Introduction

Patients hospitalized in the intensive care unit (ICU) have a higher risk of upper gastrointestinal bleeding (UGIB) especially due to stress ulcers. ^1,2 Its appearance is associated with adverse outcomes including 2 to 4 times higher mortality rates and 4 to 8 day longer ICU stays. ^3,4 The incidence of gastrointestinal bleeding in ICUs ranges from 0.17% to 7.0%. ^3,5 An upper gastrointestinal endoscopy (esophagogastroduodenoscopy or EGD) is the test of choice for patients with UGIB, including those in the ICU. EGD can be both diagnostic, macroscopic examination of lesions and taking of biopsies, and therapeutic, various methods of achieving hemostasis. ^5,6 Its performance is well demonstrated in patients with UGIB. ^7-10 However, EGDs are more controversial when they are performed in patients admitted to the ICU for other reasons. In addition, the discovery of gastritis, esophagitis, or other GI problems may not require any endoscopic or pharmacological treatment. ^5,7,11,12

Among the causes of UGIB in ICU patients are esophageal varices, gastric varices, esophagitis, ulcers and stress-induced gastritis. ¹³ It has been estimated that up to 90% of critically ill patients may present gastroduodenal mucosal damage after three days in an ICU and that damage may progress to ulcers and cause bleeding,. Bleeding from erosive gastritis is another potential danger. ¹⁴ Nevertheless, UGIB is clinically important bleeding in only 2% to 3% of these cases, and stress ulcers are identified endoscopically as the source of bleeding in less than 50% of these patients. ^4,15,16 These data have raised a discussion about the real need for EGD given its costs and given the real impact of indiscriminate use on this type of patients, especially since only a small percentage progresses towards manifest and clinically important gastrointestinal bleeding. ¹⁷

Other less common conditions responsible for UGIB are Mallory-Weiss syndrome and vascular lesions. ¹³ The principal upper gastrointestinal risk factors in an ICU include mechanical ventilation for more than 48 hours, active coagulopathy, liver disease, and kidney disease. ^5,18,19 Other risk factors are shock, liver failure, kidney failure, sepsis, multiple traumas, burns of more than 35% of the body surface, organ transplantation, skull or spinal cord trauma, history of previous ulcerative disease and hypoalbuminemia. ^20-24

We found no publications on the prevalence of bleeding lesions of the upper digestive tract in ICU patients in Colombia, nor did we find literature on the frequency of endoscopic hemostasis in these patients. Taking into account the limited information available, we decided to perform this study in the Gastroenterology Unit of the Clínica Fundadores in Bogotá by identifying ICU patients who had developed UGIB and had undergone EGD.

Materials and methods

This is a cross-sectional study based on EGD findings from ICU patients at the Clínica Fundadores who underwent EGD because of UGIB. Adult patients over 18 years of age who were treated during the period from January 1, 2003 to December 31, 2017 were included.

Results

During the study period, 246 EGDs were performed on ICU patients. Ninety-two of these were excluded:

We included 154 patients who underwent EGDs (Figure 1) including 99 men (64.29%) and 55 women (35.71%). Their median age was 68, and half of them were between 59 and 76 years of age.

Figure 1 Diagram of patients included in the study. 

EGD findings in the esophagus, stomach and duodenum are shown in Figures 2, 3 and 4.

Figure 2 Esophageal findings of EGDs. 

Figure 3 Endoscopic findings in the stomach. 

Figure 4 Endoscopic findings in the duodenum. 

Twenty-four patients (15.58%) were treated endoscopically, but it was not necessary in the remaining 130 patients (84.4%) (Figure 5). Among the diagnoses recorded in the medical records for esophageal varices, Child Pugh C cirrhosis with encephalopathy, mostly secondary to diabetes mellitus and alcohol was the most common. Other findings were presented with ambivalent information due to the large number of pathologies and comorbidities of these critical patients. They included sepsis, shock, heart failure, renal failure, respiratory failure, and coagulopathy.

Figure 5 Endoscopic treatment 

Modalities of therapeutic endoscopy used are shown in Table 1. Hemostasis of bleeding was achieved in all patients.

The prevalences of the primary pathologies responsible for ICU admissions are shown in Table 2.

Table 1 Modalities of therapeutic endoscopic used to treat UGIB 

Table 2 Prevalence of pathologies responsible for ICU admissions 

The prevalence of mechanical ventilation was 20% (n = 30), and the prevalence of prophylaxis with proton pump inhibitors (PPI) was 32% (n = 49). The distribution of lesions by age group is shown in Table 3.

Table 3 Prevalence of endoscopic findings by age group. 

Discussion

Fifteen percent of the ICU patients with UGIB in this study required endoscopic treatment. The need for therapeutic EGD is similar to a study of 66 ICU patients from the University of Pennsylvania. That study found that 15% of those patients merited hemostatic endoscopic therapy. ²⁵ Another study by Kim et al. of 66 patients at the University of Seoul found that endoscopic management of bleeding had been needed in 19% of their patients. ²⁶

EGDs are frequently requested for critically ill patients with gastrointestinal bleeding for both diagnostic and therapeutic purposes. However, not all mucosal lesions identified by this means require endoscopic treatment which increases the costs of patient care and the likelihood of complications. ^7,27 In general, these patients die due to the severity of the underlying medical condition or due to multiorgan dysfunction rather than due to bleeding. ²⁸

Lee et al. conducted a prospective study of 105 patients with gastrointestinal bleeding who were in a critical care unit and found that the prevalence of erosive disease was 21.9%. ¹¹ In our study, erosive gastritis had a prevalence of 47.4% (n = 73), but none of these lesions required endoscopic treatment. The clinical relevance of these lesions is controversial, since only a small percentage progress towards overt and clinically important gastrointestinal bleeding. ^17,22

The most frequent esophageal finding in our study was erosive esophagitis which was found in 59% of the patients, a higher rate than found in previously published studies. ^25,29-31 This finding demonstrates once again that the majority of EGDs performed in ICU patients with UGIB will find lesions that do not merit therapeutic endoscopy. Consequently, there will be no impact on the treatment of these patients. ⁷ Mallory-Weiss syndrome was found in 8 patients (4.55%), none of whom required endoscopic therapy. Mallory-Weiss syndrome is related to a sudden increase in intragastric or intra-abdominal pressure which is transmitted to the esophagogastric junction. ³² Risk factors include severe nausea, vomiting, closed abdominal hypo-trauma, cardiopulmonary resuscitation, coughing, shouting, barotrauma and seizures. In our series, this syndrome occurred less frequently than reported in other publications. The multiple predisposing factors could explain this difference. ^10,33 However, it would be important to minimize related factors since Mallory-Weiss syndrome can deepen and produce transmural rupture which leads to Boeerhäve syndrome. ³⁴.

Gastric ulcers were found in 18.18% (n = 28) of the patients in our study. In 2005, Skok et al. conducted a prospective cohort study of 486 patients in Slovenia which found gastric ulcers in 84 patients (17.3%). ³⁵ However, only 5.8% (9 patients) in our study required endoscopic treatment for gastric ulcers. A retrospective observational study of 88 French patients between 2007 and 2012 evaluated the clinical impact of EGD in critically ill patients with suspected bleeding. It found that only 3.5% of patients required endoscopic management for gastric ulcers. ⁷ These results are important, since EGDs are relatively expensive and are not risk-free. Both costs and risks increase as the procedure becomes generalized. ³⁵

Our study found that non-varicose causes occurred twice as frequently as varicose causes (11.03% vs. 4.55%) which is similar to other studies. ³⁶ There were 17 patients with duodenal ulcers (11.04%) which is half of what has been reported in the literature. ²⁹ We do not know the reason for this discrepancy, but it could be related to the prevalence of Helicobacter pylori, the frequency of use of prophylactic PPIs and the type of critical pathology of patients, especially mechanical ventilation, active coagulopathies, liver disease and renal disease. In our study, many patients had these predisposing pathologies, and 32% of the patients used prophylactic PPIs. Prophylactic PPIs have been shown to decrease the rate of clinically significant bleeding below the rate of patients taking placebos (2.5% vs. 4.2%). PPIs are recommended in these circumstances. ^2,37

Many doctors are still afraid to use PPIs prophylactically because of the theoretical risks of possible adverse effects such as pneumonia, myocardial ischemia, and C. difficile. ¹⁵ Nevertheless, the evidence that supports these fears is very weak, and so far only association and non-causality have been established. ³⁸ In our study, the rate of erosive duodenitis was higher than that reported in the literature (21.43% vs. 6%), but we identified no active bleeding related to this pathology in any of our patients. ^29,34

This study’s limitations include its retrospective nature and changing diagnoses due to multiple concomitant pathologies. Also, it was not easy to determine the risk factors that predispose to UGIB which would merit endoscopic treatment and thus avoid the 85% of EGDs which are unnecessary.

In conclusion, 15% of our ICU patients with UGIB needed endoscopic therapy. Prospective studies, preferably multicenter, are needed to identify risk factors that can predict the need for therapeutic EGDs in patients with UGIB. Patients who do not have these predictors should be treated empirically with PPIs which will avoid unnecessary expenses of diagnostic EGDs. To date, the published literature has identified ICU patients at risk of UGIB but has not identified those whose bleeding may require therapeutic EGD.