A prospective, multicenter cohort study to validate a simple performance status-based survival prediction system for oncologists.

Abstract

Survival prediction systems such as the Palliative Prognostic Index (PPI), which includes the Palliative Performance Scale (PPS), are used to estimate survival for terminally ill patients. Oncologists are, however, less familiar with the PPS in comparison with the Eastern Cooperative Oncology Group (ECOG) performance status (PS). This study was designed to validate a simple survival prediction system for oncologists, the Performance Status-Based Palliative Prognostic Index (PS-PPI), which is a modified form of the PPI based on the ECOG PS. This multicenter, prospective cohort study enrolled all consecutive patients who were referred to 58 palliative care services in Japan. The primary responsible physicians rated the variables required to calculate the PS-PPI and the PPI. Patient survival in these risk groups was compared, and the sensitivity and specificity of the PS-PPI and the PPI were evaluated. Patients were subclassified as patients receiving care from in-hospital palliative care teams, palliative care units, or home-based palliative care services. Subsets of patients receiving chemotherapy were also analyzed. This study included 2346 patients. Survival predictions based on the PPI and the PS-PPI differed significantly among the 3 risk groups (P < .001). The PS-PPI was more sensitive, whereas the PPI was more specific. All areas under the receiver operating characteristic curves of both indices were >0.78 for predicting survival at all times, from 3 weeks to 180 days. In predicting the prognosis of patients with advanced cancer, the PS-PPI was as accurate as the PPI. The PS-PPI was useful for short- and long-term survival prediction and for the prediction of survival for patients undergoing chemotherapy. Cancer 2017;123:1442-1452. © 2016 American Cancer Society.