{
  "abstract": "Background Risk stratification is central to managing upper gastrointestinal bleeding (UGIB). Although several scores predict mortality and re-bleeding, most were derived in mixed or non-variceal populations. Given the distinct pathophysiology of variceal bleeding, we aimed to evaluate the discriminatory performance of established UGIB risk scores in this specific population to determine their clinical applicability for risk stratification.Methods We performed a retrospective analysis of patients presenting with oesophagogastroduodenoscopy (OGD)-proven variceal bleeding at a UK tertiary centre between 2015 and 2023. The ABC score, CHAMPS score, MAP(ASH) score, and Glasgow–Blatchford Score (GBS) were calculated using available clinical and laboratory data; missing values were handled by complete-case analysis, yielding score availability of 83.6% - 91.3%. The primary and secondary outcomes were 30-day inpatient mortality and re-bleeding within 6 weeks, respectively. Discriminatory performance was assessed using receiver operating characteristic (ROC) curves and area under the curve (AUROC).Results Among 323 included patients (inpatient mortality 36%; mean MELD 15.4), discriminatory performance for mortality was limited overall.For mortality, overall discriminatory performance was limited. The ABC score demonstrated the highest discrimination (AUROC 0.62 [poor discrimination]), followed by CHAMPS (0.61), MAP(ASH) (0.55), and GBS (0.52 [no discrimination])For re-bleeding, discriminatory performance was uniformly poor: GBS (AUROC 0.53), MAP(ASH) (0.52), CHAMPS (0.49), and ABC (0.46). No score achieved clinically useful prediction for re-bleeding.Conclusion In patients with endoscopy-proven variceal bleeding, commonly used UGIB risk scores demonstrate limited discriminatory performance for both mortality and re-bleeding. Although ABC and CHAMPS showed comparatively higher discrimination for mortality, and GBS performed marginally better for re-bleeding, overall prognostic accuracy remained modest. These findings highlight the challenges of applying existing UGIB risk scores to variceal bleeding populations and suggest that current tools are best suited for contextual risk stratification rather than definitive prognostic prediction, underscoring the need for the development and validation of variceal-specific risk prediction models.",
  "authors": [
    {
      "affiliations": [
        "Royal Stoke University Hospital, University Hospitals North Midlands, Stoke-on-Trent, United Kingdom"
      ],
      "name": "Mohamed Jalloh"
    },
    {
      "affiliations": [
        "Royal Stoke University Hospital, University Hospitals North Midlands, Stoke-on-Trent, United Kingdom"
      ],
      "name": "Hareesha Bharadwaj"
    },
    {
      "affiliations": [
        "Royal Stoke University Hospital, University Hospitals North Midlands, Stoke-on-Trent, United Kingdom"
      ],
      "name": "Maryam Javed"
    },
    {
      "affiliations": [
        "Royal Stoke University Hospital, University Hospitals North Midlands, Stoke-on-Trent, United Kingdom"
      ],
      "name": "Humberto Gomez"
    },
    {
      "affiliations": [
        "Royal Stoke University Hospital, University Hospitals North Midlands, Stoke-on-Trent, United Kingdom"
      ],
      "name": "Jessica Lightfoot"
    },
    {
      "affiliations": [
        "Royal Stoke University Hospital, University Hospitals North Midlands, Stoke-on-Trent, United Kingdom"
      ],
      "name": "Vanessa Carr"
    },
    {
      "affiliations": [
        "Royal Stoke University Hospital, University Hospitals North Midlands, Stoke-on-Trent, United Kingdom"
      ],
      "name": "Shannon Lewis"
    },
    {
      "affiliations": [
        "Royal Stoke University Hospital, University Hospitals North Midlands, Stoke-on-Trent, United Kingdom"
      ],
      "name": "Sushanth Shankar"
    },
    {
      "affiliations": [
        "Royal Stoke University Hospital, University Hospitals North Midlands, Stoke-on-Trent, United Kingdom"
      ],
      "name": "Thomas Eardley"
    },
    {
      "affiliations": [
        "Royal Stoke University Hospital, University Hospitals North Midlands, Stoke-on-Trent, United Kingdom",
        "School of Medicine, Keele University, Stoke-on-Trent, United Kingdom"
      ],
      "name": "Balamrit-Singh Sokhal"
    },
    {
      "affiliations": [
        "Royal Stoke University Hospital, University Hospitals North Midlands, Stoke-on-Trent, United Kingdom"
      ],
      "name": "Farhan Gohar"
    }
  ],
  "title": "P99 Evaluation of UGIB risk scores in a variceal bleeding cohort: discriminatory performance for mortality and re-bleeding",
  "uid": "7bae5a77-e975-5515-87e8-d3b8094aa987"
}