Modern Prognostics: How Accurate are the Predictions?

To "buoy up a dying man with groundless expectations of recovery is really cruel." It may lead to "overlooking the important concerns of futurity and involve families in confusion and distress."

Samuel Bard, 1769

How we choose to live our lives largely depends on our perception of the future. Should we strive for maximum comfort "today" or make a sacrifice anticipating a later reward? What should we focus on: career, family, a dying relative, children, spirituality, finances? Should a cancer patient accept a rigorous treatment hoping to recover or a palliative care knowing the end is near? Is it time to fight or time to reflect?

Traditionally, patients and their families rely on physicians to make a prognosis of a life-threatening illness. Unfortunately, the accuracy of these predictions is generally rather poor. According to a recent study, which assessed survival estimates of 343 doctors for 468 patients, "only 20% predictions were accurate". "Accurate" means within 33% of actual survival. From the remaining 80%, 63% of predictions were overoptimistic and 17% were overly pessimistic. "Overall, doctors overestimated survival by a factor of 5.3".

This lack of predictive accuracy had little to do with the doctor or patient. Generally, more experienced doctors were more accurate. "Non-oncology medical specialists were 326% more likely than general internists to make overly pessimistic predictions". Also, if a doctor-patient relationship was longer, the error in prediction tended to be greater. In addition, "male patients were 58% less likely to have overly pessimistic predictions".

The reasons for this undue optimism may be related to the physician's natural desire to be reassuring and to avoid being the bearer of bad news. Physicians may be more concerned with the effects of prognostication on patients and their families than with empirical accuracy. As well, modern medicine is oriented around diagnosis and accepted therapy; prognosis is presumed to be dependent upon the symptoms and complications and not on the individual characteristics of a patient.

Whatever the reason, this optimism in estimating life expectancy undermines the quality of care given to a patient at the end of life. For example, dying patients are not referred to hospice care in time. A US study showed that seven percent of patients referred to hospice care died within hours of admission. Also, a majority of patients have only one month of such care even though three months are recommended. The question of whether or not admission to hospice care may shorten a patient's life is not an easy one to answer. It is generally believed to be of benefit to the patient.

According to Colin Murray Parkes, a psychiatrist, "prognoses should be based on proved indices, not intuition". He praises recent research instruments such as Morita's palliative prognostic index and Maltoni's palliative prognostic score in his commentary on the above-described study. "These short and simple instruments make use of a mixture of performance measures and systemic symptom assessments rather than relying on intuition and clinical judgement alone". These recent scales have been shown to be fairly good predictors of short term survival.

In conclusion, prognoses of survival for terminally ill patients are extremely inaccurate. This adversely affects the final care offered to the patient as well as relevant family management. It is important for physicians to try not to be overly optimistic. Ideally, new assessment scales should be used instead of clinical judgement.

Suggested Reading

1. Christakis NA, Lamont EB. Extent and determinants of error in doctor's prognoses in terminally ill patients: prognosis of cohort study. BMJ 2000, 320(7255):459.
2. Christakis NA. The elipsis of prognosis in modern medical thought. Soc. Sci. Med. 1997, 44(3):301.
3. Forster LE, Lynn J. Predicting life span for applicants to inpatient hospice. Arch Intern Med 1988, 148:2540.