CHARLES NEVER knew that his doctors and nurses had made the most important decision of his life for him, without his knowledge or consent. When he started to die, sitting in a chair in his hospital room, the medical staff assumed that he wanted to be revived with life-support equipment.

According to his wife, no one had ever asked the 72-year-old under what circumstances he would want life support started -- and, more importantly, stopped.

Yet for 10 days, medical personnel used an array of equipment to keep Charles alive even as his wife pleaded that he be allowed to die unencumbered. Charles thus became one more example of a tragic and increasingly common medical dilemma: The technology that saves viable patients is also sometimes a grueling trap for the elderly or those in the final stages of a chronic disease. And the Supreme Court did nothing to resolve the dilemma by ruling in June that a state can prevent family members from ending a patient's life support unless there is "clear and convincing proof that the patient wants it ended."

Tall and bone-thin, Charles had been admitted to the hospital in early February because he had refused to eat for almost two weeks, saying that he wanted to die and that he felt too sick to eat. Now, in the hospital, one of his doctors had written "suspected neoplasm" -- cancer -- in his chart.

On Feb. 17, at 10:15 a.m., his nurse found him sitting in a chair in his room, pulseless and not breathing. She quickly pushed a blue button on the wall. Immediately, the hospital operator announced Charles's approaching death over the intercom in the same tone that she announced license numbers of cars in the parking lot with their lights on: "Code blue, Room 420. Code blue, Room 420."

Other nurses ran to the room and helped lift Charles onto his bed. They struggled his dead weight into a sitting position so that they could lay a hard board under his back; he flopped down on it when they let him go. His open eyes never blinked, and his big, black pupils did not shrink from the fluorescent lights above him.

A nurse put her hands over his heart and, with arms straight, elbows locked, leaned her whole weight into his chest, then let it spring back, to force blood through his heart and blood vessels. She counted each compression to herself in rhythm, like a chant: "One-one-thousand, two-one-thousand . . . ." A respiratory therapist covered Charles's mouth and nose with a black rubber mask and punctuated every fifth chest compression with a hiss of oxygen squeezed from a green plastic bulb into the dying man's lungs. After each "breath," Charles's chest heaved unnaturally back into place.

A nurse from the intensive-care unit and an emergency-room doctor arrived within seconds. The ICU nurse attached wires to Charles's chest from a special briefcase-size heart monitor. A green line blipped across the black screen every few seconds. Soon it became a rapid succession of peaks and valleys, then a coarse line.

At 10:18 a.m., the ICU nurse, on the doctor's signal, pressed a metal paddle on each side of Charles's chest. She pushed a button on one paddle to deliver an electric shock through his heart, to try to jolt it back into a normal rhythm. The shock made a loud, dull thump as his upper body lurched. The green line on the monitor again showed peaks every few seconds. His glassy eyes reflected the fluorescent tubes in the ceiling.

Meanwhile, another nurse had managed after several tries to slip a needle into a vein in Charles's arm so that IV drugs and fluid could be poured into his bloodstream. His blood pressure, a measurement of how much blood is being pumped to the body's tissues, was zero over zero. Starved of blood and the oxygen it carries, his brain, kidneys and other organs could not survive more than a few minutes. By 10:22 a.m., the doctor had tilted back Charles's limp head to open his mouth and forced one end of a stiff plastic tube, about 12 inches long and an inch-and-a-half in circumference, down Charles's throat, through his vocal cords and into the entrance of his lungs. The other end of the tube stuck out of his mouth and was connected to a ventilator, a waist-high metal box that pumped oxygen and air into his lungs at a rhythm set by the doctor. Charles's complexion remained grayish, but his lips lost their blue tinge and became dusky beige. Long strips of adhesive tape were stuck to his cheeks, then wound around the protruding end of the tube to keep it anchored in his throat.

At 10:23 a.m., the green line on the monitor had deteriorated into a series of rough waves, and the ICU nurse sent another shock through Charles's heart. This time, it went into a fast, almost normal rhythm. In the past eight minutes, he had had three injections of adrenaline to stimulate his heart and two injections of sodium bicarbonate to counteract the acid that had accumulated in his blood. Lab technicians and the doctor repeatedly drew blood from Charles's veins and arteries to check his blood's acid and oxygen levels.

At 10:37 a.m., his blood pressure still read zero over zero, but Dopamine, a powerful blood-pressure-stimulating drug, had been dripping into his vein. By 10:40 a.m., his blood pressure was 40 over zero. (Normal blood pressure varies, but a common standard is 120 over 80.) His wrists were tied to the bedrail in case he woke up enough to try to pull out the tube in his throat or the IVs in his veins.

Charles was taken, still comatose, to the intensive care unit, where a catheter was passed into his bladder to drain his urine, and a long, thin tube was passed down his right nostril and into his stomach to drain his gastric fluids. Both tubes were taped in place to allow continuous drainage.

At about 11 a.m., violent seizures began to jerk Charles's body. Such seizures, continuing for long periods as Charles's did, change the body's metabolism and starve the brain of glucose, the sugar that feeds it, causing brain damage.

At about 1 p.m., almost three hours after Charles had been started on life support, Charles's wife told the doctor that she didn't think her husband would want to be kept alive "as a vegetable, by machines." Based on her request and Charles's poor prognosis, the doctor wrote a do-not-resuscitate order on his chart. But it was a pointless gesture: Do-not-resuscitate orders mean that if a patient becomes gravely ill and is not expected to survive, he won't be started on life-support equipment. For Charles, it was too late: Once life-support equipment has been started, legal and ethical issues make it hard to withdraw, even if the patient himself requests that it be stopped.

Almost a week later, Charles's condition had not changed. A technician injected a radioactive liquid into his bloodstream and used a special detector to trace the substance; it showed radioactivity flowing in his brain, meaning that some blood was getting through. Despite a neurologist's note that Charles's prognosis was "extremely grave," despite the fact that Charles was in a profound coma, unable to breathe on his own, or to maintain his blood pressure without a powerful IV drug, and despite his wife's request that he not be kept alive "as a vegetable," the life support could not be removed, according to his doctor, because the radioactivity study had shown that some blood was getting through to Charles's brain.

On Feb. 26, 10 days after his nurse had found him cold and pulseless in his chair, Charles died. All of his life-support was still in place. He never roused from his coma. His doctors still are not sure what caused his death. Charles's case is not uncommon in today's high-technology intensive-care units, where the burden is on patients and their families to know beforehand which life-sustaining measures they do or do not want used on them. Health-care personnel must assume that, unless specified otherwise in writing by his or her doctor, every patient wants every available technology to be used under every circumstance. As a result, the same life-support measures used on a young accident victim with a good chance of survival also will be used on an elderly patient in the final stages of heart failure who may have come to the hospital hoping only for help to die more comfortably.

The exception is the patient who has given the doctor a document called an advance directive, such as a living will. But even these documents do not ensure that patients can avoid life support. Moreover, many doctors do not discuss advance directives with their patients. Patients are expected somehow to know not only that they have entered into an unspoken agreement that they want every type of life support used on them in every situation, but they also are expected to know about sophisticated medical technology so that they can choose which treatments they don't want.

A study in the Journal of the American Medical Association (JAMA) last Nov. 3 reported that only 28.6 percent of physicians surveyed had discussed advance directives with 25 or more patients. Only 14.5 percent of the physicians had 25 or more patients who had documented advance directives. Almost half of the doctors said that at least half of the time, the patient or his family had initiated the discussion rather than the physician; only 6 percent said they had initiated all discussions.

Living wills, which are considered legal documents in 39 states, are the most common form of advance directive; a less-used, although a more comprehensive and potentially more effective document, the durable power of attorney for health care, is considered legal in 13 states. Yet an article in the same JAMA issue reported that a University of Colorado at Denver study "found that 23 percent of Colorado physicians are not familiar with living wills and that 74 percent are not familiar with durable powers of attorney; further, the study showed that 74 percent of these physicians do not generally discuss advance directives with patients."

In 1988, the organization that sets standards for hospitals, the Joint Commission on the Accreditation of Health Care Organizations, began requiring hospitals to adopt policies recognizing do-not-resuscitate orders. But a study in the November 1989 JAMA showed that less than two-thirds of the nation's hospitals had complied. Into this thicket stepped the Supreme Court on June 25, ruling 5 to 4 in Cruzan v. Director, Missouri Department of Health, that competent patients have a constitutional right "to refuse lifesaving hydration and nutrition" but that -- absent proof of the patient's wishes -- the state's interest in preserving life takes precedent over "the views of close family members about what their relative would want." Yet that surely is not the last word on what Chief Justice William Rehnquist called "a perplexing question with unusually strong moral and ethical overtones."

For one thing, even advance directives are not foolproof. Consider the example of a patient who is living at home or in a nursing home during the final stages of a chronic or terminal disease, who has decided that he does not want life-support equipment used on him and who is brought to an emergency room during an acute episode of illness. Unless the patient brings his written advance directive with him to show to the unfamiliar emergency-room doctors, he is likely to end up on life support -- if those doctors judge that he needs it and that he is too confused to make decisions for himself. Yet ironically, at this very time when he needs most to make his wishes known, the dying patient is almost certain to be confused, frequently because of low oxygen levels.

Once a patient is put on life support, it is not likely to be withdrawn, even if he has a written advance directive, is not confused and demands to have the equipment removed. Even states where living wills and durable powers of attorney for health care are legal, doctors still can override the wishes of the patient. Most physicians will not withdraw life support unless tests show, beyond doubt, that the patient is brain dead. Ironically, such patients, unaware of pain, are least likely to suffer from the discomfort caused by invasive life-support equipment; it is the conscious or semiconscious patient who can feel its discomfort, and when he tries to pull the tubes out himself, it is this patient who will have his hands tied down, adding to the discomfort he already feels.

Another problem with advance directives is that life support may have started as routine medical intervention; it does not always have a clear starting point where a decision can be made. For example, if a doctor wants to supplement a non-critical patient's nutrition with artificial feedings, he can order the nurses to insert a tube down the patient's nose and into his stomach, tape it in place and start the feedings. Or if a patient's breathing becomes labored and the oxygen content of his blood drops to dangerous levels, the doctor can insert a breathing tube -- the same kind Charles had -- down the patient's throat and into his lungs and connect the tube to a ventilator to breathe for him. And a doctor can start a Dopamine IV on a patient if his blood pressure starts to drop. All of those measures can be started without the signed consent of the patient or his family; they can be started on a confused or comatose patient without notifying the family first. If the patient's situation becomes hopeless, these treatments, which began as routine measures, are given another name: life support.

For people who undergo resuscitation, the statistics are grim. A study in the July 1989 issue of The American Journal of Medicine showed that 79 percent of the patients resuscitated in one hospital survived only three months or less -- excluding patients with advanced cancer, severe brain injury or end-stage heart disease. If they had been included, the percentage of nonsurvivors almost certainly would have been higher. The study also showed that only 10.5 percent of resuscitated patients 80 years or older were alive three months later.

Despite a 545-page report on life-support concerns published in 1983 by a presidential commission, scenarios such as Charles's are replayed over and over in intensive-care units. What legislation has been passed to protect patients from unwanted life support is nonbinding, and doctors, in an age when medical lawsuits abound, sometimes supercede advance directives in favor of the legally safer path of "doing everything" for their patients.

For Charles, at least, it could have been worse. "He was lucky," his nurse said after he died. "He never woke up."

Susan Cahill is an intensive-care nurse in suburban Washington.