Fax: (630) 275-2390
Myth: We can tailor palliative care to match life expectancy. Reality: No we can't†
Article first published online: 19 APR 2010
Copyright © 2010 American Cancer Society
Volume 116, Issue 13, pages 3078–3079, 1 July 2010
How to Cite
Hartsell, W. F. (2010), Myth: We can tailor palliative care to match life expectancy. Reality: No we can't. Cancer, 116: 3078–3079. doi: 10.1002/cncr.25118
See referenced original article on pages 3251-6, this issue.
- Issue published online: 18 JUN 2010
- Article first published online: 19 APR 2010
- Manuscript Accepted: 15 JAN 2010
- Manuscript Received: 29 NOV 2009
The goal of palliative treatment is to return the patient to an optimal functional level in the shortest period of time with the fewest side effects. As Nora Janjan wrote, palliative therapy should “efficiently provide comfort using antineoplastic therapies or supportive care approaches to the patient with the fewest treatment-related side effects, recognizing that the patient will die of the disease.”1 For patients with terminal cancer, quality of life is more important than tumor control. Treatment courses given at the end of life are frequently too protracted, keeping the patients on treatment and in the clinic—rather than at home with friends and family. That is one conclusion that Gripp et al have drawn from their series of patients, reported in this issue of Cancer.2 Unfortunately, they seem to have drawn their conclusions first and then tried to find the data to match those conclusions. Although it is an admirable (and, in my view, appropriate) conclusion, it cannot be justified from their data.
Gripp et al prospectively evaluated 216 patients referred to the University Hospital Dusseldorf Department of Radiation Oncology for palliative radiation therapy. They used 3 methods to estimate survival time for these patients and have previously reported those results.3 The further analysis in the current issue of Cancer reviewed the 33 patients who survived for 30 days or less.2 Not surprisingly, the median survival in this group was only 15 days. Because the typical prescribed radiation therapy treatment course was for 2 or more weeks, a substantial number of these patients were not able to complete treatment. Gripp et al conclude that palliative radiation therapy is not appropriately customized for patients at the end of life, especially not for those patients who have prolonged irradiation schedules.
What is wrong with this conclusion? In Gripp et al's previous study, they showed that there is a poor correlation between predicted survival and actual survival for individual patients.3 There is a correlation between the predicted and actual survivals for the entire cohort of patients, but this correlation is rather weak. There have been multiple other attempts to correlate some factor (physician prediction, performance status, symptoms, biochemical markers such as albumin levels, quality of life scores, etc) with survival times.4, 5 All of these attempts have come to the same conclusion—we can predict the general survival trend, but we are lousy at predicting the survival time for any individual patient. In addition, physician predictions of survival tend to be overly optimistic, overestimating survival time for the majority of patients. The problem from the Gripp et al study is not that the therapy was not appropriately customized; instead, it is the implication that we can appropriately customize the treatment. If we can't accurately predict survival times, how can we attempt to fit shorter- or longer-course treatments based on predicted survival times?
In the current study, Gripp et al collected these survival predictions in a prospective fashion, but no apparent attempt was made to tailor the treatment prospectively based on those predictions. They then retrospectively took the poorest-performing group of patients (those who died in the first 30 days) to tell us that treatment was not effective. Well, taken in retrospect, this seems obvious—but is this analysis of any use to us? Not particularly. This bit of information would be much more useful if we knew it before treatment, not after the fact. Unfortunately, they have already told us that they were not able to define this group of patients before treatment. Only 16% of this group of patients had a predicted survival of <30 days, and 21% were predicted to live longer than 6 months!
This study would have been more useful if they had taken the group of patients for whom a very short survival was predicted and then determined the effectiveness of treatment. Selecting the group of patients who don't do well to tell us that they didn't do well is not particularly helpful. This group composed only 15% of their total patient sample. If the other 85% of patients did very well—ie, longer than predicted survivals, good quality of life—would we draw the same conclusions? No, we would tout this as a very effective treatment. We are not given enough information in the study to determine whether these patients were typical or just an unfortunate minority.
Although their data do not justify the conclusions, the authors do raise an important point. Patients who are at the end-of-life do not need a lengthy course of treatment when a shorter course will suffice. There are multiple randomized trials that show that a single treatment with radiation therapy is as effective as 10 or more treatments in palliating bone pain.6, 7 Short courses of treatment are equally effective in palliating many other symptoms.8-10 The side effects from these short-course treatments may occur less frequently than with the longer course of treatment. The concerns about long-term side effects are not substantiated by randomized controlled trials.4
In summary, the goal of palliative therapy is to palliate, not to cure. For the best palliative results for our patients (and their caregivers), this should be done as rapidly as possible. This is not only evidence-based medicine, it is the right treatment for our patients.
CONFLICT OF INTEREST DISCLOSURES
The author made no disclosures.