J Korean Soc Emerg Med.  2019 Oct;30(5):401-410. 10.0000/jksem.2019.30.5.401.

Validation of Glasgow-Blatchford score, Pre-Rockall score, and AIMS65 score to predict active bleeding in patients with upper gastrointestinal bleeding in normotensive patients and suggestion for developing new predictors

• Kim D
• Ha YR
• Ahn JH
• Kim YS
• Shin TY
• Jung RB
• Lee KH
• Yu W
• Yoon YT

• KMID: 2470559
• DOI: http://doi.org/10.0000/jksem.2019.30.5.401

Abstract

OBJECTIVE
The aim of this study was to validate the Glasgow-Blatchford score (GBS), Pre-Rockall score (PRS), and AIMS65 score to predict active bleeding in patients with normotension and upper gastrointestinal bleeding (UGIB), and analyze the variables that can predict active bleeding to help develop new predictive factors.
METHODS
Data were collected retrospectively from January 2015 to December 2017. A systolic blood pressure ≥90 mmHg were defined as normotension, and the patients were divided into active bleeding and not-active bleeding groups based on an esophagogastroduodenoscopy and levin-tube irrigation. The GBS, PRS, and AIMS65 of each group were calculated. The receiver operator characteristic (ROC) curve and area under the curve (AUC) were also calculated to obtain the predictive power for active bleeding. Furthermore, the factors that can predict active bleeding were analyzed by multivariate logistic regression. The ROC curve and AUC were calculated using the variables that were adopted as useful factors.
RESULTS
Of the 250 patients included, 85 were active bleeding and 165 were not-active bleeding. The ROC curve showed GBS (AUC, 0.54; 95% confidence interval [CI], 0.47-0.61), PRS (AUC, 0.58; 95% CI, 0.50-0.65), and AIMS65 (AUC, 0.51; 95% CI, 0.43-0.59) to have low predictive power for active bleeding. Multivariate logistic regression revealed the lactate (odds ratio [OR], 1.10; 95% CI, 1.01-1.20) and shock indices (OR, 4.15; 95% CI, 1.12-15.40) to be significant predictors of active bleeding. When calculating the probability of predicting active bleeding through these variables, AUC 0.64 (95% CI, 0.57-0.71) showed higher prediction power than the previous scores.
CONCLUSION
The conventional scoring systems that predict the prognosis of UGIB showed low predictability in predicting active bleeding in UGIB patients with a systolic blood pressure ≥90 mmHg. Further study suggests the development of new score using factors, such as the lactate and shock indices.

Keyword

Gastrointestinal hemorrhage; Emergency medicine; Predictive value; Lactic acid

MeSH Terms

Area Under Curve
Blood Pressure
Emergency Medicine
Endoscopy, Digestive System
Gastrointestinal Hemorrhage
Hemorrhage*
Humans
Lactic Acid
Logistic Models
Prognosis
Retrospective Studies
ROC Curve
Shock
Lactic Acid