In their report, Roglic and colleagues note that most international mortality statistics rely on the underlying cause of death as recorded on the death certificate. However, this approach has limitations and more sophisticated methods have been developed and applied to estimate cause-specific deaths for AIDS, tuberculosis and other diseases, but not diabetes.

To estimate cause-specific deaths for diabetes, the researchers used a computerized disease model that assessed disease burden using a variety of relationships, including age, sex and disease-specific mortality. They also used population-based observations and other independent methods.

The researchers estimated that overall, 7.5 million people with diabetes died in 2000. This included 4.6 million people with diabetes who died from other causes and 2.9 million deaths due to diabetes. Of these deaths, 1 million occurred in developing countries and 1.9 million occurred in developed nations.

The lowest proportion of deaths (2.4 percent) was seen in the poorest African countries and in Cambodia, Laos, Myanmar and Vietnam. It was highest (9 percent) in the Arabian Peninsula and the Americas (8.5 percent).

Overall in 2000, the proportion of deaths from diabetes worldwide was 5.2 percent. An earlier WHO estimate using routine statistics came up with a corresponding figure of 1.7 percent for 2002.