Objectives In palliative care settings, predicting prognosis is important for patients and clinicians. The Palliative Prognostic Index (PPI), a prognostic tool calculated using clinical indices alone has been validated within cancer population. This study was to further test the discriminatory ability of the PPI (ie, its ability to determine whether a subject will live more or less than a certain amount of time) in a larger sample but with a palliative care context and to compare predictions at two different points in time.

Methods Multicentre, prospective, observational study in 10 inpatient hospices in the UK. The PPI score was calculated on the day of admission (PPI₁) and again once on days 3–5 of inpatient stay (PPI₂). Patients were followed up for 6 weeks or until death, whichever was earlier.

Results Of the 1164 patients included in the study, 962 had both scores available. The results from PPI₂ showed improved sensitivity, specificity, positive predictive value and negative predictive value compared with PPI₁. For PPI₁versus PPI₂, area under receiver operator character curve (ROC) for <21 days were 0.73 versus 0.82 and for ≥42 days prediction 0.72 versus 0.80. The median survival days for patients with PPI₁ ≤4, 4.5–6 and >6 were 38 (31 to 44), 17 (14 to 19) and 5 (4 to 7).

Conclusion This study showed improved discriminatory ability using the PPI score calculated between day 3and day5 of admission compared with that calculated on admission. This study further validated PPI as a prognostic tool within a palliative care population and showed recording at two time points improved accuracy.