The study revealed that Medicare patients in the northern regions of the nation spent more for Part D drugs for the two conditions and tended to adhere better to taking them than did patients in southern regions.

All of the 10 lowest-spending areas were in southern states, and all of the 10 highest-spending areas were in northern or central states. Of the 50 lowest-cost regions, 43 were south of the 37th parallel, which extends roughly from the Virginia-North Carolina border to central California, and 43 of the 50 highest-cost regions were above that line.

"Is it because of the patients or the physicians?" lead researcher Bruce Stuart, PhD, asks rhetorically. "Well, patients can't get the drugs without the physicians. How well they take the drugs is up to the patients. We think it is more patient behavior than physician behavior. We are trying to find out what those factors might be. Why would there be regional differences in terms of patient behavior?"

Stuart is the Parke-Davis chair in geriatric pharmacotherapy at the UM School of Pharmacy, executive director of the Peter Lamy Center for Drug Therapy and Aging, and a professor in the Department of Pharmaceutical Health Services Research.

Researchers did not find any strong evidence of Medicare savings in treating diabetes and heart failure -- savings such as lower hospital costs or fewer medical services -- as a result of higher Part D spending.

"However, this is only half a story because there are several plausible conjectures behind these findings," says Stuart. "The (Part D) program came into play in 2006 and for many of the people who got the benefit, this could be the first time they were using the drugs." Also, these drugs are primarily for long-term benefit. "The answer is likely that it takes a while for these drugs to work," Stuart says.

Stuart says analysis of subsequent years of Part D will be needed to make firm conclusions as to the payoff in hospital costs and services for people using and adhering to the heart and diabetes drugs regimens. Stuart hopes to begin a follow-up study to replicate the first one.

For the first study, Stuart and his team analyzed a 5 percent random sample of the Medicare population from a database compiled by the Centers for Medicare and Medicaid Services. They chose heart failure and diabetes because managing those chronic conditions is heavily dependent on drugs, they are very common chronic diseases, and the drugs have been proved effective in clinical trials.

Stuart says the study team formed two preliminary "bottom lines." First, although the researchers couldn't find much difference in who was taking the drugs, they clearly found that among people who used them, regimen adherence was higher in the north and that made drug spending higher. "Then we asked, 'Do people who are spending more and having higher adherence have lower spending on Part A and Part B services to treat diabetes and heart failure?' Stuart explains. The researchers did not see that relationship, but when they looked at total Medicare costs, they found that regions in the South with lower adherence had higher average Medicare spending for all A and B services compared to northern regions.

"Discovering which regional factors are responsible for differences in medication practices should be a high priority," the researchers wrote in Health Affairs.