Introduction

Nonvariceal upper gastrointestinal bleeding (NVUGIB) is a well-known major healthcare problem with significant mortality and morbidity [1]. Nevertheless, its incidence and mortality have been decreasing in the last years, largely attributable to the widely use of proton pump inhibitors (PPI) and investigation and eradication of Helicobacter pylori infection, which changed the natural history of peptic ulcer disease (PUD), the most common cause of NVUGIB [2].

Anticoagulants are a well-recognized risk factor for upper gastrointestinal bleeding (UGIB), including direct oral anticoagulants (DOAC) and vitamin K antagonists (VKA) [3, 4]. The use of acid suppressants [PPI and histamine 2 receptor antagonists (H2RA)] significantly reduce the risk of UGIB in patients taking anticoagulants [5]. DOAC are currently the first-line option for the prevention of stroke and systemic embolism in non-valvular atrial fibrillation (AF) and for the treatment and secondary prevention of venous thromboembolism (VTE) [6, 7]. Although its use is associated with a significantly lower all-cause mortality and intracranial bleeding compared with VKA, the risk of major and clinically relevant non-major UGIB seems to be not different, irrespective of indication [8, 9].

Upper gastrointestinal endoscopy remains the modality of choice for identifying the etiology of UGIB and achieve therapeutic hemostasis. According to the recently developed European Society of Gastrointestinal Endoscopy (ESGE) guideline, following hemodynamic resuscitation, endoscopy should be performed within 24 h from the time of patient presentation because it is associated with better outcomes [10].

Patients taking anticoagulants who develop NVUGIB represent a clinical challenge since different pharmacokinetic and pharmacodynamic profiles of the anticoagulants must be considered, and this issue needs to be addressed both prior to and following endoscopy [11].

Nevertheless, few data exist specifically addressing the optimal timing of endoscopy in patients with NVUGIB receiving anticoagulants [10, 12]. Therefore, the objective of this study was to analyze if timing of endoscopy, in anticoagulants users admitted with NVUGIB, may influence outcomes, such as endoscopic or surgical treatment, length of hospital stay, intermediate/intensive care unit admission, recurrent bleeding, and 30-day mortality.

Material and Methods

Study Design

This was a retrospective cohort study that consecutively included all patients admitted to the Emergency Department (ED) for UGIB and underwent endoscopy in a University-Affiliated Hospital between January 2011 and June 2020. The present study was approved by the local Ethics Committee for Health.

Patient Selection and Hospital Management Protocol

All adult patients (≥ 18 years old) with suspicion of having UGIB at presentation and submitted to endoscopy were included, namely those who presented with hematemesis, melena, hematochezia with hemodynamic instability, or detection of hematic content in a nasogastric tube. In the situation that more than one UEG was performed during the same admission, only the findings of the index endoscopy were included. In addition, patients with one or more of the following items were excluded from the study: not receiving anticoagulants; gastrointestinal variceal bleeding source; documentation of a bleeding cause in the lower or mid gastrointestinal tract; UGIB in an inpatient setting; and ED presentation due to recurrent bleeding from a previous episode of acute UGIB.

Only NVUGIB patients, including those with a definite diagnosis (namely peptic ulcer disease, mucosal erosive disease of the esophagus/stomach, malignancy, Mallory–Weiss syndrome, Dieulafoy’s lesion, angiodysplasias) or without an identifiable cause in endoscopy and taking anticoagulants were considered in the final sample. Patients were then divided into two groups according to the timing of endoscopy: the early endoscopy group (≤ 24 h after presentation) and the delayed endoscopy group (> 24 h after presentation).

According to the recommendations of our department and to the emergency department protocol of our hospital, all patients started intravenous PPI as soon as UGIB was suspected. Following endoscopy, patients with high-risk endoscopic stigmata according to Forrest classification continued on intravenous infusion of PPI for 72 h [13]. Moreover, Forrest IIb stigmata was treated with clot removal followed by endoscopic hemostasis if underlying high-risk stigmata was found [13]. Otherwise, patients received oral PPI, if needed, as indicated by endoscopic findings. Immediate assessment of hemodynamic status was performed with prompt intravascular volume replacement using crystalloid fluids if hemodynamic instability existed. In hemodynamically stable patients with NVUGIB and no history of cardiovascular disease, a restrictive red blood cell transfusion strategy with a hemoglobin threshold of ≤ 7 g/dL prompted red blood cell transfusion. A post-transfusion target hemoglobin concentration of 7–9 g/dL was considered. In patients with a history of acute or chronic cardiovascular disease, a more liberal red blood cell transfusion strategy was followed with a hemoglobin threshold of ≤ 8 g/dL leading to transfusion. A post-transfusion target hemoglobin concentration of ≥ 10 g/dL was considered. Intravenous erythromycin was administered before endoscopy in patients with clinically severe or ongoing active UGIB [10]. The antithrombotic management policy was carried out in accordance with the protocol of the hematology department of our hospital, ensuring that the patient was optimized for the endoscopy without unnecessarily delaying the procedure. In patients taking low-dose aspirin as monotherapy for primary cardiovascular prophylaxis, aspirin was temporarily interrupted and restarted after careful re-evaluation of its clinical indication. For patients taking low-dose aspirin as monotherapy for secondary cardiovascular prophylaxis, aspirin was not interrupted. Patients on dual antiplatelet therapy for secondary cardiovascular prophylaxis had aspirin uninterrupted, while the second antiplatelet agent was interrupted and restarted within 5 days. Patients taking VKA or enoxaparin had these anticoagulants withheld. In patients with hemodynamic instability, low-dose vitamin K was supplemented with intravenous prothrombin complex concentrate (PCC) or fresh frozen plasma if PCC was not available. In patients taking DOAC, the anticoagulant was withheld. For patients with severe ongoing bleeding, the use of a DOAC reversal agent or intravenous PCC was considered.

Data Collection

Several data were retrospectively collected from electronic medical records.

On presentation, demographic data (age and sex), clinical presentation (hematemesis, melena, hematochezia), Glasgow Coma Scale (GCS) score, history of syncope, vital signs [heart rate (HR), systolic and diastolic blood pressure (SBP and DBP, respectively)], laboratory values [i.e., hemoglobin, platelets, creatinine, urea, albumin, and international normalized ratio (INR)], as well as comorbidities (namely, chronic liver disease (CLD), congestive heart failure (CHF), chronic kidney disease (CKD) and disseminated neoplasia] were retrieved. Data were also obtained regarding the use of certain drugs that could be related to bleeding, such as acetylsalicylic acid (ASA) and other antiplatelet agents, nonsteroidal anti-inflammatory drugs (NSAID), anticoagulants (including DOAC, VKA and enoxaparin), steroids, and selective serotonin reuptake inhibitors (SSRI), as well as could drugs that could decrease the probability of bleeding (PPI or H2RA). Date and time of presentation and endoscopy, endoscopic findings, namely etiology of bleeding and stigmata of recent bleeding (for PUD, Forrest classification), and length of hospital stay (days) were recorded [13]. Forrest Ia, Ib, and IIa were considered high-risk stigmata according to Forrest Classification [10].

Different clinical outcomes were analyzed, namely the need for endoscopic or surgical treatment, length of hospital stay, intermediate/intensive care unit admission, recurrent bleeding, and 30-day mortality. Surgery was performed in patients with persistent bleeding refractory to all modalities of endoscopic hemostasis.

Based on these data, many clinical and endoscopic-based scores were calculated for each patient, namely the pre- and the post-endoscopic Rockall score (RS), the Glasgow-Blatchford (GBS) score, and the AIMS65 [14].

Definition of Variables, Comorbidities, and Other Covariates

Timing of endoscopy was defined as the time elapsed between presentation in the ED and beginning of endoscopy. In patients with PUD as the source of NVUGIB, Forrest classification was assessed [13].

CLD was considered in patients with a known history, or clinical, laboratory, and imaging evidence of liver disease [15], while CHF was defined as a known history or clinical evidence and echocardiography of CHF [16]. CKD was defined in patients with an estimated glomerular filtration rate < 60 mL/min/1.73 m2 for ≥ 3 months calculated using the Modification of Diet in Renal Disease Study equation [17]. Disseminated neoplasm was considered in patients with a malignant tumor in whom metastasis was identified. Therapeutic endoscopic intervention included one or more of the following hemostatic strategies: use of hemoclips, endoscopic elastic ligation, argon plasma coagulation, bipolar electrocoagulation, sclerotherapy, with or without a previous injection of adrenaline. Mortality was defined as death for any cause (overall mortality) occurring within 30 days of hospital admission. The need for surgery was defined as the necessity of undergoing laparotomy/laparoscopic intervention if persistent bleeding occurred, defined as ongoing active bleeding (spurting, arterial pulsatile bleeding, or oozing) that was present at the end of index endoscopy and refractory to standard hemostasis modalities [10]. Recurrent bleeding was defined as the observation of an overt bleeding event (hematemesis or melena) or a decrease of at least 2 g/dL in hemoglobin levels after initial treatment and stabilization [18]. Pre- and post-endoscopic RS, GBS, and AIMS65 scores were calculated according to their description in the original articles [14, 19, 20].

Statistical Analysis

Statistical Package for Social Sciences program version 26 (IBM Corporation, Armonk, NY) was used. A univariable analysis was performed to assess differences between the two groups (early and delayed endoscopy). Categorical variables were described using absolute frequencies and percentages. χ2 Test or Fisher’s exact test was used to compare categorical variables. Depending on the normality tests, continuous variables were expressed as mean ± standard deviation (SD) or median and interquartile range (IQR). Means were compared between the distinct groups using independent samples t-test. When applicable, non-parametric tests were performed. In addition, multivariable analysis using a logistic or linear regression analysis was performed to assess if timing of endoscopy had any influence on the outcomes that in univariable analysis were significantly different between early and delayed groups. Statistical significance was defined as P < 0.05.

Results

From a total of 759 patients admitted for UGIB and submitted to endoscopy, 123 had UGIB secondary to bleeding varices and 636 patients had NVUGIB. Among patients with NVUGIB, 138 (21.7%) were taking anticoagulants. Vitamin K antagonists (n = 88, 63.8%) were the most frequently used anticoagulants, followed by enoxaparin (n = 27, 19.6%) and DOAC (n = 23, 16.7%).

In patients on anticoagulants, the overall mean time to endoscopy was 23.8 ± 19.3 h. Eighty-two (59.4%) patients were submitted to early endoscopy (median time to endoscopy: 9.0 [IQR 11] h) and 56 patients (40.6%) to delayed endoscopy (median time to endoscopy: 44.0 [IQR 27] h).

Table  1 presents the demographic and clinical characteristics of both study groups.

Table 1 Baseline demographic, clinical, laboratory, and endoscopic findings according to the time from presentation to endoscopy in patients with nonvariceal upper gastrointestinal bleeding that were taking anticoagulants
Full size table

Outcomes

Table 2 displays the outcome variables according to the time from presentation to endoscopy. Of the outcomes considered, only endoscopic treatment [31 (37.8%) vs. 11 (19.6%), P = 0.025] and intermediate/intensive care unit admission [42 (51.2%) vs. 15 (26.7%), P = 0.005] were significantly different between early and delayed groups, respectively.

Table 2 Outcome variables according to the time from presentation to endoscopy in patients with nonvariceal upper gastrointestinal bleeding and taking anticoagulants
Full size table

After adjusting for confounders, on multivariable analysis, the timing of endoscopy (early/delayed) was still identified as an independent predictive factor for endoscopic treatment, length of hospital stay, and intermediate/intensive care unit admission. Patients who underwent early endoscopy received endoscopic therapy more frequently (OR 2.4; 95% CI 1.1–5.4; P = 0.034), had shorter length of hospital stay [7 (IQR 6) vs. 9 (IQR 7) days, P = 0.042] and higher rate of intermediate/intensive care unit admission (OR 2.7; 95% CI 1.3– 5.9; P = 0.01] than patients having delayed endoscopy (Tables 3, 4, 5). In addition, the GCS at admission was found to be an independent predictive factor for intermediate/intensive care unit admission. Patients with lower GCS score at presentation were more likely to be admitted to the intermediate/intensive care unit (OR 0.6; 95% CI 0.37–0.96; P = 0.032).

Table 3 Predictive factors for endoscopic treatment in patients with nonvariceal upper gastrointestinal bleeding and taking anticoagulants, according to univariable and multivariable analyses
Full size table
Table 4 Results of a multiple regression analysis with length of hospital stay as the outcome variable
Full size table
Table 5 Predictive factors for intermediate/intensive care unit admission in patients with nonvariceal upper gastrointestinal bleeding and taking anticoagulants, according to univariable and multivariable analyses
Full size table

Discussion

To the best of our knowledge, this is the first study to specifically investigate the relationship between the timing of endoscopy in patients with NVUGIB receiving anticoagulants and critical clinical outcomes. We found that patients on anticoagulants who underwent early (≤ 24 h) endoscopy, had a higher rate of endoscopic treatment, shorter length of hospital stay, and higher intermediate/intensive care unit admission. However, other outcomes, such as the need for surgical intervention, recurrent bleeding, and 30-day mortality, did not significantly differ between the two groups.

The optimal timing to perform endoscopy in patients with NVUGIB has been a subject of intense debate and the focus of many studies. In 2021, the ESGE guideline addressing the management of NVUGIB patients recommended that, following hemodynamic resuscitation, early (≤ 24 h from the time of patient presentation) endoscopy should be performed (strong recommendation, high-quality evidence), being associated with lower in-hospital mortality, shorter length of stay, and lower total hospital costs. In addition, performing urgent endoscopy (≤ 12 h) is not recommended because, when compared to early endoscopy, patient outcomes are not consistently improved [10]. While some studies concluded that urgent endoscopy was an independent predictor of lower mortality in high-risk patients [21, 22], others have shown that urgent endoscopy was a predictor of worse patient outcomes [23, 24], or that clinical outcomes were not significantly different between urgent and early endoscopy [25, 26]. On the other hand, the American College of Gastroenterology guideline, also published in 2021, is more conservative in their recommendations, only “suggesting” that patients admitted in hospital for NVUGIB should undergo endoscopy within 24 h of presentation, being stated as a conditional recommendation supported with very low-quality evidence [27].

Unfortunately, this controversy becomes more evident regarding the optimal timing of endoscopy in anticoagulant users admitted with NVUGIB [10, 12].

No systematic reviews, randomized trials, or observational studies specifically addressing the impact of the timing of endoscopy on clinically relevant outcomes in patients receiving anticoagulants have been published [10, 12]. This is particularly relevant because, although the overall incidence of NVUGIB has decreased in recent years and is stabilizing globally, some risk factors for gastrointestinal bleeding are becoming more common. These factors include the aging of the population, resulting in an increasing prevalence of cardiovascular diseases, and the progressive broadening of indications for anticoagulants use in the clinical practice [11, 28]. In addition, it is known that the use of anticoagulants in patients for NVUGIB is associated with increased admitted second-look endoscopy and recurrent bleeding rates compared to patients who were not using anticoagulants [29]. Therefore, optimizing the management of these patients is of paramount importance.

In our study, we found that a significant proportion (21.7%) of the patients admitted for UGIB and submitted to endoscopy were taking anticoagulants. Interestingly, Vitamin K antagonists were the most commonly used anticoagulants (63.8%), followed by enoxaparin (19.6%), and DOAC (16.7%). Although DOACs have been increasingly prescribed worldwide in recent years, they were not the most used anticoagulants in our sample. This might be due to a substantial proportion of the patients experiencing the bleeding event in the first half of the 2010s, reflecting the clinical practice at that time.

Endoscopy was performed within 24 h after presentation in most patients (59.4%). We believe that endoscopy was not performed within 24 h of presentation in a substantial proportion of the patients (40.6%) because, in our unit, endoscopy is only available between 8 AM and 8 PM on working days which is the Gastroenterology Department’s schedule. Therefore, it is probable that some patients admitted with UGIB during the weekend and who were clinically stabilized after initial management in the ER might not have been transferred to another center. Consequently, they were submitted to endoscopy only in the morning of the following working day, possibly explaining this delay.

Regarding clinical outcomes, only endoscopic treatment, duration of hospital stay, and intermediate/intensive care unit admission were significantly different between the early and delayed groups. In fact, even after adjusting for confounders, the timing of endoscopy (early/delayed) was still identified as an independent predictive factor for all outcomes. Although studies that evaluated the influence of timing of endoscopy in NVUGIB on clinical outcomes did not specifically address the impact of anticoagulants use, their results are in line with our findings, showing higher rates of endoscopic treatment and shorter hospital stay in the early endoscopy group [30, 31]. On the other hand, Freitas et al. [32] found that performing endoscopy within 24 h of presentation was associated with a higher need of endoscopic treatment only in low-risk (GBS < 12) patients but not in high-risk patients (GBS ≥ 12). While we observed higher rates of endoscopic treatment in the early endoscopy group, this did not significantly influence the potential benefits in terms of need of subsequent surgical treatment, intermediate/intensive care unit admission, recurrent bleeding and 30-day mortality, compared to patients who underwent early endoscopy and did not receive endoscopic therapy.

In regard to the rate of intermediate/intensive care unit admission, contrary to our results, most studies did not find a different rate of intermediate/intensive care unit admission between early and delayed endoscopy groups, even when considering high and low-risk GBS groups and different cut-off values of timing of endoscopy from presentation [22, 23, 25, 31,32,33]. We speculated that this finding could be potentially explained by the longer time to medical optimization that patients in the delayed group had when compared with the early group. In line with our findings, Guo et al. [34] observed that urgent (< 6 h from presentation) vs. early (6–24 h from presentation) endoscopy was associated with higher ICU admission rates. Finally, we found that GCS was an independent predictive factor for intermediate/intensive care unit admission. Therefore, patients with lower GCS score at presentation were more likely to be admitted in intermediate/intensive care unit. This is not surprising as altered mental status is one of the variables used in the AIMS65 score, a simple risk stratification score for upper gastrointestinal bleeding capable of predicting in-hospital mortality and the need for ICU admission [35].

Our study presents some limitations. This was a single-center study, with retrospective design with its possible inherent bias. Although we relied on medical records to measure the data and outcomes of interest, missing data were minimal. Even though we chose to use the time from presentation to endoscopy, we did not record the time from presentation-to-endoscopy referral, nor the time from endoscopy referral to endoscopy, both of which could have influenced the overall presentation-to-endoscopy time. Despite these limitations, we believe our study is relevant because we considered potential confounders in the analysis, and it includes a nonnegligible sample of patients.

Despite current guidelines recommending performing endoscopy within the first 24 h after initial stabilization and not using anticoagulants as a determinant of the timing of endoscopy in patients with acute UGIB, the literature regarding this topic in this specific population is scarce. According to our results, outcomes such as need for surgery, recurrent bleeding, and 30-day mortality did not differ between anticoagulated patients submitted to early or delayed endoscopy. However, as those submitted to early endoscopy had shorter length of hospital stay, performing early endoscopy might be considered the mainstay strategy in patients with acute UGIB taking anticoagulants. However, further data are needed to clarify the impact and the optimal timing of endoscopy in patients with NVUGIB using anticoagulants, in different-risk patients, as evidence remains scarce.