GENERAL MEDICINE/ORIGINAL RESEARCH

Erythromycin Infusion or Gastric Lavage for Upper Gastrointestinal Bleeding: A Multicenter Randomized Controlled Trial Dominique Pateron, MD, Eric Vicaut, MD, PhD, Erwan Debuc, MD, Karima Sahraoui, MD, Nicolas Carbonell, MD, Xavier Bobbia, MD, Dominique Thabut, MD, Frédéric Adnet, MD, PhD, Pierre Nahon, MD, Rolland Amathieu, MD, Mounir Aout, PhD, Nicolas Javaud, MD, Patrick Ray, MD, Jean Claude Trinchet, MD, PhD, for the HDUPE Collaborative Study Group* From the Emergency Department, Saint-Antoine Hospital, Assistance Publique-Hôpitaux de Paris, Pierre et Marie Curie-Paris 6 University, Paris, France (Pateron, Debuc); the Clinical Research Unit, Assistance Publique-Hôpitaux de Paris, Fernand Widal Hospital, Paris 7 University, Paris, France (Vicaut, Aout); the Emergency Department, Argenteuil Hospital, Argenteuil, France (Sahraoui); the Hepatogastroenterology Department, Saint-Antoine Hospital, Assistance Publique-Hôpitaux de Paris, Paris, France (Carbonell); the Emergency and Intensive Care Department, Nîmes Hospital, Nîmes, France (Bobbia); the Hepatogastroenterology Department, Pitiè-Salpétrière Hospital, Assistance Publique-Hôpitaux de Paris, Paris, France (Thabut); Emergency Medical Services, SAMU 93, Avicenne Hospital, Assistance Publique-Hôpitaux de Paris, Paris 13 University, Bobigny, France (Adnet); the Hepatogastroenterology Department, Jean Verdier Hospital, Assistance Publique-Hôpitaux de Paris, Unité de Formation et de Recherche Santé-Médecine-Biologie Humaine, Paris 13 University, Bondy, France (Nahon, Trinchet); the Intensive Care Unit (Amathieu) and Emergency Department (Javaud), Jean Verdier Hospital, Assistance Publique-Hôpitaux de Paris, Bondy, France; and the Emergency Department, Pitiè-Salpétrière Hospital, Assistance Publique-Hôpitaux de Paris, Pierre et Marie Curie-Paris 6 University, Paris, France (Ray).

Study objective: The quality of endoscopy depends on the quality of upper gastrointestinal tract preparation. We determine whether in acute upper gastrointestinal bleeding the frequency of satisfactory stomach visualization was different after intravenous erythromycin, a nasogastric tube with gastric lavage, or both. Methods: We performed a prospective, randomized, multicenter (6 emergency departments) study in patients with acute upper gastrointestinal bleeding presenting with hematemesis or melena. The patients were randomized into 3 groups: (1) intravenous erythromycin infusion without nasogastric tube placement (erythromycin group), (2) nasogastric tube placement without erythromycin (nasogastric group), and (3) intravenous erythromycin infusion combined with nasogastric tube placement (nasogastric-erythromycin group). The main outcome measure was the proportion of satisfactory stomach visualization. Results: Two hundred fifty-three patients (181 men, mean age 61 years [SD 15 years], 84 with cirrhosis) were randomized: 84 (erythromycin group), 85 (nasogastric group), and 84 (nasogastric-erythromycin group). Overall, there was 85% satisfactory stomach visualization; between-group differences were not significant: ⫺4% (95% confidence interval [CI] ⫺15% to 6%) for the erythromycin group and nasogastric-erythromycin group, 2% (95% CI ⫺14% to 9%) for the erythromycin group and nasogastric group, and ⫺6.5% (95% CI ⫺17% to 4%) for the nasogastric group and nasogastric-erythromycin group. The duration of the endoscopic procedure, rebleeding frequency, the need for a second endoscopy, the number of transfused blood units, and mortality at days 2, 7, and 30 did not differ significantly between groups. Conclusion: In acute upper gastrointestinal bleeding, administration of intravenous erythromycin provides satisfactory endoscopic conditions, without the need for a nasogastric tube and gastric lavage. [Ann Emerg Med. 2011;xx:xxx.] Please see page XX for the Editor’s Capsule Summary of this article. 0196-0644/$-see front matter Copyright © 2011 by the American College of Emergency Physicians. doi:10.1016/j.annemergmed.2011.01.001

INTRODUCTION Background Improvements in endoscopic diagnosis and hemostasis have radically changed the management of acute upper *All members are listed in the Appendix. Volume xx, . x : Month 

gastrointestinal bleeding, a common medical emergency with a high incidence of morbidity and mortality.1,2 Endoscopy is used to determine the cause of bleeding and for hemostasis and should be performed within 12 hours of the first clinical signs of bleeding.3 The reliability of the procedure depends on the good visualization of the gastrointestinal tract.4 A nasogastric tube is placed to establish that bleeding is from the upper Annals of Emergency Medicine 1

Erythromycin Infusion for Upper Gastrointestinal Bleeding

Editor’s Capsule Summary

What is already known on this topic Erythromycin has promotility properties and has been used in the setting of acute upper gastrointestinal bleeding to improve conditions for endoscopy. Nasogastric tube placement with lavage is commonly used for the same indication. What question this study addressed In the setting of acute upper gastrointestinal bleeding, which intervention most frequently provides satisfactory endoscopy conditions: intravenous erythromycin, nasogastric tube with lavage, or both? What this study adds to our knowledge In this 253-patient trial, the frequency of satisfactory endoscopy conditions was similar among groups. How this is relevant to clinical practice In acute upper gastrointestinal bleeding, administration of intravenous erythromycin provides satisfactory endoscopy conditions, potentially obviating the need for a nasogastric tube.

gastrointestinal tract, to monitor bleeding on repeated gastric lavage, and to clear the gastrointestinal tract.4-7 However, nasogastric tube placement is highly unpleasant, rather painful, and time consuming.8 Complications such as aspiration pneumonia, misplacement, or digestive perforation may occur, although they are reported to be rare.9 Furthermore, guidelines for nasogastric tube placement are based on a low level of scientific evidence.10 Erythromycin is a motilin receptor agonist that accelerates gastric emptying by inducing antral contractions.11 Randomized controlled studies showed that erythromycin was more effective than placebo in cleaning the gastrointestinal tract before endoscopy and that erythromycin associated with gastric lavage was more effective than gastric lavage alone.12-14 However, there are no head-to-head comparisons of erythromycin versus gastric lavage or versus combination treatment in the management of acute upper gastrointestinal bleeding. Goals of This Investigation We determined whether in acute upper gastrointestinal bleeding the frequency of satisfactory stomach visualization was different after intravenous erythromycin, a nasogastric tube with gastric lavage, or both. 2 Annals of Emergency Medicine

Pateron et al

MATERIALS AND METHODS Study Design This was a prospective, randomized, multicenter, clinical trial for patients with acute upper gastrointestinal bleeding and presenting with hematemesis or melena. The emergency departments (EDs) of 6 hospitals participated in the study. Patients were randomized in 3 parallel groups. The study was approved by the regional ethics committee and registered with the Clinical Trials registry. Patients gave written informed consent for nasogastric tube placement or erythromycin infusion before endoscopy. Our study complied with the CONSORT reporting guidelines for randomized controlled trials. Setting and Selection of Participants All patients older than 18 years and who were referred to the 6 EDs for acute upper gastrointestinal bleeding between October 2005 and December 2007 were eligible for inclusion. Acute upper gastrointestinal bleeding was defined as presentation with hematemesis or melena either at or during the 12 hours before admission to the ED. A clinical examination and standard laboratory tests were performed on patient arrival. Routinely performed ECG eliminated patients with a contraindication to erythromycin (QTc interval ⬎0.45 seconds). Other criteria of noninclusion were (1) Glasgow Coma Scale score less than 15; (2) shock, defined as a persistent decrease in systolic arterial pressure (⬍90 mm Hg) and a pulse rate above 110 beats/min, unresponsive to fluid replacement; (3) known allergy to erythromycin; (4) treatment with drugs that might interact with erythromycin (eg, carbamazepine, theophylline, ergot alkaloids); (5) pregnancy; and (6) previous gastrectomy. Interventions As soon as the inclusion criteria had been checked in the ED, an interactive voice response system at the data management center randomized patients into one of 3 groups: (1) erythromycin group: patients received an intravenous infusion of erythromycin (250 mg during 20 minutes) and underwent endoscopy 30 minutes after the end of the infusion; (2) nasogastric group, control group: after nasogastric tube placement (16 to 20 French), gastric lavage was carried out with 500 mL of water at room temperature and repeated every hour until the aspirated gastric fluid was clear, and endoscopy was performed 15 minutes after the last irrigation; and (3) nasogastricerythromycin group: nasogastric tube placement and gastric lavage were performed as in the nasogastric group, followed by erythromycin infusion as in the erythromycin group. All patients underwent upper gastrointestinal endoscopy when hemodynamically stable (pulse rate ⬍100 beats/min, systolic arterial pressure ⬎100 mm Hg). Endoscopists were unaware of treatment assignation because it was not mentioned either verbally or in written reports (patient’s medical record) and because the emergency physicians removed the nasogastric tube and erythromycin infusion before patient arrival in the endoscopy room. In the event of active bleeding from a peptic Volume xx, . x : Month 

Pateron et al

Erythromycin Infusion for Upper Gastrointestinal Bleeding

Figure. Trial flow chart. ER, Erythromycin; NG, gastric lavage; NGER, erythromycin plus gastric lavage.

ulcer during the endoscopic procedure, the endoscopist could choose the method of hemostasis (injection therapy, mechanical treatment, or thermal treatment). The patient received intravenous omeprazole for 72 hours and then oral antibiotics to eradicate Helicobacter pylori. In the event of variceal bleeding, endoscopic sclerotherapy or ligature was performed. Patients with a diagnosis of cirrhosis (known cirrhosis or 2 of the following signs: jaundice, hepatomegaly, spider angiomas, cutaneous collateral portocaval circulation, palmar erythrosis) received somatostatin, octreotide, or terlipressin as soon as possible for 48 hours to 5 days and norfloxacin (400 mg orally twice a day) for 7 days.15 Volume xx, . x : Month 

A note was made of the hospital admission department (ICU, ED, medical department) or of whether the patient returned home. The results for standard laboratory tests (WBC count, platelets, hemostasis, ionogram, and blood urea nitrogen), the number of transfused blood units, and any rebleeding were recorded on days 2, 7, and 30. Discharged patients who failed to come to their follow-up visit were contacted by telephone. Methods of Measurement The primary endpoint was the quality of gastrointestinal tract visualization, as given by the scoring system by Frossard et Annals of Emergency Medicine 3

Erythromycin Infusion for Upper Gastrointestinal Bleeding Table 1. Baseline clinical and biological characteristics in each group.* Characteristics Mean age, y Men/women Mean pulse rate, beats/min Mean systolic arterial BP, mm Hg Mean diastolic arterial BP, mm Hg Cirrhosis, No. (%) Known peptic ulcer, No. (%) Previous gastrointestinal bleeding, No. (%) NSAID during last 15 days, No. (%) Mean hemoglobin, g/100 mL Mean leucocytes, No./mm3 Mean platelets, No.⫻103/ mm3 Mean prothrombin time, % Mean urea, mmol/L Mean AST, IU/L Mean ALT, IU/L

ER (Nⴝ84)

NG (Nⴝ85)

NGER (Nⴝ84)

61 (14) 58/26 97 (20) 124 (21)

61 (15) 62/23 95 (19) 128 (24)

60 (17) 61/23 100 (20) 124 (23)

69 (15)

70 (15)

68 (15)

28 (33) 16 (19) 29 (35)

29 (34) 11 (13) 26 (31)

27 (33) 13 (16) 27 (33)

32 (38)

26(31)

22 (27)

8.9 (2.8)

8.5 (3.0)

8.4 (3.2)

10.2 (5.8) 198 (128)

11.6 (12.5) 217 (126)

11.3 (7.3) 225 (141)

65 (20) 12 (8) 71 (82) 46 (58)

61 (24) 15 (23) 54 (63) 35 (31)

63 (24) 12 (9) 97 (174) 57(85)

BP, Blood pressure; NSAID, nonsteroidal anti-inflammatory drug; AST, aspartate aminotransferase; ALT, alanine aminotransferase. *All means are given with SD.

al.12 The endoscopist scored each of 4 areas of the stomach and duodenum (fundus, corpus, antrum, and bulbus) from 0 to 2 (0: less than 25% of the surface visible; 1: 25% to 75% visible; 2: more than 75% visible). The total score was the sum of the 4 individual scores and ranged from 0 to 8. A total score of 6 or more indicated a satisfactory stomach visualization and a score below 6 an unsatisfactory visualization. Secondary endpoints were patient outcomes until day 30, duration of the endoscopic procedure, number of endoscopic hemostasis procedures, ability to identify the source of bleeding, adverse effects related to erythromycin infusion or nasogastric tube placement, number of transfused blood units, rebleeding, and death. Pain after nasogastric tube placement was assessed by a visual analog scale (0: no pain; 100: maximum pain). Primary Data Analysis Our study was a superiority trial designed to demonstrate a difference between the 3 groups. The sample size needed to detect a 15% absolute difference in the proportion of satisfactory stomach visualization between 2 groups with 80% power (if the proportion in one group was 80) was 270 (90 patients per group) when Tukey’s adjustment of ␣ for multiplicity (2-sided ␣ risk of 5%) was used. Two-by-two comparisons of satisfactory stomach visualization frequencies between groups were made with the Mantel-Haentzel test, after stratification by center. The Breslow-Day method was used to test for odds-ratio heterogeneity among centers. Normally 4 Annals of Emergency Medicine

Pateron et al distributed variables were compared by a t test and nonnormally distributed quantitative variables by the MannWhitney test. All tests were 2-sided at a significance level adjusted for multiplicity by Tukey’s method. The analysis was intention to treat. Missing data were considered to be primary endpoint failures. All P values were 2-tailed, and P⬍.05 was considered significant. A post hoc analysis of the most severe cases was planned (ICU admissions, transfused patients, and patients with cirrhosis who are known to have a poor prognosis). Any benefit of erythromycin administration, either alone or combined with gastric lavage, might differ in these patients.

RESULTS Characteristics of Study Patients A total of 270 patients were included and assessed for eligibility between October 2005 and December 2007. They were followed up until January 2008. Of these patients, 253 presented at the ED with hematemesis or melena and met the inclusion criteria. Eighty-four patients (33%) were diagnosed with cirrhosis. Patients were randomized as follows: 84 (erythromycin), 85 (nasogastric), and 84 (nasogastricerythromycin) (Figure). Ten patients did not undergo endoscopy because of early death (n⫽2), esophageal stenosis (n⫽1), and lack of cooperation (n⫽7). The main clinical and biological characteristics at patient admission and the interval between the onset of bleeding and endoscopy did not differ significantly between groups (Tables 1 and 2). Main Results The mean endoscopic score did not differ significantly between groups (Table 2). There was a satisfactory stomach visualization in 85% of all patients, with no significant difference between groups (84%, 82%, and 88% for erythromycin, nasogastric, and nasogastric-erythromycin, respectively). The differences between groups were as follows: ⫺4.3% (95% confidence interval [CI] ⫺14.9% to 6.3%) for erythromycin and nasogastric-erythromycin, 2.2% (95% CI ⫺13.7% to 9.2%) for erythromycin and nasogastric, and ⫺6.5% (95% CI ⫺17.4% to 4.4%) for nasogastric and nasogastric-erythromycin. When patients with a missing score were considered as having failed treatment (n⫽2 for both erythromycin and nasogastric, n⫽6 for nasogastricerythromycin), the percentages were similar (82%, 80%, and 82%, respectively) and the differences were 0% (95% CI ⫺11.6% to 11.6%) for erythromycin and nasogastricerythromycin, 2.1% (95% CI ⫺9.7% to 14%) for erythromycin and nasogastric, and ⫺2.1% (95% CI ⫺14% to 9.7%) for nasogastric and nasogastric-erythromycin groups. In no case was the Breslow-Day test for heterogeneity significant. The percentage of patients with a satisfactory stomach visualization did not differ between groups in patients with cirrhosis or admitted to an ICU (Table 3). However, the percentage of transfused patients with a satisfactory stomach visualization was significantly higher in the nasogastricVolume xx, . x : Month 

Pateron et al

Erythromycin Infusion for Upper Gastrointestinal Bleeding

Table 2. Endoscopic features and treatment in each group.* No. (%) or Median (IQR)

Endoscopic Features and Treatment

ER (Nⴝ84)

Endoscopic score Interval between onset of bleeding and endoscopy, h Endoscopic findings Esophageal varices Ulcer Gastritis Mallory-Weiss syndrome Other Empty stomach Hemostatic treatment Ability to identify the source of bleeding Duration of endoscopy (min) Need for a second endoscopy

NG (Nⴝ85)

Difference (95% CI)

NGER (Nⴝ84)

8 (8, 8) 8 (7, 8) 8 (7, 8) 5.3 (3, 12.5) 6.4 (2.8, 13.3) 5. 1 (3.2, 11.2)

25 (30) 36 (22) 8 (10) 4 (5) 11 (13) 69 (84) 31 (38) 64 (78)

30 (36) 24 (14) 20 (24) 3(4) 9 (11) 68 (82) 28 (34) 65 (78)

31 (39) 31 (20) 19 (24) 3 (4) 8 (10) 69 (89) 34 (44) 66 (85)

10 (7, 16)

12 (7, 15)

14 (10, 20)

14 (20)

20 (26)

17 (23)

ER vs NG 0 ⫺1.1 (⫺4.2 to 2.2)

ER vs NGER

NG vs NGER

0 0.2 (⫺2.1 to 2.4)

0 1.3 (⫺1.8 to 4.3)

⫺5.7 (⫺20.0 to 8.7) ⫺9.3 (⫺24.0 to 5.5) 7.5 (⫺0.8 to 15.8) 2.1 (⫺6.8 to 11.0) ⫺14.3 (⫺25.6 to ⫺3.1) ⫺14.6 (⫺26.1 to ⫺3.1) 1.3 (⫺4.9 to 7.4) 1.0 (⫺5.3 to 7.4) 2.6 (⫺7.4 to 12.5) 3.2 (⫺6.8 to 13.2) 2.2 (⫺9.2 to 13.7) ⫺4.3 (⫺14.9 to 6.3) 4.5 (⫺10.1 to 19.2) ⫺5.3 (⫺20.6 to 10) 0.3 (⫺12.3 to 12.9) ⫺6.6 (⫺18.6 to 5.5) ⫺2 (⫺5 to 4)

⫺4 (⫺5 to 3)

⫺6.3 (⫺19.7 to 7.2)

⫺3.6 (⫺17.0 to 9.8)

⫺3.6 (⫺18.6 to 11.4) ⫺5.4 (⫺13.7 to 2.8) 0.3 (⫺13.0 to 13.5) 0.2 (⫺5.6 to 6.1) 0.6 (⫺8.9 to 10.1) ⫺6.5 (⫺17.4 to 4.4) ⫺9.9 (⫺24.8 to 5.1) ⫺6.3 (⫺18.3 to 5.6) ⫺2 (⫺5 to 5) 2.7 (⫺11.1 to 16.5)

IQR, Interquartile range. *There were no significant differences between groups for any variable.

Table 3. Patients with a clean gastrointestinal tract in subgroup analyses. No. (%) Patients With cirrhosis Transfused Admitted to ICU

Difference (95% CI)

ER (Nⴝ84)

NG (Nⴝ85)

NGER (Nⴝ84)

ER vs NG

ER vs NGER

NG vs NGER

23 (85.2) 47 (82.5) 38 (84.4)

20 (71.4) 46 (76.7)* 39 (78)

22 (95.7) 52 (92.9)* 39 (88.6)

13.8 (⫺7.7 to 35.2) 5.8 (⫺8.8 to 20.4) 6.4 (⫺9.2 to 22.1)

⫺10.5 (⫺26.3 to 5.3) ⫺10.4 (⫺22.4 to 1.6) ⫺4.2 (⫺18.3 to 10.0)

⫺24.2 (⫺42.9 to ⫺5.5) ⫺16.2 (⫺28.8 to ⫺3.5) ⫺10.6 (⫺25.5 to 4.2)

*The percentage of transfused patients with a clean gastrointestinal tract was significantly higher in the NGER group than the NG group (93% versus 77%, P⫽.021; Breslow-Day test, P⫽.021).

erythromycin than nasogastric group (Table 3). On the other hand, there was no significant difference between the nasogastric and erythromycin groups (P⫽.53; Breslow-Day test, P⫽.98) or between the nasogastric-erythromycin and erythromycin groups (P⫽.11; Breslow-Day test, P⫽.07). Of the 253 patients, 232 were followed up until day 30. The mean duration of the endoscopic procedure, need for hemostasis, ability to identify the source of the bleeding, and need for a second endoscopy did not differ significantly between groups (Table 2). There were no complications associated with endoscopy or nasogastric tube placement in any group. However, 6 patients (2 nasogastric, 4 nasogastric-erythromycin) did not tolerate gastric tube placement. The mean visual analog scale score for pain after tube placement was 42 (SD 32). The score exceeded 60, indicative of severe pain,16 in 28% of patients in the nasogastric group and 24% in the nasogastricerythromycin group (P⫽.59). The mean number of blood units transfused, rebleeding, and mortality during follow-up did not differ significantly between groups (Table 4). Rebleeding occurred in 22 patients (9%) within 24 hours of endoscopy, in 27 further patients (10%) Volume xx, . x : Month 

within 1 week, and in 7 further patients (3%) within 1 month (Table 4). A similar proportion of patients in the 3 groups underwent second-look endoscopy for rebleeding or poor gastric visualization (Table 2). Overall mortality rate was 7% (9 erythromycin, 7 nasogastric, and 3 nasogastric-erythromycin) (Table 4). Most patients were admitted to an ICU.

LIMITATIONS Our study was designed to detect a difference in the occurrence of satisfactory stomach visualization between groups. However, the observed absence of a significant difference does not imply equivalence of patient preparation methods. There was a 2.2% difference (95% CI ⫺9.2% to13.7%) in the proportion of satisfactory stomach visualization gastrointestinal tracts in favor of the erythromycin infusion alone group versus the nasogastric tube placement group. The lower limit of this CI indicates that any potential difference between the 2 groups is unlikely to exceed 9%. This clinically acceptable difference is much smaller than the 15% difference we used to calculate sample size. Annals of Emergency Medicine 5

Erythromycin Infusion for Upper Gastrointestinal Bleeding

Pateron et al

Table 4. Outcomes. No. (%) or Median (IQR)

Difference (95% CI)

ER (Nⴝ84) NG (Nⴝ87) NGER (Nⴝ84) Number of blood units transfused First 24 h First week First month Rebleeding First 24 h First week First month Deaths First 24 h First week First month Orientation from the ED ICU ED Medical department Return home

2 (0, 3) 2 (0, 4) 2 (0, 4)

2 (0, 3) 2 (0, 4) 2 (0, 4)

2 (0, 3) 2.5 (0, 4.5) 3 (0, 5)

6 (7) 13 (16) 19 (23)

8 (10) 17 (20) 18 (21)

8 (10) 19 (23) 19 (23)

3 (4) 7 (9) 9 (12)

2 (2) 5 (6) 9 (12)

3 (4) 3 (4) 3 (4)

46 (55) 22 (26) 14 (17) 2 (2)

52 (61) 23 (27) 10 (12) 0

48 (58) 18 (21) 17 (20) 0

DISCUSSION Our study has shown that gastrointestinal visualization by endoscopy in patients with acute upper gastrointestinal bleeding is not influenced by the method of patient preparation (nasogastric tube placement, erythromycin infusion, or their combination). A satisfactory stomach visualization was achieved in approximately 85% of patients, regardless of preparation method, and outcomes in the month after endoscopy did not differ significantly. Nasogastric tube placement provided no additional clinical benefit over intravenous erythromycin infusion in the management of patients with acute gastrointestinal bleeding. A high occurrence of good gastrointestinal tract preparation by erythromycin infusion before endoscopy has already been observed in patients with acute upper gastrointestinal bleeding.12-14 The seminal randomized study by Frossard et al12 revealed that, compared with placebo, erythromycin markedly increased the proportion of satisfactory stomach visualization (82% versus 33%). Combining erythromycin infusion with gastric lavage also led to an increase and improved the quality of endoscopy over gastric lavage alone.14 Our study did not detect any significant difference in satisfactory stomach visualization frequency with erythromycin either alone or combined with gastric lavage, but our values were higher than those reported so far for all 3 preparation methods, and in particular for gastric lavage alone (82% in our study versus 44% in the study by Carbonell et al14). The characteristics of our patient population with acute upper gastrointestinal bleeding might have accounted for this discrepancy. We included all patients presenting with hematemesis or melena even if their clinical condition was good, thus covering all patient types. Only 55% of our patients were admitted to an ICU, whereas in previous reports, all patients were admitted to an ICU or to a specialized unit.12,14 6 Annals of Emergency Medicine

ER vs NG 0 (⫺2 to 2) 0 (⫺1.0 to 0.5) 0 (⫺1 to 1)

ER vs NGER 0 (⫺2 to 1) ⫺0.5 (⫺1.5 to 0) ⫺1.0 (⫺2.0 to 0.5)

NG vs NGER 0 (⫺2 to 0.5) ⫺0.5 (⫺1.5 to 1) ⫺1 (⫺2 to 1)

⫺2.4 (⫺10.9 to 6.0) ⫺2.9 (⫺11.6 to 5.8) 0.5 (⫺8.7 to 9.7) ⫺4.5 (⫺16.0 to 7.0) ⫺7.1 (⫺19.0 to 4.7) ⫺2.6 (⫺15.0 to 9.7) 1.4 (⫺11.0 to 13.9) 0 (⫺12.7 to 12.7) ⫺1.4 (⫺13.9 to 11.0) 1.2 (⫺3.9 to 6.3) 3.1 (⫺5.4 to 11.5) 0.3 (⫺10.0 to 10.6)

0 (⫺5.6 to 5.6) 5.5 (⫺2.3 to 13.4) 8.1 (⫺0.4 to 16.6)

⫺1.2 (⫺6.3 to 3.9) 2.5 (⫺4.3 to 9.2) 7.8 (⫺0.6 to 16.2)

⫺6.4 (⫺21.3 to 8.4) ⫺2.4 (⫺17.4 to 12.6) ⫺4.0 (⫺18.8 to 10.8) ⫺0.9 (⫺14.2 to 12.5) 4.8 (⫺8.1 to 17.6) 5.6 (⫺7.3 to 18.5) 4.9 (⫺5.6 to 15.4) ⫺3.6 (⫺15.3 to 8.2) ⫺8.5 (⫺19.5 to 2.5) 2.4 (⫺0.9 to 5.6) 2.4 (⫺0.9 to 5.6) 0

In addition, we confirmed the benefit of combining erythromycin with lavage in the most severe cases. Transfused patients undergoing combined treatment rather than lavage alone had a higher proportion of satisfactory stomach visualization. A similar benefit of adding erythromycin has already been observed in patients with cirrhosis.12,14 The use of erythromycin before endoscopy has been reported to be costeffective.17 However, because we did not evaluate management of acute gastrointestinal bleeding without nasogastric tube placement or erythromycin, we cannot assert definitively that any intervention is required before endoscopy. The use of promotility agents in nonvariceal bleeding before endoscopy was not endorsed in the 2010 international consensus guidelines.5 Rebleeding and mortality rates were similar in all 3 groups and were not influenced by nasogastric tube placement. Our overall 7% mortality rate was in line with that cited in epidemiology studies.18-20 We encountered no adverse effects in any group, although severe but rare complications have been described with nasogastric tube placement.8,9,21 Nasogastric tube placement was associated with significant pain, as already reported.6,22 A quarter of our patients experienced severe pain. The safety profile of erythromycin is excellent.23 It was easy to use, without severe adverse effects, and painless. Although nasogastric aspiration has been described in several studies as useful, an observational study has shown that aspiration immediately before endoscopy has low sensitivity and low specificity in predicting active bleeding at endoscopy.24-27 In our study, nasogastric tube placement and gastric lavage did not influence outcomes even in the most severely ill patients. One can therefore conclude that erythromycin infusion might be a good substitute for gastric lavage, avoiding nasogastric tube placement before endoscopy, in ED patients with acute upper Volume xx, . x : Month 

Pateron et al gastrointestinal bleeding presenting with hematemesis or melena. Supervising editor: Steven M. Green, MD Author contributions: DP initiated and coordinated the research. DP, EV, ED, FA, and JCT designed the study. EV prepared the study grant application. DP and ED coordinated the participating centers. EV managed all logistic aspects. EV and MA monitored the budget and data. DP, EV, KS, NC, DT, and PR managed and analyzed data and wrote the article. ED, XB, FA, PN, RA, MA, NJ, and JCT participated in the study design, management, data collection, interpretation, and writing of the article. All authors contributed substantially to the study and approved the final version of the article. DP takes responsibility for the paper as a whole. Funding and support: By Annals policy, all authors are required to disclose any and all commercial, financial, and other relationships in any way related to the subject of this article that might create any potential conflict of interest. See the Manuscript Submission Agreement in this issue for examples of specific conflicts covered by this statement. Funded by the French Ministry of Health’s “Programme Hospitalier de Recherche Clinique” for 2004 (PHRC 2004 No. AOM-04093). Publication dates: Received for publication September 23, 2010. Revisions received October 14, 2010, and December 15, 2010. Accepted for publication January 3, 2011. Reprints not available from the authors. Address for correspondence: Dominique Pateron, MD, Emergency Department, Hôpital Saint-Antoine, APHP, UMR S 707, Université Paris 6, 184 rue du Faubourg Saint-Antoine, 75012 Paris, France; 33-1-49-28-30-70, fax 33-1-49-28-42-25; E-mail [email protected].

REFERENCES 1. Blatchford O, Davidson LA, Murray WR, et al. Acute upper gastrointestinal haemorrhage in west of Scotland: case ascertainment study. BMJ. 1997;315:510-514. 2. Cappell MS, Friedel D. Initial management of acute upper gastrointestinal bleeding: from initial evaluation up to gastrointestinal endoscopy. Med Clin North Am. 2008;92:491-509. 3. Cooper GS, Chak A, Way LE, et al. Early endoscopy in upper gastrointestinal hemorrhage: association with recurrent bleeding, surgery, and length of hospital stay. Gastrointest Endosc. 1999; 49:145-152. 4. Spiegel BMR, Vakil NB, Ofnman JJ. Endoscopy for acute nonvariceal upper gastrointestinal tract hemorrhage: is sooner better? Arch Intern Med. 2001;161:1393-1404. 5. Barkun A, Bardou M, Kuipersl EJ. International consensus recommendations on the management of patients with nonvariceal upper gastrointestinal bleeding. Ann Intern Med. 2010;152:101-113. 6. Witting MD, Mager L, Heins AE, et al. Usefulness and validity of diagnostic aspiration in patients without hematemesis. Ann Emerg Med. 2004;43:525-532. 7. Basuk PM, Isenberg JI. Gastric lavage in patients with gastrointestinal hemorrhage: yea or nay? Arch Intern Med. 1990; 150:1379-1380.

Volume xx, . x : Month 

Erythromycin Infusion for Upper Gastrointestinal Bleeding 8. Wolfe TR, Fosnocht DE, Linscott MS. Atomized lidocaine as topical anesthesia for nasogastric tube placement : a randomized, double-blind, placebo-controlled trial. Ann Emerg Med. 2000;35:421-425. 9. Gough D, Rust D. Nasogastric intubation: morbidity in an asymptomatic patient. Am J Emerg Med. 1986;4:511-513. 10. Cappell MS. Safety and efficacy of nasogastric intubation for gastrointestinal bleeding after myocardial infarction: an analysis of 125 patients at two tertiary cardiac referral hospitals. Dig Dis Sci. 2005;50:2063-2070. 11. Lin H, Sanders S, Gu Y, et al. Erythromycin accelerates solid emptying at the expense of gastric sieving. Dig Dis Sci. 1994;39:124-128. 12. Frossard JL, Spahr L, Queneau PE, et al. Erythromycin intravenous bolus infusion in acute upper gastrointestinal bleeding: a randomized, controlled, double-blind trial. Gastroenterology. 2002;123:17-23. 13. Coffin B, Pocard M, Panis Y, et al. Erythromycin improves the quality of EGD in patients with acute upper GI bleeding: a randomised controlled study. Gastrointest Endosc. 2002;56:174-179. 14. Carbonell N, Pauwels A, Serfaty L, et al. Erythromycin infusion prior to endoscopy for acute upper gastrointestinal bleeding: a randomized, controlled, double-blind trial. Am J Gastroenterol. 2006;101:1211-1215. 15. De Franchis R. Updating consensus in portal hypertension: report of the Baveno III Consensus Workshop on definitions, methodology and therapeutic strategies in portal hypertension. J Hepatol. 2000;33:846-852. 16. Aubrun F, Langeron O, Quesnel C, et al. Relationships between measurement of pain using visual analog score and morphine requirements during postoperative intravenous morphine titration. Anesthesiology. 2003;98:1415-1421. 17. Winstead NS, Wilcox CM. Erythromycin prior to endoscopy for acute upper gastrointestinal haemorrhage: a cost-effectiveness analysis. Nat Clin Pract Gastroenterol Hepatol. 2008;5:358-359. 18. Van Leerdam ME. Epidemiology of acute upper gastrointestinal bleeding. Best Pract Res Clin Gastroenterol. 2008;22:209-224. 19. Czernichow P, Hochain P, Nousbaum JB, et al. Epidemiology and course of acute upper gastrointestinal haemorrhage in four French geographical areas. Eur J Gastroenterol Hepatol. 2000;12:175-181. 20. Pastatis GA, Matrella E, Kapsoritakis A, et al. An epidemiological study of acute upper gastrointestinal bleeding in Crete, Greece. Eur J Gastroenterol Hepatol. 2000;12:1215-1220. 21. Dees G. Difficult nasogastric tube insertions. Emerg Med Clin North Am. 1989;7:177-182. 22. Singer AJ, Richman PB, Kowalska A, et al. Comparison of patient and practitioner assessments of pain from commonly performed emergency department procedures. Ann Emerg Med. 1999;33:652658. 23. Rubinstein E. Comparative safety of the different macrolides. Int J Antimicrob Agents. 2001;18:71-76. 24. Aljebreen AM, Fallone CA, Barkun AN. Nasogastric aspirate predicts high-risk endoscopic lesions in patients with acute upperGI bleeding. Endoscopy. 2004;59:172-178. 25. Perng CL, Lin HJ, Chen CJ, et al. Characteristics of patients with bleeding peptic ulcer requiring emergency endoscopy and aggressive treatment. Am J Gastroenterol. 1994;89:1811-1814. 26. Gilbert DA, Silverstein FE, Tedesco FJ, et al. The national ASGE survey on upper gastrointestinal bleeding: III. Endoscopy in upper gastrointestinal bleeding. Gastrointest Endosc. 1981; 27:94-103. 27. Cuellar RE, Gavaler JS, Alexander JA, et al. Gastrointestinal tract hemorrhage. The value of a nasogastric aspirate. Arch Intern Med. 1990;150:1381-1384.

Annals of Emergency Medicine 7

Erythromycin Infusion for Upper Gastrointestinal Bleeding

Appendix. HDUPE Group members. Frederic Adnet, MD, PhD, Avicenne University Hospital, Bobigny, and Léonard de Vinci-Paris 13 University Roland Amathieu, MD, Jean Verdier University Hospital, Bondy and Léonard de Vinci-Paris 13 University Mounir Aout, PhD, Fernand-Widal University Hospital, Paris and Paris Diderot-Paris 7 University Caroline Birlouez, MD, Gonesse Hospital, Gonesse Xavier Bobbia, MD, Nîmes University Hospital, Nîmes Jean François Bourgaux, MD, Nîmes University Hospital, Nîmes Nicolas Carbonell, MD, Saint-Antoine University Hospital and Pierre & Marie Curie-Paris 6 University Erwan Debuc, MD, Jean Verdier University Hospital, Bondy (currently Saint-Antoine University Hospital) Gilles Dhonneur, MD, PhD, Jean Verdier University Hospital, Bondy and Léonard de Vinci-Paris 13 University François Durand, MD, PhD, Beaujon University Hospital, Clichy, and Paris Diderot-Paris 7 University Nicolas Javaud, MD, Jean Verdier University Hospital, Bondy Frederic Lapostolle, MD, PhD, Avicenne University Hospital, Bobigny

8 Annals of Emergency Medicine

Pateron et al Jean-Yves Lefrant, MD, Nîmes University Hospital, Nîmes Laurent Muller, MD, Nîmes University Hospital, Nîmes Pierre Nahon, MD, Jean Verdier University Hospital, Bondy and Léonard de Vinci-Paris 13 University Dominique Pateron, MD, Jean Verdier University Hospital, Bondy (currently Saint-Antoine University Hospital and Pierre & Marie Curie-Paris 6 University) Patrick Ray, MD, Pitié-Salpêtrière University Hospital, Paris and Pierre & Marie Curie-Paris 6 University Philippe Renard, MD, Argenteuil Hospital, Argenteuil Bruno Riou, MD, PhD, Pitié-Salpêtrière University Hospital, Paris and Pierre & Marie Curie-Paris 6 University Karima Sahraoui, MD, Argenteuil Hospital, Argenteuil Jean-Luc Sebbah, MD, Gonesse Hospital, Gonesse Dominique Thabut, MD, Pitié-Salpêtrière University Hospital, Paris Jean Claude Trinchet, MD, PhD, Jean Verdier University Hospital, Bondy and Léonard de Vinci-Paris 13 University Loic Tual, MD, PhD, Jean Verdier University Hospital, Bondy Eric Vicaut, MD, PhD, Fernand-Widal University Hospital, Paris, and Paris Diderot-Paris 7 University

Volume xx, . x : Month 