Evaluation of the Palliative Prognostic Score (PaP) in a Cancer Center-Based Palliative Care Clinic.

Abstract

Effective prediction of short survival allows clinicians to coordinate care appropriately. The Palliative Prognostic Score (PaP) is a multivariable prognostication tool to identify shorter survival. To test predictive accuracy of PaP to identify patients with four weeks or less of survival. A retrospective study of 855 consecutive patients with advanced cancer seen in initial consultation at a palliative care clinic in a tertiary cancer center (November 2016-March 2019). The data were assessed by Cox proportional survival statistics, area under the receiver operating curve (AUROC) statistics and cross tabulation of Clinical Prediction of Survival (CPS) scores against the actual survival of patients in the cohort. Median survival was 85 days (95% CI 75-95). The AUROC of the total PaP score was 0.810 (95% CI 0.774-0.847), while that of the CPS domain alone was 0.800 (95% CI 0.760-0.840); thus, other PaP elements (laboratory values, clinical symptoms, Karnofsky Performance Status) contributed little to the prognostic discrimination of the tool. Of 177 patients who died in 4 weeks or less, only 21 (11.9%) were assigned the correct CPS sub-categories; in the rest, survival was over-predicted. CPS is a highly weighted element of the PaP. Consequently, the prognostic discrimination of the tool is reliant on correct CPS. The other elements of the score do not have enough cumulative weight to correct over-predictive CPS. Shorter survival CPS categories were underutilized, undermining the PaP for the intended outcome of identifying patients with 4 weeks or less of survival in the study setting.