Prognostication in advanced cancer: An ancient art in modern medicine

“I know you don’t have a crystal ball, but … how much time do I have?”

Physicians who care for patients with advanced cancer hear this question frequently. Predicting the future sounds more like mystery than medicine, but discussing prognosis challenges all of us. Patients and families often preface the question stating that they recognize the limits of physicians’ predictive powers, but ask nonetheless. For physicians, these are complicated conversations that involve breaking bad news and answering questions regarding life expectancy.

In 2010, Keating and colleagues conducted a study of more than 4,000 physicians that found that 65% of them would inform terminally ill patients (prognosis of 4 to 6 months or less) of their prognosis, once they were aware of it. However, there were still 15% who would only inform patients and families of prognosis if asked.

Stephanie Harman

An approach for discussing prognosis

The actual act of discussing prognosis is a difficult conversation. Historically in medical education, there has been little formal training in conducting these discussions. Finlay and Casarett recently reviewed prognostication for patients with advanced cancer and applied the SPIKES protocol to discuss prognosis. SPIKES is a technique for delivering bad news developed by Walter Baile, MD, a member of the HemOnc Today Editorial Board, and colleagues.

The first step, “S”, refers to the setting of the discussion, both the physical setting as well as the participants and the physician’s own preparation. The second step, “P”, stands for perception, to prompt assessment of the patient’s understanding of their disease. The “I” represents invitation and information. In the context of prognosis, this involves eliciting the patient’s preferences regarding how much and what information they would like to receive. The fourth step is “K” for knowledge. It refers to the actual delivery of the information to the patient and family. As patients and families respond to the prognosis, the “E” stands for emotion/empathy. The physician must acknowledge the patient and family’s reactions and respond to their immediate questions. Finally, the last step is “S” for summarize and strategize to ensure the patient and family’s understanding of the discussion as well as next steps.

Tools for prognostication

Multiple studies have demonstrated that physicians’ tendency with their clinical predictions is to overestimate survival.

Lamont and Christakis noted in their 2003 Journal of the American Medical Association article that although physicians overestimate, there is still a correlation between their predictions and patients’ actual survival, making clinical predictions by physicians another prognostication tool. Combining one’s own clinical prediction with consultation of other physicians can also help improve prognostic accuracy, as the average of several physicians’ prognoses is often more accurate than that of a single physician.

Patient-related factors — like signs and symptoms of disease and performance status — also contribute to estimating prognosis.

Performance status in particular has consistently predicted survival in patients with cancer, not only in advanced disease, but also in its use for enrollment of clinical trials. This includes both the ECOG performance scoring system as well as the Karnofsky score: Patients with higher ECOG scores (>2) and lower Karnofsky scores (<50%) have shorter survival.

The rate of decline of performance status also has some predictive power, as a more rapid decline in performance status is indicative of a shorter prognosis.

In 2005, the European Association for Palliative Care released a study of evidence-based prognostic factors for advanced cancer patients. It recommended that clinical predictions of survival should be used in conjunction with other prognostic factors or scores. It also highlighted the most important signs and symptoms to include performance status, cancer-related anorexia-cachexia syndrome, dyspnea and delirium/cognitive failure.

Although it did find some biologic factors such as leukocytosis, lymphocytopenia and high C-reactive protein to be associated with a poor prognosis, these factors have been less studied than the clinical signs and symptoms.

In addition, the association found that there were several prognostic scoring systems, including the Palliative Prognostic Score and the Palliative Prognostic Index, that have been validated and incorporate performance status, biologic and clinical factors, and clinical predictions (by the physician) to group patients into broad estimates of survival.

Do patients want to be told? The short answer is yes.

Although almost all patients want some information about their future, they also prefer an individualized approach that takes into account their information needs. A 2009 review by Innes and Paynes looked at patients’ preferences for prognostic information and found that patients do desire honesty from their physicians, although that does not mean being given all information. Most prefer some prognostic information and find such discussions helpful in maintaining control and planning.

Stephanie Harman, MD, is a palliative care physician at Stanford University Medical Center and director of its Inpatient Palliative Care service.

For more information:

• Finlay E. CA Cancer J Clin. 2009;59:250-263.
• Lamont EB. JAMA. 2003;290:98-104.
• Maltoni M. J Clin Oncol. 2005;23:6240-6248.