Introduction
Several scoring systems have been established to identify upper gastrointestinal bleeding (UGIB) patients at high risk of adverse outcomes, being the best ones MAP (ASH) and ABC (1). Traditional scores(Glasgow-Blatchford, Rockall, AIMS65) were validated and compared mostly in nonvariceal-UGIB. On the contrary, MAP (ASH) and ABC included variceal bleed (VB) in their validation and comparison cohorts.
Also, to determine the prognosis of liver cirrhosis there are two frequently used scores, Child-Pugh score (CP) and MELD, and recently, ALBI score. Previous studies have reported the ability of these in predicting VB outcomes. A recent metanalysis show MELD and CP performed better predicting mortality than traditional scores (2). But there are no studies comparing liver scores with MAP (ASH) or ABC.
Aims & Methods
To compare the predictive value of MAP (ASH) and ABC scores vs. liver scores stratifying outcomes of VB.
Retrospective analysis of a prospective registry of UGIB patients admitted to ‘Virgen de las Nieves University Hospital’ (2013-2020). VB was defined as that coming from gastric or esophageal varices. Main outcomes were in-hospital (intervention,rebleeding, misbalance of chronic condition, mortality) and delayed-6months (UGIB, cardiovascular, mortality) ones. Intervention is a composite endpoint including need of transfusion, endoscopy therapy, interventional radiology or surgery. Statistical analysis was performed by SPSS and MedCalc.
Results
We included 1345 UGIB patients, 241 with VB, mean age 60.70+/-12.62, 26.1% females. Main differences between deaths and survivors were hepatocellular carcinoma, (57.1%vs.16.1%,p=0.001, OR 6.97,95%CI 2.28-21.34), misbalance of chronic condition (40.7% vs. 5.5%, p<0.001, OR 11.83,95 % CI 5.15-27.22), intervention (21.1% vs. 3%, p=0.013, OR 8.56,95 % CI 1.14-64.49), rebleeding (50%vs.12.7%, p<0.001, OR 6.88, 95% CI 3.12-15.20) and length hospital stay (15.91vs.9.23,p=0.031). In the multivariate analysis, misbalance of chronic condition was an independent risk factor for mortality (OR 4.23, 95 % CI 1.11-16.06, p=0.034).
Comparisons of ROC curves for relevant outcomes are shown in the table, with statiscal differences only for intervention (MAP(ASH) >CP, p=0.0195), misbalance of chronic condition (MAP(ASH) >ALBI, p=0.023; ABC > CP, p=0.020; ABC > ALBI, p=0.001) and delayed UGIB/liver mortality (MELD > Child, p=0.042).
AUROC (95%CI)
Intervention | MAP(ASH)
0.744*
(0.674 - 0.806) | ABC
0.701
(0.628 - 0.767) | MELD
0,622
(0,546 - 0.693) | Child-Pugh
0.582
(0.506 - 0.655) | ALBI
0.668
(0.594 - 0.737) |
| Rebleeding | 0.553
(0.477 - 0.627) | 0.522
(0.446 - 0.597) | 0.602
(0.526 - 0.674) | 0.546
(0.470 - 0.620) | 0.543
(0.467 - 0.617) |
| Misbalance chronic condition | 0.721*
(0.649 - 0.785) | 0.770*
(0.702 - 0.830) | 0.692
(0.619 - 0.759) | 0.662
(0.587 - 0.730) | 0.612
(0.536 - 0.684) |
| In-hospital mortality | 0.678
(0.604 - 0.747) | 0.677
(0.604 - 0.747) | 0.734
(0.662 - 0.798) | 0.741
(0.669 - 0.804) | 0.677
(0.602 - 0.746) |
| UGIB in-hospital mortality | 0.656
(0.568 - 0.736) | 0.633
(0.545 - 0.714) | 0.770
(0.689 - 0.838) | 0.774
(0.694 - 0.842) | 0.711
(0.626 - 0.786) |
| Other cause in-hospital mortality | 0.751
(0.666 - 0.823) | 0.762
(0.666 - 0.823) | 0.696
(0.607 - 0.774) | 0.722
(0.635 - 0.798) | 0.638
(0.547 - 0.721) |
| Delayed cardiovascular events | 0.634
(0.553 - 0.709) | 0.534
(0.452 - 0.614) | 0.566
(0.484 - 0.645) | 0.588
(0.506 - 0.666) | 0.658
(0.506 - 0.666) |
| Delayed UGIB | 0.534
(0.453 - 0.615) | 0.582
(0.501 - 0.660) | 0.536
(0.455 - 0.616) | 0.516
(0.434 - 0.596) | 0.519
(0.437 - 0.599) |
| Delayed Mortality | 0.581
(0.499 - 0.659) | 0.683
(0.603 - 0.755) | 0.555
(0.473 - 0.635) | 0.617
(0.535 - 0.694) | 0.676
(0.596 - 0.749) |
| Delayed UGIB and liver mortality | 0.600
(0.340 - 0-824) | 0.733
(0.468 - 0-914) | 0.850*
(0.596 - 0-974) | 0.542
(0.289 - 0-780) | 0.650
(0.386 - 0-860) |
Conclusion
In our study, MAP (ASH) and ABC scores performed better at prediction intervention and misbalance of chronic condition. On the other hand, MELD is superior to CP in predicting delayed UGIB and liver mortality. None of them performs well in predicting rebleeding or other delayed outcomes. Although we see some differences in AUROCs for the other outcomes, no statical differences were found. Therefore, MAP (ASH) and ABC have a good performance not only in patients with UGIB but also in VB.
Independent risk factor for in-hospital mortality was misbalance of chronic condition, showing once again that it is not UGIB itself what determines the evolution if not the imbalance that it produces.
References
- Jimenez-Rosales R, Lopez-Tobaruela JM, Lopez-Vico M, Ortega-Suazo EJ, Martinez-Cara JG, Redondo-Cerezo E. Performance of the New ABC and MAP(ASH) Scores in the Prediction of Relevant Outcomes in Upper Gastrointestinal Bleeding. Journal of Clinical Medicine. 2023; 12(3):1085.
- Yang L, Sun R, Wei N, Chen H. Systematic review and meta-analysis of risk scores in prediction for the clinical outcomes in patients with acute variceal bleeding. Ann Med. 2021 Dec;53(1):1806-1815.