The new Medicare payment system substantially overpays some hospitals and underpays others because it ignores differences in the severity of an illness, paying the same for every patient with a given diagnosis, according to Johns Hopkins University researchers.

Susan Horn, principal author of a study published today in the New England Journal of Medicine, said the current study and one to be published later show that under the Medicare system, one hospital would receive 59 percent more per case than it should, based on its real costs for patients. Another would receive 38 percent more.

On the other hand, a third hospital the researchers looked at would receive 26 percent less, and a fourth 22 percent less.

According to the Hopkins team, the distortion results from the fact that the new Medicare prospective payment system requires that Medicare recipients who are hospitalized be classified into one of 468 disease categories. Then, with some adjustments, the hospitals are paid the same amount for each patient in one category. The payment rates are set annually.

As a result, Horn said in a briefing Tuesday, a hospital with a large number of seriously ill Medicare patients in a given category -- heart attack, for example -- who stay a long time and require extra nursing care, procedures and medicines, receives no more per patient than a hospital whose heart-attack patients are much less severely ill.

Horn said this produces unintended windfalls for some hospitals that could be reduced by using a severity-of-illness index worked out by the Hopkins team.

Following the index, hospital personnel would rate each admission for severity (on a scale of 1 to 4) in seven categories, including stage of the illness, other medical problems than the one the patient is being admitted for, the amount of extra care the patient is likely to need, and possible complications from treatment.

The Hopkins team used this index for 57,245 patients at six varying hospitals.

It found, according to the New England Journal of Medicine article, that instead of all hospitals having about the same proportion of severe cases in each category -- or teaching hospitals having the sickest patients -- there were wide and unexpected variations. For example, in one category the team examined, "heart failure and shock," the severity rating (on the 1-to-4 scale) was 1.5 for one hospital and 2.17 for another with more severely ill patients.

Looking at all the categories, the Hopkins group found that at one university teaching hospital, normally thought of as handling the toughest cases, 66 percent of cases were in the least-severe category.

At the same time, a community nonteaching hospital of the type normally thought to handle less complicated cases had 64 percent of its cases in the least-severe category.

Gregory Bulkley, a physician at Hopkins who participated in the study, used two of his patients, Jack and Eleanore Baldwin, to illustrate how failure to consider severity could distort payments.

Jack Baldwin, who was hospitalized for a bowel tumor, had numerous complicating conditions, including emphysema, a history of stroke, heart attack and a bleeding ulcer. The cost of treating him was $22,447, Bulkley said; the Medicare payment would have been only $10,265.

On the other hand, Eleanore Baldwin, whom Bulkley treated for gall-bladder problems requiring surgery, was in such excellent general health that the Medicare payment did not exceed costs.

Dr. Robert J. Rubin, a former assistant secretary of health and human services who helped design the Medicare prospective payment system, said Horn's conclusions are a worthwile research finding and "the absence of a severity index was always understood to be a serious flaw" in the payment plan.

But he said her findings "may be difficult to translate into a practical solution to the problem" she has outlined, because "at present there is no severity measure that is administratively simple and inexpensive, including hers."