Abstract

BACKGROUND & AIMS: In-vitro and rodent studies, and short-term human trials suggest that compounds in chocolate can enhance insulin sensitivity. Also, a recent prospective Japanese epidemiological analysis found that long-term chocolate consumption was inversely associated with diabetes risk. The objective of the present analysis was to test the epidemiological association between long-term chocolate consumption and diabetes risk in a U.S. cohort. METHODS: Multivariable prospective Cox Regression analysis with time-dependent covariates was used to examine data from 7802 participants in the prospective Atherosclerosis Risk in Communities Cohort. The data included 861 new diabetes cases during 98,543 person-years of follow up (mean = 13.3 years). RESULTS: Compared to participants who ate 1 oz of chocolate less often than monthly, those who ate it 1-4 times/month, 2-6 times/week and ≥ 1 time/day had relative risks of being diagnosed with diabetes that were lower by 13% (95% confidence interval: -2%, 25%), 34% (18%, 47%) and 18% (-10%, 38%). These relative risks applied to participants without evidence of preexisting serious chronic disease that included diabetes, heart attacks, stroke or cancer. In conclusion, the risk of diabetes decreased as the frequency of chocolate intake increased, up to 2-6 servings (1 oz) per week. Consuming ≥ 1 serving per day did not yield significantly lower relative risk. CONCLUSIONS: These results suggest that consuming moderate amount of chocolate may reduce the risk of diabetes. Further research is required to confirm and explore these findings.