Nonvariceal upper gastrointestinal bleeding (NVUGIB) is the most prevalent gastrointestinal emergency in the hospital environment.1 NVUGIB management has improved over the past few decades, thanks to the introduction of faster and more effective diagnostic and therapeutic elements, such as the administration of proton pump inhibitors,2 the advent of esophagogastroduodenoscopy (EGD) with different types of endoscopic hemostatic treatments,3 and the establishment of eradication treatment of Helicobacter pylori – a bacterium closely linked to the development of peptic ulcer.4

Despite those advances, the morbidity and mortality due to NVUGIB is not insignificant, even in advanced specialty hospitals.5,6 Therefore, efforts made to identify diagnostic and therapeutic elements that can reduce the morbidity and mortality in those patients are currently extremely relevant.

EGD performed on the patient with NVUGIB after his/her hospital admission potentially enables the identification and consequent treatment of lesions that significantly increase the risk for recurrent bleeding. Said intervention has been shown to reduce the risk for rebleeding and the need for surgical treatment.7 An international consensus by the Asia-Pacific working group recommends the performance of EGD in those patients, within the first 24 h of their hospital admission.8 Nevertheless, current evidence fails to clearly show whether the performance of that endoscopic procedure within that suggested time interval is clinically beneficial. Some controlled clinical trials have failed to demonstrate a reduction in hospital mortality or short-term mortality in patients with NVUGIB that underwent urgent EGD within 6 or 12 h from emergency room admission.9 Other cohort studies have reported contradictory results: some fail to identify a modification in the risk for death or rebleeding, according to the time the EGD was carried out,10 whereas others describe an increased risk for death if the EGD is performed after 6 h from patient admission.11

A clinical trial was recently published, describing the time to EGD in subjects admitted for upper gastrointestinal bleeding, that failed to show a reduction in the mortality rate in patients that underwent EGD in fewer than 6 h from hospital admission.12 The latest systematic review was published in 2001.13 Given the conflicting evidence in the previously published studies, as well as the lack of a recent systematic review, we decided to conduct a systematic review, with a meta-analysis, to determine whether urgent endoscopic intervention could reduce the risk for mortality and rebleeding, compared with elective EGD, in subjects with NVUGIB.

Our aim was to describe the risk for mortality, rebleeding, and red blood cell transfusion, in relation to urgent EGD in cases of NVUGIB.

The protocol of our systematic review was registered with the International Prospective Register of Systematic Reviews (PROSPERO http://www.crd.york.ac.uk/prospero). We followed the recommendations of how to report a systematic review, described in the Cochrane Handbook, MOOSE proposal, and PRISMA.

According to our findings, there was no significant difference in relation to short-term mortality in subjects hospitalized due to NVUGIB, whether or not they underwent urgent EGD. However, there was an overall difference in the risk for rebleeding, even though the performance of urgent EGD significantly reduced the need for red blood cell transfusion.

To clearly explain our findings in relation to mortality, the differences in the methodologies of the studies analyzed must be emphasized. Even though the interval of time utilized to define urgent EGD varied among the randomized clinical trials, the comparison carried out in the present study was as free as possible from possible confounders and showed no significant difference in mortality between the subjects that underwent urgent EGD and those that underwent elective EGD. In contrast, in our separate sensitivity analysis of the cohort studies, the risk for mortality was increased in the subjects exposed to urgent EGD. Most likely, that discrepancy is due to the fact that certain residual confounding factors cannot be avoided in cohort studies. In fact, patients that undergo urgent EGD clinically present with some kind of aggravating condition. Thus, their underlying condition per se, not the performance of urgent EGD, could increase the risk for mortality. Indeed, there are clinical variables that increase the risk of mortality during the hospitalization of patients with NVUGIB that were not taken into account in the majority of the studies evaluated in the present systematic review. In a study by González-González,33 there was a 10.2% mortality rate in patients with NVUGIB, but only 3.1% of those cases were directly associated with bleeding. Moreover, the number of comorbidities and albuminemia upon admission (among other variables) were significantly associated with mortality. Such a patient is already at higher risk for mortality, when he/she undergoes EGD, and so the urgency of the EGD cannot be determined as the factor having a detrimental effect on that patient’s outcome. Even though it is difficult to calculate the protective effect of EGD in the context of NVUGIB, population cohort studies, such as the one by Kohn et al.,34 showed that access to EGD and NVUGIB managed by specialists were variables that significantly reduced the risk for mortality in patients hospitalized due to NVUGIB.

With respect to rebleeding, something similar to the mortality results occurred. When exclusively considering the randomized clinical trials, there was no difference in the risk for rebleeding between the subjects exposed to urgent EGD and those exposed to elective EGD. Nevertheless, in the cohort studies, urgent EGD was significantly associated with a higher risk of rebleeding. Numerous clinical variables could contribute to the risk for rebleeding in the context of NVUGIB. Travis et al.35 found that active bleeding at the time of EGD, as well as the type of hemostatic treatment employed or the use of low-molecular-weight heparin, among other variables, increased the risk for rebleeding in patients with NVUGIB, during their hospitalization. Those variables were not contemplated in the majority of the cohort studies, and so their unequal distribution could partially explain the finding of the increased risk in one group versus the other. Likewise, the risk for rebleeding in the case of peptic ulcer lesions is associated with the presence or absence of active bleeding, at the time of examination. The bleeding activity classification described by Forrest et al.36 is the most frequently used, but it was largely absent in the studies included in our review, which could explain the conflicting results observed in the risk for rebleeding in the cohort studies.

Even though fewer studies described the outcome of number of transfused red blood cell units, we found that urgent EGD was significantly associated with a lower number of units per patient. In recent years, controlled randomized clinical trials, such as the study by Villanueva et al.,37 have shown that a conservative strategy of red blood cell transfusion has a positive impact on the progression of patients hospitalized due to gastrointestinal bleeding, in general, making the number of transfused red blood cell units a relevant outcome that should be observed in patients with NVUGIB. A shorter time interval from patient admission to the performance of EGD (and eventual hemostatic therapy) should stop gastrointestinal bleeding early, and in turn, reduce total blood loss, which could explain the association observed in our meta-analysis. The length of hospital stay is another similar outcome, but given the relative scarcity of studies that evaluate it, we did not include it in our analysis. As demonstrated by Lee et al.11 hospital stay was significantly shorter in subjects that had urgent EGD, compared with those that had elective EGD. If, as shown in our study, the 2 procedures have a similar incidence of mortality, then urgent EGD becomes relevant, with respect to reducing the use of hospital resources that significantly increase the cost of healthcare in patients with NVUGIB.

The present systematic review has limitations that must be addressed. First, there are relatively few controlled clinical trials in our review, and more such studies are needed to reach the best quality conclusions possible. In addition, the heterogeneity in the definitions of urgent EGD is striking, given that the performance of EGD ranged from 2 h to 24 h after hospital admission. Even though the separate analyses of the cohort studies and clinical trials resulted in a more homogeneous definition, it reflects the relative lack of consensus in determining the cutoff interval for establishing a specific description. No distinction was made in any of the studies included in the review, regarding the manner in which NVUGIB appeared (i.e., the presence of melena, hematochezia, or hematemesis) and its impact on the outcomes, nor were the risk factors related to NVUGIB described, such as finding an actively bleeding lesion at the time of EGD. There were not enough details in the studies analyzed to carry out a meta-regression analysis to determine the effect of variables, such as age or hemoglobin level upon admission, on the outcomes studied.

In conclusion, the performance of urgent EGD in subjects hospitalized due to NVUGIB did not have a significant impact on all-cause mortality, in the short term. The clinical trials failed to demonstrate a significant difference in the risk for rebleeding, but when the cohort studies were added, the incidence of rebleeding was higher in the subjects that underwent urgent EGD. More evidence of higher quality is needed to determine the true impact of urgent EGD on the risk for those relevant clinical variables.

No specific grants were received from public sector agencies, the business sector, or non-profit organizations in relation to this study.

The authors declare that there is no conflict of interest.