week of red meat, eating at least two servings a day of whole grain, drinking two or more cups of coffee a day, drinking five or more grams of alcohol a day, and drinking less than one serving of soda a week.
Results: The 24-year risk of diabetes was 9.6% under no intervention and 4.3% when all interventions were imposed (55% lower risk [95% confidence interval = 47 to 63%]). The most effective interventions were weight loss (24% lower risk), physical activity (19%), and moderate alcohol use (19%). Overweight/obese women would benefit the most, with 10.8 percentage point reduction in 24-year risk of diabetes. The validity of these estimates relies on the absence of unmeasured confounding, measurement error, and model misspecification.
Conclusion: A combination of dietary and nondietary lifestyle modifications, begun in midlife or later in relatively healthy women, could have prevented at least half of the cases of type 2 diabetes in this cohort of U.S. women.