Health Care Policy

Community Physician Input Could
Improve HMO Practice Guidelines

A five-state study by Harvard researchers suggests that in certain cases, medical practice guidelines and the treatment recommended by practicing physicians may be two different things.

The researchers say that in other cases, for which clinical research is more complete, the practicing physicians closely agreed with clinical experts, who develop the majority of these guidelines.

The study found that practicing physicians and clinical experts (who may be more familiar with published research) disagreed on how to treat the most elderly heart attack patients. The inconsistency suggests a potential conflict between the treatments practicing physicians recommend for this patient population and the practice guidelines that insurance companies use to determine which treatments will be covered.

To remedy the situation, the researchers say that evaluations of "best" medical practices should be founded on the beliefs not only of expert panels but also of physicians practicing in the community. Including the practitioners' perspective would increase the breadth of clinical experience the guidelines represent. The researchers also recommend that more clinical research be done on the most elderly patients--those 75 and older--so medical literature can reflect more complete clinical results.

John Ayanian suggests that community physicians contribute to practice guidelines.

"This is one of the first large-scale studies on practice patterns to take into account the beliefs of practicing physicians," says lead author John Ayanian, assistant professor of medicine and health care policy at HMS and Brigham and Women's Hospital. "It is important for policymakers to understand practicing physicians beliefs if they want to improve care by altering physician behavior and decision-making," he says.

The study appears in the June 25 New England Journal of Medicine along with a second article and an editorial on quality of care. The second article, co-authored by Lucian Leape of the Harvard School of Public Health, compares the beliefs of different expert panels.

The Physician Raters

The Medical School researchers presented a mix of cases to 1,058 practicing physicians and a panel of nine clinical experts. The physicians rated the appropriateness of coronary angiography for 20 categories of patients who had recently had an acute myocardial infarction. This procedure, x-ray imaging of the coronary arteries after injection of a contrast material, helps in diagnosing the cause of the heart attack and in planning further treatment. The practicing physicians included internists, family practitioners, and cardiologists in California, Florida, New York, Pennsylvania, and Texas. The nine experts also represented a range of specialties.

Though there was general agreement between the practicing physicians and expert panel for most case categories, those in which the patient was 75 or older and had no complications resulted in significant variation between the two groups.

Ratings of the Appropriateness of Coronary
Angiography After Acute Myocardial Infarction
by Surveyed Physicians and an Expert Panel

Clinical Indication	Patient's Age (Yr)	Onset of Symptoms (Hr)	Thrombolytic Therapy	Complications	Rating by Surveyed Physicians	Rating by Expert Panel
					Median (interquartile range)	Median (interquartile range)
K	>=75	<6	Contraindicated	None	6 (3-8)	3 (3-5)
L	>=75	<6	Contraindicated	Persisten Chest Pain	8 (7-9)	7 (7-8)
M	>=75	<6	Not Contraindicated or Administered	None	5 (3-7)	2 (2-4)
N	>=75	<6	Not Contraindicated or Administered	Persistent Chest Pain	7 (5-9)	7 (6-8)
O	>=75	<6	Administered	None	2 (1-4)	1 (1-3)
P	>=75	<6	Administered	Persistent Chest Pain	8 (7-9)	7 (4-8)
Q	>=75	>12	Not Administered	None	4 (2-6)	4 (2-5)
R	>=75	>12	Not Administered	Persistent Chest Pain	8 (7-9)	8 (7-8)
S	>=75	>12	Not Administered	Persistent Pulmonary Edema	8 (6-9)	8 (5-9)
T	>=75	>12	Not Administered	Stress-induced Ischemia	8 (7-9)	6 (6-7)

Note: The ratings were on a 9-unit scale ranging from very inappropriate (1) to uncertain (5) and very appropriate (9).

The chart shows only that half of the hypothetical cases studied in which the patient was 75 or older.

"One potential explanation is that practicing physicians have more experience in seeing positive benefits of angiography for older patients, while the expert panel may be following the medical literature more closely. We don't have an abundant set of data to guide us for older patients," Ayanian says. "We need to improve the evidence available through clinical research to guide us in the care of these patients."

New Tool for Health Services

One of the noteworthy aspects of the study was the use of hierarchical regression modeling, a statistical technique that was developed more than a decade ago but only recently applied to health services research. It identifies the source of variation in treatment recommendations by establishing a "hierarchy," or nested set of effects associated with physician characteristics. The method enabled the researchers first to identify which patient characteristics (such as the presence or absence of complications) influenced physicians' recommendations. Then the researchers could look across physicians to determine how the relationship between physician beliefs about the procedure's benefit and patient characteristics varied by physician specialty, state, age, and sex. "The effects were estimated using a computer simulation technique that cycles through all of the ratings and determines whether there is a propensity for a particular physician to rate high, for example," says co-author Sharon-Lise Normand, associate professor of biostatistics in the HMS Department of Health Care Policy. This propensity could be correlated to the physician's characteristics.

The technique showed that disagreement also existed within the group of practicing physicians. Their varying expertise and practice environment seemed to influence their beliefs. For complicated cases, cardiologists rated coronary angiography as more appropriate than did primary care physicians. Even among cardiologists, marked variation existed for uncomplicated cases: those who performed invasive procedures gave higher ratings for the appropriateness of angiography than did cardiologists who did not. Furthermore, physicians from hospitals providing invasive treatments like coronary angioplasty and bypass surgery rated angiography as more appropriate in uncomplicated cases than did physicians from other hospitals. Physicians working in health maintenance organizations rated angiography as less appropriate than did other physicians.

The 20 case categories in the study are defined by sets of clinical indications: patient age, time of heart attack, status of thrombolytic therapy, and presence of complications such as chest pain. The physicians rated the appropriateness of coronary angiography for patients in each category using a nine-unit scale with 1 being very inappropriate; 5, uncertain; and 9, very appropriate.

Other study co-authors, all from HMS, were Mary Beth Landrum, assistant professor of biostatistics; Edward Guadagnoli, associate professor of health care policy; and Barbara McNeil, chair of the Department of Health Care Policy. The study was supported by a grant from the federal Agency for Health Care Policy and Research.

--Robert Neal