ABSTRACT This study aimed to assess whether admission plasma lipopolysaccharide‐binding protein (LBP), procalcitonin (PCT), and lactate could improve detection of nosocomial infection in cirrhotic patients presenting with upper gastrointestinal bleeding (UGIB). A retrospective analysis was conducted in 196 consecutive adults with cirrhotic UGIB admitted between January 2021 and January 2025, in whom index biomarkers were defined as the first blood samples obtained within 24 h of hospital arrival and before any diagnosis of nosocomial infection. Nosocomial infection within 28 days of admission occurred in 58 of 196 patients. Compared with noninfected patients, those with nosocomial infection had higher WBC, CRP, PCT, LBP, lactate, international normalized ratio (INR), and total bilirubin (TB), lower albumin and sodium, a higher neutrophil‐to‐lymphocyte ratio (NLR), and a lower lymphocyte‐to‐monocyte ratio (LMR). Individual discrimination was excellent for LBP (area under the curve AUC 0.967), CRP (0.918), WBC (0.914), lactate (0.910), and PCT (0.901). In multivariable analysis, LBP, CRP, and albumin remained independently associated with nosocomial infection. A WBC + CRP model achieved an AUC of 0.975, whereas LBP + CRP + albumin and LBP + PCT + lactate panels yielded AUCs of 0.997 and 0.999, respectively; both LBP‐based panels significantly outperformed WBC + CRP. An admission LBP + CRP + albumin model provides a pragmatic, high‐performing tool for early risk stratification, while an LBP + PCT + lactate panel offers near‐perfect discrimination as an expanded option; these LBP‐based tri‐marker models may help refine early risk stratification and targeted management in cirrhotic patients with UGIB.
Chen et al. (Fri,) studied this question.