In the hospital cafeteria and the medical school amphitheater, it is a frequent question: What if a surgeon gets AIDS? Could he or she continue to practice or would patients be in danger if, for example, an infected surgeon cut himself and bled into the patient's wound?

Now, according to a report in today's Journal of the American Medical Association, there is evidence that the risk of a patient's being infected by a general surgeon who carried the human immunodeficiency virus (HIV) is extremely low.

The study, the largest to look at patients of an HIV-positive doctor, involved a general surgeon in Nashville who performed about 300 operations a year during the estimated seven years he was infected. Researchers offered free testing for the virus to the doctor's patients. Of the 616 patients who agreed to be tested, only one, an intravenous drug user who may have had AIDS before his surgery, was HIV positive.

The study was done by William Schaffner, a specialist in infectious disease at Vanderbilt School of Medicine in Nashville. Schaffner and others say the study should help the federal Centers for Disease Control and groups such as the American Medical Association decide what to recommend to hospitals with HIV-positive doctors on their staffs. Current CDC recommendations specify only that decisions regarding what procedures an HIV-positive surgeon may perform "must be determined on an individual basis" -- a recommendation some say is too vague, given that there may be hundreds of practicing surgeons infected with HIV.

"Most hospitals have not really formally addressed it," said Frank Rhame, director of the HIV clinic at the University of Minnesota Hospital, who wrote an editorial accompanying Schaffner's report. The CDC statement "is one of those phrases that sounds like you're saying something but there's no content to it; they don't tell you how you're supposed to do it."

To date, there are no reported cases of AIDS being transmitted from a health care worker to a patient in a hospital setting.

The Nashville surgeon died in January 1989. Researchers say he may have been infected as early as 1982. They were able to identify 2,160 people he had operated on between then and September 1988, when he stopped practicing. Of those, 264 had died and 244 could not be reached. None who died had AIDS or an AIDS-related condition listed on their death certificate.

The study is limited because only 37 percent of the people contacted agreed to be tested for the virus, meaning researchers were unable to study a significant proportion of the people who may have been exposed to the surgeon's blood. Also, the researchers were uncertain as to when the surgeon became HIV positive.