UB Researcher Observes Strong Statistical Correlation Between Prevalence of Diabetes, Air Pollution

Pollutants may be concentrated in body fat of those who are obese.

BUFFALO, N.Y. - August 2002 - A dramatic statistical correlation between the prevalence of diabetes and air pollution levels has been demonstrated by a University at Buffalo researcher who publishes his observations in the August issue of the journal, Diabetes Care.

The correlation appears in the "Letters: Observations" column of the journal in which researchers report information often as a way of suggesting areas in which further, rigorous studies should be done.

Alan Lockwood, M.D., professor of neurology and nuclear medicine in the School of Medicine and Biomedical Sciences at the University at Buffalo, and the author of the letter, stated that while the statistical analysis does not prove a cause-and-effect relationship between diabetes and air pollution, the correlation is strong enough to warrant further research.

"The significance of this relationship demands attention," he said. "The correlation between the two was striking. The probability that these two variables are not related is approximately five chances in 100,000."

It's estimated that diabetes affects more than 15 million Americans, one-third of whom are undiagnosed. The cost of the disease has been put at more than $98 billion; with diabetes accounting for 1 out of every 7 health-care dollars spent in the U.S.

Lockwood compared data showing the Toxic Release Inventory (TRI) for each U.S. state with data showing the prevalence of diabetes in the states. He found a significant relationship between TRI emissions and the prevalence of diabetes. Some heavily industrialized states, such as Ohio, for example, had high levels of TRI emissions --147 million pounds of emissions -- and a high prevalence of diabetes -- 7.5 percent of the state's population. Conversely, states with low TRI emissions -- such as Alaska, which had 2.6 million pounds of emissions and 4.4 percent prevalence of diabetes -- showed a low prevalence of the disease.

Diabetes prevalence has risen substantially over the past 10 years, according to Lockwood's letter, which cites the 49 percent increase in the disease reported in the 2000 Behavioral Risk Factor Surveillance System published by the Centers for Disease Control. Many recent studies and reports have investigated the connection between the increase in obesity and the increase in the disease. In addition, Lockwood's letter notes, "environmental toxins, notably dioxins, also have been suggested as contributing factors.

"The nature of the relationship between air pollutants and diabetes remains to be determined, but there is a potentially plausible link," he said. "What may happen is that some pollutants, such as dioxins, are being concentrated in the fat of obese people and these then may contribute to the development of diabetes. That is the hypothesis."

Lockwood became interested in a possible correlation between diabetes and air pollution while examining the Toxics Release Inventory (TRI), a list of the discharge of air pollutants over certain levels that industry must report, and which are available on the Web site of the Environmental Protection Agency.

He was studying the TRI data as part of his duties as a member of the national group, Physicians for Social Responsibility. Lockwood chairs its committee on the environment and health.

Around the same time, he said he happened to read an article in The New York Times about the geographic distribution of diabetes in the U.S. The article was accompanied by a map that showed areas in the U.S. where the prevalence of the disease was highest in 2000.

"It occurred to me that the map in The Times looked a lot like the map of the TRI data," said Lockwood.

He then tracked down the source of the data in The Times map, which was the Journal of the American Medical Association; downloaded the 1999 TRI data off of the Environmental Protection Agency Web site, and plugged these data into a popular statistics software package.

Source: University at Buffalo