Published online Jul 6, 2023. doi: 10.12998/wjcc.v11.i19.4513
Peer-review started: March 28, 2023
First decision: April 26, 2023
Revised: May 14, 2023
Accepted: May 30, 2023
Article in press: May 30, 2023
Published online: July 6, 2023
Processing time: 93 Days and 17.8 Hours
Upper gastrointestinal (GI) bleeding patients require immediate assessment at the time of arrival in the emergency department (ED). A comprehensive, however fast approach regarding haemodynamic status, transfusion strategy and need for intervention should be performed. This can be achieved by calculating GI bleeding risk scores which should be able to predict several outcomes such as need for intervention, mortality, rebleeding rate or death. Pre-endoscopy risk scores have proved to be a reliable tool which may allow timely sequential decisions. Glasgow-Blatchford score (GBS) has been validated to identify low risk patients which may be managed as outpatients. Pre-endoscopic Rockall score (PERS) evaluates the risk of rebleeding and mortality, while albumin, international normalized ratio, altered mental status, systolic blood pressure, age older than 65 (AIMS65), determines the risk of death. Age, blood tests and comorbidities (ABC) is a relatively new risk score used to predict mortality in patients with both upper and lower GI bleeding. There is a certain variability among these risk scores, potentially due to the differences in population included. Moreover, there is limited data for variceal upper GI bleeding. Venous lactate is another important tool in critically ill patients, such as those with shock, trauma, or heart failure. It has been shown to predict in-hospital mortality, need for intensive care or surgical intervention, as well as rebleeding rate in patients with upper GI bleed. It may be used to improve performance of existing scoring systems and guide clinicians towards early triage of patients.
As emergency services are struggling with high patient flow, clinicians must promptly decide appropriate management in patients with upper GI bleeding. A standardized approach and protocols should attempt to quickly assess the need for admission, timing of endoscopy and level of care. It is generally recommended to perform endoscopy within 24 h of presentation as it plays a pivotal role in identifying the source of bleeding and it can achieve haemostasis in most cases. Unfortunately, it may not be available in all centers or, if performed in low-risk patients, it may overcrowd the service with unnecessary urgent interventions. Hence, we need a standardized tool to guide the emergency medicine clinician for appropriate referral and management of patients. This should reduce the burden and costs on the healthcare system and on-call physicians.
To evaluate the performance of pre-endoscopic risk scores (GBS, PERS, AIMS65, and ABC) in patients with variceal and non-variceal upper GI bleeding for predicting the following primary outcomes: In-hospital mortality, type of intervention (endoscopic or surgical) and length of admission (≥ 7 d). We will further evaluate whether the addition of venous lactate improves the score performance in predicting the determined outcomes.
We retrospectively analyzed all patients above 18 years old presenting to the emergency department (ED) with upper GI bleeding from January 2020 to December 2021. Each patient presenting with exteriorized upper GI bleeding was fully assessed by the emergency medicine physician. Immediate venous catheterization and fluid resuscitation was performed and full work-up with blood tests (full blood count, coagulation parameters, liver, kidney function, venous lactate), and other investigations were performed within 24 h from presentation. Patients with a Hb ≤ 7 g/dL had at least one unit of red blood cell concentrate transfused, with a higher Hb threshold (Hb ≤ 8 g/dL) for patients with associated cardiovascular disease. Post-transfusion target Hb was between 7-9 g/dL. Endoscopy was performed within 24 h of ED arrival in all patients included in analysis. Forrest classification was used to describe peptic ulcer disease, with Baveno and Sarin’s classification for gastroesophageal varices. Patients with non-variceal upper GI bleeding received an infusion with PPIs, while those with variceal bleeding were treated with Somatostatin. Endoscopic treatment was performed depending on the cause of bleeding. A combined approach with injection therapy (dilute epinephrine) and mechanic therapy (thermal coagulation or haemostatic clip) was used for FIa, FIb, and FIIa, with clot removal in FIIb lesions. In variceal bleeding, endoscopic ligation was the main approach. Surgical treatment was performed in cases where endoscopic treatment failed. The need for admission was established by the gastroenterology and general surgery teams on-call.
The final study included 363 patients with upper GI bleeding with a mean age of 60 years old and a predominance of male sex. Non-variceal bleeding was the main cause of presentation, liver cirrhosis the most frequently associated comorbidity in the entire group. The main symptom of presentation was haematemesis in patients with variceal bleeding and melena in the non-variceal group. Approximately 9% of our patients had chronic treatment with antiplatelets or oral anticoagulation. Gastric/duodenal ulcer was the main cause of GI bleeding. Most patients in variceal bleeding group required mechanical endoscopic therapy with band ligation and only 2 patients had variceal sclerotherapy. In the non-variceal bleeding group, dual therapy with thermal anticoagulation and local administration of dilute Adrenaline was the main type of endoscopic intervention. Failed endoscopy was recorded in approximately 4.7% of patients. Only 9 patients in the non-variceal group required surgical intervention. In-hospital mortality had an overall rate of 9.4%, most cases were in the variceal group. All scores had discriminative ability in predicting in-hospital mortality irrespective of study group. AIMS65 score had the best performance for the variceal bleeding group and ABC in the non-variceal bleed. The optimal cut-off value for predicting in-hospital mortality was calculated for each score depending on the type of bleeding. For variceal bleeding, an AIMS65 score above 1 with sensitivity of 90% and specificity of 73% and ABC score with a cut-off value of 5.5 and specificity. In the non-variceal bleeding group, the optimal cut-off value for in-hospital mortality was 1.5 for AIMS65 score, with sensitivity of 66.7% and specificity of 57%. We have determined the best scoring system for in-hospital mortality in the included population, both variceal and non-variceal bleeding, with ABC being the best predictor. For variceal bleeding patients, only PERS score, at a cut-off value above 3.5 was a good predictor for endoscopic treatment with a sensitivity of 76.5% and specificity of 40%. Venous lactate did not show good performance in predicting variceal bleeding, due to low sensitivity (64.3%) and specificity (53.2%). However, logistic regression model showed it is an independent predictor for the determined outcomes. For the variceal bleeding population, addition of lactate to AIMS65 score, leads to a 5-fold increase in probability of in-hospital mortality. For non-variceal bleeding, addition of lactate to GBS score showed a 12-fold increase in probability of in-hospital mortality. In terms of intervention, higher level of venous lactate increases by 5.5 times the probability of endoscopic intervention and by 2 the probability for surgical intervention.
To our knowledge, this is the first study to compare four of the most representative pre-endoscopic risk scores, AIMS65, PERS, GBS and the relatively new ABC score in variceal and non-variceal GI bleeding cohort. ABC score is the most accurate in predicting in-hospital mortality in both mixed and non-variceal bleeding population. AIMS65 had the best performance in predicting in-hospital mortality in patients with variceal upper GI bleeding, however, for practicality, we advise the use of ABC score for both populations. In terms of intervention, PERS and GBS should be used to determine the need for endoscopic and surgical intervention. Lactate can be used in conjunction to AIMS65 and GBS score to predict in-hospital mortality and intervention.
Although GBS is currently largely used, further studies are needed to investigate the relatively new ABC score regarding its role in daily clinical practice and possible implementation in guidelines.