You've probably seen this before on television or the movies. The grim-faced, gray-haired doctor delivers the bad news to a well-appearing but worried patient: “The test results aren't good — you've got two months left, maybe three if you're lucky. I'd suggest you get your affairs in order.” The way these doctors could look seemingly healthy people in the eye and tell them how long they had to live always seemed to me so powerful, bordering on omniscient (and often a bit insensitive). Since graduating from medical school, I have never encountered this situation and I have yet to meet a physician willing to make such predictions to patients, especially those who looked as well as those in the movies.
While delivering bad news and helping patients and families deal with terminal illness are vitally important functions that all patients should expect of their physicians, is it possible to expect an accurate prediction of how much time a person has left? Or, on the other hand, is it possible to predict with accuracy who will make a complete recovery while in the throes of a serious medical illness? The answers depend on the situation, but rarely is it like the movies.
Predicting the Likelihood of Death or Recovery
The science of predicting the outcome of an illness swings between two extremes: complete recovery and death. In general, predicting the outcome for people with severe illness or minor illness is much easier than for all of those in between. In the last 20 years, researchers have made significant advances in taking information available at the time a person seeks medical attention and predicting how well things will go, whether the focus is on the chances of complete recovery, partial recovery, or survival for one, five or 10 years. Other studies address “medical futility,” when, even with the best available therapy, the chances of imminent death are high.
The way researchers devise ways to predict the course of an illness usually involves the collection of specific information (such as age, current illness, examination findings and test results) followed by a calculation of the chances of survival based on similar patients studied in the past. A formula can be developed that uses patient information to calculate risk among groups of people with a particular condition. In describing what will happen to those patients over time, such predictions about the group can be quite accurate (although, as discussed below, predictions about any given individual are not nearly as accurate).
For example, one study of more than 350 hospital admissions of patients with chronic obstructive pulmonary disease (COPD) who required admission to an intensive care unit found that 24% died in the hospital. However, of those who were over 65, 30% died before discharge, 41% had died by three months and 59% by one year. Besides advanced age, those who had more severe lung disease, had other major medical problems or a longer stay in the hospital before coming to the intensive care unit also had lower survival rates. Another study found that among similar patients, only 26% were alive and reporting a good or excellent quality of life at six months.
Another study found that of more than 26,000 patients in the hospital requiring cardiopulmonary resuscitation (CPR), only one-third could be revived, and only 15% survived to leave the hospital. Other studies have shown almost no chance of survival when a person with multiple medical problems and advanced age requires CPR for cardiac or respiratory arrest — that is, CPR was nearly futile in those situations.
Predicting Outcomes in Groups vs. Individuals
Although studies of groups of patients can accurately predict the likelihood of a certain outcome, those predictions break down if applied to people who do not have that specific illness or if applied to one person with that condition. For example, among everyone in a high-risk group, picking out the few who will do well is generally difficult or impossible. That’s why if you ask your doctor what is going to happen to you, you will often get an answer to a slightly different question such as: “Most people with your condition do well.” Most physicians will admit they cannot tell you with absolute confidence what is going to happen in your individual case — there is simply too much variability between people and too many other factors that matter. And there’s a lot we do not understand — factors that affect outcome but are not yet recognized may lead to a better, or worse, recovery than expected.
The Image vs. Reality
In addition to the overestimation of most doctors’ predictive abilities, at least two other explanations play a role in why real-life and movie versions of predicting death are so different: the setting and the element of surprise. Most critically ill patients for whom death is imminent are in a hospital (often an intensive care unit), in a hospice program or in a similar setting when such pronouncements are made, not a doctor's office. People who feel generally well at an office visit would very rarely hear from their doctor that death should be expected in a short period of time. Similarly, for persons with less than six months' life expectancy, the poor prognosis is rarely a surprise; people are usually quite ill for some time before that judgment is made.
The notion that a physician can confidently predict how much time a person has left to live is largely myth, perpetuated by the image of wise, all-knowing physicians in the model of Marcus Welby and Dr. Kildare. How good are doctors at predicting death? Other than the far extremes of excellent health or severe illness, they are not very good at all. Many families have an example they love to mention of a feisty, elderly grandparent “given” three months to live who then outlives her doctors. Despite the temptation to credit resolve and will power, this phenomenon is probably best explained by understanding how poorly doctors predict longevity.
Robert H. Shmerling, M.D. is associate physician at Beth Israel Deaconess Medical Center and associate professor at Harvard Medical School. He has been a practicing rheumatologist for over 20 years at Beth Israel Deaconess Medical Center. He is an active teacher in the Internal Medicine Residency Program, serving as the Robinson Firm Chief. He is also a teacher in the Rheumatology Fellowship Program.