There have been few reports of associations between coffee consumption and diabetes. In healthy subjects given an oral glucose tolerance test (OGTT), the 1-h glucose concentration was lower when the glucose load was given with coffee (1). There were no significant differences in the other time points in the test, nor in serum insulin concentration at any time point. In contrast, others have reported a deterioration of glucose tolerance after coffee ingestion (2,3). In healthy subjects a rise in blood glucose levels after caffeine intake was detected at the 2nd, 3rd, and 4th hours in comparison to those taking a placebo (4). Blood insulin levels were comparable after caffeine or placebo ingestion during the entire OGTT. Ingestion of coffee or the injection of caffeine is associated with elevated plasma catecholamines and free fatty acids, as well as decreased insulin sensitivity (5), all of which might be expected to increase the incidence of diabetes. Recently, coffee consumption was reported to decrease the incidence of type 2 diabetes in the Netherlands (6). Prompted by these inconsistent findings, we attempted to replicate the Dutch finding in the longitudinal population-based study of diabetes among the Pima Indians (7).
The average daily use of coffee was assessed by questionnaire from 1978 to 1992 in 2,680 nondiabetic individuals aged ≥15 years (mean age 27 years; 60% women) who had at least one follow-up examination by September 2002. Consumption was recorded as never, occasionally (less than one a day), one or two a day, three or more a day, or occasionally heavy. Diabetes was diagnosed by OGTT using the 1985 World Health Organization criteria (8).
During an average follow-up time of 11 years (range 1–23), 824 individuals developed diabetes. After adjustment for age, sex, and BMI in a time-dependent proportional hazards model, the risk of developing diabetes for those who reported drinking coffee compared with those who never drink coffee (referent category) was as follows: occasionally 1.09 (95% CI: 0.89–1.34), one or two a day 0.92 (0.74–1.13), three or more a day 1.01 (0.82–1.26), and occasionally heavy 0.81 (0.55–1.03). Overall, adjusted for age, sex, and BMI, there was no significant association between consumption categories and incidence of diabetes (P = 0.6). Tea consumption was also unrelated to the incidence of diabetes (not shown).
One strength of our study was the periodic use of an OGTT, as compared with the subjective self-reporting of diabetes used in the Dutch study (6). In that study, coffee consumption was associated with lower socioeconomic status and less healthy behaviors, factors that might be associated with a lower likelihood of being tested for diabetes. Our study included nearly three times as many incident cases of diabetes as the Dutch study (824 vs. 306), resulting in a narrow 95% CI (0.84–1.18) around our estimate of the hazard rate ratio of 0.99 for any coffee consumption. In conclusion, our data provide no evidence for a relationship of coffee consumption and risk of type 2 diabetes.
