Original article:
Coffee consumption and risk of type 2 diabetes mellitus. van Dam RM, Feskens EJM. Lancet 2002; 360: 1477–8.

Summary
A prospective epidemiological study involving 17,111 Dutch men and women aged 30–60 years has found that a high coffee consumption appears to reduce the risk of type 2 diabetes. After adjusting for potential confounders, the risk of subsequently developing diabetes was halved in those consuming seven or more cups per day compared with those who drank two cups per day or fewer.
Although the findings appear not to be attributable to confounding, they require confirmation and interpretation in the light of potential adverse cardiovascular effects of high consumption of some types of coffee. They should not, at this stage, be extrapolated to advice to increase coffee consumption in the hope of reducing risk of developing type 2 diabetes.

Comment
This short paper will arouse considerable interest amongst coffee drinkers. Coffee contains substantial amounts of magnesium and other micronutrients. The phenol chlorogenic acid component has been shown in vitro to reduce glucose absorption and oxidative stress [1], and inhibit hydrolysis of glucose-6-phosphate, which could reduce glucose output by the liver [2]. On the other hand, caffeine has been shown to decrease sensitivity to insulin, but little clinical or epidemiological information is available regarding the effects of coffee in type 2 diabetes or the risk of diabetes associated with coffee drinking.
Dietary data derived from a food frequency questionnaire were related to rates of type 2 diabetes determined from a subsequent self-administered questionnaire in a large sample of Dutch men and women. When compared with those drinking fewer than two cups of coffee per day, relative risks in those drinking three to four cups, five to six cups and seven or more cups per day were 0.71, 0.73 and 0.60, respectively. Consumption of decaffeinated coffee was too low to be studied separately. After adjusting for intake of tea, other foods and lifestyle measures, BMI, cardiovascular disease and some cardiovascular risk factors, the inverse association between coffee consumption and risk of diabetes was not altered.
These rather surprising results immediately raise the possibility that they are explained by selective non-response (23% did not complete the follow-up questionnaire), underdiagnosis of diabetes resulting from self-reporting, or residual confounding. However, coffee consumption was similar in responders and non-responders and results were similar after exclusion of the first 4 years of follow-up. The fact that high coffee consumption was associated with mostly unfavourable lifestyle and dietary factors makes incomplete adjustment for at least the known confounding factors unlikely. Clinical studies do not provide clear confirmation or rejection of these epidemiological data. Although caffeine lowers insulin sensitivity acutely, increased coffee consumption for 14 days has been shown to reduce fasting glucose. Substitution of regular coffee for decaffeinated coffee for 20 days did not influence plasma glucose levels.
These impressive epidemiological data are intriguing. Despite in-depth analyses using the data available, it is still possible that some unmeasured confounding factor explains the results. Despite the impressive levels of statistical significance, they might be due to chance. They will almost certainly be tested in other data sets in the near future and other publications are likely to follow. Food frequency questionnaires were obtained and good clinical data are available in the several cohort studies of health professionals in the United States and in other comparable studies. Although interesting, these results certainly do not provide a justification for a public health recommendation to increase coffee consumption to reduce the epidemic of type 2 diabetes. Not only do the results require confirmation, they also need to be interpreted in the light of other more convincingly demonstrated adverse effects of high intakes of coffee. High intakes of boiled coffee have been clearly shown to increase total and LDL cholesterol. There is clearly little point in recommending a measure which may reduce the risk of type 2 diabetes but increase cardiovascular risk.

References
1. Clifford MN. Chlorogenic acid and other cinnamates — nature, occurrence, dietary burden, absorption and metabolism. J Sci Food Agric 2000; 80: 1033–43.
2. Keijzers GB, De Galan BE, Tack CJ, Smits P. Caffeine can decrease insulin sensitivity in humans. Diabetes Care 2002; 25: 364–9.

Summary and Comment:
Jim Mann, Dunedin, New Zealand