Upper gastrointestinal bleeding (UGIB) has an incidence of 500,000 hospital admissions per year1 and a mortality rate from 3 to 14%, depending on etiology. Variceal bleeding and peptic ulcer disease are the most common conditions.2

After adequate resuscitation, upper endoscopy should be indicated within the first 24 h of UGIB.1,3 However, the presence of fresh blood, clots, and/or blood debris may limit adequate visualization of the upper gastrointestinal tract, especially the gastric fundus. The American College of Gastroenterology guidelines for the management of UGIB state that the use of prokinetic drugs, such as erythromycin, may improve endoscopic visualization, reduce the need for a repeat endoscopy, reduce hospital stay, and lower costs.1,4,5

Few prospective clinical studies on the use of metoclopramide as a prokinetic drug to improve visualization during endoscopy in patients with UGIB have been published, and their results are heterogeneous. Our primary aim was to prospectively evaluate the usefulness of metoclopramide at a dose of 20 mg/IV versus placebo in the visualization of the upper gastrointestinal tract in patients with acute UGIB, using the modified Avgerinos score.6 The secondary aim was to evaluate the need for a second endoscopy and the length of hospital stay.

A prospective, longitudinal, triple-blind, randomized controlled trial was carried out from March 1, 2023, to June 1, 2024. We adhered to the CONSORT guidelines for reporting our results. All consecutive patients diagnosed with acute UGIB at the emergency department of the Hospital Universitario “Dr. José Eleuterio González” were evaluated.

The study protocol was submitted to ClinicalTrials.gov with identification number NCT06297954.

Inclusion criteria: patients with acute UGIB (hematemesis, melena, and/or vomiting of coffee-ground-like material) within 24 h prior to hospital admission, age ≥ 18 years, an ASA rating < 3, with or without oral anticoagulant use, and patients who were hemodynamically stable prior to upper endoscopy.

Exclusion criteria: patients who were pregnant, allergic to metoclopramide, had pacemakers, presented with uncontrolled arrhythmia before the endoscopic study, had a previous diagnosis of gastroparesis, or presented with blood glucose > 200 mg/dL upon admission.

Patients were randomly assigned in a 1:1 ratio to receive metoclopramide or placebo. Randomization was performed using computer-generated random allocation. Twenty-five ampoules were assigned to group A (placebo) and another 25 ampoules to group B (20 mg metoclopramide, Carnotprim Lab). The metoclopramide and placebo ampoules were made of the same material and were identical in size and color. Subsequently, the 50 ampoules were numbered and delivered to the research staff. The active and placebo ampoules were administered intravenously, and endoscopy was performed 30−120 min later (Fig. 1).

Figure 1.

Patient randomization protocol.

The contents of the ampoules were never revealed to the treating endoscopist, the staff who administered the medication, the patient, the person in charge of the statistical analysis, or the expert endoscopist (GGJA, with more than 20 years of experience), who reviewed all the videos to compare the previous modified Avgerinos score results. The contents of the ampoules were revealed at the end of the protocol.

A Pentax gastroscope and an EPK-1000 Video Processor were used on all patients. Propofol, midazolam, and fentanyl were administered as sedation. Orotracheal intubation was performed, as necessary, under the instructions of the anesthesiologist. All studies were recorded from inception to completion. The treating endoscopist applied the modified Avgerinos score for gastric and duodenal visibility. The modified Avgerinos score is graded from 0 to 2, depending on the visualized surface area of the gastric fundus, body, and antrum and the duodenal bulb. A score of 0 signifies < 25% visibility of the surface area, 1 signifies 25%–75%, and 2 signifies >75%. A score ≥ 6 was considered adequate visualization, and a score ≤ 5 as inadequate visualization.7 The agreement between the treating endoscopist and the score reported by the expert endoscopist who reviewed the videos was evaluated using the Kappa analysis. The Kappa values were interpreted as follows: 0.0–2.0 no agreement, 2.1–3.9 minimum agreement, 4.0–5.9 poor agreement, 6.0–7.9 moderate agreement, 8.0–8.9 strong agreement, and >9.0 near-perfect agreement.8

Demographic data, laboratory tests, clinical presentation, etiology, time from onset of bleeding to hospital admission, and time from hospital admission to the endoscopic procedure, blood transfusion units, the Glasgow-Blatchford bleeding score, length of hospital stay, need for repeat endoscopy, overall complications, metoclopramide-related complications, such as electrocardiographic or neurological alterations, and mortality, were obtained.

We calculated the sample size based on the reported efficacy of erythromycin in a previous study.7 We hypothesized that metoclopramide would achieve a comparable effect. Using the difference between the assumed metoclopramide efficacy (82%) and the placebo rate (33%), a two-sided test with an alpha of 0.05 and 95% power (beta of 0.05) yielded a required sample size of 24 patients per group.

Patient data were recorded on an Excel spreadsheet and were de-identified by eliminating names, dates of birth, and medical record serial numbers. The SPSS version 25 statistical package was used for the data analysis. Frequencies and percentages were reported as categorical variables, and measures of central tendency and dispersion (mean/median; standard deviation/interquartile range) were reported as quantitative variables for the descriptive statistics. The distribution of the sample was evaluated through the Kolmogorov-Smirnov test.

Categorical variables were compared using the Pearson’s chi-square test or Fisher’s exact-test. The Student’s t-test and/or the Mann-Whitney U test were used, as appropriate, to compare independent groups. Logistic and linear regression models were performed to predict the length of hospital stay and the need for repeat endoscopic procedures in both groups. A p value < 0.05 and a 95% confidence interval were considered statistically significant. Likewise, the median modified Avgerinos scores were compared, using the ANOVA statistical test, in relation to the bleeding etiologies.

The trial was conducted in accordance with the principles of the Declaration of Helsinki and Good Clinical Practice guidelines. The protocol was reviewed and approved by the Ethics Committee of the School of Medicine of the Universidad Autónoma de Nuevo León, with registration number GA22-00011. All participants gave written informed consent to participate in the trial.

A total of 52 patients were analyzed, and 50 were included in the study. Twenty-seven patients (54%) were men, 23 (46%) were women, and the mean patient age was 60 years (range: 48–70). The most frequent bleeding presentation was melena, at 50% (25 patients), and the most frequent etiology was variceal bleeding, at 42% (21 patients), followed by peptic ulcer disease, at 32% (16 patients) (Table 1).

The mean time from hospital admission to endoscopy was 14:37 h (SD ± 5.54); 14:21 h (SD ± 5.35) in group A and 14:54 h (SD ± 5.83) in group B, with no statistical significance (p = 0.836).

Nineteen patients (38%) had esophageal bleeding. This bleeding was due to esophageal varices in 18 (94%) patients, and one patient (6%) had erosive esophagitis. The most frequent bleeding site was the gastric area, with 26 patients (52%), and 16 of them (32%) had gastric ulcers. When classifying the ulcers by re-bleeding risk, we found: Forrest IA in one patient (2%), Forest IB in one patient (2%), Forrest IIA in 8 patients (16%), and Forrest III in 6 patients (12%). Five patients (10%) had erosive gastritis, 3 patients (6%) had Sarin Type I gastric varices, one patient (2%) had Dieulafoy’s lesion, and one patient (2%) had a malignant gastric ulcer. The duodenal area had the fewest lesions, in a total of 5 patients (10%); 4 (8%) with Dieulafoy’s lesions and one (2%) with a malignant ulcer.

The median modified Avgerinos score for group A was 7.00 (range: 5.00−8.00) versus 7.00 (range: 6.50−8.00) for group B (p = 0.282). Regarding bleeding etiology, the median modified Avgerinos score for patients with variceal bleeding was 7.50 (range: 5.00−8.00) in group A versus 7.00 (range: 6.50−8.00) in group B, with no statistical significance (p = 0.705). The patients with peptic ulcer had a median modified Avgerinos score of 7.50 (range: 7.00−8.00) in group A vs 7.50 (range: 6.25−8.00) in group B, with no statistical significance (p = 0.721) (Table 2).

There was moderate agreement regarding the modified Avgerinos score between the treating endoscopist and the expert endoscopist who reviewed the endoscopies (a Kappa value of 0.684) (p < 0.001).

When classifying patients into those with and without adequate gastroduodenal visualization, a higher number of patients with a full stomach was found in group A (7 patients), compared with group B (2 patients) (p = 0.066), and the difference was not statistically significant.

Patients were divided according to the time from hospital admission to endoscopy, with their modified Avgerinos scores: < 8 h, group A 5.00 (range: 4.00−6.00), group B 7.00 (range: 6.00–7.00) (p = 0.923); 8−12 hours, group A 7.00 (range: 4.00–8.00), group B 7.00 (range: 7.00–7.50) (p = 0.960); and 12−24 hours, group A 7.00 (range: 5.25–8.00), group B 7.00 (range: 6.00–8.00) (p = 0.683). None of these data reached statistical significance.

In our sub-analysis of the median modified Avgerinos score for each anatomical area (fundus, body, antrum, and duodenal bulb), we found significantly improved visualization of the gastric body and antrum in group B (metoclopramide). The median modified Avgerinos score was 2.00 (range: 2.00−2.00) for the gastric body in group B vs 2.00 (range: 1.00−2.00) in group A (p = 0.004) and was 2.00 (range: 2.00−2.00) for the antrum in group B vs 2.00 (range: 1.50−2.00) in group A, (p = 0.010) (Table 3).

Regarding bleeding etiology, for variceal bleeding in group A (placebo), the median modified Avgerinos score was 2.00 (range: 0.75–2.00) in the fundus, 1.50 (range: 1.00−2.00) in the body, 2.00 (range: 2.00−2.00) in the antrum, and 2.00 (range: 2.00−2.00 in the duodenal bulb, and in group B (metoclopramide) was 2.00 (range: 1.00−2.00) in the fundus, 2.00 (range: 2.00−2.00) in the body, 2.00 (range: 2.00−2.00) in the antrum, and 2.00 (range: 2.00−2.00) in duodenal bulb. The differences in the anatomic areas between groups were not statistically significant (p = 0.973, p = 0.223, p = 0.705, and p = 0.973, respectively).

For non-variceal bleeding in group A (placebo), the median modified Avgerinos score was 2.00 (range: 0.00–2.00) in the fundus, 2.00 (range: 1.00−2.00) in the body, 2.00 (range: 1.00−2.00) in the antrum, and 2.00 (range: 2.00−2.00) in the duodenal bulb. In group B (metoclopramide), the scores were 2.00 (range: 1.00−2.00) in the fundus, 2.00 (range: 2.00−2.00) in the body, 2.00 (range: 2.00−2.00) in the antrum, and 2.00 (range: 2.00−2.00) in duodenal bulb. The differences in the anatomic areas between groups were not statistically significant (p = 0.310, p = 0.070, p = 0.134, p = 0.561, respectively).

We observed no statistically significant difference in terms of the need for a second endoscopy (3 patients in group A vs 4 in group B [p = 0.590]), days of hospital stay (7.13 [SD ± 3.33] in group A vs 7.57 [SD ± 3.70] in group B [p = 0.747]), or the need for blood transfusion (11 patients in group A vs 11 in group B [p = 0.742]). The mean Glasgow-Blatchford bleeding score in group A was 11.55 (SD ± 2.65) vs 9.36 (SD ± 4.63) in group B (p = 0.191).

The need for a second endoscopy in the next 72 h for any reason was indicated in 3 (12%) patients in group A and 4 (16%) patients in group B (p = 0.684). Hospital stay was similar in both groups, at 7.0 days (SD ± 3.08) in group A and 7.32 days (SD ± 4.31) in group B (p = 0.764).

The mean overall hemoglobin level before endoscopy was 8.28 g/dL (SD ± 2.27); a mean of 7.92 g/dL (SD ± 2.06) in group A and 8.63 g/dL (SD ± 2.45) in group B (p = 0.276). Blood transfusion was required in 11 (38%) patients; 6 in group A and 5 in group B (p = 0.733). There were 6 deaths over the study period, 3 in each group (p = 0.100). All deaths occurred during hospitalization and none of them were directly related to the bleeding episode. All patients were followed for possible adverse effects, and no complications, such as anaphylaxis, extrapyramidal reaction, or cardiac arrhythmia, were observed during hospitalization.

In our randomized, triple-blind clinical trial in patients with UGIB, metoclopramide at a dose of 20 mg IV versus placebo did not improve visualization of the upper gastrointestinal tract during endoscopy, according to the modified Avgerinos score, nor did metoclopramide prevent the need for a repeat endoscopy or reduce the length of hospital stay.

Since 1994, the benefit of prokinetics, such as erythromycin, on gastroduodenal visualization during endoscopy in patients with UGIB has been reported.9 Erythromycin is the drug with the best profile so far. In a recent systematic review, erythromycin was shown to improve visualization of the gastric mucosa and reduce the need for blood transfusion in patients with UGIB.10

To date, six studies (two published in full, one brief report, and three abstracts) have evaluated the usefulness of metoclopramide on gastroduodenal visualization during endoscopy in patients with UGIB. However, their results are heterogeneous11–16 (Table 4).

In contrast to our results, the double-blind, controlled clinical trial by Manupeeraphant et al.11 reported a superior effect of metoclopramide versus placebo on gastroduodenal visualization. This superiority was observed even in patients undergoing gastric lavage, in whom fresh blood, clots, or coffee-ground-like material were found. However, despite the significant number of patients (n = 284), that study has limitations. For example, gastric lavage to assess bleeding severity may modify the amount of blood content, blood content was not detected during gastric lavage in 20% of the patients, and the clinical presentation of melena was significantly higher in the metoclopramide group. Given that patients with melena would likely have lower blood content in the proximal stomach than patients with fresh blood, such a difference could benefit the metoclopramide group.

In 2010, Daram et al.15 published a double-blind abstract study on metoclopramide (10 mg) use versus placebo before endoscopy for UGIB. They reported that no statistical difference was found between the placebo group and metoclopramide group, according to visual scores.

Similar to our work, 5 controlled clinical trials (four compared the use of metoclopramide versus placebo and one compared the use of metoclopramide versus no drug) found that metoclopramide did not improve gastroduodenal visualization during endoscopy12–16 (Table 4).

In 2023, Vimonsuntirungsri et al.12 conducted a double-blind controlled clinical trial at two centers in Thailand. The primary aim was to evaluate whether metoclopramide had an impact on the visualization of the upper gastrointestinal tract in patients with UGIB. Adequate endoscopic visualization was reported in the metoclopramide group (77.4%) compared with the placebo group (61.6%) (p = 0.16). However, when solely analyzing patients with gastric lesions, adequate visualization was found in 92.9% of patients with metoclopramide versus 50% of patients with placebo, (p = 0.03). There was better visualization in the fundus in the metoclopramide group, with an endoscopic visualization score of 1.79 (SD ± 0.42) in the metoclopramide group vs 1.29 (SD ± 0.72) in the placebo group (p = 0.03). As a secondary endpoint, the frequency of a repeat endoscopy at 72 h was lower with metoclopramide (3.2%), compared with placebo (22.6%), (p = 0.02). In contrast, gastric/duodenal visualization did not differ significantly between patients receiving metoclopramide and those given placebo. Furthermore, metoclopramide had no clinical effect on hospital stay duration (p = 0.747) or the requirement for a repeat endoscopy (p = 0.590).

Likewise, Estes et al.13 reported their results in a controlled clinical trial on 52 patients with UGIB, who received metoclopramide or placebo 30−120 min before endoscopy, using the modified Avgerinos score. Metoclopramide did not improve mucosal visualization, reduce the duration of endoscopy, decrease the number of transfusions, reduce the need for repeat endoscopy, or decrease the length of hospital stay.

Sussman et al.14 published an abstract, in which 26 patients were randomized to receive intravenous metoclopramide 10 mg or no medication. Endoscopy was performed at 30−120 min, finding no statistically significant difference in endoscopic visualization (p = 0.64). In addition, Habashi et al.16 published an abstract, in which 45 patients were assigned to 3 groups: metoclopramide, erythromycin, and placebo. Each group consisted of 15 patients. They found no statistically significant difference when evaluating the quality of gastroduodenal visualization.

Among the secondary aims of our study, we evaluated the need to perform a repeat endoscopy within the following 72 h for any reason, finding no statistical significance (p = 0.684) between metoclopramide (4 patients) and placebo (3 patients).

A meta-analysis carried out in 2010 evaluated the effect of prokinetics on patients with UGIB and reported that the use of prokinetics, such as erythromycin or metoclopramide, reduced the need for a second endoscopy (OR 0.55; 95% CI 0.32−0.94). However, no benefit was found when only the studies with metoclopramide were analyzed.17

Although both metoclopramide and erythromycin have prokinetic action, each one acts through different pathways. A possible explanation of why erythromycin has been shown to improve endoscopic visualization, compared with metoclopramide, could be due to its greater efficacy in generating an increase in gastric motility. This has been found in studies evaluating gastric emptying in critically ill patients.18,19

On the other hand, we observed no influence of metoclopramide on the number of units of blood transfused or on days of hospital stay.

In relation to our secondary aims, only the work of Vimonsuntirungsri et al.12 described a reduced need for a repeat endoscopy using metoclopramide. The other studies reported no differences, as in our study11,13–16 (Table 4).

The limitations of our study are due to the fact that it is a single center study.

The time from onset of bleeding to endoscopy could influence the visualization result. When analyzing patients who underwent endoscopy within the first 8 h of hospital admission versus patients who underwent endoscopy at 12−24 hours, no differences in mucosal visualization, need for a new endoscopy, or length of hospital stay were found. However, the plasma concentration of metoclopramide is known to be maintained for up to 8 h.20

We consider one of the strengths of our study to be the fact that it is the first triple-blind, randomized, controlled clinical trial evaluating the efficacy of metoclopramide at a dose of 20 mg, higher than the 10 mg dose analyzed in previous trials.

Metoclopramide at a dose of 20 mg/IV vs placebo administered before endoscopy did not improve endoscopic visualization, the need for repeat endoscopy, or the length of hospital stay in patients with acute upper gastrointestinal bleeding of any etiology.

No financial support was received in relation to this study.