DURHAM, N.C. - Insurance providers face significant challenges in reliably assessing the quality of individual surgeons' patient outcomes, and need better ways to measure physician quality when selecting the best surgeons for their members.

This is according to a statistical analysis of more than 220,000 coronary artery bypass graft (CABG) procedures performed in 75 Florida hospitals between 1998 and 2006. Marco Huesch, a physician and assistant professor of strategy at Duke University's Fuqua School of Business, conducted the analysis, which was published in the December issue of the American Journal of Managed Care.

"Under the American managed care system, patients have little choice between providers for major surgeries, as most of us are limited to surgeons contracted by our insurance plans," Huesch said. "While it might be natural to assume that health plans select surgeons based on the quality of their patient outcomes, as it turns out, it's generally almost impossible for an individual company to do so."

Huesch's analysis of rates of in-hospital mortality did find differences in outcomes between surgeons. However, it also confirmed that the "law of small numbers," a statistical term for a situation where there are too few items in a sample to draw reliable conclusions, would prevent insurers from accurately assessing the quality of the Florida surgeons' care.

"While we found differences in mortality rates across physicians, no single insurance provider had enough patients undergoing surgeries by the same physicians in order to generate statistically significant data against which it could judge outcomes," Huesch said. He said the current effort to reform U.S. health care is likely to exacerbate the situation. "As we create new insurance entities to cover previously uninsured Americans, we'll simply have even more data being collected by separate companies, none of which is being combined in a way that it can be usefully evaluated."

"These findings are distressing because they show how far removed we still are from having a health care marketplace where patients and their insurance companies can make purchasing decisions based on the quality of physicians' work," Huesch said. "If a particular model of car is unreliable, consumers are able to learn that and then weigh the benefits of owning that car against the reliability concerns, but we don't currently have information available to make a similar process possible in health care."

Huesch suggests that there are other ways to aggregate data to inform patient and insurance company decisions.

"The simplest way to address this would be to have insurance companies pool their surgical outcome data, creating a larger picture of each individual surgeons' patient population and aggregate outcomes," Huesch said. "The current focus on improving health information technology should, in theory, make it easier to collect and analyze this type of information in an organized way that still preserves patient confidentiality."

Huesch received no external funding for this research.