Duke Study Finds Evidence-Based Medicine had Little Influence on Patient Care
ANAHEIM, CA -- The ability of evidence-based medicine to change physicians' practice habits was tested in a novel little study - and lost.
Researchers at the Duke Clinical Research Institute wanted to see if findings from an important trial on the best way to treat heart disease in diabetic patients actually changed the course of medicine. It didn't, due to physicians believing they knew the best way to treat their patients.
The Duke researchers prepared the results of their experiment for presentation Monday at the annual scientific meeting of the American College of Cardiology. They said the findings could hint at a more widespread problem in medicine.
"Physicians knew about the study and its results, but it had no measurable impact on their treatment choice," said cardiologist Dr. Darren McGuire, who performed the analysis with his Duke colleagues, Drs. Kevin Anstrom and Eric Peterson. "It appears that they continue to believe what they are doing is right, even in the face of evidence to the contrary."
The researchers looked at results of a five-year-old, $30 million, federally funded BARI (Bypass Angioplasty Randomized Investigation) trial, considered a milestone study in the treatment of diabetic patients. While the overall trial looked at the difference in outcomes between heart bypass surgery and angioplasty, a sub-analysis found striking differences in how those procedures affected diabetic patients. It concluded that these patients did best if they received heart bypass surgery instead of angioplasty to open clogged heart arteries.
On Sept. 21, 1995, the National Institutes of Health, which had sponsored the trial, released a "clinical alert" to physicians nationwide saying that bypass surgery "should be the preferred treatment for patients with diabetes on drug or insulin therapy who have multi-vessel coronary artery disease and need a first coronary revascularization."
It is theorized that diabetic patients have more diffuse disease as well as smaller blood vessels, so offering them a new artery works better than trying to eliminate blockages, and the study supported the notion, McGuire said.
The next step for the Duke researchers was to study care patterns at 13 leading hospitals to find out if treatment had changed as a result of the study. The hospitals all belong to the National Cardiovascular Network (NCN), a voluntary collaboration of 22 medical centers that all perform "high volume" of heart procedures and share clinical data on their outcomes in an effort to continually improve quality.
The researchers found that, all in all, there was no increase in use of bypass surgery for diabetic patients in the years after the BARI trial. Eight hospitals had about the same mixture of surgery and angioplasty as before BARI; three centers increased the number of surgeries they offered to diabetic patients, relative to angioplasties; and two actually increased the number of angioplasty procedures for diabetic patients in direct conflict with the clinic alert, McGuire said.
The good news is that the centers, as a whole, had already offered surgery more often for diabetic patients even before the BARI results were known, McGuire said. However, the Duke researchers also documented "tenfold variation" among centers in how likely angioplasty was used in these patients. "The bad news is that hospitals didn't do better after the NIH's clinical alert," said Peterson.
McGuire then phoned cardiologists and surgeons at the 13 centers to ask why that was. He found out that everyone he called knew about the study and believed the results were valid. They also agreed that the results could apply to their patients, "but yet it did not affect their own personal treatment choices," he said.
Surgeons told McGuire that while they wholeheartedly embraced the BARI findings ("understandable because it endorses surgery," McGuire said) cardiologists, who see patients first, were not referring them on to surgeons. McGuire said that cardiologists explained their reluctance to give up angioplasty for these patients because newer techniques, such as the use of stents and glycoprotein platelet inhibitors, have likely lessened the negative impact these procedures have on diabetics.
"But there is no evidence that says these new techniques improve the outcomes of angioplasties in diabetic patients," McGuire pointed out. "They don't wait for new data to emerge, but just go forth, believing what they believe.