Survival prediction in terminally ill cancer patients by clinical estimates, laboratory tests, and self-rated anxiety and depression.

PURPOSE: To study how survival of palliative cancer patients relates to subjective prediction of survival, objective prognostic factors (PFs), and individual psychological coping. PATIENTS AND METHODS: Survival was estimated according to three categories (< 1 month, 1 to 6 months, and > 6 months) by two physicians (A and B) and the institutional tumor board (C) for 216 patients recently referred for palliative radiotherapy. After 6 months, the accuracy of these estimates was assessed. The prognostic relevance of clinical symptoms, performance status, laboratory tests, and self-reported emotional distress (Hospital Anxiety and Depression Scale) was investigated.
RESULTS: In 61%, 55%, and 63% of the patients, prognoses were correctly estimated by A, B, and C, respectively. kappa statistic showed fair agreement of the estimates, which proved to be overly optimistic. Accuracy of the three estimates did not improve with increasing professional experience. In particular, the survival of 96%, 71%, and 87% of patients who died in less than 1 month was overestimated by A, B, and C, respectively. On univariate analysis, 11 of 27 parameters significantly affected survival, namely performance status, primary cancer, fatigue, dyspnea, use of strong analgesics, brain metastases, leukocytosis, lactate dehydrogenase (LDH), depression, and anxiety. On multivariate analysis, colorectal and breast cancer had a favorable prognosis, whereas brain metastases, Karnofsky performance status less than 50%, strong analgesics, dyspnea, LDH, and leukocytosis were associated with a poor prognosis.
CONCLUSION: This study revealed that physicians' survival estimates were unreliable, especially in the case of patients near death. Self-reported emotional distress and objective PFs may improve the accuracy of survival estimates.