Original Contribution

High-risk ED patients with nonvariceal Upper gastrointestinal hemorrhage undergoing emergency or urgent endoscopy: a retrospective analysis

Chi-Ming Tai MD^a, Shih-Pei Huang MD, PhD^b, Hsiu-Po Wang MD^c, Tsung-Chun Lee MD^d, Chi-Yang Chang MD^a, Chia-Hung Tu MD^a, Ching-Tai Lee MD^a, Tsung-Hsien Chiang MD^a,

Jaw-Town Lin MD, PhD^b, Ming-Shiang Wu MD, PhD^b,*

^aDepartment of Internal Medicine, E-Da Hospital/I-Shou University, Kaohsiung 824, Taiwan

^bDepartment of Internal Medicine, National Taiwan University Hospital and College of Medicine, National Taiwan University,

Taipei 100, Taiwan

^cDepartment of Emergency Medicine, National Taiwan University Hospital and College of Medicine, National Taiwan University, Taipei 100, Taiwan

^dDepartment of Internal Medicine, National Taiwan University Hospital Yun-Lin Branch, Yun-Lin 640, Taiwan

Received 13 June 2006; revised 27 June 2006; accepted 2 July 2006

Abstract

Objectives: The optimal timing of interventional endoscopy within the initial 24 hours remains controversial. We designed a retrospective study to compare the outcomes between Emergency endoscopy (EE) and urgent endoscopy (UE) for high-risk patients with nonvariceal upper gastrointestinal hemorrhage presenting to the emergency department (ED).

Methods: The medical records of 189 patients with nonvariceal upper gastrointestinal hemorrhage who underwent endoscopy within 24 hours of admission to the ED were reviewed. Patients were divided into 2 groups: EE group (b8 hours) or UE group (8-24 hours). We compared the endoscopic findings, hemostatic procedures, rate of hemostasis, rebleeding, need for transfusion, length of hospitalization, and mortality between the 2 groups.

Results: There were 88 patients (47%) in the EE group and 101 patients (53%) in the UE group. Ulcers with active bleeding or exposed vessel were found more frequently in the EE group than in the UE group (19% vs 8%, P = .03; 34% vs 12%, P b .001). Fifty patients had blood retention in the stomach, especially in the EE group (40% vs 15%, P b .001). Forty-four (50%) patients in the EE group and 21 (21%) patients in the UE group received endoscopic interventions. Combination modalities of endoscopic hemostasis were more commonly used in the EE group than in the UE group (40% vs 15%, P b .001). Primary hemostasis was achieved at a rate of 95% in both groups. There was no statistical difference regarding the rate of recurrent bleeding, total amount of transfusion, length of hospital stay, and mortality rate in both groups.

* Corresponding author. Tel.: +886 2 23123456×5410; fax: +886 2 23947899.

E-mail address: [email protected] (M.-S. Wu).

0735-6757/$ – see front matter D 2007 doi:10.1016/j.ajem.2006.07.014

Conclusions: Although more active lesions were detected and more therapeutic attempts were performed in the EE group, the outcome showed no difference in both groups. Emergency endoscopy performed less than 8 hours after arrival to the ED showed no definite benefit in comparison with UE performed within 8 to 24 hours.

D 2007

Introduction

Upper gastrointestinal hemorrhage (UGIH) is commonly encountered in emergency medical practice. Despite consid- erable advances in endoscopic hemostatic modalities and pharmacologic treatment, UGIH still causes significant mor- bidity and mortality, as well as substantial financial costs. The annual incidence of hospital admissions for UGIH in the United States and Europe is 0.1%, with a mortality rate of 10% [1].

Esophagogastroduodenoscopy plays a key role in both diagnosis and treatment of UGIH. It identifies the site of bleeding in more than 95% of cases of UGIH [2]. Currently, endoscopy within the initial 24 hours is the standard of therapy for the management of UGIH [3]. However, there is no general agreement on the definition of bearlyQ regarding endoscopy in UGIH. It varies widely among studies, ranging from 2 to 24 hours after presentation to the emergency department (ED). Choudari and Palmer [4] reported no difference in endoscopic completeness, findings, injection intervention, and outcome for patients under- going bearlyQ (b6 hours), bintermediateQ (N6-12 hours), and b delayedQ (N12-24 hours) endoscopy. Lin et al [5] found that bearlyQ endoscopy (b12 hours) resulted in decreased demand of blood transfusion only for patients with bloody nasogas- tric aspirate. On the other hand, Yen et al [6] found that bearlyQ endoscopy (b2 hours) resulted in more frequent and significant Oxygen desaturation during the procedure.

A practical problem is that immediate interventional endoscopy is not readily available at anytime in most hospitals. Because the best timing and the benefit of endoscopy within the first 24 hours after admission are undetermined, we reviewed the medical records of high-risk patients with UGI bleeding who presented to the ED to understand the differences of endoscopic findings, hemo- static procedures, need for transfusion, and outcomes bet- ween emergency endoscopy (EE) (b8 hours) and urgent endoscopy (UE) (N8-24 hours).

Methods

From July 2004 to December 2004, 406 patients presenting to our ED underwent EGD for hematemesis and/or melena. High-risk patients were defined by age more than 60 years, severe comorbidity, active bleeding (wit- nessed hematemesis, red blood per nasogastric tube, hematochezia), hypotension or shock, red blood cell transfusion of more than 6 U, and severe coagulopathy [7]. A total of 189 patients who underwent EGD within

24 hours and met the previously mentioned criteria for the high-risk group were enrolled in this retrospective study.

Inclusion criteria were as follows: (a) age at least 18 years; (b) clinical suspicion of UGIH demonstrated by hematemesis, bcoffee groundQ emesis, melena, or hema- tochezia; (c) EGD performed within 24 hours of arrival at the ED; (d) nonvariceal UGIH demonstrated by EGD; and

(e) patients with features of high-risk group. Exclusion criteria were as follows: (a) UGIH developed during hospitalization; (b) transferred from other hospital; (c) bleeding related to varices or tumors; (d) neither blood nor bleeder detected by EGD; (e) only minor lesions (gastritis or erosions) detected by EGD; and ( f) incomplete or unavailable chart record.

The optimal timing of interventional endoscopy within the initial 24 hours remains undetermined. Because the working hours are about 8 hours per day, we chose 8 hours as the cut point. Emergency endoscopy was defined as endoscopic examination performed within 8 hours of arrival at the ED. Urgent endoscopy was defined as endoscopic examination performed between 8 and 24 hours of arrival at the ED.

Endoscopic therapy included diluted epinephrine (1:10000) injection, heater-probe thermocoagulation, hemo- clip placement, argon plasma coagulation (APC), and alcohol injection. For lesions with active bleeding or large exposed vessels, we used combined interventional therapy of diluted epinephrine injection followed by heater-probe thermocoagulation, hemoclip placement, or APC. For small exposed vessel, single modality was chosen to arrest the hemorrhage. For lesions that are difficult to approach, only diluted epinephrine injection was used. Successful primary hemostasis was defined as endoscopically verified cessation of bleeding for at least 1 minute after the first endoscopic hemostasis. Recurrent bleeding was defined as fresh hematemesis or hematochezia, instability of vital sign, bloody nasogastric drainage, and reduction in hemoglo- bin level in excess of 2 g/dL within 24 hours after successfully achieving primary hemostasis [8]. Treatment of recurrent bleeding included repeated EGD for most patients and, far less commonly, angiography and surgi- cal intervention. Surgical intervention and angiography were considered only for patients with unstable and deteriorated hemodynamics.

Statistical analysis

Analyses were based on the categorization of the 2 groups of EE and UE. Factors including demographic

	189	88	101
Age	65.6 F	64.0 F	67.0 F 14.6	.139
(y, mean F SD)	14.6	14.5
Sex
Male-female	119:70	57:31	62:39	.653
Comorbidity
Cardiac (%)	47 (25)	24 (27)	23 (23)	.503
Pulmonary (%)	4 (2)	1 (1)	3 (3)	.625
Hepatic (%)	20 (11)	6 (7)	14 (14)	.155
Renal (%)	10 (5)	5 (6)	5 (5)	1.000
Neurologic (%)	17 (9)	9 (10)	8 (8)	.618
Coagulopathy (%) 19 (10) 8 (9) Medication			11 (11)	.810
NSAID (%) 59 (31) 31 (35)			28 (28)	.276

characteristics, clinical characteristics, endoscopic findings, therapeutic modalities, and outcomes between the 2 groups were compared. Categorical data were analyzed by Fisher exact test, and continuous data were compared by Mann- Whitney U test. All tests were 2-tailed, and a P value less than 0.05 was considered statistically significant. All analyses were performed by SPSS 11.0 version (Chicago, IL) for WINDOWS.

Table 1 Demographic characteristics

Total, n (%) EE, n (%) UE, n (%) P

NSAID indicates nonsteroidal anti-inflammatory drug.

	189	88	101
Bleeder
Duodenal	83 (44)	42 (48)	41 (41)	.379
ulcer (%)
Gastric	72 (38)	33 (38)	39 (39)	.882
ulcer (%)
MWS (%)	13 (7)	6 (7)	7 (7)	1.000
Esophageal	8 (4)	4 (5)	4 (4)	1.000
ulcer (%)
SRMD (%)	6 (3)	0 (0)	6 (6)	.031
Hemorrhagic	1 (1)	0 (0)	1 (1)	1.000
gastritis (%)
Dieulafoy	1 (1)	1 (1)	0 (0)	.466
ulcer (%)
Angiodysplasia	1 (1)	0 (0)	1 (1)	1.000
(%)
Bleeding stigmata Active bleeding	25 (13)	17 (19)	8 (8)	.030
(%) Exposed vessel	42 (22)	30 (34)	12 (12)	b.001
(%)
Adherent clot	7 (4)	3 (3)	4 (4)	1.000
(%)
Ulcer oozing (%)	11 (6)	8 (9)	3 (3)	.117
Flat spots (%)	23 (12)	11 (13)	12 (12)	1.000
No bleeding 81 (43)		19 (22)	62 (61)	b.001

Results

There were 88 patients (47%) in the EE group and 101 patients (53%) in the UE group enrolled for analysis. Demographic characteristics are shown in Table 1. Age, sex, comorbidity, and use of medication did not differ between

Table 2 Clinical characteristics

Total, n (%) EE, n (%) UE, n (%) P

189 88 (47) 101 (53)

Major presentation

both groups. Clinical characteristics, including bloody emesis, coffee ground emesis, melena, hematochezia, systo- lic blood pressure, heart rate, and hematocrit also showed no significant difference (Table 2).

The most frequently encountered endoscopic diagnoses were Duodenal ulcer (44%) and gastric ulcer (38%), but their frequencies between the EE and UE groups were not significantly different. Stress-related mucosal disease was detected less frequently in the EE group compared with the UE group (0% vs 6%, P = .03), but the case number was small.

Stages of ulcers were classified according to the modified Forrest classification [7]. Ulcers with active bleeding or

Table 3 Endoscopic findings and therapeutic modalities

Total, n (%) EE, n (%) UE, n (%) P

therapy (%)

MWS indicates Mallory-Weiss syndrome; SRMD, stress-related mucosal disease.

N100 (%) 79 (42) Hematocrit (%)	40 (45)	39 (39)	.377
b30 136 (72)	66 (75)	70 (69)	.420

Bloody emesis (%)	19 (10)	10 (11)	9 (9)	.633	stigmata Blood retention (%)	50 (26)	35 (40)	15 (15)	b.001
Coffee ground	48 (25)	20 (23)	28 (28)	.504	Single modality (%)	30 (16)	16 (18)	14 (14)	.433
emesis (%)					Diluted epinephrine	17 (9)	9 (10)	8 (8)	.444
Melena	119 (63)	55 (63)	64 (63)	1.000	injection (%)
(only) (%)					Heater-probe	3 (2)	2 (2)	1 (1)	.599
Hematochezia (%)	3 (2)	3 (3)	0 (0)	.099	thermocoagulation
Systolic blood					(%)
pressure					Hemoclip (%)	7 (4)	4 (5)	3 (3)	.707
(mm Hg) b100 (%)	15 (8)	4 (5)	11 (11)	.176	Alcohol injection (%)	2 (1)	1 (1)	1 (1)	1.000
Heart rate (beats/min)					APC (%) 1 (1) 0 (0) 1 (1) 1.000 Combination 35 (19) 28 (32) 7 (7) b.001

exposed vessel were found more frequently in the EE group than in the UE group (19% vs 8%, P = .03; 34% vs 12%, P b .001). On the other hand, ulcers without active or recent bleeding stigmata were found more frequently in the UE group than in the EE group (22% vs 61%, P V .001) (Table 3). Fifty patients had blood retention in the stomach, especially in the EE group (40% vs 15%, P b .001).

A total of 65 patients (35%) received endoscopic therapy. Forty-four patients (50%) in the EE group and 21 patients (21%) in the UE group underwent endoscopic therapy. Treatment with single modality showed no significant difference in both groups. Diluted epinephrine injection was the most commonly used method. Combination modalities were more commonly used in the EE group than in the UE group (40% vs 15%, P b .001), and diluted epinephrine injection was most commonly used to combine with other modalities (Table 3).

Outcomes of 65 EGDs with therapeutic attempt are shown in Table 4. The rate of primary hemostasis was high in both groups (95%). Among the 3 patients with persistent bleeding after the first endoscopy, 2 patients belonged to the EE group and received single modality, and 1 belonged to the UE group and received combination modalities. Five patients (8%) had rebleeding, including 3 in the EE group and 2 in the UE group (7% vs 10%, P = .654). One patient in either group underwent second EGD; hemostasis was achieved in the patient of the EE group. Three other patients with rebleeding received angiography.

Transfusion was carried out in 144 patients (76%). The transfusion rate and total amount of transfusion were not significantly different between these 2 groups (3.5 F 4.8 U vs 3.4 F 3.9 U, P = .765) (Table 5). All 6 patients receiving second modality underwent angiography. The lengths of hospital stay were similar in both groups (5.1 F 5.0 days vs 6.0 F 7.7 days , P = .522). In our series, 7 pa- tients eventually died. One in the UE group died of continuous bleeding, and surgery was not performed

Table 4 Outcomes of 65 EGDs with therapeutic attempt

2nd hemostasis (%)

Total (n = 65), n (%)		EE (n = 44), n (%)	UE (n = 21), n (%)	P
Primary	62 (95)	42 (95)	20 (95)	1.000
hemostasis (%)
Persistent	3 (5)	2 (5)	1 (5)	1.000
bleeding (%)
One modality	28 (43)	14 (32)	14 (67)	.015
(%)
Combination	34 (52)	28 (64)	6 (29)	.016
therapy (%)
Recurrent	5/62 (8)	3/42 (7)	2/20 (10)	.654
bleeding (%)
2nd EGD (%)	2/5 (40)	1/3 (33)	1/2 (50)	1.000
Successful	1/2 (50)	1/1 (100)	0/1 (0)	–

Total (%)	144 (76)	65 (74)	79 (78)	.499
Total (U)	3.5 F 4.3	3.5 F 4.8	3.4 F 3.9	.765
Need for 2nd modality	6 (3)	2 (2)	4 (4)	.687
Angiography (%)	6 (3)	2 (2)	4 (4)	.687
Surgery (%)	0 (0)	0 (0)	0 (0)	–
Days in hospital	6.1 F	6.3 F	6.0 F	.440
(d) Days in hospitala	10.1 5.6 F 6.6	12.4 5.1 F 5.0	7.7 6.0 F 7.7	.522
(d)
Inhospital mortality	7 (4)	1 (1)	6 (6)	.124
(%)

because of comorbidity. The other 6 patients died of Severe comorbidities, including pneumonia with respiratory failure or septic shock.

Table 5 Outcomes of total cases

Total, n (%)

189

Transfusion requirements

EE,

n (%)

88

UE, P

n (%)

101

^a One patient in the emergency group with hospitalization for 112 days was excluded.

Discussion

In the present study, we find that the outcomes including transfusion rate, length of hospitalization, and mortality rate were not different in both the EE and UE groups of high-risk UGIH patients, although more active lesions were detected and more therapeutic approaches were attempted in the EE group. For acute nonvariceal UGIH, EGD remains the most effective modality for diagnosing and treating most causes of UGIH [7]. Most of the earlier studies confirmed the advantage of early endoscopy, but data are limited regarding whether much earlier therapeutic endoscopy within the initial 24 hours can offer more benefit for UGIH patients [9]. However, a practical problem is that emergency EGD within 8 hours may not be readily available in many hospitals. The reasons for this condition included lack of consensus due to controversial results in this issue, limited therapeutic endoscopists, the need for time to stabilize hemodynamics before endoscopy, and the need for time to correct underlying comorbidities for high-risk UGIH patients. Our retrospective study does not support the need for endoscopy within 8 hours of ED presentation in high- risk patients with nonvariceal, nontumor UGIH.

Peptic ulcer disease represents the most common cause of UGIH, accounting for a third to half of all episodes in earlier report [10]; but 82% of UGIH cases were noted in this study. The possible explanation may be that we enrolled patients of high-risk, nonvariceal UGIH only. In addition, gastritis and erosions are not responsible for major, life- threatening hemorrhage because they are breaks that are confined to the mucosa and no blood vessels of significant

size are present in the mucosa [11]. Therefore, they were also excluded from this study. For these reasons, the percentage of peptic ulcers was higher than that in the previous study.

In our series, more actively bleeding lesions and more blood were retained in the EE group, which may cause EGD to be more difficult to complete. In addition, these will increase the risk of EGD and necessitate more therapeutic modalities. The most bothering problem regarding performing emergency EGD is blood retained in the stomach. It may obscure the endoscopic view or lead to potential bronchoaspiration. Furthermore, repeated endos- copy must be performed in 5% to 12% of cases [12-14]. Approximately 80% of all UGIH episodes stop bleeding spontaneously, and the amount of blood left in the stomach will decrease after spontaneous hemostasis [15,16]. Johnston [16] proposed the evolutionary scheme of bleeding ulcer to clean-based ulcer, and Lau et al [17] found that completion of the process generally takes less than 72 hours. Therefore, less actively bleeding lesions and less blood retention may be detected if EGD was performed later, as seen in our UE group.

In the EE group, more combined modalities were used to treat the more active lesions. Endoscopic therapy for bleeding ulcers has been demonstrated to be effective in patients with high-risk endoscopic findings (active bleeding, nonbleeding visible vessels, and, perhaps, adherent clots) in decreasing rebleeding, blood transfusion, duration of hospital stay, mortality, and cost [18]. Endoscopic therapy can be performed by a variety of methods, such as injection, thermal coagulation, and clip placement. They are generally considered equivalent, with a combination of methods being superior to an individual therapy [3]. It is reasonable to use combination modalities to treat high-risk stigmata. However, although more combined modalities were used in the EE group, the outcomes showed no significant difference. From this viewpoint, EE may be more costly; but it does not benefit high-risk patients more when compared with UE.

The transfusion rate and total amount of transfusion showed no significant difference in both groups. Because spontaneous hemostasis was noted in most patients (87%) when EGD was performed, the total transfusion amount may show no difference in both groups. Although 7 patients (4%) died in the hospital in this study, only 1 patient in the UE group (1%) died of continuous bleeding; and surgery was not performed because of comorbidity. The death of the remaining 6 patients was due to severe comorbidities and was not directly related to bleeding. Therefore, bleeding-related inhospital mortality rate remained low in our study.

Because of the advance of therapeutic modalities and the additional use of proton-pump inhibitors (PPIs) after endoscopy, the rate of primary hemostasis was high and the recurrent bleeding rate was approximately 10% in an earlier report [19]. We also found that the recurrent bleeding

rate was 8% in our high-risk patients and that there was no difference between the 2 groups. Patients with recurrent bleeding respond favorably to repeated endoscopic therapy [20], but the location of bleeders may influence the result of endoscopic therapy. In our series, 2 of 5 patients with recurrent bleeding received repeated EGD, and the remain- ing 3 received angiography because the location of the lesion is difficult for the endoscopic therapy approach.

In interpreting the current findings, several potential limitations should be noted. First, in this study, we focused on the difference between EE and UE to patients with definite and significant hemorrhage. By definition, gastritis and erosions cannot be responsible for major hemorrhage because they are breaks that are confined to the mucosa and no blood vessels of significant size are present in the mucosa. Besides, the diagnosis of gastritis or erosions by EGD in some patients was not the cause of bleeding. Therefore, we excluded these minor lesions that might not cause significant hemorrhage or might not be the cause of hemorrhage. Besides, the percentage of bleeding caused by gastritis or erosions that belonged to the high-risk patients was low (b3%) in our database. Second, the role of PPIs was not discussed here. Proton-pump inhibitor therapy was warranted in all patients with UGIH severe enough to require endoscopic therapy; and PPI therapy was recom- mended in patients with suspected Peptic ulcer bleeding associated with hemodynamic instability, patients in whom endoscopic evaluation is delayed or unavailable, and/or those who require blood transfusion [21]. Actually, some patients in the EE groups who could receive EGD within 1 to 2 hours did not receive PPI before EGD. Although the percentage was low, we did not know whether this would influence the outcome. Third, because the medical or EGD records did not completely record the complication includ- ing aspiration, oxygen desaturation, or hypotension during EGD, the rate of EGD-related complication, which might influence the outcomes, was not discussed in this study.

In conclusion, EE performed less than 8 hours after admission showed no definite benefit in comparison with UE performed within 8 to 24 hours. More blood retention in the EE group may reduce the completeness and increase the risk of EGD, although more patients underwent combined endoscopic therapeutic modalities. However, further prospective randomized studies should be taken to verify the benefits and costs of early EGD for high-risk UGIH patients.

References

1. Braunwald E, Fauci AS, Kasper DL, editors. Harrison’s principles of internal medicine. 15th ed. New York7 McGraw-Hill; 2001. p. 252.
2. Peterson WL, Barnett CC, Smith HJ, et al. Routine early endoscopy in upper-gastrointestinal-tract bleeding: a randomized, controlled trial. N Engl J Med 1981;304:925 – 9.
3. Barkun A, Bardou M, Marshall JK, Nonvariceal Upper GI Bleeding Consensus Conference Group. Consensus recommendations for

managing patients with nonvariceal upper gastrointestinal bleeding. Ann Intern Med 2003;139:843 – 57.

1. Choudari CP, Palmer KR. Timing of endoscopy for severe peptic ulcer hemorrhage: out of hours emergency endoscopy is unnecessary. [abstract]Gastroenterology 1993;104:A55.
2. Lin HJ, Wang K, Perng CL, et al. Early or delayed endoscopy for patients with peptic ulcer bleeding. A prospective randomized study. J Clin Gastroenterol 1996;22:267 – 71.
3. Yen D, Hu S, Chen L, et al. Arterial oxygen desaturation during emergent nonsedated Upper gastrointestinal endoscopy in the emer- gency department. Am J Emerg Med 1997;15:644 – 6.
4. ASGE Standards of Practice Committee. ASGE guideline: the role of endoscopy in acute non-variceal upper-GI hemorrhage. Gastrointest Endosc 2004;60:497 – 504.
5. Lee YC, Wang HP, Wu MS, et al. Urgent bedside endoscopy for clinically significant upper gastrointestinal hemorrhage after admission to the intensive care unit. Intensive Care Med 2003; 29:1723 – 8.
6. Spiegel BM, Vakil NB, Ofman JJ. Endoscopy for acute nonvariceal upper gastrointestinal tract hemorrhage: is sooner better? A systematic review. Arch Intern Med 2001;161:1393 – 404.
7. Rockall TA, Logan RF, Devlin HB, et al. Incidence of and mortality from acute upper gastrointestinal haemorrhage in the United Kingdom. Steering Committee and members of the National Audit of Acute Upper Gastrointestinal Haemorrhage. BMJ 1995;311:222 – 6.
8. Laine L. Treatment of nonvariceal upper GI hemorrhage: what’s happening in the real world. Am J Gastroenterol 2004;99:1247 – 9.
9. Choudari CP, Rajgopal C, Elton RA, et al. Failures of endoscopic therapy for bleeding peptic ulcer: an analysis of risk factors. Am J Gastroenterol 1994;89:1968 – 72.
10. Brullet E, Calvet X, Campo R, et al. Factors predicting failure of endoscopic injection therapy in bleeding duodenal ulcer. Gastrointest Endosc 1996;43:111 – 6.
11. Laine L, Peterson WL. Bleeding peptic ulcer. N Engl J Med 1994; 331:717 – 27.
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. Johnston JH. Endoscopic risk factors for bleeding peptic ulcer. Gastrointest Endosc 1990;36:S16-S20.
14. Lau JY, Chung SC, Leung JW, et al. The evolution of stigmata of hemorrhage in bleeding peptic ulcers: a sequential endoscopic study. Endoscopy 1998;30:513 – 8.
15. Cook DJ, Guyatt GH, Salena BJ, et al. Endoscopic therapy for acute nonvariceal upper gastrointestinal hemorrhage: a meta-analysis. Gastroenterology 1992;102:139 – 48.
16. Spiegel BM, Ofman JJ, Woods K, et al. Minimizing recurrent peptic ulcer hemorrhage after endoscopic hemostasis: the cost-effectiveness of competing strategies. Am J Gastroenterol 2003;98:86 – 97.
17. Saeed ZA, Cole RA, Ramirez FC, et al. Endoscopic retreatment after successful initial hemostasis prevents ulcer rebleeding: a prospective randomized trial. Endoscopy 1996;28:288 – 94.
18. Erstad BL. Proton-pump inhibitors for acute peptic ulcer bleeding. Ann Pharmacother 2001;35:730 – 40.