Comparison of risk scores in upper gastrointestinal bleeding in western India: A prospective analysis

Development and validation of a machine learning model to predict hemostatic intervention in patients with acute upper gastrointestinal bleeding

BACKGROUND

Acute upper gastrointestinal bleeding (UGIB) is common in clinical practice and has a wide range of severity. Along with medical therapy, endoscopic intervention is the mainstay treatment for hemostasis in high-risk rebleeding lesions. Predicting the need for endoscopic intervention would be beneficial in resource-limited areas for selective referral to an endoscopic center. The proposed risk stratification scores had limited accuracy. We developed a machine learning model to predict the need for endoscopic intervention in patients with acute UGIB.

METHODS

A prospectively collected database of UGIB patients from 2011 to 2020 was retrospectively reviewed. Patients older than 18 years diagnosed with UGIB who underwent endoscopy were included. Data comprised demographic characteristics, clinical presentation, and laboratory parameters. The cleaned data was used for model development and validation in Python. We conducted 80%-20% split sample training and test sets. The training set was used for supervised learning of 15 models using a stratified 5-fold cross-validation process. The model with the highest AUROC was then internally validated with the test set to evaluate performance.

RESULTS

Of 1389 patients, 615 (44.3%) of the cohorts received the endoscopic intervention (293 variceal- and 336 nonvariceal-bleeding interventions). Eighteen features, including demographic characteristics, clinical presentation, and laboratory parameters, were selected as input for 15 machine learning models. The result revealed that the linear discriminant analysis model could achieve the highest AUROC of 0.74 to predict endoscopic intervention. The model was validated with the test set, in which the AUROC was increased from 0.74 to 0.81. Finally, the model was deployed as a web application by Streamlit.

CONCLUSIONS

Our machine learning model can identify patients with acute UGIB who need endoscopic intervention with good performance. This may help primary care physicians prioritize patients who need referrals and optimize resource allocation in resource-limited areas. Further development and identification of more specific features might improve prediction performance.

TRIAL REGISTRATION

None (Retrospective cohort study) PATIENT & PUBLIC INVOLVEMENT: None.

Comparative diagnostic utility of Rockall and Glasgow-Blatchford scores in non-variceal upper gastrointestinal bleeding: a systematic review and meta-analysis

INTRODUCTION

The Rockall score and Glasgow-Blatchford score (GBS) are two scoring systems validated in the evaluation of upper gastrointestinal bleeding (UGIB). However, no meta-analysis exists to summarize the current data and clarify the use of Rockall score and GBS focusing on non-variceal UGIB. We aimed to evaluate and compare the utility of the Rockall score and GBS in predicting clinical outcomes in non-variceal UGIB.

METHODS

We conducted a systematic review and meta-analysis, searching the MEDLINE and EMBASE databases for all peer-reviewed articles using the terms including 'Glasgow-Blatchford', 'Rockall', and 'gastrointestinal bleed' from their inception to 22 March 2023. Outcomes included mortality, rebleeding, need for blood transfusion, and need for surgical intervention.

RESULTS

Seven studies with 755 participants with non-variceal bleeding were included in the analysis. Pooled analysis demonstrated no difference in the area under the receiver operating characteristic (AUROC) between GBS and Rockall score to predict mortality [weighted mean difference (WMD) = 0.01, 95% CI: -0.06 to 0.08] or rebleeding (WMD = 0.04, 95% CI: -0.03 to 0.11). GBS had a higher AUROC to predict the outcomes compared to Rockall score for the needs for transfusion (WMD = 0.09, 95% CI: 0.01-0.16) and surgical intervention (WMD = 0.21, 95% CI: 0.14-0.29).

CONCLUSION

The GBS could be superior to the Rockall score in predicting the needs for transfusion and surgical intervention in non-variceal UGIB. However, both scores demonstrate low performance for predicting mortality or rebleeding.

Clinical and Biochemical Differences in Patients Having Non-Variceal Upper Gastrointestinal Bleeding on NSAIDs, Oral Anticoagulants, and Antiplatelet Therapy

Introduction: Upper gastrointestinal bleeding (UGIB) is among the most common causes of morbidity and mortality worldwide, accounting for major resource allocation and increasing incidence. This study aimed to evaluate the severity of non-variceal bleeding in patients at risk of bleeding through the use of NSAIDs, oral anticoagulants, and antiplatelet therapy. Material and Method: The study included 296 patients admitted in the Gastroenterology Department of the Municipal County Emergency University Hospital, Timisoara, between 01.01.2018 and 01.04.2020, and diagnosed via gastroscopy with non-variceal gastrointestinal bleeding. The patients were divided among four groups based on their use of different drugs known to induce UGIB, i.e., aspirin and clopidogrel, NOACs, NSAIDs, and anti-vitamin K drugs, respectively. Statistical analyses were performed based on ANOVA one-way tests for continuous variables and Chi-square tests for categorical variables with pairwise comparisons based on Bonferroni adjusted significance tests. Results: The results showed several parameters having statistical significance among the different groups of patients. Patients on NOACs had statistically significant lower hemoglobin levels, lower hematocrit values, lower erythrocytes, lower RDW and higher fibrinogen levels compared to patients on VKA. Discussion: Surprisingly, the results from our study suggest that the use of NOACs was associated with a higher risk of bleeding when compared to VKA, which differs from the existing literature. Conclusions: One of the important factors causing upper non-variceal bleeding can be iatrogenic, either due to antiplatelet drugs or anticoagulants, to which NSAID treatment is additionally associated for various reasons. In our study, the use of NOACs seemed to have a more severe bleeding spectrum with higher morbidity compared to VKA.

At admission hemodynamic instability is associated with increased mortality and rebleeding rate in acute gastrointestinal bleeding: a systematic review and meta-analysis

BACKGROUND

Acute gastrointestinal bleeding (GIB) is a life-threatening event. Around 20-30% of patients with GIB will develop hemodynamic instability (HI).

OBJECTIVES

We aimed to quantify HI as a risk factor for the development of relevant end points in acute GIB.

DESIGN

A systematic search was conducted in three medical databases in October 2021.

DATA SOURCES AND METHODS

Studies of GIB patients detailing HI as a risk factor for the investigated outcomes were selected. For the overall results, pooled odds ratios (ORs) with 95% confidence intervals (CIs) were calculated based on a random-effects model. Subgroups were formed based on the source of bleeding. The Quality of Prognostic Studies tool was used to assess the risk of bias.

RESULTS

A total of 62 studies were eligible, and 39 were included in the quantitative synthesis. HI was found to be a risk factor for both in-hospital (OR: 5.48; CI: 3.99-7.52) and 30-day mortality (OR: 3.99; CI: 3.08-5.17) in upper GIB (UGIB). HI was also associated with higher in-hospital (OR: 3.68; CI: 2.24-6.05) and 30-day rebleeding rates (OR: 4.12; 1.83-9.31) among patients with UGIB. The need for surgery was also more frequent in hemodynamically compromised UGIB patients (OR: 3.65; CI: 2.84-4.68). In the case of in-hospital mortality, the risk of bias was high for 1 (4%), medium for 13 (48%), and low for 13 (48%) of the 27 included studies.

CONCLUSION

Hemodynamically compromised patients have increased odds of all relevant untoward end points in GIB. Therefore, to improve the outcomes, adequate emergency care is crucial in HI.

REGISTRATION

PROSPERO registration number: CRD42021285727.

Predicting mortality in patients with nonvariceal upper gastrointestinal bleeding using machine-learning

BACKGROUND

Non-endoscopic risk scores, Glasgow Blatchford (GBS) and admission Rockall (Rock), are limited by poor specificity. The aim of this study was to develop an Artificial Neural Network (ANN) for the non-endoscopic triage of nonvariceal upper gastrointestinal bleeding (NVUGIB), with mortality as a primary outcome.

METHODS

Four machine learning algorithms, namely, Linear Discriminant Analysis (LDA), Quadratic Discriminant Analysis (QDA), logistic regression (LR), K-Nearest Neighbor (K-NN), were performed with GBS, Rock, Beylor Bleeding score (BBS), AIM65, and T-score.

RESULTS

A total of 1,096 NVUGIB hospitalized in the Gastroenterology Department of the County Clinical Emergency Hospital of Craiova, Romania, randomly divided into training and testing groups, were included retrospectively in our study. The machine learning models were more accurate at identifying patients who met the endpoint of mortality than any of the existing risk scores. AIM65 was the most important score in the detection of whether a NVUGIB would die or not, whereas BBS had no influence on this. Also, the greater AIM65 and GBS, and the lower Rock and T-score, the higher mortality will be.

CONCLUSION

The best accuracy was obtained by the hyperparameter-tuned K-NN classifier (98%), giving the highest precision and recall on the training and testing datasets among all developed models, showing that machine learning can accurately predict mortality in patients with NVUGIB.

Gastrointestinal Emergencies and the Role of Endoscopy

Many gastrointestinal (GI) disorders present to the emergency room with acute clinical presentations, some even life threatening. Common emergencies encountered that require urgent endoscopic interventions include GI hemorrhage (variceal and nonvariceal), foreign body ingestion, obstructive jaundice, postprocedure-related complications such as postpolypectomy bleed or perforation, etc. A major advantage of emergency endoscopy is that it is cost effective and, on many occasions, can be life-saving. The present review will highlight a practical approach on various endoscopic modalities and their use in the GI emergencies.

A Comparative Analysis of Risk Scoring Systems in Predicting Clinical Outcomes in Upper Gastrointestinal Bleed

Background

Upper gastrointestinal bleed (UGIB) is a life-threatening condition that presents as hematemesis (fresh blood), coffee-ground vomiting, or melena. Multiple scoring systems are developed to predict different clinical outcomes, which are important to managing UGIB and are essential to determining low and high-risk patients. The study aimed to compare the sensitivity and specificity of risk scoring systems and their optimum cut-off values in the assessment of UGIB.

Methods

The prospective cross-sectional study included patients (N = 81) with acute UGIB. Four different proposed scores [Glasgow-Blatchford score (GBS), AIMS65, pre-endoscopic Rockall, and full Rockall scoring system] were used for evaluating patients with UGIB. The optimum cut-off values of these risk scores were used to predict the clinical outcomes.

Results

The AIMS65 score [Area Under the Receiver Operating Characteristic curve (AUROC): 0.91, cut-off: >1, sensitivity: 100%, specificity: 76.62%] and pre-Rockall were similar (AUROC: 0.91, cut-off: >0, sensitivity: 100%, specificity: 93.51%) at predicting mortality. The GBS (cut-off: >9, AUROC: 0.79, sensitivity: 69.23, specificity: 87.50) and AIMS65 scores (cut-off: >0, AUROC: 0.67, sensitivity: 72.31, specificity: 62.5) were good predictors of need for ICU care.

Conclusion

GBS was superior in predicting categorization into high risk and low risk, and endoscopic intervention, blood transfusion, and intensive care unit (ICU) care in UGIB patients. Pre-Rockall score and AIMS65 score were similar in predicting the mortality rate in UGIB.

The risk factors for the recurrent upper gastrointestinal hemorrhage among acute peptic ulcer disease patients in Syria: A prospective cohort study

BACKGROUND

Upper gastrointestinal bleeding (UGIB) is a life-threatening medical emergency characterized by bleeding from the esophagus, stomach, or duodenum. This study aims to analyze the risk factors for upper gastrointestinal tract rebleeding among acute peptic ulcer patients.

METHODS

This is a cohort clinical study conducted between July 2018 and June 2020. Patients admitted or hospitalized because of UGIB or developed it during their hospital stay were included.s The patients were divided into two groups for the statistical analysis using Forrest's ulcer rebleeding risk classification. Group 1: Forrest 1a+1b+2a+2b, and group 2: Forrest 2c+3. The fasting time before the endoscopic procedure was from 12 to 24 hours. Follow-ups were conducted for 30 days after the treatment.

RESULTS

The total number of included subjects was 152, out of which 57.89% (n = 88) were male patients. The mean SD for patients' age was 52.63 16.89±; more than 40% (n = 62) of subjects were using antiplatelet medications, while only 13.15% (n = 20) used NSAIDs, and the mean SD for the transferred units was 2.32 ± 1.88, 7.24% (n = 11) of patients died. After 30 days of the treatment, 6.57% (n = 10) of patients suffered from recurrent bleeding. The most common presentation was melena 67.95% (n = 103), 53% (n = 81) of patients had hematemesis, 69.73% (n = 106) patients had gastric ulcer and 30.26% (n = 46) had duodenal ulcers.

CONCLUSION

Age, NSAIDs, altered mental capacity, Forrest classification (Ia,Ib, and IIa), and blood transfusion were associated with a higher risk of rebleeding. Furthermore, patients who needed 3.83 blood units were at higher risk of recurrent bleeding.

Prediction model of emergency mortality risk in patients with acute upper gastrointestinal bleeding: a retrospective study

Background

Upper gastrointestinal bleeding is a common presentation in emergency departments and carries significant morbidity worldwide. It is paramount that treating physicians have access to tools that can effectively evaluate the patient risk, allowing quick and effective treatments to ultimately improve their prognosis. This study aims to establish a mortality risk assessment model for patients with acute upper gastrointestinal bleeding at an emergency department.

Methods

A total of 991 patients presenting with acute upper gastrointestinal bleeding between July 2016 and June 2019 were enrolled in this retrospective single-center cohort study. Patient demographics, parameters assessed at admission, laboratory test, and clinical interventions were extracted. We used the least absolute shrinkage and selection operator regression to identify predictors for establishing a nomogram for death in the emergency department or within 24 h after leaving the emergency department and a corresponding nomogram. The area under the curve of the model was calculated. A bootstrap resampling method was used to internal validation, and decision curve analysis was applied for evaluate the clinical utility of the model. We also compared our predictive model with other prognostic models, such as AIMS65, Glasgow-Blatchford bleeding score, modified Glasgow-Blatchford bleeding score, and Pre-Endoscopic Rockall Score.

Results

Among 991 patients, 41 (4.14%) died in the emergency department or within 24 h after leaving the emergency department. Five non-zero coefficient variables (transfusion of plasma, D-dimer, albumin, potassium, age) were filtered by the least absolute shrinkage and selection operator regression analysis and used to establish a predictive model. The area under the curve for the model was 0.847 (95% confidence interval [0.794–0.900]), which is higher than that of previous models for mortality of patients with acute upper gastrointestinal bleeding. The decision curve analysis indicated the clinical usefulness of the model.

Conclusions

The nomogram based on transfusion of plasma, D-dimer, albumin, potassium, and age effectively assessed the prognosis of patients with acute upper gastrointestinal bleeding presenting at the emergency department.

The Clinical Impact of Rockall and Glasgow-Blatchford Scores in Nonvariceal Upper Gastrointestinal Bleeding

Introduction

Risk stratification in patients with nonvariceal upper gastrointestinal bleeding (NVUGIB) is crucial for proper management. Rockall score (RS; pre-endoscopic and complete) and Glasgow-Blatchford score (GBS) are some of the most used scoring systems. This study aims to analyze these scores' ability to predict various clinical outcomes and possible cutoff points to identify low- and high-risk patients. Secondarily, this study intents to evaluate the appropriateness of patients' transfers to our facility, which provides a specialized emergency endoscopy service.

Methods

This study was retrospectively conducted at Centro Hospitalar Universitário do Porto and included patients admitted to the Emergency Department with acute manifestations of NVUGIB between January 2016 and December 2018. Receiver operating characteristic (ROC) curves and corresponding areas under the curve (AUC) were calculated. Transferred patients from other institutions and nontransferred (directly admitted to this institution) patients were also compared.

Results

Of a total of 420 patients, 23 (5.9%) died, 34 (8.4%) rebled, 217 (51.7%) received blood transfusion, 153 (36.3%) received endoscopic therapy, 22 (5.7%) had surgery, and 171 (42.3%) required hospitalization in the Intermediate or Intensive Care Unit. Regarding mortality prediction, both complete RS (AUC 0.756, p < 0.001) and pre-endoscopic RS (AUC 0.711, p = 0.001) showed good performance. In the prediction of rebleeding, only complete RS (AUC 0.735, p < 0.001) had discriminative ability. GBS had good performance in the prediction of transfusion (AUC 0.785, p < 0.001). No score showed discriminative capability in the prediction of other outcomes. Transferred and nontransferred patients had similar pre-endoscopic RS (3.41 vs. 3.34, p = 0.692) and GBS (13.29 vs. 12.29, p = 0.056). Only patients with GBS ≥6 were transferred to our facility. There were no adverse outcomes recorded in any group when GBS was ≤3.

Discussion/Conclusion

Complete RS and pre-endoscopic RS are effective at predicting mortality, but only complete RS showed good performance at predicting rebleeding. GBS is better at predicting transfusion requirement. Our study suggests that a transfer can possibly be reconsidered if GBS is ≤3, although current recommendations only propose outpatient care when GBS is 0 or 1. Patients' transfers were appropriate, considering the high GBS scores and the outcomes of these patients.