A study of terminally ill cancer patients and their physicians found that in only 37 percent of cases was the doctor willing to give the patient his or her best estimate of how long the patient could expect to live. Without such knowledge, patients cannot make informed choices about how to spend their remaining time or prepare themselves or their families for the kind of death they would have chosen, given the opportunity.

The study, conducted by two physicians at the University of Chicago Medical Center and published in June 19 issue of the Annals of Internal Medicine, found that in 40 percent of cases, doctors would knowingly provide an inaccurate estimate of survival time, usually an overestimate. In 23 percent of cases doctors would refuse to provide an estimate.

"Although nearly everyone agrees that frank, open and honest communication between a patient and his doctor is optimal, on this one absolutely crucial issue it remains very much the exception," said study co-author Nicholas Christakis, M.D., Ph.D., professor of medicine and sociology at the University of Chicago. "As a consequence, two out of three patients may have to make important medical and personal decisions based on missing or unreliable information."

The authors worry that without reliable information patients with no chance for recovery may delay gathering their families and friends until it is too late, or chose to undergo costly and invasive but predictably ineffective therapies. Recent studies, for example, confirm that many terminally ill cancer patients with unresponsive disease nevertheless receive chemotherapy.

The investigators conducted a four-minute telephone survey with 258 Chicago-area physicians who referred 326 patients to hospice care in 1996. They asked each referring physician for his or her best estimate of how long that patient was likely to live. They also asked what he would say if the patient insisted he provide an estimate of probable survival time.

In 23 percent of the cases, physicians said they would not give the patient a precise prognosis, even if asked. In only 37 percent of cases would the doctor communicate his best guess at probable survival. In 40 percent of the cases, physicians said they would intentionally provide an inaccurate estimate, usually suggesting that the patient would live much longer than the doctor really expected.

"Physicians want to give patients hope," said cancer specialist Elizabeth Lamont, M.D., an instructor of medicine at the University of Chicago and co-author of the study. "They may imagine they are being kind or encouraging or even protective by withholding bad news, but we think that many patients need information about their survival and that when they ask they deserve a frank response."

This tendency to exaggerate survival time is particularly troubling, note the authors, in light of previous studies showing that even experienced doctors making their best guesses tend to overestimate survival times by a wide margin. Adding an intentional error only compounds this inherent prognostic bias.

In addition, this survey-- which focused only on cancer patients who had already been referred to hospice for palliative care, a turn of events with obvious prognostic implications -- may underestimate the extent of error and misrepresentation. "If physicians infrequently provide frank disclosure to hospice patients with cancer who request it," note the authors, "they may be even less likely to provide it to non-hospice patients, with or without cancer."

Three factors that correlated with frank communication were the age of the patient, the doctor's confidence in his ability to make predictions, and the age of the physician. Doctors were more willing to be frank about death with older patients. They were more willing to share their predictions if they felt certain of their own prognostic skills. Older physicians, however, were less willing to offer patients a prognosis.

It may be that the "wisdom born of experience might discourage physicians from frank disclosure," suggest the authors. More likely, however, is the fact that older physicians, who trained during in 1950s and 60s, were taught to "protect" patients from disheartening news. "Forty years ago," said Lamont, many physicians would not even tell patients they had cancer, much less predict the outcome."

"How long have I got, Doc?" is a fundamental question that terminally ill patients frequently ask. The question deserves a straightforward answer, say the researchers. "When physicians can't or won't make predictions about a patient's future," said Christakis, "patients may die deaths they deplore in locations they despise."

If physicians are to enhance the care of the dying, he added, "they need to start viewing the death of patients as normal and unavoidable and not as a personal or professional failure, to be avoided not only clinically but also rhetorically."

"Communicating bad news can be an unpleasant and painful process for physicians," said Lamont, "but that doesn't make it any less necessary." Physicians should be trained to do it tactfully and respectfully, she said. They should face the difficulties involved when seriously ill patients request accurate information, no matter how hard this information may be to acquire or communicate.

"At some point," conclude the authors, "patients might benefit more from having their doctors shift the focus from providing hope for recovery to hope for a good death."

Funding for the study was provided by the Soros Foundation Project on Death in America, the Robert Wood Johnson Clinical Scholars Program, and the National Institutes of Health.