Brent James, M.D., likes to tell people how sharing data with physicians can transform and improve the way they practice medicine.
Once, for example, Intermountain Health Care in Salt Lake City, where James is vice president for medical research, studied transurethral prostatectomies and discovered that one physician's approach to the surgery was more than twice as expensive as another's.
By looking at clinical data that chronicled how the physicians cared for patients, James found that one of the physicians kept patients in the hospital almost twice as long as the other. The reason for the variation was that the physicians had different ideas about when to remove a catheter following surgery.
To settle on one timetable, Intermountain ran a small clinical trial to test each physician's theory. It found that removing catheters earlier -- and sending patients home sooner -- had no negative effect on care. That's when the physician with the more costly method changed the way he practiced, James says.
"What we do is look for differences in practice and use them to drive clinical discussions," James says. "My job is to create an environment where physicians can learn from one another."
James is a pioneer of physician profiling, a term that means using data to compare physicians to one another or to a certain standard of care, such as a practice guideline. He's been at it since 1987, the year he conducted the prostate surgery study. Since then, he's led a charge at Intermountain -- a chain of 23 hospitals and 75 clinics in Utah, Idaho and Wyoming -- to use profiling as a way to improve quality of care. Intermountain employs 400 physicians and is affiliated with another 1,900 who practice in its hospitals.
Variations on physician profiling have caught on across the country. Organizations ranging from insurers to providers to professional associations are talking about it or doing it with differing levels of sophistication, says Donald Steinwachs, director of the Health Services Research and Development Center and chairman of the Department of Health Policy and Management at Johns Hopkins University in Baltimore.
Using data gathered mostly from claims, groups are putting together reports on physician performance that cover a wide range of factors, such as how many hospital days they use, and then feeding the statistics back to the physicians.
The goal is to identify those doctors who are using more resources than others and to urge them to change their behavior to save costs. But when profiling is done right, it also focuses on improving quality of care, James and Steinwachs say.
Since James first focused on profiling, his work has changed from comparing doctors to doctors to comparing doctors to clinical guidelines or protocols. He meets with clinical teams of doctors and nurses to establish appropriate step-by-step guidelines for specific procedures and diagnoses. Current profiling efforts at Intermountain are being conducted in community-acquired pneumonia, chest surgery, labor and delivery, and antibiotic use.
The goal of profiling, James says, is to identify variables -- what physicians do differently -- and then try to understand and learn from the reasons behind their decisions.
From a larger perspective, profiling is part of several general healthcare trends, including:
"This represents a shift from focusing just on people coming into the office to focusing on the whole population enrolled," says Gail Amundson, M.D., HealthPartners' associate medical director for quality and utilization.
For example, while some organizations send out stacks of paper containing detailed physician reporting, HealthPartners' medical groups have access to their profiles through an intranet, or private computer network. The groups not only see general reports, but also can use the system to get information on other statistics that interest them, Amundson says.
Reports that profile physician behavior come from a variety of sources. Insurance companies often profile physicians and set up programs to encourage them to meet certain statistical goals.
HealthPartners, for example, profiles group practices by measuring them against established practice guidelines. It then ties financial incentives to meeting target goals in areas such as care of diabetics, mammography and tobacco-use counseling, Amundson says.
The HealthPartners data offer medical groups a complete picture of one set of the practice's patients. For example, if a woman gets a mammogram at a mobile unit rather than scheduling it through her primary-care provider, the insurer has that information and it will show up on the profile.
But data from one insurer represents only a fraction of a physician's practice, so it's important for physicians to get wider perspectives as well, experts say. A few physicians are compiling their own small databases of information to profile themselves for marketing and contract negotiation purposes, says Susan K. Bellile, president of Q3, a Westchester, Ill.-based consulting firm that specializes in managing information for physicians.
Other organizations, such as the Medical Group Management Association and physician practice management companies, are stepping in to profile physicians' entire practices. The MGMA recently completed a 42-month demonstration project comparing 77 practices in four states based on size. Some of the practices studied have as many as 300 doctors. The association hopes to offer a national profiling system next year, says Donna L. Burman, research database administrator.
PPMs can offer a full view of a practice's performance, says Terry Fouts, M.D., senior vice president of medical management company PHP Healthcare Corp, Reston, Va. "We have a lot better access to data that can help physicians be more efficient because we do their billing, have an electronic medical record and see all of their pharmacy encounters," he says.
Many organizations strive to get physicians involved in deciding which statistics profiles should include. The possibilities are endless, says J. Leonard Lichtenfeld, M.D., chief medical officer and senior vice president for physician services at Cornerstone Physicians Corp., an Irvine, Calif.-based PPM.
Cornerstone manages more than 130 physicians in four group practices and recently began a profiling program.
One of the first steps has been to ask practice managers what statistics they would like to see, Lichtenfeld says. "We want to provide physicians with as much information as we possibly can, but there are so many things that you can measure."
Deciding which statistics to include in a profile depends on the physician and the practice, he says. Among the options: Profile the number of outpatient encounters a physician has or how many hospital days are being used. If a group has a capitated contract that covers pharmacy costs, physicians would benefit from seeing their prescription data, Lichtenfeld says.
Another consideration is that information on a profiling report isn't useful unless it's easy to understand. When the MGMA compared practices based on relative value units -- a government measurement intended to assign standard values to such things as physician office visits -- the association was surprised to discover it had to explain the RVU concept to physicians because they didn't understand how it worked, Burman says.
Still, RVUs, particularly a subset called "work RVUs," are a good way to see who is pulling his or her weight in a group practice and who is not, Lichtenfeld says. Cornerstone was part of the MGMA project.
Bellile advises physicians to do their best to decipher profiles, even when the numbers are difficult to understand. Sometimes the profile reports are wrong or, at the least, misleading, she says. For example, one urologist received a report that showed he had a higher than average hospital readmission rate. He investigated and found that the higher number was coming from patients with kidney stones, whom he had managed correctly by sending them home after treatment to see if they would pass the stones on their own. Though it was expected that a few would need to return to the hospital to have the stones removed surgically, when they did, they showed up as "readmitted" in the physician's profile.
"The report suggested that readmission was bad, when in this particular specialty and in this particular case, patients were actually being treated conservatively," Bellile says. "So the doctor wrote a note to the payer who created the report and said, 'Well I guess you'd just rather I'd operate on everybody right away when they first show up because then I'd have no readmissions.' "
Sometimes there isn't enough time or data to put together accurate profiles, Intermountain's James says. Many organizations rely solely on claims data, which can be a good starting point for profiling but has many limitations.
Insurance companies and other organizations find claims useful because they have compiled a lot of them, Bellile says. But very little information is actually available on a claim. For instance, a claim may include just one diagnosis, a procedure code and the amount charged. That leaves out a lot of clinical detail, she says.
When using claims for profiling, it's important to adjust the data to allow for complex cases, Johns Hopkins' Steinwachs points out. The university has done several studies on case-mix adjustment.
If a patient has multiple conditions, but only one diagnosis is listed on the bill, profiling may be skewed; it may appear that the physician is using more resources than necessary. Adjusting for difficult cases increases the chance that profiling comparisons are fair.
As more and more clinical functions become automated, and electronic medical records systems become more widely used, physician profiling will become more useful and complete, Intermountain's James predicts. Intermountain's quality initiatives often are based on data from clinical information systems, for example.
But a better understanding of how best to use the data needs to evolve, too, James says. Organizations have to think about data as a way to start discussions about care delivery, not just to identify practices that deviate from the norm, he says.
Profiling is here to stay. The hope, James says, is to eventually empower physicians to hold one another accountable for the care they provide.
"Physicians need to be in control of this," he says. "It's not appropriate for a health plan or an integrated delivery system to tell physicians how to practice medicine. That's the job of the medical profession.
"The aim is to have the health system or health plan support the medical profession in that role. If you set (profiling) up right, it becomes a professional exercise in finding the best care for patients. That's how it ought to work."
MargaretAnn Cross is an Allentown, Pa.-based freelance writer.