Mustafa Önder GÖNEN, Havva Serin YİĞİT
Journal of Medicine and Palliative Care - 2026;7(4):676-681
Aims: To evaluate the discriminative performance of the PATHOS score for in-hospital mortality risk stratification among patients admitted to a palliative care ward from the emergency department (ED). Methods: This single-center retrospective cohort study included consecutive adult patients admitted from the ED to a palliative care ward between January 2023 and January 2025. The PATHOS score (platelets, age, troponin, heart rate, oxygen saturation, systolic blood pressure) was calculated from parameters recorded at ED arrival. Patients were grouped as survivors or non-survivors. Univariate and multivariable logistic regression and receiver operating characteristic (ROC) curve analyses were performed against in-hospital mortality, and the score's discrimination was internally validated by bootstrap resampling. Results: Of 489 patients, 118 (24.1%) died during the index admission. Non-survivors had higher mean PATHOS scores than survivors (3.31+/-1.03 vs. 2.01+/-0.85; p<0.001). The score yielded an area under the curve (AUC) of 0.824 (95% CI: 0.788-0.857); bootstrap internal validation indicated negligible optimism (optimism-corrected AUC 0.824). At the optimal cutoff of >=3, sensitivity was 77.97%, specificity 76.82%, positive predictive value 51.69%, and negative predictive value 91.64%. In multivariable analysis, the PATHOS score (OR: 3.48, 95% CI: 2.29-5.38; p<0.001), malignancy (OR: 2.71, 95% CI: 1.98-3.72; p<0.001), and serum creatinine (OR: 1.49, 95% CI: 1.09-2.03; p=0.012) emerged as independent predictors. Conclusion: Among palliative ward admissions from the ED, the PATHOS score, computed from parameters recorded at ED arrival, showed good, internally validated discrimination for in-hospital mortality. Because the score was measured in patients already admitted to palliative care, it is best positioned as an aid to risk stratification at the point of admission rather than as an ED triage tool; external validation in independent multicenter cohorts is warranted before broader clinical adoption.
