Background:

Many studies have found an inverse association between consumption of filtered coffee and incident type 2 diabetes. The effect of boiled coffee has been less studied.

Methods:

Information on self-reported coffee consumption was available from health surveys conducted from 1985 to 1999. We estimated type 2 diabetes incidences from redeemed prescriptions of oral antidiabetic drugs in the period 1 January 2004 to 1 January 2008.

Results:

With less than 1 cup/day as the reference, the relative risks associated with 1–4, 5–8, and 9 or more cups of boiled coffee per day were 0.87 (95% confidence interval = 0.80–0.95), 0.65 (0.59–0.72), and 0.65 (0.57–0.74), respectively, after adjusting for confounders. The corresponding relative risks associated with other types of coffee (mainly filtered) were 0.84 (0.79–0.90), 0.67 (0.62–0.71) and 0.62 (0.56–0.68).

Conclusions:

A moderate inverse association was found between consumption of both boiled and other types of coffee at the age of 40–45 years and the risk of being prescribed oral antidiabetic drugs 5–20 years later.