© Masson, Paris, 2005. Gastroenterol Clin Biol 2005;29:140-144

140

Management of bleeding peptic ulcer in France : a national inquiry

Gilles LESUR (1), Bruno BOUR (2), Philippe AEGERTER (3), Association Nationale des Hépatogastroentérologues des Hôpitaux Généraux

(1) Service d’Hépatogastroentérologie, Hôpital Ambroise Paré, 92104 Boulogne Cedex ; (2) Département d’Hépatogastroentérologie, Centre Hospitalier, 72000 Le Mans ; (3) Département d’Informatique Médicale, Hôpital Ambroise Paré, 92104 Boulogne Cedex.

SUMMARYAims of the study — To evaluate and compare management prac-tices in France for bleeding peptic ulcers using a national inquiry ofuniversity and non-university hospitals.

Method — Responses to questionnaires sent to 812 gastroenterolo-gists, 496 practicing in non-university hospitals and 316 in univer-sity hospitals, were compared.

Results — An analysis was possible in 279 (34% response rate) ofthe questionnaires. Forrest classification was used more frequentlyin university hospitals (83% vs 60%, P < 0.01). Endoscopic hemo-static therapy was used more frequently in university hospitals forForrest Ib (92% vs 81%, P = 0.02), IIa (93% vs 73%, P < 0.001),and IIb (58% vs 29%, P < 0.001) ulcers. Injection therapy, mainlyepinephrine, was the first-intention treatment for 99% of theresponding gastroenterologists. Proportions of clinicians employinghemoclips (27%) or argon plasma coagulation (21%) were similarin both types of practice. Anti-secretory treatment included mainlyomeprazole (82%), given intravenously (76%), sometimes as bolusiv doses followed by iv high-dose continuous infusion (15%) withsome variations according to the type of hospital. In the event ofrecurrent or persistent bleeding, surgery was more frequent in non-university hospitals. When rebleeding occurred, a second endo-scopic treatment was performed in about one quarter of patients.

Conclusion — In France, management practices for bleeding pepticulcer vary between university and non-university hospitals.

RÉSUMÉ

Gilles LESUR, Bruno BOUR, Philippe AEGERTER, Association Nationale des Hépatogastroentérologues des Hôpitaux Généraux

But — Évaluer et comparer par une enquête déclarative de pratiquela prise en charge des hémorragies ulcéreuses dans les hôpitauxgénéraux et universitaires français.

Méthode — Les réponses de 496 hépato-gastroentérologues exer-çant en hôpital général et de 316 exerçant en hôpital universitaireont été comparées.

Résultats — Deux cent soixante dix-neuf questionnaires étaient ana-lysables (34 %). La classification de Forrest était plus utilisée en hôpi-tal universitaire (83 % vs 60 %, P < 0,01). L’hémostase endoscopiqueétait plus souvent pratiquée en hôpital universitaire en cas d’ulcèresForrest I b (92 % vs 81 %, P = 0,02), II a (93 % vs 73 %, P < 0,001)et II b (58 % vs 29 %, P < 0,001). Les injections, avant tout d’adréna-line, étaient la première modalité thérapeutique dans 99 % des cas.L’utilisation des clips (27 %) et de la coagulation au plasma argon(21 %) était comparable dans les 2 types de structure. Le traitementantisécrétoire était le plus souvent de l’oméprazole (82 %), principa-lement par voie intraveineuse (76 %) et parfois selon le schémabolus-perfusion continue (15 %) mais avec des modalités différentesd’utilisation selon les centres. En cas d’hémorragie persistante ourécidivante, un traitement chirurgical était plus souvent réalisé dansles hôpitaux généraux. En cas de récidive hémorragique un nouveautraitement endoscopique était proposé environ une fois sur 4.

Conclusion — La prise en charge des hémorragies ulcéreuses enFrance est différente en hôpital général et en hôpital universitaire.

Introduction

Bleeding peptic ulcer (BPU) disease is the leading cause ofupper gastrointestinal bleeding. In France, one-third of all uppergastrointestinal bleeding episodes are considered to arise for BPUwith an estimated incidence of about 25,000 episodes per year[1]. Mortality remains high, estimated at about 15% in the mostrecent reports [1, 2]. In 1999, the French Society of Gastrointesti-nal Endoscopy (SFED) published guidelines for the management ofBPU [3]. In France, nine out of ten patients with BPU are treated inpublic hospitals but no data concerning current practices areavailable [4]. The purpose of our work was to 1) study manage-ment practices for BPU in French public hospitals; 2) comparepractices between university hospitals and non-university hospitals.

Patients and methods

A questionnaire was developed using the usual methodology for cli-nical practice inquiries [5]. The questions concerned the different steps ofmanagement procedures for upper gastrointestinal bleeding. Yes-no ormultiple choice questions were used. The items were divided into fourcategories: 1) organization of management practices; 2) conditions ofthe first endoscopic procedure; 3) treatment modalities; 4) type of hospi-tal, its organization, and medical personnel. The target population wasdefined from the national registries of hepatogastroenterologists maintai-ned by the ANGH (Association Nationale des Hépato-gastroentérolo-gues des hôpitaux non universitaires) and the SNFGE (Société NationaleFrançaise de Gastro-entérologie) and from phone conversations with thesecretaries of the university hospital units to establish the list of qualifiedpractitioners (senior hospital physicians, non-titularized senior physi-cians). The questionnaire was tested before the inquiry with five hospital

Prise en charge des hémorragies digestives ulcéreuses en France en CHU et en CHG : une enquête nationale de pratique

(Gastroenterol Clin Biol 2005;29:140-144)

Reprints : G. LESUR, address above.E-mail : gilles.lesur@ apr.ap-hop-paris.fr This work was supported by a grant from SFED-AstraZeneca.

Management of bleeding peptic ulcer in France : a national inquiry

141

physicians performing emergency gastrointestinal endoscopy procedu-res. The questionnaire was then sent by mail, at the end of 2001, to812 hepatogastroenterologists, 496 practicing in 211 non-universitygeneral hospitals and 316 practicing in 54 university hospitals (106 in20 centers of the Public Parisian Hospitals [AP-HP] and 210 in 34 tea-ching centers of 25 University Hospitals). The practitioners were expectedto respond within two months. The practitioners were not contactedagain if no response was received.

The chi-square test was used for qualitative variables and Student’st test or a non-parametric test as appropriate for quantitative variables.The physician and not the hospital was the analysis unit. The significancethreshold was set at P < 0.05. Certain items could not be analyzed dueto insufficient sample size.

Results

An analyzable questionnaire was returned by 269 gastroen-terologists (34% response rate), 191 practicing in non-universityhospitals and 88 in 40 university hospitals (38% vs 28%,P < 0.01).

Organization of management practices

Patients were more frequently hospitalized in gastroenterol-ogy units and in specialized sectors in university hospitals(43.2% vs 27.7%, P = 0.016). One-third of university hospitalshad a specialized sector for this type of patient. In non-universityhospitals, 61.3% of patients were hospitalized in a unit otherthan a gastroenterology or intensive care unit. Endoscopy wasperformed by a senior operator for 64.5% of patients (67% innon-university hospitals and 63% in university hospitals). Thesenior operator had 10.88 5.54 years experience in universityhospitals and 16.15 5.79 in non-university hospitals(P = 0.01). A 24-h endoscopy service was available in 84.9% ofthe hospitals with no difference by type of hospital. A gastroen-terologist was on call for emergency procedures in 82.1% and77.3% of the non-university and university hospitals respectivelywhile emergency endoscopists were on duty permanently in onlythe university hospitals (14.8% vs 0%, P < 0.001).

Conditions of the first endoscopic procedure (table I)

The first endoscopic procedure was performed within the firstsix hours of the bleeding episode in 28% of patients and within24 h in 96%. For 83.2%, sedation was not used. Generalanesthesia was given in 8.2% of patients and controlled or non-controlled sedation in 8.6%. Gastric aspiration was employed in21.5% of patients with hematemesis, in 2.5% of patients withmelana, and systematically in 23.7% of cases. The procedurewas conducted without prior aspiration in 40.1% of patients. Anerythromycin infusion was administered in 21.9% of patients,with a 250 mg dose for half of them and in association with gas-tric aspiration in 36.1%. The emergency procedure was per-formed in an endoscopy room with video-endoscopy for 78.5%of patients. When performed elsewhere, optic equipment wasused for two-thirds of patients. An assistant was present for97.8% of the procedures (an endoscopy nurse for 29%, availa-ble personnel for 69%). The operative conditions were not differ-ent between university and non-university hospitals.

Treatment modalities

A prognostic score (e.g., Baylor, Rockall) was not used forclassification in 94% of patients. The Forrest classification wasused in 67% of patients and more frequently in university hospi-tals (83% vs 60%, P < 0.01). An endoscopic hemostatic treat-ment was delivered in 98%, 84%, 79%, 38%, 2%, and 1% ofpatients with Forrest grade Ia, Ib, IIa, IIb, and III ulcers, respec-tively. Endoscopic treatment was performed significantly moreoften in university hospitals for diffuse oozing bleeding, visiblenon-hemorrhagic vessels, and adherent clots (figure 1). Clotmobilization was attempted more often in non-university hospi-tals (87% vs 65%, P < 0.01). Hemostatic injections wereattempted in 99% of patients using diluted epinephrine (63%),saline solution (14%), polidocanol (8%), ethanol (5.4%), otheragents (2.5%) or a combination of agents (1.4%). Other tech-niques employed were hemoclips (27%), argon plasma coagula-tion (21%), bipolar electrocoagulation (9%), thermocoagulation(5%) and Nd Yag laser (0.4%). Epinephrine injection was usedmore often in university hospitals (86% vs 51%, P < 0.001) andpolidocanol in non-university hospitals (11% vs 2%, P < 0.02)(figure 2). A combination was used for 47% of patients, gener-ally injections and hemoclips (34% of the combination treat-ments), injections and argon plasma coagulation (21%), orinjections and bipolar electrocoagulation (11%). Use of thermalor mechanical hemostatic techniques was not significantly differ-ent between university and non-university hospitals.

Antisecretory medications were administered in associationwith the endoscopic treatment in 99% of patients. Omeprazolewas generally used (82%), mainly by intravenous infusion (76%)and occasionally with a bolus-continuous infusion scheme(15%). Omeprazole was used more often than other protonpump inhibitors in general hospitals (88% vs 67%, P < 0.001).Bolus omeprazole followed by intravenous infusion was morefrequent in university hospitals (34% vs 9%, P < 0.001) wherethe mean dose of omeprazole was higher (84.51 8.88 mg/dvs 64.02 2.68 mg/d, P = 0.03) (figure 3). If bleeding per-sisted, surgical treatment was undertaken more often in generalhospitals (60% vs 43%, P = 0.02). Surgical treatment was alsoemployed for recurrent bleeding more often in general hospitals(51% vs 31%, P = 0.001); 27% of the physicians proposed anew endoscopic treatment (32% and 25% in university and non-university hospitals, P = NS). When overt bleeding did not recur,a new endoscopic procedure was performed within 72 hours in43% of patients. When a second procedure was performed, dis-covery of a visible non-hemorrhagic vessel did not lead to anew endoscopic treatment in 40% of patients. The same hemo-

ABBREVIATIONS :BPU : bleeding peptic ulcersSFED : Société Française d’Endoscopie Digestive (French Society of

Gastrointestinal Endoscopy)

Table I. – Modalities of emergency endoscopy for non variceal uppergastrointestinal bleeding in all the hospitals.Conditions de réalisation de l’endoscopie initiale dans l’ensem-ble des centres hospitaliers.

Without gastric inspiration 40 %

Erythromycin infusion 22 %

Endoscopy within 24 hours 96 %

General anesthesia 8 %

Performed by a senior operator 64 %

In an endoscopy room 78 %

With video-endoscopy 85 %

With an assistant 98 %

G. Lesur et al.

142

static treatment used initially was applied again in nine out often patients.

At the time of endoscopy, 55.2% of the physicians performedgastric biopsies to detect Helicobacter pylori infection ; a rapidurease test was used by 18.3%. Serological tests forHelicobacter pylori were preferred by 28% of physicians. Aneradication treatment was instituted for 83.2% of patients afterthe bleeding episode was controlled. Eradication was proposedfor 92% of patients. A repeat endoscopy was planned for 50.5%of patients. No significant difference was found between univer-sity and non-university practitioners concerning management ofHelicobacter pylori infection.

Discussion

This is the first clinical inquiry conducted in France concern-ing management practices for BPU. High prevalence and persist-ent gravity of BPU seem to justify our inquiry. The study wasbased on the physicians’ statements and did not focus on objec-tive measurements of clinical practices and results. This approachled to several biases: imperfectly representative sample since theresponse rate was 34%, absence of information concerning thepatients and the characteristic features of the health care centers,particularly the number of beds and number of patients treated.

Despite these limitations, our results provide interesting infor-mations. The initial endoscopic procedure was performed withinthe first 24 hours in nearly all patients, in line with reports show-ing the importance of early endoscopy for prognosis [6]. Mostendoscopic procedures for upper gastrointestinal bleeding wereperformed without general anesthesia. Sedation controlled by ananesthesist and general anesthesia were exceptional. Use ofsedation without proper control by an anesthesist, an unaccepta-ble practice according to the current French regulations, wasvery rare. Since endoscopy for gastrointestinal bleeding is a longdifficult procedure, we considered that the specialist societies,namely the SFED and the SFAR, should promote more wide-spread use of anesthesia for emergency procedures. At the timeof this inquiry, erythromycin was not widely used in France forthis indication, the two trials demonstrating its usefulness foremergency upper endoscopic procedures being published in2002 [7, 8]. Use of erythromycin, which does not appear tohave any inconvenience, can be expected to become routinepractice. Lavage-aspiration, which was not often used even incase of hematemesis, will undoubtedly be used even less in thefuture. The Baylor [9] and Rockall [10] scores, which are widelyused in Great Britain, are not commonly employed in Francedespite several publications which have demonstrated their prog-nostic value. The fact that these scoring systems are “relatively”complex cannot, alone, explain this situation. A recent Frenchpublication demonstrated that diffusion of practice guidelines

92%81%

93%

73%58%

29%

0

20

40

60

80

100University

General

Forrest I b II a II b

* P=0,02 ** P<0,001

***

**

Fig. 1 – Indications of endoscopic therapy in ulcers Forrest I a, II a and II b in university and general hospitals.Indications du traitement endoscopique pour les ulcères I a, II a et II b de la classification de Forrest dans les CHU et les CHG.

86 %

51 %

2 %11 %

0

20

40

60

80

100

University

General

epinephrine

polidocanol

* p <0.001 ** p <0.02

*

**

Fig. 2 – Injection therapy with epinephrine and polidocanol in universityand general hospitals.Injections hémostatiques d’adrénaline et de polidocanol en CHU etCHG.

88 %

67 %

34 %

9 %

0

20

40

60

80

100

UniversityGeneral

* P<0.001

All procotols

Bolus + perfusion *

*

Fig. 3 – Modalities of iv omeprazole utilization in university and generalhospitals.Modalités d’utilisation de l’oméprazole par voie veineuse en CHUet en CHG.

Management of bleeding peptic ulcer in France : a national inquiry

143

including the Rockall score can contribute to improved progno-sis of upper gastrointestinal bleeding [11]. Widespread, thoughnot universal (67%), use of the Forrest classification, particularlyin non-university hospitals where it was used significantly lessoften, was a rather unexpected finding. This is an old systempublished in 1975 and is known to be operator-dependent [12].It is however useful for management of BPU, as correct use per-tains to indications of hemostatic treatment [6] or even type ofendoscopic treatment [13]. Moreover, a detailed description oflesions observed (active or non-active bleeding, visible non-hemorrhagic vessel …) can be used to choose endoscopic treat-ment without determining the Forrest classification. The results ofthe current study and observations of daily practice suggest thatsome physicians are not fully informed about this classificationsystem. A point which might be improved during future training.

Regarding indications for hemostatic treatment, nearly allspurting BPU (98%) were treated endoscopically, while manage-ment of oozing BPU varied significantly between university andnon-university hospitals (treatment in 92% and 81% of patientsrespectively). According to the literature where the distinctionbetween spurting and oozing BPU is made, the risk of rebleed-ing is much lower for oozing bleeding. For oozing BPU, the riskof persistent bleeding or recurrence is approximately 10% [14].For this reason, certain authors consider that endoscopic treat-ment offers little or no benefit for patients with oozing bleeding,suggesting that endoscopic treatment is not warranted for thesepatients [14]. This difference in attitude, expressed by more“active” intervention in university hospitals than in non-university hospitals, was also found for patients with a visiblenon-hemorrhagic vessel (treatment in 58% and 29% of patients,respectively). The risk of rebleeding for patients with a visiblenon-hemorrhagic vessel or an adherent clot who do notundergo endoscopic treatment is about 50% and 20% respec-tively [6]. Combining endoscopic treatment with medical treat-ment for patients with visible non-hemorrhagic vesselssignificantly reduces the risk of recurrence compared with medi-cal treatment alone [15]. For adherent clots, endoscopic treat-ment [16, 17] and high-dose proton pump inhibitorsadministered intravenously [17] both significantly reduce therisk of rebleeding. The longer experience of the non-universityhospital senior physicians who responded to the questionnaire(16 years versus 12 years for university hospital physicians) anddifferent recruitment practices (more seriously ill patients beingreferred to university hospitals) might explain in part this differ-ence.

Injections were clearly the preferred treatment modality at thetime of the inquiry. A large number of publications have demon-strated that epinephrine injections, which have proven efficacyand no risk of morbidity, are a simple-to-use low-cost methodgenerally considered as the gold standard [6]. It was recentlysuggested that high-dose epinephrine (> 10 mL) could be signifi-cantly more effective in terms of recurrence than the usual doses[18]. Sclerosing agents are used much less often, but were never-theless employed more often in non-university hospitals (11% vs2%, P < 0.001). There is no satisfactory explanation for thisobservation. Use of sclerosing agents, particularly at a highdose, has been incriminated as a cause of rare but sometimesserious local complications [6]. Most of the reported trials havenot demonstrated any superiority of the epinephrine-sclerosingagent combination over epinephrine alone [6]. Bipolar electroco-agulation and thermoregulation, widely used in other countries,appeared as marginal options in our inquiry. Combinations ofhemostatic methods with different modes of action were used forabout half of patients, the predominant combinations beinginjections plus hemoclips or argon plasma coagulation. There isno real proof of the therapeutic superiority of combinationsexcept for patients with spurting bleeding [19]. The use of hemo-

clips and argon plasma coagulation is widespread but the litera-ture provides conflicting data concerning hemoclips [20, 21],and is more favorable for argon plasma coagulation which canbe associated with epinephrine injections [22, 23].

The use of proton pump inhibitors, mainly omeprazole, incombination with endoscopic treatment is common practice inFrance for BPU. Intravenous infusion is preferred. The bolus pluscontinuous infusion scheme is not infrequent and is in line withdata in the literature demonstrating that high-dose proton pumpinhibitors administered intravenously have a synergetic effectwith endoscopic treatment and reduce the risk of rebleeding forForrest I and IIa ulcers [24].

Conservative surgery was not an exceptional option for per-sistent or recurrent bleeding. The surgical option was more fre-quent in non-university hospitals for persistent bleeding. Asecond endoscopic treatment was rarely proposed in either typeof hospital (27%). These data contrast with those from a rand-omized trial which demonstrated that definitive hemostatis canbe achieved with a second endoscopic procedure in the event ofrecurrence after a first procedure in the majority of patients andthat this can be achieved without increasing mortality or riskinghigher surgical morbidity [25].

Regarding tests for identification of Helicobacter pyloriinfection and treatment, the results of the current study were inagreement with data in the literature, particularly concerningtreatment chronology after the hemorrhagic episode.Helicobacter pylori infection increases the risk of mid- and long-term rebleeding, but does not appear to modify the risk of earlyrebleeding [26].

The only study comparable with ours is a publication fromThe Netherlands reported in 2000 [27]. The main differences inthe results with our study concern the high response rate of theDutch physicians (73% response rate with 90 analyzable ques-tionnaires), the much higher frequency of the epinephrine-polidocanol combination (60%), use of H2 receptor antagonistin one quarter of patients, and the frequency of second proce-dures for rebleeding (76%). It is interesting to note that in TheNetherlands where BPU is managed either by gastroenterolo-gists or internists performing endoscopic procedures, the gastro-enterologists were more “active” in their interventions for ForestIb, IIa and IIb ulcers. An observation also made in our inquiry.The use of the Forrest classification was also similar in the Dutchinquiry (61%) where injections were used as the first intentiontreatment for 93% of patients.

In conclusion, our study, which is the first inquiry of this typeconducted in France, shows that management practices for BPUvary between university and non-university hospitals. The differ-ences probably arise from the different types of institutionalorganization and patients recruitment as well as experienceamong the physicians who responded to the questionnaire.Nevertheless, management practices in France are largely inagreement with the SFED recommendations. More widespreadaccess to anesthesia for emergency endoscopy as well asimproved operator training and more uniform managementpractices for BPU could be beneficial. Further studies based onobjective criteria instead of subjective statements are necessaryto evaluate management practices for this frequent and seriousdisease.

REFERENCES

1. Czernichow P, Hochain P, Nousbaum JB, Raymond JM, Rudelli A,Dupas JL, et al. Epidemiology and course of acute upper gastro-intes-tinal haemorrhage in four French geographical areas. Eur J Gastroen-terol Hepatol 2000;12:175-81.

2. Bourienne A, Pagenault M, Heresbach D, Jacquelinet C, Faroux R,Lejean-Colin I, et al. Etude prospective multicentrique des facteurs

G. Lesur et al.

144

pronostiques des hémorragies ulcéreuses gastro-duodénales. Réeva-luation des facteurs cliniques et endoscopiques à l’ère de l’hémostaseendoscopique. Gastroenterol Clin Biol 2000;24:193-200.

3. Bour B. Technique d’hémostase endoscopique des ulcères gastro-duodénaux. Recommandations de la Société Française d’EndoscopieDigestive. 1999, 27-32.

4. Amouretti M, Czernichow P, Kerjean A, Hochain P, Nousbaum JB,Rudelli A, et al. Prise en charge des hémorragies digestives hautescommunautaires. Trajectoire des malades dans 4 départements fran-çais. Gastroenterol Clin Biol 2000;24:1003-11.

5. Pazart L. Les enquêtes de pratique. In: Matillon Y, Durieux P, eds.L’évaluation médicale : du concept à la pratique. Paris: Médecine-Sciences Flammarion, 1994:43-7.

6. Lesur G, Artru P, Mitry E. Hémorragies digestives ulcéreuses : histoi-re naturelle et place de l’hémostase endoscopique. Gastroenterol ClinBiol 2000;24:656-66.

7. Coffin B, Pocard M, Panis Y, Riche F, Laine MJ, Bitoun A, et al.Erythromycin improves the quality of EGD in patients with acute up-per GI bleeding: a randomized controlled study. Gastrointest Endosc2002;56:174-9.

8. Frossard JL, Spahr L, Queneau PE, Giostra E, Burckhardt B, Ory G,et al. Erythromycin intravenous bolus infusion in acute upper gas-trointestinal bleeding: a randomized, controlled, double-blind trial.Gastroenterology 2002;123:17-23.

9. Saeed ZA, Ramirez FC, Hepps KS, Cole RA, Graham DY. Prospec-tive validation of the Baylor bleeding score for predicting the likeli-hood of rebleeding after endoscopic hemostasis of peptic ulcers.Gastrointest Endosc 1995;41:561-5.

10. Rockall TA, Logan RF, Devlin HB, Northfield TC. Risk assessmentafter acute upper gastrointestinal haemorrhage. Gut 1996;38:316-21.

11. Latrive JP, Guilgot M, Barbare JC, Bensoussan TA, Diab G,Duchmann JC, et al. Trois années d’observation des hémorragiesdigestives hautes en hôpital général. Evaluation du pronostic et despratiques médicales après diffusion de recommandations (abstract).Gastroenterol Clin Biol 2003;27:A15.

12. Bour B, Person B, Cales P, Blanchi A, Burtin P, Oberti F, et al. Inter-observer agreement on endoscopic diagnosis of bleeding peptic ul-cers. Gastrointest Endosc 1997;46:27-32.

13. Lesur G, Bour B. Vers un traitement à la carte des hémorragies diges-tives ulcéreuses? Gastroenterol Clin Biol 2003;27:394-7.

14. Machicado GA, Jensen DM. Thermal probes alone or with epine-phrine for the endoscopic haemostasis of ulcer haemorrhage. Bail-lieres Best Pract Res Clin Gastroenterol 2000;14:443-58.

15. Sung JJ, Chan FK, Lau JY, Yung MY, Leung WK, Wu JC, et al. Theeffect of endoscopic therapy in patients receiving omeprazole forbleeding ulcers with nonbleeding visible vessels or adherent clots: arandomized comparison. Ann Intern Med 2003;139:237-43.

16. Jensen DM, Kovacs TO, Jutabha R, Machicado GA, Gralnek IM,Savides TJ, et al. Randomized trial of medical or endoscopic therapyto prevent recurrent ulcer hemorrhage in patients with adherent clots.Gastroenterology 2002;123:407-13.

17. Bleau BL, Gostout CJ, Sherman KE, Shaw MJ, Harford WV, Keate RF,et al. Recurrent bleeding from peptic ulcer associated with adherentclot: a randomized study comparing endoscopic treatment with medi-cal therapy. Gastrointest Endosc 2002;56:1-6.

18. Lin HJ, Hsieh YH, Tseng GY, Perng CL, Chang FY, Lee SD. A pro-spective, randomized trial of large- versus small-volume endoscopicinjection of epinephrine for peptic ulcer bleeding. Gastrointest Endosc2002;55:615-9.

19. Chung SS, Lau JY, Sung JJ, Chan AC, Lai CW, Ng EK, et al. Ran-domised comparison between adrenaline injection alone and adrena-line injection plus heat probe treatment for actively bleeding ulcers.BMJ 1997;314:1307-11.

20. Gevers AM, De Goede E, Simoens M, Hiele M, Rutgeerts P. A rand-omized trial comparing injection therapy with hemoclip and with in-jection combined with hemoclip for bleeding ulcers. GastrointestEndosc 2002;55:466-9.

21. Chung IK, Ham JS, Kim HS, Park SH, Lee MH, Kim SJ. Comparisonof the hemostatic efficacy of the endoscopic hemoclip method withhypertonic saline-epinephrine injection and a combination of the twofor the management of bleeding peptic ulcers. Gastrointest Endosc1999;49:13-8.

22. Chau CH, Siu WT, Law BK, Tang CN, Kwok SY, Luk YW, et al.Randomized controlled trial comparing epinephrine injection plusheat probe coagulation versus epinephrine injection plus argon plas-ma coagulation for bleeding peptic ulcers. Gastrointest Endosc2003;57:455-61.

23. Cipolletta L, Bianco MA, Rotondano G, Piscopo R, Prisco A,Garofano ML. Prospective comparison of argon plasma coagulatorand heater probe in the endoscopic treatment of major peptic ulcerbleeding. Gastrointest Endosc 1998;48:191-5.

24. Lau JY, Sung JJ, Lee KK, Yung MY, Wong SK, Wu JC, et al. Effectof intravenous omeprazole on recurrent bleeding after endoscopictreatment of bleeding peptic ulcers. N Engl J Med 2000;343:310-6.

25. Lau JY, Sung JJ, Lam YH, Chan AC, Ng EK, Lee DW, et al. Endo-scopic retreatment compared with surgery in patients with recurrentbleeding after initial endoscopic control of bleeding ulcers. N Engl JMed 1999;340:751-6.

26. Schilling D, Demel A, Nusse T, Weidmann E, Riemann JF. Helico-bacter pylori infection does not affect the early rebleeding rate in pa-tients with peptic ulcer bleeding after successful endoscopic hemostasis:a prospective single-center trial. Endoscopy 2003;35:393-6.

27. van Leerdam ME, Rauws EA, Geraedts AA, Tytgat GN. Managementin peptic ulcer hemorrhage: a Dutch national inquiry. Endoscopy2000;32:935-42.