Comparison of five commonly used survival prediction tools (KPS, PPS, ECOG, PPI, and PaP) for 7‑day, 14‑day, and 30‑day mortality risk in children with advanced cancer receiving palliative care: a single‑center retrospective cohort study

Background Survival prediction is a core issue in pediatric palliative care, yet no specific tool has been developed for this population. The predictive value of five adult‑derived tools—KPS(Karnofsky Performance Scale), PPS(Palliative Performance Scale), ECOG(Eastern Cooperative Oncology Group Performance Status), PPI(Palliative Prognostic Index), and PaP (Palliative Prognostic Score), have not been adequately validated in pediatric end‑of‑life populations. Objective To systematically compare the predictive value of these five tools for 7-day, 14-day, and 30-day mortality risk in children with advanced cancer receiving palliative care, and to explore the independent prognostic value of clinician-predicted survival (CPS), a core component of PaP. Methods This single-center retrospective cohort study included children with stage IV malignant tumors admitted to the palliative care unit of West China Fourth Hospital, Sichuan University, between January 2020 and March 2026. Survival differences were analyzed using Kaplan-Meier and Cox proportional hazards regression, with the proportional hazards assumption verified by Schoenfeld residuals and time-dependent interaction terms. ROC curves were constructed for 7-, 14-, and 30-day mortality as binary outcomes; AUCs were compared using DeLong’s test. Precision-recall (PR) curves and AUPRCs were additionally used to address outcome imbalance. CPS was analyzed as a sensitivity analysis. Calibration was assessed using the Hosmer-Lemeshow test and Brier scores. All cutoffs were exploratory and data-driven. Results A total of 90 eligible children were included. Median survival was 7.0 (IQR 2.75–16.0) days. The 7-, 14-, and 30-day mortality rates were 52.2%, 72.2%, and 92.2%, respectively. KM and multivariable Cox analyses confirmed KPS and PPS as independent protective factors (KPS: HR = 0.492, P = 0.002; PPS: HR = 0.504, P = 0.002). For KPS and PPS, we used a reversed score (100 minus the original score) solely to align the direction of the predictor with mortality risk, yielding 7-day AUCs of 0.691 and 0.720, respectively. After multivariable adjustment, PaP was an independent risk factor (HR = 1.724, P = 0.048) with a 30-day AUC of 0.705. In a sensitivity analysis, CPS ≥ 8.5 was independently associated with mortality after adjustment (adjusted HR = 2.35, P < 0.001), with AUCs of 0.650, 0.627, and 0.651 for 7-, 14-, and 30-day mortality; its discriminative ability was comparable to PaP ( P > 0.05 for all DeLong comparisons). ECOG and PPI demonstrated limited discriminative ability in this highly homogeneous end-of-life cohort. PH assumptions were satisfied for all variables. Calibration analyses for PaP were performed using predicted probabilities derived from separate multivariable logistic‑regression models for 7‑, 14‑, and 30‑day mortality. Non‑significant Hosmer‑Lemeshow test results were observed for PaP ( P = 0.523, 0.800, and 0.399 for 7‑, 14‑, and 30‑day mortality). Given the limited sample size in this cohort, these non‑significant P‑values cannot serve as definitive evidence of adequate model calibration. For KPS, PPS, ECOG, and PPI, the Hosmer‑Lemeshow test was not applicable, and Brier scores ranged from 0.070 to 0.249. Conclusion In this single-center cohort of children with advanced cancer in the final days of life, KPS and PPS show promise for macro‑level prognostic stratification within this single‑centre terminally‑ill paediatric cohort, while PaP demonstrates preliminary potential for 30‑day individualised mortality risk prediction in this cohort. CPS, a core component of PaP, showed independent prognostic value and may partly account for the predictive performance of PaP. ECOG and PPI showed limited discriminative performance among this highly homogeneous end‑of‑life cohort, but this finding should not be generalized to broader pediatric palliative care populations without further validation. All proposed exploratory cutoffs require external validation before clinical translation.