Prognostic factors for survival prediction in advanced cancer patients and development of a simple survival prediction tool for application in a community palliative care setting

Survival prediction is crucial in advanced cancer populations for clinical, ethical and organization reasons. Among a myriad of tools, Edmonton Symptom Assessment Scale (ESAS) is a previously-validated tool for prognostication. Objective: To review the effectiveness and limitations of different survival prediction tools, report our single center experience of using ESAS as survival predictor, and develop a simple prognostic tool suitable for clinical practice in a community palliative care center. Methods: All deceased patients aged 18 years of age or older, having a diagnosis of advanced, metastatic or incurable malignancies, and an ESAS completed in their first encounter at the Tuen Mun Hospital in Hong Kong between September 2011 and October 2011 were included. Patient demographic and clinical data was collected retrospectively based on a rigorous chart review. Overall survival and a multivariate analysis evaluating prognostic factors was conducted. Results: ESAS records of 85 patients were included in our analysis. Median Karnofsky Performance Scale (KPS) score was 60 (range 30-100). Median ESAS score was 10 (range: 0-51), and median survival was 6.4 weeks (95% Cl: 5.3-9.6 weeks). Multivariate analysis suggested ESAS (hazard ratio [HR] 3.94) and KPS (HR 1.95) independently predicted overall survival. Three groups were stratified based on these prognostic factors: median survival of group 1 (ESAS < 25 and KPS ≥ 60), group 2 (either ESAS ≥ 25 or KPS < 60) and group 3 (ESAS ≥ 25 and KPS < 60) were 105 days, 20 days and 10 days, respectively (p < 0.0001). Conclusions: Our finding suggest that ESAS offers promising potential as a prognostication tool predicting survival. Its ease of use and dual role as both a symptom assessment and a survival estimation tool allows the ESAS to be seamlessly integrated into daily clinical practice.