Physicians are not delving into medical records to reduce the large variances in clinical practice that persist in every community, despite plenty of written documentation at their disposal to explain the differences.
That's a missed opportunity doctors can rectify by sharing their data with each other and developing objective ways to investigate variances in a nonjudgmental manner, says Steven Berkowitz, M.D., president and medical director of Quality Care Network, an Austin, Texas-based physician-practice foundation.
In a presentation from 11: 45 a.m. to 1 p.m. Tuesday, Oct. 23, Berkowitz will draw on his experiences in directing clinical outcomes improvement and practice efficiency efforts at four-hospital St. David's Healthcare Partnership, Austin, and elsewhere. (The session repeats at 3 p.m. on Oct. 23.)
Whether they're multispecialty medical groups, independent practice associations or hospital-based staffs, physician organizations are tolerating complication rates three to four times the norm for certain procedures or treatments, Berkowitz says. Often the variances are traceable to a handful of otherwise good doctors who need to modify the way they're performing a particular procedure.
One Austin-area hospital was recording 44% more complications of normal vaginal deliveries than expected, and a team of clinicians searched through medical records to find out why. They found that four of 12 physicians on staff accounted for two-thirds of the complications, and most of the problems were the result of a specific procedure. The department chairman reviewed the procedures with those doctors.
Not all variances indicate problems, Berkowitz says. "A variance is not good or bad-it's simply an observation," he says. For example, the advanced age of a patient or involvement in a serious auto accident can explain a poor outcome. But physicians have to lead the effort to explain big differences in outcomes, Berkowitz says. "The unexplained variance is the enemy of the quality- improvement process."
The first step in identifying valid variances is to share data throughout the practice or healthcare organization, he says. Too small a sample will result in inaccurate conclusions.
A study seven years ago of procedures related to enlarged prostates revealed a 3% to 4% complication rate at one Austin hospital-but a 33% rate for one doctor who happened to have one complication among the three procedures done at that facility. If the study had been expanded to other hospitals where the doctor performed scores of such procedures, his record would not have been statistically skewed, Berkowitz says.
But to set the stage properly for valid studies of local variances, "physicians have to feel comfortable having their records reviewed by other physicians," he says. That's the only way to learn from each other and allow best practices to settle in among the region's doctors.