Her report, published in this week's JAMA Internal Medicine, looked at Medicare data from the Dartmouth Atlas between October 1, 2008 and September 30, 2009, to see hospital referral region patterns of variation across the nation as a whole. For the state of Texas, Sheffield used claims data from smaller hospital service areas, so she could see practices of individual physicians who performed colonoscopies.
She discovered that Medicare beneficiaries were much less likely to have a "potentially or probably inappropriate" colonoscopy if they lived in a non-metropolitan or rural area. Practitioners who were more likely to perform potentially or probably inappropriate colonoscopies were more likely to have been graduated from medical school before 1990 rather than after, and were more likely to perform a higher volumes of the procedure on Medicare beneficiaries each year.
The data was de-identified, so as not to reveal the practice pattern of an individual physician by name.
"Our purpose was not to point fingers at individual physicians or specialties. We just wanted to examine patterns in potentially inappropriate colonoscopy, because patterns can illustrate issues in everyday practice. It can help illuminate and show the range of practice in terms of the range of inappropriate colonoscopies.
Sheffield says that it may be that colonoscopists were simply slow to adapt the recommendations to their practices in certain parts of the country. In a subset of cases, she acknowledges, there may have been legitimate reasons why a physician recommended the procedure in a patient, and perhaps failed to code it properly for the claims database.