Rathmann and Giani (1) point out that there is a greater diversity of diabetes prevalence in Europe than suggested by our recent article (2) on the global prevalence of diabetes. We agree with this point. Unfortunately, the data from their study (3) in Augsburg, Germany, had not been published when the data for our article were assembled in 2002. As described in the article, we plan to update the estimates of diabetes prevalence and will include the Augsburg data in the next revision.
The global burden of disease study (4) only included studies describing the prevalence of diabetes defined using oral glucose tolerance tests. Prevalence studies based on self-reported diabetes were excluded because, as Rathmann and Giani found, ∼50% of diabetes in European populations is undiagnosed. (This proportion varies with age, sex, and ethnicity).
We had similar concerns as Rathmann and Giani about applying diabetes prevalence rates from the Netherlands to other European countries, given the higher levels of physical activity and lower prevalence of obesity in the Netherlands compared with many other countries. For example, total diabetes prevalence, as derived from the study in the Netherlands, was lower in middle-aged people even when compared with self-reported diabetes prevalence from the Health Survey for England (5). A recent article (6) on the prevalence of known diabetes in eight European countries was published using data from sentinel general practices. The results suggest that the estimated prevalence of 2.7% for both diagnosed and undiagnosed diabetes for both sexes at all ages presented in our article for the Netherlands may be an underestimate, as the estimate for diagnosed diabetes for the Netherlands was ∼2.6%. In Belgium, the prevalence of diagnosed diabetes at all ages was higher at ∼3.3% than the estimate of 3.1% for both diagnosed and undiagnosed diabetes given in the global prevalence article.
The estimates of diabetes prevalence presented in the global burden article (2) are likely to represent conservative estimates for most regions as a consequence of the inclusion criteria and the need to extrapolate the limited available data, which may not reflect current patterns of diabetes prevalence. Even these conservative estimates have major public health implications. We hope that their publication will provide an incentive for better data collection on diabetes prevalence around the world, ideally as part of diabetes prevention programs.