Development and Internal Validation of the Palliative Metabolic Risk Score (PMRS) for Predicting Critical Outcome in Palliative Inpatients.

Background/

Objectives

In-hospital critical outcome among palliative inpatients remains high, often driven by acute physiological instability rather than chronic comorbidities. Although diabetes mellitus (DM) is common in this population, its independent impact on critical outcome is unclear. This study aimed to determine whether acute metabolic and inflammatory markers-specifically glucose, C-reactive protein (CRP), albumin, and oxygen requirement-better predict short-term outcomes, defined as in-hospital critical outcome or ICU transfer during the same hospitalization period, than DM status alone.

Methods

This retrospective study included 200 palliative inpatients admitted to the Internal Medicine Clinic of Kestel State Hospital, Bursa, Turkey, between January 2024 and January 2025. Demographic, clinical, and laboratory data were obtained from electronic records. The primary outcome was in-hospital critical outcome or ICU transfer ("critical outcome"). Logistic regression and receiver-operating characteristic (ROC) analyses identified independent predictors. The study was approved by the Bursa Yüksek İhtisas Training and Research Hospital Ethics Committee (ethics approval: protocol code 2024-TBEK 2025/05-12).

Results

The mean age was 77.7 ± 12.3 years, and 47% were male. DM was present in 30.5% but did not independently predict critical outcome (p = 0.904). In contrast, oxygen requirement (OR = 4.08, p = 0.002), mean glucose (OR = 1.01, p = 0.001), and cancer (OR = 3.28, p = 0.016) were significant predictors. ROC analysis identified CRP > 64.1 mg/L and albumin < 25 g/L as optimal thresholds, and these two markers formed the basis of the low-, intermediate-, and high-risk stratification, with critical-outcome rates of 39.0%, 45.1%, and 85.4% (p < 0.001).

Conclusions

Acute metabolic and inflammatory disturbances-particularly hyperglycemia, elevated CRP, hypoalbuminemia, and oxygen requirement-are stronger prognostic indicators than DM. A simple bedside model incorporating these parameters may improve prognostic accuracy and communication in palliative care.