Palliative Prognostic Index accuracy of survival prediction in an inpatient palliative care service at a Brazilian tertiary hospital

Purpose: The Palliative Prognostic Index (PPI) was developed to improve survival prediction for advanced cancer patients. However, there is limited data about the PPI application in a real-world scenario. This study aimed to assess the accuracy of PPI > 6 in predicting survival of cancer inpatients. Methods: A prospective observational cohort in an inpatient palliative care service at a tertiary hospital in Sao Paulo-SP, Brazil, between May 2011 and December 2018. Results: We included 1,376 critically ill cancer inpatients. Patients were divided into three PPI subgroups: PPI <= 4, PPI 4-6, and PPI >= 6. Their respective medium overall survival values were 44 days (95% confidence interval [CI] 35.52-52.47), 20 days (95% CI 15.40-24.59), and 8 days (95% CI 7.02-8.98), (p < 0.001). PPI = 6 predicted survival of <3 weeks with a positive predictive value (PPV) of 72% and an negative predictive value (NPV) of 68% (sensitivity 67%, specificity 72%). PPI > 4 predicted survival of <6 weeks with a PPV of 88% and an NPV of 36% (sensitivity 74%, specificity 59%). When PPI was <4, the mortality rate over 3 weeks was 39% with a relative risk (RR) of 0.15 (95% CI 0.11-0.20; p < 0.001), and the 6-week mortality rate was 63% with a RR of 0.18 (95% CI 0.13-0.25; p < 0.001) compared to PPI >= 4. Conclusions: PPI was a good discriminator of survival among critically ill cancer inpatients and could assist in hospital discharge decision. PPI may help healthcare policymakers and professionals in offering high-quality palliative care to patients.