Time Is of the Essence: Evaluating Short-Term Prognosis Tools in Cancer and Noncancer Patients

Accurate end-of-life prognostication supports patient, family, clinician, and health system decision-making. The Palliative Prognostic Index (PPI), Palliative Performance Scale (PPS), and Short-Term Prognosis Signs (SPS) have been used in palliative care populations, particularly advanced cancer, but their utility for short-term prognostication in noncancer patients is less established.

To compare PPI, PPS, and SPS, alone and in combination, for predicting death within three days among cancer patients, then evaluate the best-performing tools in noncancer patients.

This retrospective study reviewed records of 290 in-unit decedents admitted to palliative care beds at a tertiary referral center and categorized by terminal diagnosis as cancer or noncancer. Admission PPI, PPS, SPS, and combined models were evaluated using logistic regression in the cancer training dataset. Model performance was assessed using area under the curve (AUC) and Akaike Information Criterion (AIC). Fitted model parameters were applied to the noncancer validation dataset.

In 166 cancer patients, PPI + SPS performed best, with AUC 0.72 (95% CI, 0.64-0.80) and AIC 207.56. Among individual predictors, SPS performed best, with AUC 0.70 (95% CI, 0.62-0.77) and AIC 208.18. Among 124 noncancer patients, PPI + SPS and SPS had similar AUCs of 0.69 (95% CI, 0.59-0.79) and 0.69 (95% CI, 0.59-0.78), respectively.

PPI + SPS performed best in cancer patients, although improvement over SPS alone was modest. In noncancer patients, SPS and PPI + SPS demonstrated similar, modest discrimination, supporting prospective validation in broader palliative care populations.