During his 20 years working in the intensive care unit at George Washington University Medical Center in Washington, Dr. William Knaus had watched hundreds of patients die. Many of them, suffering great pain, were kept alive, against their wishes, by extraordinary measures. "At times it was hard to get the family close enough to the patients to hold their hands," says Knaus. "There are all these machines and all this noise. It used to be that death was the worst thing that could happen to you, but medical technology now has allowed doctors to create a situation that's even worse."

Dismayed by what he saw as unnecessary suffering and indignity, both for the patients and their families, Knaus codirected a four-year study of hospital deaths, funded by the Robert Wood Johnson Foundation. The results, published late last year in the Journal of the American Medical Association, confirmed what doctors and nurses had observed but had never documented: the nightmarish world in which patients' requests to halt aggressive treatment in their final days often went unheard.

A graduate of West Virginia University Medical School, Knaus, 49, recently joined the University of Virginia School of Medicine in Charlottesville to head up a new department that will teach doctors, among other things, to be more-humane in treating the terminally ill. Knaus, who lives with his wife, Janet, and their three children in Arlington, spoke with correspondent Jennifer Mendelsohn at his office.

What does the study suggest about the final days of the average patient's life?

It tells us that we have a very big problem. We found that half the patients died in moderate-to-severe pain in the last three days of their lives—and those were people who were conscious and could tell people around them what was going on. We found that half the time the patients and their families had told us that they wanted to avoid aggressive care—resuscitation and other measures—but the physicians did not know because there was no communication going on.

Are the doctors at fault?

No. First of all, the health-care system is oriented toward giving us access to very expensive, high-technology care. Joanne Lynn (a George Washington University gerontologist and codirector of the terminal-care study) says that she can get intensive care for her patients easier than she can get them eyeglasses.

Secondly, we haven't really taught physicians how to take care of seriously ill and dying patients. Death, in too many respects, is still the enemy; we still teach that, and doctors now feel that if somehow they don't treat, they're not helping. But they are. We need to recognize that we can do a better job of helping people live well while they're dying.

Wouldn't better communication between doctors and patients solve this problem?

That's what everyone assumed would work, but when we tried it, it didn't. In the second phase of our research, we assigned each patient a full-time nurse clinician whose only responsibility was to assist the doctor and patient in communication on these issues. Compared with the control group, there was absolutely no change in the course of treatment relative to what a patient wanted—not one bit. Trying to solve this on a one-to-one basis probably isn't going to be sufficient.

Did living wills signed by patients ward off unwanted treatment?

No. Living wills are too general, too nonspecific. They don't hold anyone to any standard. If they say you don't want any extraordinary measures, what does that mean? It doesn't really communicate anything specific; an explicit plan has to be made.

Can the family intervene?

Not very easily. One of the shortcomings we saw in the study is that patients and their families are told the extremes: "We can treat you, or we can let you die." The system makes it scary to make choices, because the alternative sounds like abandonment.

What about physician-assisted suicide, as promoted by Dr. Jack Kevorkian?

These suicides are a strong symptom of all that's going wrong. Why do physician-assisted suicides even exist? Because people are afraid of the system, afraid that it doesn't know when to stop. They're afraid of how they're going to die. Few people talk about these issues in public—except for those who promote assisted suicide.

How would this change if doctors altered their treatment of the terminally ill patients?

Patients who turn to people like Kevorkian are not saying, "I want to commit suicide." They are saying, "I don't want to die in pain. I don't want to die isolated. I don't want to have my dying prolonged and drawn out." That's what it's about. If we help them avoid those situations, then I think the whole issue of physician-assisted suicide would go away. The number of people who really want to commit suicide with a physician's help is vanishingly small. But the number of people who don't want to die in pain or die prolonged deaths is much larger.

Is there anything people can do to ensure that they will be in control of their terminal care?

None of us really sits down and says, "Some day I'm going to die. How do I want to do that as well as possible?" But that is exactly what we need to do. Ask yourself some difficult questions: "Is it important for me to die at home, surrounded by friends and family?" If so, find out in advance from your hospital and insurance company how that will be possible. Talk to your doctor. Ask, "What happens when I am in pain? Will I be able to get medication to make me comfortable? Is my family going to be able to obtain services that will allow me to stay out of a hospital? What does my insurance cover?"

Is it ever too early for a healthy person to begin thinking about this?

Death is universal; it's never too early to start thinking about it. We need to recognize that dying is one of the more important things we'll do in life. It will be the last memory we leave to those we love. And if we want to do it better, then we're going to have to talk about it.