The relative accuracy of the clinical estimation of the duration of life for patients with end of life cancer

BACKGROUND, Although the prediction of the duration of life of patients wit h end of life cancer most often relies on the clinical estimation of surviv al (CES) made by the treating physician, the accuracy and practical value o f CES remains controversial. METHODS. The authors prospectively evaluated the accuracy of CES in an ince ption and population-based cohort of 233 cancer patients who were seen at t he onset of their terminal phase. They also systematically reviewed the lit erature on CES in advanced or end-stage cancer patients in MEDLINE, CANCERL IT, and EMBASE data bases, using two search strategies developed by a resea rch librarian. RESULTS. CES had low sensitivity in detecting patients who died within shor ter time frames (less than or equal to 2 months), and a tendency to overest imate survival was noted. A moderate correlation was observed between actua l survival (AS) and CES (Pearson correlation coefficient = 0.47, intraclass correlation coefficient = 0.46, weighted kappa coefficient = 0.42). CONCLUSIONS. Treating physicians appear to overestimate the duration of lif e of end of life ill cancer patients, particularly those patients who die e arly in the terminal phase and who may potentially benefit from earlier par ticipation in palliative care programs. CES should be considered one of man y criteria, rather than a unique criterion, by which to choose therapeutic intervention or health care programs for patients in the end of life cancer phase. Cancer 1999;86:170-6. (C) 1999 American Cancer Society.