Timing of endoscopy in patients with upper gastrointestinal bleeding

The optimal timing of endoscopy in patients with acute upper gastrointestinal bleeding (UGIB) remains controversial. In this study, we investigated the clinical outcomes of urgent endoscopy in patients with UGIB compared with elective endoscopy. From January 2016 to December 2018, consecutive patients who visited the emergency department and underwent endoscopy for clinical manifestations of acute UGIB, including variceal bleeding, were eligible. Urgent endoscopy (within 6 h) and elective endoscopy (after 6 h) were defined as the time taken to perform endoscopy after presentation to the emergency department. The primary outcome was mortality rate within 30 days. A total of 572 patients were included in the analysis. Urgent endoscopy was performed in 490 patients (85.7%). The 30-day mortality rate did not differ between the urgent and elective endoscopy groups (5.3% and 6.1%, p = 0.791). There was no difference regarding the recurrent bleeding rate, total amount of transfusion, or length of hospital between the groups. In multivariate analysis, age and the amount of transfusion were associated with mortality. Urgent endoscopy was not associated with a lower 30-day mortality rate compared with elective endoscopy in patients with acute UGIB.


Ethics information.
Approval for accessing patient information was granted from the Institutional Review Board of Gangneung Asan Hospital (2020-03-009). As this retrospective data collection was considered anonymized by the Ethics committee, the need for patient consent was waived by the Institutional Review Board of Gangneung Asan Hospital. The study was conducted according to good clinical and scientific practices and following the ethical principles of the Declaration of Helsinki.

Results
Characteristics of the study population. The baseline characteristics of the study population are summarized in Table 1. The median age of the 572 patients was 63 years (range, 19-95 years) and 64.0% were male. More than 80% of the patients had comorbidities, including diabetes mellitus, cerebrovascular accident, chronic kidney disease, or liver cirrhosis. The proportion of patients who were taking antithrombotic agents at the time of admission was 23.8%, and 25 patients (4.4%) were on dual antiplatelet therapy.
Regarding the presenting symptoms, 51.9% of the patients had hematemesis, and 40.7% had melena. When comparing the urgent and elective endoscopy group, the median systolic blood pressure was significantly lower in the urgent group (p = 0.015). The GBS was 11 in the urgent group and 10 in the elective group, respectively (p = 0.334). The proportion of patients with GBS greater than 7 was 75.1% (368/490) in the urgent group and 67.1% (55/82) in the elective group (p = 0.135).
Endoscopic findings and clinical outcomes. The median time to endoscopy was 2.4 h (range, 0.4-45.0 h). Endoscopic diagnoses included peptic ulcer (n = 310, 54.2%), gastroesophageal varix (n = 142, 24.8%), Mallory-Weiss syndrome (n = 49, 8.6%), and malignancy (n = 31, 5.4%) ( Table 2). Endoscopic treatment was performed in 338 (59.1%) patients, which was successful in 334 patients. Two patients with duodenal ulcers required trans-arterial embolization after failed endoscopic treatment. One patient who had a gastric gastrointestinal stromal tumor underwent surgery, and the remaining patient with a duodenal ulcer had conservative management after confirmation of the absence of contrast extravasation on computed tomography scan with angiography. In one patient, duodenal varix was found at endoscopic examination and embolization was performed immediately as the first-line treatment.  www.nature.com/scientificreports/ The mortality rate within 30 days was 5.4% in total (31/572), 5.3% (26/490) in the urgent endoscopy group and 6.1% (5/82) in the elective endoscopy group, respectively (p = 0.791) (Fig. 2). In a subgroup analysis of patients with variceal bleeding (n = 142), mortality rate did not differ between the two groups (Supplementary Table 1). In-hospital death occurred in 23 patients; the cause of death was hypovolemic shock associated with uncontrolled bleeding in 4 patients, and hepatic failure in 12 patients.
The rebleeding rate was assessed in 571 patients, excluding the patient who underwent embolization immediately after endoscopy. The rebleeding rate within 30 days was 6.0% (34/571) in total, 5.9% (29/489) in the urgent group, and 6.1% (5/82) in the elective group (p = 1.000). The median duration of hospital stay and the amount of transfused RBC did not differ between two groups.

Discussion
In this study, we compared the clinical outcomes of 572 patients with acute UGIB by the timing of emergency endoscopy. The mortality rate within 30 days and the recurrent bleeding rate did not differ between the patients who underwent urgent endoscopy within 6 h of presentation and those examined later. Logistic regression analysis also showed that urgent endoscopy was not associated with lower mortality rate, while age and the amount of transfusion were associated with increased mortality.
Most guidelines recommend performing endoscopy within 24 h of presentation among patients with acute UGIB 3-6 . Endoscopy performed within 24 h was associated with a reduced length of hospital stay, and delayed endoscopy was associated with higher mortality in patients with UGIB 8,15 . However, previous studies have shown conflicting results as to whether an earlier endoscopy within 6 to 12 h after presentation can offer more benefit for patients with UGIB. A retrospective study of 934 high-risk patients with GBS ≥ 12 found that endoscopy performed within 13 h resulted in a lower mortality rate and shorter hospital stays than later endoscopy 9 . In addition, a cohort study showed that urgent endoscopy within 6 h was an independent predictor of a lower mortality rate compared with elective endoscopy (6-48 h) in high-risk patients with GBS > 7 10 .
In contrast to these results, a retrospective study of 169 patients with acute non-variceal UGIB revealed no significant difference in mortality rate, rebleeding rate, or hospital stay between patients receiving endoscopy within 6 h and within 6-24 h 16 . Similarly, another study showed that early endoscopy within 12 h was not  www.nature.com/scientificreports/ associated with reduced mortality compared with later endoscopy within 12-24 h 17 . In a recent randomized controlled trial, a total of 516 high-risk patients (GBS ≥ 12) with overt signs of acute UGIB, including variceal bleeding, were randomized to urgent (within 6 h after gastroenterology consultation) and early (within 24 h) endoscopy groups 7 . Of note, some patients were not treated as assigned because of hemodynamic instability or changes in their medical conditions. There was no significant difference between the urgent and early groups in mortality rate (8.9% vs 6.6%, hazard ratio, 1.35; 95% CI 0.72-2.54) and rebleeding rate (10.9% vs 7.8%, hazard ratio, 1.46; 95% CI 0.83-2.58). In addition, another recent retrospective study of 6474 patients presented with acute UGIB showed that urgent endoscopy (within 6 h) showed worse outcomes compared with early (between 6 and 24 h) and late (between 24 and 48 h) endoscopy groups 18 . In the present study, we included patients with acute UGIB regardless of the final diagnosis. The 30-day mortality rate and rebleeding rates did not differ between the patients who underwent endoscopy within 6 h and those who underwent endoscopy after 6 h. These results support recent guidelines that recommends performing endoscopy within 24 h following hemodynamic resuscitation in patients with UGIB and that do not encourage emergent (within 12 h) endoscopy 4,5 . Endoscopy performed early in the clinical course is considered useful to triage patients based on the endoscopic findings. Patients with a low risk of recurrent bleeding could be discharged earlier. In addition, early intervention within 24 h was associated with a shorter length of hospital stay 19 . In contrast, some studies have suggested that earlier endoscopy may yield more high-risk endoscopic stigmata that would have been resolved with proton pump inhibitor therapy. A previous study showed that more retained blood and more actively bleeding lesions were found in patients who underwent endoscopic examination within 8 h from presentation compared with those who underwent endoscopy between 8 and 24 h 8 . A recent randomized trial also showed that patients with active bleeding or visible vessels and those who required endoscopic treatment were more frequently found in the urgent endoscopy (within 6 h after gastroenterology consultation) group than in the early endoscopy (within 24 h) group 7 . In the present study, we found that the proportion of patients who underwent endoscopic treatment was higher in the urgent group than in the elective group, and the need for endoscopic treatment was not associated with reduced recurrent bleeding. These results suggest that initial hemodynamic resuscitation and proton pump inhibitor therapy are more beneficial than earlier endoscopy within 6 h for improving clinical outcomes 20 .
Several factors have been reported to be associated with mortality in patients with UGIB, including comorbidities, vital signs, and failed endoscopic treatment 10,[21][22][23][24] . In the present study, the clinical factors associated with mortality were age and RBC transfusion. A previous observational study that investigated 186 patients in Korea also showed that age (≥ 65 years) was an independent predictive factor for mortality 21 . Worse outcomes in the elderly might be attributable to a tendency to having multiple comorbidities and the susceptibility to physiological changes of elderly patients. There were also discrepancies regarding the impact of RBC transfusion on mortality in previous studies 21,22,24 . This may be due to the demographics of the patients and the study design, and additional studies considering various demographic characteristics are necessary to identify clinical factors affecting outcomes of patients with acute UGIB.
There are several limitations of this study. First, this is a retrospective study. Although it is desirable to design randomized control trial, performing such a study would be difficult and may have ethical issue since a deliberate delay in endoscopic procedures may lead to fatal outcomes. Future prospective studies would be beneficial to confirm the limited role of urgent endoscopy. Second, because this is a single-center study, the results may not represent the overall national demographics. Indeed, in our center, most endoscopic examinations were performed within 6 h, and this clinical practice pattern may influence the lack of differences in mortality rates between groups. Since the resources required to perform emergency procedures are limited, the results of this study cannot be readily generalized. Another possible limitation of this study is that our study population consists of heterogeneous patients, including those with gastroesophageal variceal bleeding. However, in clinical practice, it is often challenging to discriminate patients with variceal bleeding from non-variceal bleeding at the time of presentation, even after a thorough history taking. Since the decision whether to perform endoscopy immediately or not is made based on the information obtained at presentation, it is reasonable to include all patients with UGIB, including variceal bleeding, to determine the role of urgent endoscopy in a real-world setting.
In conclusion, urgent endoscopy within 6 h was not significantly associated with lower mortality and rebleeding rates. We also found that age and RBC transfusion were the factors related to mortality. Based on these findings, we suggest performing elective endoscopy after hemodynamic stabilization in patients with suspected acute UGIB. www.nature.com/scientificreports/