Exploring laboratory and clinical factors associated with sudden unexpected death in patients receiving palliative care

Sudden unexpected death (SUD) in palliative care can disrupt care planning and cause substantial emotional distress for patients and their families. This study aimed to identify clinical and laboratory factors associated with SUD in patients receiving palliative care. We retrospectively analyzed 530 patients receiving palliative care between March 2019 and December 2022. SUD was defined as death occurring within 24 h after the clinical determination of the terminal stage. Multivariable logistic regression and receiver operating characteristic (ROC) curve analyses were performed to evaluate factors associated with SUD, including clinical characteristics, laboratory parameters, and the modified Glasgow Prognostic Score (mGPS). Among the 530 patients, 113 (21.3%) were classified as having SUD. In multivariable analysis, a history of chemotherapy (OR, 0.597; P = 0.031), serum magnesium level (OR, 2.272; P < 0.001), mGPS (OR, 0.681; P = 0.048), and Palliative Performance Scale (PPS) score (OR, 0.955; P < 0.001) were independently associated with SUD. In ROC analysis, PPS (AUC, 0.633; 95% CI, 0.578–0.688), serum magnesium (AUC, 0.593; 95% CI, 0.533–0.652), and chemotherapy history (AUC, 0.565; 95% CI, 0.516–0.614) demonstrated statistically significant but modest discriminatory ability. A composite model integrating these variables demonstrated modestly improved discrimination (AUC, 0.692; 95% CI, 0.634–0.749; P < 0.001). SUD in patients receiving palliative care was associated with serum magnesium, mGPS, PPS, and a history of chemotherapy. Although a composite model integrating these factors demonstrated modestly improved discrimination, its clinical utility remains uncertain. These findings should therefore be considered exploratory and hypothesis-generating. Prospective multicenter studies are warranted to validate these associations and determine whether these factors provide incremental prognostic information beyond established clinical assessment.