Development and Validation of a Scoring System to Predict Severe Acute Lower Gastrointestinal Bleeding in Vietnamese

Early prognosis prediction for non-variceal upper gastrointestinal bleeding in the intensive care unit: based on interpretable machine learning

INTRODUCTION

This study aims to construct a mortality prediction model for patients with non-variceal upper gastrointestinal bleeding (NVUGIB) in the intensive care unit (ICU), employing advanced machine learning algorithms. The goal is to identify high-risk populations early, contributing to a deeper understanding of patients with NVUGIB in the ICU.

METHODS

We extracted NVUGIB data from the Medical Information Mart for Intensive Care IV (MIMIC-IV, v.2.2) database spanning from 2008 to 2019. Feature selection was conducted through LASSO regression, followed by training models using 11 machine learning methods. The best model was chosen based on the area under the curve (AUC). Subsequently, Shapley additive explanations (SHAP) was employed to elucidate how each factor influenced the model. Finally, a case was randomly selected, and the model was utilized to predict its mortality, demonstrating the practical application of the developed model.

RESULTS

In total, 2716 patients with NVUGIB were deemed eligible for participation. Following selection, 30 out of a total of 64 clinical parameters collected on day 1 after ICU admission remained associated with prognosis and were utilized for developing machine learning models. Among the 11 constructed models, the Gradient Boosting Decision Tree (GBDT) model demonstrated the best performance, achieving an AUC of 0.853 and an accuracy of 0.839 in the validation cohort. Feature importance analysis highlighted that shock, Glasgow Coma Scale (GCS), renal disease, age, albumin, and alanine aminotransferase (ALP) were the top six features of the GBDT model with the most significant impact. Furthermore, SHAP force analysis illustrated how the constructed model visualized the individualized prediction of death.

CONCLUSIONS

Patient data from the MIMIC database were leveraged to develop a robust prognostic model for patients with NVUGIB in the ICU. The analysis using SHAP also assisted clinicians in gaining a deeper understanding of the disease.

Prediction of gastrointestinal active arterial extravasation on computed tomographic angiography using multivariate clinical modeling

AIMS

To evaluate the ability of logistic regression and machine learning methods to predict active arterial extravasation on computed tomographic angiography (CTA) in patients with acute gastrointestinal hemorrhage using clinical variables obtained prior to image acquisition.

MATERIALS AND METHODS

CT angiograms performed for the indication of gastrointestinal bleeding at a single institution were labeled retrospectively for the presence of arterial extravasation. Positive and negative cases were matched for age, gender, time period, and site using Propensity Score Matching. Clinical variables were collected including recent history of gastrointestinal bleeding, comorbidities, laboratory values, and vitals. Data were partitioned into training and testing datasets based on the hospital site. Logistic regression, XGBoost, Random Forest, and Support Vector Machine classifiers were trained and five-fold internal cross-validation was performed. The models were validated and evaluated with the area under the receiver operating characteristic curve.

RESULTS

Two-hundred and thirty-one CTA studies with arterial gastrointestinal extravasation were 1:1 matched with 231 negative studies (N=462). After data preprocessing, 389 patients and 36 features were included in model development and analysis. Two hundred and fifty-five patients (65.6%) were selected for the training dataset. Validation was performed on the remaining 134 patients (34.4%); the area under the receiver operating characteristic curve for the logistic regression, XGBoost, Random Forest, and Support Vector Machine classifiers was 0.82, 0.68, 0.54, and 0.78, respectively.

CONCLUSION

Logistic regression and machine learning models can accurately predict presence of active arterial extravasation on CTA in patients with acute gastrointestinal bleeding using clinical variables.

Safe discharge for patients admitted for lower gastrointestinal bleeding (LGITB): derivation and validation of a novel scoring system

AIM

Bleeding from the lower gastrointestinal tract (LGITB) is a common clinical presentation. Recent guidelines have recommended for incorporation of clinical risk assessment tools in the management for LGITB. We derived and validated a novel clinical scoring system to predict safe discharge after LGITB admission, and compared it to other published scoring systems in current literature.

METHODS

A retrospective cohort of 798 patients with LGITB from August 2018 to March 2021 was included in the derivation cohort. Multivariate binary logistic regression was performed to identify significant clinical variables predictive of safe discharge. A clinical scoring system was developed based on the results, and validated on a prospective cohort of 312 consecutive patients with LGITB from April 2021 to March 2022. The performance of the novel scoring system was compared to other LGITB clinical risk assessment scores via area under the receiver operating characteristics curve (AUROC) analysis.

RESULTS

Variables predictive of safe discharge included the following; absence of previous LGITB admission, absence of ischemic heart disease, absence of blood on digital rectal examination, absence of dizziness or syncope at presentation and the systolic blood pressure and haemoglobin levels at presentation. The novel score had an AUROC of 0.907. A cut-off point of 4 provided a sensitivity of 41.9%, specificity of 97.5%, positive predictive value of 96.4% and negative predictive value of 51.5% for prediction of safe discharge. The score performs comparably to the Oakland score.

CONCLUSION

The novel LGITB clinical risk score has good predictive performance for safe discharge in patients admitted for LGITB.

Risk factors for major gastrointestinal bleeding in the general population in Finland

BACKGROUND

Data on non-drug related risk-factors for gastrointestinal bleeding (GIB) in the general population are limited, especially for life-style factors, clinical measurements and laboratory parameters.

AIM

To identify and investigate non-drug risk factors for major GIB in the general population of Finland.

METHODS

We performed a retrospective cohort study using data from the FINRISK health examination surveys, which have been conducted every 5 years across Finland from 1987 to 2007. Participants were adults aged 25 years to 74 years, excluding those with a previous hospitalization for GIB. Follow-up from enrollment was performed through linkage to national electronic health registers and ended at an event of GIB that led to hospitalization/death, death due to any other cause, or after 10 years. Covariates included demographics, socioeconomic and lifestyle factors, clinical measurements, laboratory parameters and comorbidities. Variable selection was undertaken using Least Absolute Shrinkage and Selection Operator (LASSO) and factors associated with GIB were identified using Cox regression.

RESULTS

Among 33,508 participants, 403 (1.2%) experienced GIB [256 men (63.5%); mean age, 56.0 years (standard deviation (SD) ± 12.1)] and 33105 who did not experience GIB [15768 men (47.6%); mean age, 46.8 (SD ± 13) years], within 10 years of follow-up. Factors associated with a significantly increased risk of GIB were baseline age [per 10-year increase; hazard ratio (HR) 1.62, 95% confidence interval (CI): 1.42-1.86], unemployment (HR: 1.70, 95%CI: 1.11-2.59), body mass index (BMI) (HR: 1.15, 95%CI: 1.01-1.32), gamma-glutamyl transferase (GGT) (HR: 1.05, 95%CI: 1.02-1.09), precursors of GIB (HR: 1.90, 95%CI: 1.37-2.63), cancer (HR: 1.47, 95%CI: 1.10-1.97), psychiatric disorders (HR: 1.32, 95%CI: 1.01-1.71), heart failure (HR: 1.46, 95%CI: 1.04-2.05), and liver disorders (HR: 3.20, 95%CI: 2.06-4.97). Factors associated with a significantly decreased risk of GIB were systolic blood pressure (SBP) (HR: 0.78, 95%CI: 0.64-0.96), 6-10 cups of coffee a day (HR: 0.67, 95%CI: 0.46-0.99), or > 10 cups (HR: 0.43, 95%CI: 0.23-0.81).

CONCLUSION

Our study confirms established risk-factors for GIB and identifies potential risk-factors not previously reported such as unemployment, BMI, GGT, SBP and coffee consumption.

Comparison of Risk Scores for Lower Gastrointestinal Bleeding: A Systematic Review and Meta-analysis

IMPORTANCE

Clinical prediction models, or risk scores, can be used to risk stratify patients with lower gastrointestinal bleeding (LGIB), although the most discriminative score is unknown.

OBJECTIVE

To identify all LGIB risk scores available and compare their prognostic performance.

DATA SOURCES

A systematic search of Ovid MEDLINE, Embase, and the Cochrane Central Register of Controlled Trials from January 1, 1990, through August 31, 2021, was conducted. Non-English-language articles were excluded.

STUDY SELECTION

Observational and interventional studies deriving or validating an LGIB risk score for the prediction of a clinical outcome were included. Studies including patients younger than 16 years or limited to a specific patient population or a specific cause of bleeding were excluded. Two investigators independently screened the studies, and disagreements were resolved by consensus.

DATA EXTRACTION AND SYNTHESIS

Data were abstracted according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guideline independently by 2 investigators and pooled using random-effects models.

MAIN OUTCOMES AND MEASURES

Summary diagnostic performance measures (sensitivity, specificity, and area under the receiver operating characteristic curve [AUROC]) determined a priori were calculated for each risk score and outcome combination.

RESULTS

A total of 3268 citations were identified, of which 9 studies encompassing 12 independent cohorts and 4 risk scores (Oakland, Strate, NOBLADS [nonsteroidal anti-inflammatory drug use, no diarrhea, no abdominal tenderness, blood pressure ≤100 mm Hg, antiplatelet drug use (nonaspirin), albumin <3.0 g/dL, disease score ≥2 (according to the Charlson Comorbidity Index), and syncope], and BLEED [ongoing bleeding, low systolic blood pressure, elevated prothrombin time, erratic mental status, and unstable comorbid disease]) were included in the meta-analysis. For the prediction of safe discharge, the AUROC for the Oakland score was 0.86 (95% CI, 0.82-0.88). For major bleeding, the AUROC was 0.93 (95% CI, 0.90-0.95) for the Oakland score, 0.73 (95% CI, 0.69-0.77) for the Strate score, 0.58 (95% CI, 0.53-0.62) for the NOBLADS score, and 0.65 (95% CI, 0.61-0.69) for the BLEED score. For transfusion, the AUROC was 0.99 (95% CI, 0.98-1.00) for the Oakland score and 0.88 (95% CI, 0.85-0.90) for the NOBLADS score. For hemostasis, the AUROC was 0.36 (95% CI, 0.32-0.40) for the Oakland score, 0.82 (95% CI, 0.79-0.85) for the Strate score, and 0.24 (95% CI, 0.20-0.28) for the NOBLADS score.

CONCLUSIONS AND RELEVANCE

The Oakland score was the most discriminative LGIB risk score for predicting safe discharge, major bleeding, and need for transfusion, whereas the Strate score was best for predicting need for hemostasis. This study suggests that these scores can be used to predict outcomes from LGIB and guide clinical care accordingly.

Lower Gastrointestinal Bleeding

Lower gastrointestinal (LGI) bleeding indicates bleeding from colon or anorectum. Typically, patients with LGI bleeding present with bright red blood per rectum or hematochezia, although rarely they can present with melena as well. Alternatively, LGI bleeding is also defined as bleeding from a source within potential reach of a colonoscope, that is, colon and terminal ileum. LGI bleedings have more favorable outcomes than upper GI (UGI) bleeding and less common than UGI bleeding. Any patient presenting with GI bleeding should undergo a detailed history and physical examination for clues that may suggest source and possible etiology. Colonoscopy remains the most widely used and preferred instrument of choice for both diagnosis and therapy. This review will discuss in brief the causes, triaging, and role of colonoscopy in the management of LGI bleeding.

Diagnosis and management of acute lower gastrointestinal bleeding: European Society of Gastrointestinal Endoscopy (ESGE) Guideline

1: ESGE recommends that the initial assessment of patients presenting with acute lower gastrointestinal bleeding should include: a history of co-morbidities and medications that promote bleeding; hemodynamic parameters; physical examination (including digital rectal examination); and laboratory markers. A risk score can be used to aid, but should not replace, clinician judgment.Strong recommendation, low quality evidence. 2 : ESGE recommends that, in patients presenting with a self-limited bleed and no adverse clinical features, an Oakland score of ≤ 8 points can be used to guide the clinician decision to discharge the patient for outpatient investigation.Strong recommendation, moderate quality evidence. 3 : ESGE recommends, in hemodynamically stable patients with acute lower gastrointestinal bleeding and no history of cardiovascular disease, a restrictive red blood cell transfusion strategy, with a hemoglobin threshold of ≤ 7 g/dL prompting red blood cell transfusion. A post-transfusion target hemoglobin concentration of 7-9 g/dL is desirable.Strong recommendation, low quality evidence. 4 : ESGE recommends, in hemodynamically stable patients with acute lower gastrointestinal bleeding and a history of acute or chronic cardiovascular disease, a more liberal red blood cell transfusion strategy, with a hemoglobin threshold of ≤ 8 g/dL prompting red blood cell transfusion. A post-transfusion target hemoglobin concentration of ≥ 10 g/dL is desirable.Strong recommendation, low quality evidence. 5: ESGE recommends that, in patients with major acute lower gastrointestinal bleeding, colonoscopy should be performed sometime during their hospital stay because there is no high quality evidence that early colonoscopy influences patient outcomes.Strong recommendation, low quality of evidence. 6 : ESGE recommends that patients with hemodynamic instability and suspected ongoing bleeding undergo computed tomography angiography before endoscopic or radiologic treatment to locate the site of bleeding.Strong recommendation, low quality evidence. 7 : ESGE recommends withholding vitamin K antagonists in patients with major lower gastrointestinal bleeding and correcting their coagulopathy according to the severity of bleeding and their thrombotic risk. In patients with hemodynamic instability, we recommend administering intravenous vitamin K and four-factor prothrombin complex concentrate (PCC), or fresh frozen plasma if PCC is not available.Strong recommendation, low quality evidence. 8 : ESGE recommends temporarily withholding direct oral anticoagulants at presentation in patients with major lower gastrointestinal bleeding.Strong recommendation, low quality evidence. 9: ESGE does not recommend withholding aspirin in patients taking low dose aspirin for secondary cardiovascular prevention. If withheld, low dose aspirin should be resumed, preferably within 5 days or even earlier if hemostasis is achieved or there is no further evidence of bleeding.Strong recommendation, moderate quality evidence. 10: ESGE does not recommend routinely discontinuing dual antiplatelet therapy (low dose aspirin and a P2Y12 receptor antagonist) before cardiology consultation. Continuation of the aspirin is recommended, whereas the P2Y12 receptor antagonist can be continued or temporarily interrupted according to the severity of bleeding and the ischemic risk. If interrupted, the P2Y12 receptor antagonist should be restarted within 5 days, if still indicated.Strong recommendation, low quality evidence.

Overt gastrointestinal bleeding following haploidentical haematopoietic stem cell transplantation: incidence, outcomes and predictive models

Gastrointestinal bleeding (GIB) accounts for a significant proportion of life-threatening bleeding cases occurring after allogeneic haematopoietic stem cell transplantation (allo-HSCT). However, data on GIB after haploidentical HSCT (haplo-HSCT) are not available. A total of 3180 patients received haplo-HSCT at Peking University People's Hospital from January 2015 to November 2019, and GIB occurred in 188 of these patients (incidence of 5.9%). Platelet counts <30 × 10⁹/L, viral hepatitis, acute kidney injury (AKI), gastrointestinal disease or bleeding before HSCT and sinusoidal obstruction syndrome (SOS) were determined to be significant risk factors for the occurrence of GIB after haplo-HSCT. Grade III-IV acute graft-versus-host disease (aGVHD), AKI, thrombotic microangiopathy (TMA), disseminated intravascular coagulation (DIC) and gastrointestinal disease or bleeding before HSCT were significantly related to mortality in patients with GIB after haplo-HSCT. The predictive models developed for the occurrence and mortality of GIB performed well in terms of discrimination, and they might assist clinicians with personalised strategies for GIB prevention and treatment in patients after haplo-HSCT.

An External Validation Study of the Oakland and Glasgow-Blatchford Scores for Predicting Adverse Outcomes of Acute Lower Gastrointestinal Bleeding in an Asian Population

Aims

This study is aimed at (1) validating the performance of Oakland and Glasgow-Blatchford (GBS) scores and (2) comparing these scores with the SALGIB score in predicting adverse outcomes of acute lower gastrointestinal bleeding (ALGIB) in a Vietnamese population.

Methods

A multicenter cohort study was conducted on ALGIB patients admitted to seven hospitals across Vietnam. The adverse outcomes of ALGIB consisted of blood transfusion; endoscopic, radiologic, or surgical interventions; severe bleeding; and in-hospital death. The Oakland and GBS scores were calculated, and their performance was compared with that of SALGIB, a locally developed prediction score for adverse outcomes of ALGIB in Vietnamese, based on the data at admission. The accuracy of these scores was measured using the area under the receiver operating characteristic curve (AUC) and compared by the chi-squared test.

Results

There were 414 patients with a median age of 60 (48-71). The rates of blood transfusion, hemostatic intervention, severe bleeding, and in-hospital death were 26.8%, 15.2%, 16.4, and 1.4%, respectively. The SALGIB score had comparable performance with the Oakland score (AUC: 0.81 and 0.81, respectively; p = 0.631) and outperformed the GBS score (AUC: 0.81 and 0.76, respectively; p = 0.002) for predicting the presence of any adverse outcomes of ALGIB. All of the three scores had acceptable and comparable performance for in-hospital death but poor performance for hemostatic intervention. The Oakland score had the best performance for predicting severe bleeding.

Conclusions

The Oakland and SALGIB scores had excellent and comparable performance and outperformed the GBS score for predicting adverse outcomes of ALGIB in Vietnamese.