Federal health officials yesterday issued guidelines that suggest for the first time that hospital patients be tested for the AIDS virus.

The recommendation by the federal Centers for Disease Control departs from its policy of advising against widespread AIDS testing in the nation's hospitals. But in releasing new standards for protecting the nation's 6.5 million health care workers from the spread of the HIV virus, the agency "noted the need for health care workers to consider all patients as potentially infected with HIV."

"This is a compromise, really," said Dr. Harold Jaffe, chief of the epidemiology branch of the CDC's AIDS program. "There is no real data to suggest that testing would increase the safety of health care workers. But the reality is that some AIDS testing has already begun, and this document attempts to assure that it will not result in the denial of treatment to those who test positive."

Anxiety about contracting the virus has risen sharply among health care workers since the CDC reported May 21 that three women working in hospitals had been infected with the HIV virus after each was accidentally splashed with the blood of infected patients.

They were the first reported cases of health care workers contracting the virus after a single exposure to blood by some means other than a needle prick, and reports of the accidents have heightened anxiety in many hospitals.

"There is a level of fear out there that is something we needed to address," said Jaffe.

Last month in Atlanta, the CDC convened a panel of health care experts to help draft the guidelines. The effectiveness of testing patients and health care workers was hotly debated.

The guidelines announced yesterday lay out specific criteria for hospital testing programs, among them obtaining consent, informing patients of results, assuring confidentiality and guaranteeing that a positive test result will not lead to "denial of needed care." Guidelines by the CDC do not have the force of law, but they are the standard normally followed across the nation.

Opponents of testing hospital patients for the AIDS virus say that patients who test positive will invariably be neglected and mistreated, receiving poor treatment.

"This is the first major step toward a two-class health system in this country," said Jordan Barab, health and safety coordinator for the American Federation of State, County and Municipal Employees, which represents 300,000 of the nation's health care professionals. "There will be the AIDS patients with the stickers on their charts and the special wards. And then there will be all the rest."

The new rules call for health care workers to treat all blood and bodily fluids as possibly infected with HIV, the virus that causes AIDS.

The document notes that the increasing prevalence of HIV infection in the American population means an increasing risk for health care workers.

But the guidelines also make clear that the comparative risk of contracting the AIDS virus in the workplace -- particularly if workers wear proper gear and clothing, is relatively remote.

The guidelines also urge medical personnel to use mouth shields when administering mouth-to-mouth resuscitation.

In one continuing CDC study, only three of 886 workers who have had direct exposure through cuts or needle pricks have tested positive for the virus. Every day, thousands of health care workers are splashed with bodily fluids of hospital patients.

"I know many of my colleagues are scared, and their thinking is if they can pick out the dangerous patients they can protect themselves," said Dr. Quentin Stiles, who represented the American Association of Thoracic Surgeons at the meeting in Atlanta. "But we just can't divide the world that way, and we have to face that."

Many surgeons, citing the large quantity of blood that splashes on them when they operate, have called for greater testing of patients before operations. Jaffe said the risk of AIDS transmission through such means is far too small to be of serious concern to operating room personnel.

The recommendations say that no case has been reported in which a health care worker has infected a patient, but the guidelines do note "that transmission during invasive procedures remains a possibility."

"If I were about to be operated on, I would be much more interested in knowing whether my doctor was an alcoholic, had had a fight with his wife or had stayed up all night than what their antibody status was," said Jaffe. "We would rather have omitted that issue, but it was not possible to do so."

Many people who work in hospitals and with patient records say that confidentiality, while greatly to be desired, cannot be assured. In the course of a day, many people see a patient's charts.

"Medical records are never confidential," said Dr. David Henderson, hospital epidemiologist at the National Institutes of Health Clinical Center. "We do the best we can, but if there is someone who really wants those records, they can do it."