African Americans with diabetes don’t fare as well as their white counterparts, and a new study looks to physician behavior for an explanation. “Is it a few physicians with a great discrepancy, or all the physicians, each with a modest discrepancy?” asked lead author Thomas Sequist, MD, MPH, an associate professor at Harvard Medical School and an internist at the Harvard Vanguard Medical Associates. What he found was that “the disparities in diabetes outcomes were spread across all our physicians.”

The study included 90 primary care doctors, each with at least five white and five black patients with diabetes. Researchers collected data on patient health measures—A1Cs, low-density lipoprotein (LDL) cholesterol, and blood pressure—as well as sociodemographic factors, like household income and type of insurance. For all health outcomes, whites were more likely than blacks to achieve recommended health targets. Sequist and his colleagues also observed that statins—the LDL cholesterol lowering class of drugs—were prescribed more frequently to whites than to African Americans. While sociodemographics could explain a third of the observed disparities, the rest were attributed to the physicians themselves.

So, is this discrimination at work? “I think it is not racism,” says Sequist. “I think the problem is not so much that we treat patients differently, but that we treat patients the same. Maybe we should treat people differently because they have different needs.”

Now Sequist is working on solutions to this problem. In an ongoing randomized trial, half the physicians from the current study received cultural competency training and went on tours through their patients’ communities. “We will see later in the year whether or not we were able to impact these racial differences in diabetes care,” says Sequist. “One of the take home points for me is: As clinicians, we should really start taking our patients’ own background into account.”

This study was published in the June 9, 2008, issue of the Archives of Internal Medicine.