Clinicians' Prediction of Survival Is Most Useful for Palliative Care Referral.

Background: Timely palliative transition in patients with advanced cancer is essential for their improved quality of life and overall survival (OS). Most prognostic models have been developed focusing on weeks' survival. The current study aimed to compare the accuracies of several indicators, such as the Karnofsky Performance Scale (KPS), Clinicians' Prediction of Survival (CPS), and Edmonton Symptom Assessment System (ESAS), for predicting the survival of patients. Methods: Two hundred patients were enrolled at a single tertiary cancer center in South Korea between 2016 and 2019. We compared the discrimination of CPS versus KPS and ESAS total scores using the area under the receiver operating characteristic curve (AUROC) in 3-month and 6-month survival predictions. Results: The median age of patients was 66.0 years, and 128 (64%) were male. Two-thirds (66%) of the patients had an Eastern Cooperative Oncology Group performance status of 0 or 1, and 55.5% had a KPS of 80% or higher. The values of AUROC of CPS, KPS, and ESAS total score in 3-month survival prediction were 0.80 (95% confidence interval [CI]: 0.73-0.88), 0.71 (95% CI: 0.62-0.79), and 0.71 (95% CI: 0.62-0.81), respectively, whereas those in 6-month survival were 0.82 (95% CI: 0.76-0.88), 0.70 (95% CI: 0.63-0.78), and 0.63 (95% CI: 0.55-0.71), respectively. Conclusion: CPS showed the highest accuracy in predicting 3- and 6-month survival, whereas KPS had an acceptable accuracy. Experienced clinicians can rely on CPS to predict survival in months. We recommend the use of KPS with CPS to assist inexperienced clinicians.