Prognostic associations of clinical indices and laboratory-based markers with overall survival and 30-day mortality in malignant and non-malignant palliative care patients: a retrospective cohort study

Abstract Background Accurate prognostic assessment is essential in palliative care to support treatment planning, goal setting, and communication with patients and families. Although functional status, comorbidity burden, and routinely available laboratory indices have each been associated with prognosis, evidence regarding their simultaneous prognostic contribution in real-world palliative care populations including both malignant and non-malignant patients remains limited. This study primarily evaluated the associations of Eastern Cooperative Oncology Group performance status (ECOG PS), Charlson Comorbidity Index (CCI), Prognostic Nutritional Index (PNI), and neutrophil-to-lymphocyte ratio (NLR) with overall survival (OS) in a mixed inpatient palliative care cohort. Secondary objectives included evaluating 30-day mortality and examining prognostic associations separately in malignant and non-malignant subgroups. Methods This retrospective single-centre cohort included 260 consecutive eligible adults admitted to the palliative care unit between August 2023 and September 2025. Clinical, laboratory, and outcome data were obtained from medical records. Overall survival was assessed using multivariable Cox proportional hazards regression, and 30-day mortality using multivariable logistic regression. Receiver operating characteristic analysis evaluated the discriminative performance of CCI, PNI, and NLR. Analyses were performed for the overall cohort and malignant and non-malignant subgroups. Results Of 260 patients, 156 (60.0%) had malignant disease and 104 (40.0%) had non-malignant disease. In the overall cohort, ECOG PS, CCI, PNI, and NLR were independently associated with overall survival, whereas ECOG PS, CCI, and NLR were independently associated with 30-day mortality. In subgroup analyses, PNI was associated with OS in non-malignant but not malignant patients and with 30-day mortality only in non-malignant patients, whereas NLR remained associated with both outcomes in both subgroups. NLR had the highest observed area under the curve for 30-day mortality in the overall cohort (0.745; 95% confidence interval, 0.687–0.796) and an area under the curve of 0.776 (95% confidence interval, 0.683–0.852) in the non-malignant subgroup. Conclusions Functional status, comorbidity burden, and routinely available laboratory indices were associated with prognosis in palliative care patients. NLR showed the highest observed discrimination for 30-day mortality and remained associated with both outcomes in malignant and non-malignant subgroups. PNI was associated with OS in the overall cohort and the non-malignant subgroup and with 30-day mortality only in the non-malignant subgroup.