Families of critically ill patients may often take a more optimistic view of their loved one's condition than doctors do, even when they are given a specific estimate of the chances of survival, a new study suggests.

A number of studies have found that doctors and family members frequently have different opinions on critically ill patients' odds of survival. This raises the question of whether doctors are effectively communicating their estimates of patients' prognosis, that is, the course their disease is likely to take.

Many experts now recommend that doctors try to give specific numeric estimates of a patient's chances of survival — rather than "qualitative" information, such as telling families is it "very unlikely" that their loved one will survive.

So for the new study, researchers looked at whether the numeric and qualitative approaches differed in their effects on families' views.

The researchers had 169 family members of patients treated in one intensive care unit (ICU) view videos that portrayed a doctor discussing a critically ill ICU patient's prognosis with the family.

Half the family members viewed a hypothetical scenario in which the doctor told the family that their relative was "very unlikely" to survive and "very likely" to die. The doctor also said that if he did live, he would probably have to remain on a ventilator to breathe. The other half of the family members saw a video with the same scenario, with the exception that the doctor said the patient had a 10 percent chance of surviving and a 90 percent chance of dying.

In both cases, the researchers found, study participants came away with a more-positive estimate of the hypothetical patient's prognosis than the doctor on the video had given.

When asked to give their own estimates of the patient's chances of survival, study participants gave an average estimate of 26 percent after watching the video where the doctor had said survival was "very unlikely."

But even after viewing the video in which the doctor gave 10 percent survival odds, study participants still said the patient had, on average, a 22 percent chance of making it.

"The key finding is that many families don't take physicians' estimates at face value," said Dr. Douglas B. White, of the University of Pittsburgh Medical Center in Pennsylvania, who directed the study.

The findings, reported in the American Journal of Respiratory and Critical Care Medicine, also suggest that effective communication with families is not just a matter of giving numeric estimates of the chances of survival, rather than qualitative ones.

But that does not mean that the way in which doctors communicate with families is unimportant.

Instead, White said in an interview, it may be that ICU doctors need to limit the amount of the information they convey, so that family members are less likely to be overwhelmed at a time when they are distraught. They could also try explicitly asking family members if they understood the information they were just given, he said.

Trust is another key issue, the researcher noted. ICU physicians are not the patient's or family's regular doctor, which means family members are being asked to trust the judgment of a stranger.

In this study, participants who reported relatively less trust in doctors also disagreed to a greater extent with the doctor's prognosis estimate in the video.

White said that it remains unclear exactly how ICU doctors can best establish a level of trust between themselves and family members in such a short and emotionally charged time frame.

Research also suggests that families take a number of factors into consideration, other than the doctor's judgment, when it comes to their own views of a loved one's chances of survival.

In an earlier study, White and his colleagues found that families of critically ill ICU patients only rarely relied on doctors' prognostication alone.

Instead, they often considered their perceptions of their loved one's strength and "will to live," his or her history of overcoming illness, and their own trust in optimism, intuition and faith.

The current study did not look at whether discrepancies between family members' and doctors' prognostic estimates might affect families' decisions on whether to continue life-sustaining care. But past research, White noted, suggests that they do.

As for how well doctors are able to estimate prognosis, research suggests they are "fairly accurate" when estimating the general odds of patients in a given situation surviving to hospital discharge, according to White.

They are not as good, however, at predicting whether any one patient will live or die.

There is, White said, an "inherent uncertainty in medicine," and doctors need to convey that fact to family members as well.