The exercise stress test, usually included in the so-called "executive physical," provides doctors with little meaningful evidence about whether a patient has heart disease, and in many cases gives misleading results, a new study shows.

Based on case histories, stress tests and coronary angiorgrams (x-rays of the heart's blood supply) in more than 2,000 patients, who had suffered chest pains, the report, concludes that a patient's sex and the character of the chest pain are far more reliable predictors of the heart's health.

The exercise test has long been known to have high rates of both "false positives" (people with positive tests and no heart disease) and "false negatives" (people with heart disease and negative tests). Studies in 1970 showed that in patients with no chest pain (and therefore a low risk of heart disease), the test is only 35 to 40 percent accurate in predicting heart disease.

Knowing this, researchers at Boston University Medical Center asked whether the exercise test can diagnose heart disease in patients with various kinds of chest pain. They sorted more than 2,000 patients from a multi-hospital study of heart disease into four categories based on their chest pain: "definite angina" (typical cardrac pain), "probable angina," "probably not angina." and "definitely not angina."

In women with apparent "definite angina," 62 percent turn out to have heart disease. But a woman with typical chest pain and a negative stress test still has a 1 in 3 chance of heart disease, the study showed.

In patients with less typical types of chest pain, the likelihood of heart disease falls off steadily, according to the statistics. In these patients, a positive exercise test has a good chance of being a "false positive," so it again provides the physician with little hard information on which to base treatment.

Ryan said the exercise stress test has been "mistakenly used to give all-or-none answers" during the past 10 years, causing "an awful lot of unnecessary concern" to patients with few or no symptoms of heart disease.

They then looked at whether the type of chest pain correlated with heart disease, and whether a "positive" stress test (typical changes in the electrocardiogram occurring during exercise) increased the chances of heart disease showing up on an angiogram.

They found that in men with pain absolutely typical of angina, the likelihood of heart disease was so high -- 89 percent -- that a positive exercise test added little information. A negative test would have a high risk of being a "false negative," so it would be unnecessary to do the test at all, said Dr. Thomas J. Ryan, chief of gradiology at Boston University, who headed the study, published in this week's New England Journal of Medicine.

He took issue with the routine use of the test on patients with no cardiac symptoms, saying that the value of a positive test in this group as a predictor of heart disease is very low.

The exercise test can be definitive in a few patients, he said. For instance, it can serve to rule out heart disease in a middle-aged woman with chest pain not typical of angina. In such a patient, a negative exercise test can rule out heart disease to 2 or 3 percent, he said. A positive test, however, would have a high risk of being "false positive," since 53 percent of positive tests in women are false positives.