Commentary (Asch): Some Elements of Prognosis in Terminal Cancer

August 1, 1999
David A. Asch, MD, MBA

Oncology, ONCOLOGY Vol 13 No 8, Volume 13, Issue 8

You’ve got to accentuate the positive,Eliminate the negative,Latch onto the affirmative,Don’t mess with Mister In-between.-Johnny Mercer, 1944

You’ve got to accentuate the positive, Eliminate the negative, Latch onto the affirmative, Don’t mess with Mister In-between. -Johnny Mercer, 1944

Johnny Mercer may have gotten it wrong. Two important changes in medical practice over the past several decades have been the increase in patients’ involvement in decisions about their care and the growing recognition by physicians that they ought to encourage that involvement. In the prevailing model of shared decision- making, physicians and other clinicians are charged with informing patients about their condition, as well as the risks and benefits of alternative clinical approaches. Patients express their preferences based on their understanding of the possible clinical outcomes and estimates of their chances of attaining those outcomes. Then, clinician and patient work together to find the best way to achieve these patient goals.

For this system to work, however, patients need to be given accurate information about their prognosis, and they need to be able to understand and use that information. Lamont and Christakis review the apparent tendency of physicians to “accentuate the positive” when predicting the survival of cancer patients. Physicians tend to be overly optimistic when they attempt to foresee survival-ie, when they make mental guesses of life-expectancy. They also tend to be too optimistic when they foretell survival-ie, when they communicate those mental guesses to patients. According to Lamont and Christakis, these tendencies result in estimates that are twice removed from the truth; and these systematic errors may explain, in part, why some seriously ill patients have such unrealistically favorable expectations about their prognosis.

What Accounts for Patients’ Unrealistic Optimism?

Is this a problem? It may be if, as data from the SUPPORT (Study to Understand Prognoses and Preferences for Outcomes and Risk of Treatment) study suggest, patients with unrealistic optimism about their prognosis are more likely to die in the hospital on mechanical ventilation, thereby suffering the kind of death that many wish desperately to avoid.[1]

There are other interpretations of these results, however. Some patients may simply be more hopeful than others. This characteristic may explain both the tendency toward unrealistic optimism and the preference for more life-extending therapy. Hope is a defense mechanism that, for some patients, rises in direct proportion to their risk of death. For other patients, hope is something that guides decision-making but is independent of their risk of death.

Furthermore, physicians are not the only source of misinformation for patients. Patients are exposed to increasing amounts of medical information targeted directly toward them. Some of this information is disguised in other forms. For example, Diem and colleagues examined the portrayal of cardiopulmonary resuscitation (CPR) in three popular medical television shows (ER, Chicago Hope, and Rescue 911). Of 60 occurrences of CPR over 97 television episodes, 75% of patients survived the immediate arrest and 67% appeared to have survived to hospital discharge-survival rates 5 to 10 times higher than those observed in actual practice.[2] These dramas are scripted, acted, and directed by professionals whose job it is to make them appear convincing and compelling. One hopes that real physicians have more credibility, although the comments of Drs. Lamont and Christakis suggest that perhaps they ought not to.

Shared Decision-Making:Theory vs Reality

At the same time, it is worth reflecting on the possibility that the prevailing model of shared decision-making-that physicians inform patients about their clinical condition and alternative approaches and then help patients choose among them-may represent some form of wishful thinking of its own. The seemingly obvious correctness of this model contrasts sharply with observations about real-world decision-making. Patients,[3] like all human beings,[4] have terrible problems with innumeracy that limit their abilities to understand and incorporate probabilistic information. Some studies demonstrate a disconnection between patients’ desires for information and their interest in participating in their medical decisions.[5,6] Furthermore, what seems like an entire field of research focused on improving the use of advance directives and other measures to align end-of-life care with patients’ wishes has, by some accounts, been a failure.[7]

For these reasons, we need to avoid our apparent natural tendencies toward optimism as we strive to improve our ability to foresee and foretell patients’ prognoses. Improving our accuracy in these domains surely has independent value. If we are successful, we will increase that accuracy, but it may not lead to the improvements in care that we ultimately seek.


1. Weeks JC, Cook EF, O’Day SJ, et al: Relationship between cancer patients’ predictions of prognosis and their treatment preferences. JAMA 279:1709-1714, 1998.

2. Diem SJ, Lantos JD, Tulsky JA: Cardiopulmonary resuscitation on television. Miracles and misinformation. N Engl J Med 334:1578-1582, 1996.

3. Schwartz LM, Woloshin S, Black WC, et al: The role of numeracy in understanding the benefit of screening mammography. Ann Intern Med 127:966-972, 1997.

4. Paulos JA: Innumeracy: Mathematical Illiteracy and Its Consequences. New York, Hill and Wang, 1988.

5. Strull WM, Lo B, Charles G: Do patients want to participate in medical decision making? JAMA 252:2990-2994, 1984.

6. Ende J, Kazis L, Ash A, et al: Measuring patients’ desire for autonomy: Decision making and information-seeking preferences among medical patients. J Gen Intern Med 4:360-361, 1989.

7. Aronowitz RA, Asch DA: Cursing the darkness-Is there a limit to end-of-life research? J Gen Intern Med 13:495-496, 1998.