"My mom thought that if you'd sign the DNR, we would be abandoning him," she said. "My mom kept saying, 'There will be a miracle, there will be a miracle.' I felt caught in between. I am a nurse, but I am also a daughter."

She had seen chest compressions done many times. "I knew it would break my dad's bones and that he wouldn't make it," she said. "The decision was so hard for me."

After five-and-a-half months in intensive care, her father slipped into a coma and his organs began to shut down. As doctors rushed to his bedside, the family stopped them, saying, "Enough," and then, "Thank you for all you've done." He died without a DNR order in place.

The widespread misunderstanding about CPR itself can make a family's agony worse. The technique, which has been an accepted medical procedure for about 40 years, can be successful in patients who have a sudden, unexpected heart attack or severe respiratory distress. But it was not intended to be used routinely for very sick patients, for whom cardiac arrest is expected. Some studies show that the long-term survival for hospitalized patients given CPR is about 15 percent; some find even smaller percentages. But according to a 1996 article in The New England Journal of Medicine, the long-term survival rate on TV medical dramas for patients given CPR was 67 percent.

The need for policies dealing with DNR orders began to be felt in the late 1980s. CPR-related techniques had become increasingly sophisticated. Running a code, as the process is called, became more protracted. End-of-life conundrums, ethical and legal, proliferated. At the same time, with the rise of the patients' rights movement and its concomitant distrust of paternalistic doctors, patients and their families wanted a greater voice in decision-making.

In New York, doctors at one hospital had a casual way of indicating to staff, without informing families, which patients should not be resuscitated : Purple stickers were affixed to their charts. Occasionally, stickers were placed on the wrong charts or fell off. After these and other stories came to light, New York passed a law in 1987 that addressed the conditions under which a physician could write a DNR order. Patient consent was essential.

The statute did say that if resuscitation was "medically futile" and no representative could be found to consent to the DNR order, a doctor could write one, if another doctor also signed it. Under these narrow circumstances, a doctor's judgment that CPR would be useless was sufficient. But what if a "medically futile" patient's decision-maker insisted that CPR be performed anyway? Could a doctor's judgment prevail?