Outcomes in Severe Upper GI Hemorrhage from Dieulafoy's Lesion with Monitoring of Arterial Blood Flow.

BACKGROUND: Dieulafoy's lesion (DL) is a rare but increasingly recognized cause of severe upper GI hemorrhage (SUGIH). There is little consensus regarding the endoscopic approach to management of bleeding from DL. AIMS: Our purposes were to compare 30-day outcomes of patients with SUGIH from DL with Doppler endoscopic probe (DEP) monitoring of blood flow and guided treatment versus standard visually guided hemostasis (VG). METHODS: Eighty-two consecutive DL patients with SUGIH were identified in a large CURE Hemostasis database from previous prospective cohort studies and two recent RCTs at two university-based medical centers. 30-day outcomes including rebleeding, surgery, angiography, death, and severe medical complications were compared between the two treatment groups. RESULTS: 40.2% of DL bleeds occurred in inpatients. 43.9% of patients had cardiovascular disease, and 48.7% were taking medications associated with bleeding. For the entire cohort, 41.3% (26/63) of patients treated with VG had a composite 30-day outcome as compared to 10.5% (2/19) of patients treated with DEP (p = 0.017). Rebleeding occurred within 30 days in 33.3% and 10.5% of those treated with VG and DEP, respectively (p = 0.051). After propensity score matching, the adjusted 30-day composite outcome occurred in 39.0% in the VG group compared to 2.6% in the DEP group (p < 0.001). Adjusted 30-day rebleeding occurred in 25.3% in the VG group versus 2.6% in the DEP group (p < 0.001). DISCUSSION: DL patients with SUGIH were frequently inpatients and had severe cardiovascular comorbidities and recurrent bleeding. Lesion arterial blood flow monitoring and obliteration are an effective way to treat bleeding from DL which reduces negative 30-day clinical outcomes.

Severe upper gastrointestinal hemorrhage from linear gastric ulcers in large hiatal hernias: a large prospective case series of Cameron ulcers.

We report a case series of all consecutive patients hospitalized in our two tertiary referral medical centers over the past 17 years for Cameron ulcers causing severe upper gastrointestinal hemorrhage (GIH) or severe obscure GIH. Cameron ulcers were diagnosed in 25 of the 3960 screened patients with severe upper GIH or severe obscure GIH (0.6 %). Of these, 21 patients had a prospective follow-up (median time 20.4 months [interquartile range: 8.5 - 31.8]). Patients were more often elderly women with chronic anemia, always had large hiatal hernias, and were usually referred for obscure GIH. Twelve of the 21 patients (57 %) were referred for surgery while being treated with high-dose proton pump inhibitors (PPIs). The other 9 patients (43 %) continued PPIs without any rebleeding during the follow-up. Cameron ulcers in large hiatal hernias are an uncommon cause of severe upper GIH. The choice of medical vs. surgical therapy should be individualized.

The significance of gastric and duodenal histological ischemia reported on endoscopic biopsy.

Although frequently reported, it is unknown whether pathological reports of ischemia obtained from gastroduodenal biopsies suggest a diagnosis, prognosis or a requirement for additional evaluation. The aim of this study was to review the natural history, clinical presentation, endoscopic appearance, treatments, and major clinical outcomes of patients with gastroduodenal ischemia. A case series of 14 patients with variable etiologies (seven gastric and seven duodenal) was obtained from a search of our endoscopic pathological database for reports of histological ischemia. The results were as follows. The most common presentation was upper gastrointestinal bleeding (71 %). Half of the endoscopic lesions appeared very severe (large or circumferential lesions, exudative, pseudomembranous, black or pale mucosa). There were six cases of rebleeding (43 %) and four deaths (29 %). Computed tomography scanning was frequently used (12 cases, 86 %), but led to an underlying diagnosis in only three cases. In our series, patients with underlying vascular pathology have substantial 6-month mortality (29 %).

Esophageal capsule endoscopy for evaluation of patients with chronic gastroesophageal reflux symptoms: findings and its image quality.

Esophageal capsule endoscopy (ECE) may offer an alternative approach to visualize esophageal lesions associated with gastroesophageal reflux (GER) disease. The objective of this study was to report the ECE findings in patients with GER symptoms and validate a new scoring system to assess ECE video quality. Five hundred two ECE were performed in patients with GER symptoms. We devised a new grading scale called ECE Utility score to assess the quality of images using five different parameters: anatomic landmarks visualized, esophageal transit time, image quality, illumination, and artifacts. The ECE cases were independently scored by two interpreters in a randomized, blinded fashion. Reflux esophagitis was diagnosed via ECE in 254 patients (50.5%). We identified 12 cases (2.4%) with suspected Barrett's esophagus and all of them had endoscopic evidence of Barrett's esophagus on esophagogastroduodenoscopy. Histologic confirmation Barrett's esophagus was found in six patients and dysplasia was found in one patient. From the 502 cases, mean ± standard deviation total ECE Utility score was 8.89 ± 0.96 for interpreter 1 and 8.96 ± 0.93 for interpreter 2. The concordance rate between the two interpreters for the ECE Utility score ranged from 75.9-96.8% across the parameters and the Pearson correlation rate of the total score was 0.81. ECE is shown to be a simple noninvasive valuable technique for evaluating esophageal mucosa and producing high quality images in patients with GER symptoms. ECE can help as an alternative screening tool for diagnosing Barrett's esophagus.

Endothelial viral inclusions in Fuchs' corneal dystrophy.

A case of Fuchs' corneal dystrophy is presented. The patient, a 70-year-old white woman, had bilateral decreasing vision, especially while reading and driving at night. Clinical features were characteristic of Fuchs' corneal dystrophy. Ultrastructural studies demonstrated findings in Descemet's membrane typical of those previously reported with numerous tactoids of fibrous long-spacing collagen in the posterior collagenous layer and in the guttata. In the endothelial cells were encapsulated ellipsoid viral particles, 400 nm long and 225 nm in diameter. They contained an outer and inner membrane with an electron-dense intervening region and a central dense core. Nucleocapsids were present in the endothelial cells and stromal keratocytes. The authors suggest that the pathogenesis of Fuchs' corneal dystrophy may be endothelial damage, and that in this case, the etiology is a viral infection.

Management of upper gastrointestinal bleeding in the patient with chronic liver disease.

This article reviews the management of severe upper gastrointestinal bleeding in the patient with chronic liver diseases. The initial assessment, diagnostic work-up, and treatment options for variceal and nonvariceal bleeding are discussed. The role of diagnostic and therapeutic endoscopy for esophagogastric varices is reviewed with special emphasis on new endoscopic techniques including variceal band ligation and cyanoacrylate injection. Various pharmacologic, surgical, and radiologic treatment options for variceal bleeding also are discussed. In addition, nonvariceal causes of severe upper gastrointestinal bleeding are reviewed including peptic ulcer diseases, Mallory-Weiss tear, portal hypertensive gastropathy, and gastric antral vascular ectasia.

Novel external sensor array for capsule endoscopy.

Capsule endoscopy (CE) requires placement of an eight-lead sensor array over the abdomen that receives image and localization data transmitted from the capsule. The current process of applying this array to the skin has several disadvantages: firstly, it is time-consuming; secondly, it can be cumbersome for the patient; and thirdly, it is often difficult to place the leads consistently. A new external sensor array system designed to improve this process was tested. It was hypothesized that the new method would be able to receive the transmitted data adequately during CE. The new method and device were tested on an in-patient who had two sets of sensor arrays, batteries, and data recorders placed on her simultaneously. One set was placed in the standard fashion, which served as the control, while the other set was placed using the novel external method. The data provided by the two recorders were compared and the patient's preferences were noted. The quality of the CE images provided by the two methods was identical, but the CE localization tracings were different, presumably due to movement of the gown and leads during the recording period. No signal interference was noted. The patient preferred the external device. A prepositioned external sensor array is capable of transmitting CE data without any loss in image quality, but the localization tracing differs due to movement of the external array. This new method will require testing with a larger sample size in an outpatient setting to allow full assessment of the clinical value of the new approach.

Push enteroscopy for recurrent gastrointestinal hemorrhage due to jejunal anastomotic varices: a case report and review of the literature.

Small-bowel anastomotic and adhesion-related varices can form within adhesions in the setting of mesenteric venous hypertension, arising from either mesenteric venous obstruction or portal hypertension. In evaluating gastrointestinal bleeding in patients who have had previous abdominal surgery and mesenteric venous hypertension, small-bowel anastomotic varices and adhesion-related varices should be considered. For patients with recurrent, severe melena or hematochezia, we recommend that the initial diagnostic work-up should include push enteroscopy in patients with previous small-bowel surgery. Retrograde ileoscopy should also be considered these patients to look for distal small-bowel varices. Potentially, such small-bowel varices can be identified by wireless capsule endoscopy. We report a case of recurrent gastrointestinal bleeding caused by jejunal anastomotic varices which were secondary to superior mesenteric vein occlusion following an abdominal gunshot wound. Although the treatment of segmental varices has been surgical resection, for patients with overt systemic portal hypertension, a transjugular intrahepatic portal-systemic shunt or a decompressive shunting procedure are recommended.

Randomized controlled studies of injection Gold Probes compared with monotherapies for hemostasis of bleeding canine gastric ulcers.

BACKGROUND: There is a significant interest in combination therapy using endoscopic epinephrine injection and thermal coagulation for nonvariceal hemostasis. The purpose of the study was to compare the relative effectiveness, ease of use, and safety of new Injection Gold Probes to other hemostasis techniques in three randomized, controlled laboratory studies of bleeding canine gastric ulcers. METHODS: Fifteen dogs with prehepatic portal hypertension were heparinized and bleeding gastric ulcers were induced with jumbo biopsy forceps. Three different prototypes of Injection Gold Probes were compared with monotherapy (thermal, electrocoagulation, or epinephrine injection alone), control, or combination therapy with separate injector and thermal probes. The treatment times, total number of pulses or injections, volume of epinephrine injected, and ease of applications were recorded. Gastric ulcer size, ulcer healing, and complications were evaluated at 1 and 4 weeks. RESULTS: All endoscopic treatments were effective for acute hemostasis compared with control. Thermal coagulation alone was the fastest treatment to perform. The performance of the first Injection Gold Probe prototype was restricted by its small-gauge needle. The second and third Injection Gold Probe prototypes had a larger-gauge needle and irrigation channel which made them faster and easier to use than separate injection catheters and thermal probes. CONCLUSIONS: The advantages of Injection Gold Probes were the ability to irrigate, inject, and coagulate without probe removal. Combination therapy did not increase treatment-related complications compared with monotherapies.

Urgent colonoscopy for the diagnosis and treatment of severe diverticular hemorrhage.

BACKGROUND: Although endoscopy is often used to diagnose and treat acute upper gastrointestinal bleeding, its role in the management of diverticulosis and lower gastrointestinal bleeding is uncertain. METHODS: We studied the role of urgent colonoscopy in the diagnosis and treatment of 121 patients with severe hematochezia and diverticulosis. All patients were hospitalized, received blood transfusions as needed, and received a purge to rid the colon of clots, stool, and blood. Colonoscopy was performed within 6 to 12 hours after hospitalization or the diagnosis of hematochezia. Among the first 73 patients, those with continued diverticular bleeding underwent hemicolectomy. For the subsequent 48 patients, those requiring treatment received therapy, such as epinephrine injections or bipolar coagulation, through the colonoscope. RESULTS: Of the first 73 patients, 17 (23 percent) had definite signs of diverticular hemorrhage (active bleeding in 6, nonbleeding visible vessels in 4, and adherent clots in 7). Nine of the 17 had additional bleeding after colonoscopy, and 6 of these required hemicolectomy. Of the subsequent 48 patients, 10 (21 percent) had definite signs of diverticular hemorrhage (active bleeding in 5, nonbleeding visible vessels in 2, and adherent clots in 3). An additional 14 patients in this group (29 percent) were presumed to have diverticular bleeding because although they had no stigmata of diverticular hemorrhage, no other source of bleeding was identified. The other 24 patients (50 percent) had other identified sources of bleeding. All 10 patients with definite diverticular hemorrhage were treated endoscopically; none had recurrent bleeding or required surgery. CONCLUSIONS: Among patients with severe hematochezia and diverticulosis, at least one fifth have definite diverticular hemorrhage. Colonoscopic treatment of such patients with epinephrine injections, bipolar coagulation, or both may prevent recurrent bleeding and decrease the need for surgery.

Esophageal perforations masked by steroids.

Esophageal perforation is usually an acute, life-threatening event, and its diagnosis can be established on the basis of obvious clinical and radiographic findings. This article describes two cases whereby symptoms of esophageal perforations were masked by concomitant administration of steroids, thus causing marked delay in diagnosis and treatment. Esophageal rupture should be considered when patients receiving steroids develop unexplained fever with pleural effusion or pneumomediastinum, particularly following instrumentation or forceful retching.

A randomized prospective study of endoscopic bipolar electrocoagulation and heater probe treatment of chronic rectal bleeding from radiation telangiectasia.

BACKGROUND: Our purposes were to (1) evaluate efficacy and safety of bipolar or heater probe endoscopic coagulation compared to prior medical therapy for bleeding radiation telangiectasia, and (2) consider the impact of treatments on patients' impression of their overall health and activity. METHODS: Twelve months of medical management had failed in 18 men and 3 women with chronic, recurrent hematochezia and anemia after radiation treatment of pelvic malignancies. Patients had multiple rectal telangiectasias coagulated with bipolar or heater probes in a randomized, prospective study. RESULTS: Rectal bleeding stopped within four treatment sessions. During 12 months of endoscopic versus medical therapy, severe bleeding episodes diminished significantly for bipolar probe versus 12 months of prior medical therapy (75% vs 33%) and heater probe (67% vs 11%); mean hematocrits rose significantly for patients undergoing bipolar (38.2 vs 31.9) and heater probe (37.6 vs 28.4) treatments, and their impression of overall health improved. During long-term follow-up, new telangiectasias or rectal bleeding were easily controlled. No major complications resulted. CONCLUSIONS: (1) Bipolar or heater probes were safe and effective relative to medical therapy for palliation of patients with lower gastrointestinal bleeding from radiation telangiectasias, and (2) all patients improved in ability to travel and exercise and in their overall impression of their health.

Randomized, prospective study of cyanoacrylate injection, sclerotherapy, or rubber band ligation for endoscopic hemostasis of bleeding canine gastric varices.

The purpose of this randomized, nonblinded study was to compare the effectiveness, safety, and technical ease of three different endoscopic techniques for the treatment of bleeding gastric varices in a canine model. Twenty dogs with large, bleeding gastric varices underwent endoscopic hemostasis with rubber band ligation, sclerotherapy, and cyanoacrylate injection. The time and number of attempts required to achieve definitive hemostasis were evaluated for each technique, and each method was assessed for ease of use. Ulceration rates, ulcer size and depth, and stigmata of ulcer hemorrhage were assessed at 1 week. Intravariceal sclerotherapy was the fastest and easiest to perform. Rubber band ligation was intermediate in technical ease, but it caused the largest and deepest ulcers and had the highest rates of stigmata of ulcer hemorrhage and secondary bleeding. Cyanoacrylate injection was the most cumbersome endoscopic method to perform. All three treatments were effective for controlling gastric variceal bleeding. Intravariceal sclerotherapy had the most favorable results overall because of its technical ease, efficacy, and modest complication rates.

Randomized, controlled study of various agents for endoscopic injection sclerotherapy of bleeding canine gastric varices.

The purpose of this study was to compare the relative efficacy and technical ease of use of eight different agents for endoscopic hemostasis and obliteration of bleeding gastric varices in a canine model, as no comparative data are available on gastric variceal sclerotherapy. Large bleeding gastric varices in 20 heparinized dogs were randomized to endoscopic injection treatment with one of the following agents: cyanoacrylate; a 1:1:1 mixture of sodium tetradecyl sulfate 3%, ethanol 98%, and normal saline solution; ethanolamine oleate 5%; sodium morrhuate 5%; sodium tetradecyl sulfate 1.5%; polidocanol 1%; normal saline solution with epinephrine 1:10,000; or normal saline solution (control). The number and volume of injections and the time required to achieve complete hemostasis were evaluated; follow-up endoscopy was performed at 1 month to assess gastric variceal obliteration. Cyanoacrylate was the best agent overall in terms of immediate efficacy, low volume requirement, time required for initial hemostasis, and reduction of gastric variceal size. Cyanoacrylate, tetradecyl sulfate, and polidocanol were the most effective agents for reducing gastric variceal size. Epinephrine was effective for controlling induced or secondary bleeding caused by puncture of the gastric varices with the sclerotherapy needle during intravariceal injections. Ongoing studies are evaluating combinations of agents with different mechanisms of action, such as epinephrine (for vasoconstriction to minimize secondary bleeding) plus alcohol, and/or tetradecyl sulfate (for variceal thrombosis and sclerosis).(ABSTRACT TRUNCATED AT 250 WORDS)

Endoscopic treatment outcomes in watermelon stomach patients with and without portal hypertension.

BACKGROUND AND STUDY AIMS: Watermelon stomach is a source of recurrent gastrointestinal hemorrhage and anemia. The aims of this study were to describe the endoscopic appearance and treatment outcomes in watermelon stomach patients with and without portal hypertension. PATIENTS AND METHODS: All patients with watermelon stomach enrolled in a hemostasis research group's prospective studies from 1991 to 1999 were identified. Investigators collected data using standardized forms. Comparisons were made using the chi-squared test, Wilcoxon rank-sum test, and Wilcoxon signed-rank test. RESULTS: Twenty-six of 744 (4 %) consecutively enrolled patients with nonvariceal upper gastrointestinal hemorrhage had watermelon stomach as the cause. Eight of these 26 patients (31 %) also had portal hypertension. These patients had diffuse antral angiomas, as opposed to the classic linear arrays seen in those without portal hypertension. The demographic data and clinical presentations of the two groups were otherwise similar. Palliative endoscopic treatment was associated with a significant rise in hematocrit and a decrease in the need for blood transfusion or hospitalization in watermelon stomach patients with and without portal hypertension. CONCLUSIONS: Watermelon stomach patients with and without portal hypertension had similar clinical presentations. The endoscopic findings differed in that those with portal hypertension had more diffuse gastric angiomas. Bleeding was effectively palliated by endoscopic treatment, regardless of the presence of portal hypertension.

Randomized controlled study of injury in the canine right colon from simultaneous biopsy and coagulation with different hot biopsy forceps.

BACKGROUND: Monopolar hot biopsy forceps (HBF), bipolar HBF, and cold biopsy forceps (CBF) followed by bipolar electrocoagulation are used clinically to simultaneously perform a biopsy and coagulate diminutive colon polyps and angiomata. Our purpose was to conduct a randomized, controlled study to evaluate the safety of these different techniques in the canine right colon. METHODS: After right colotomy in 8 mongrel dogs, colonic mucosa was grasped en face, tented, and biopsy performed in randomized order. The dogs were sacrificed after nine days and the biopsy sites were identified and histologically examined. RESULTS: Monopolar HBF caused an overall mean rate of acute serosal whitening of 29% compared with 0% for bipolar HBF and CBF and 6% for CBF/bipolar probe. Histologically confirmed transmural injury 9 days after biopsy occurred in 44% of monopolar HBF compared with 5% of bipolar HBF, 0% of CBF, and 50% of CBF/bipolar probe. CONCLUSIONS: Monopolar HBF had significantly higher rates of acute serosal whitening and histologic transmural damage than bipolar HBF or cold biopsy alone. On the basis of these results, monopolar HBF should be avoided for coagulation of small or flat right colon lesions such as diminutive polyps or angiomata.

Prospective randomized comparative study of bipolar electrocoagulation versus heater probe for treatment of chronically bleeding internal hemorrhoids.

BACKGROUND: Our purpose was to compare the efficacy, complications, failure rates, and crossovers of heater and bipolar probe treatments of chronically bleeding internal hemorrhoids. METHODS: Eighty-one patients (31 female, 50 male) with mean age of 53 years had large (grade 2 to 3) internal hemorrhoids with bleeding for a mean of 12 years, had failed medical management, and were randomized in a prospective study of anoscopic treatments to heater versus bipolar probes. Failure was defined as a major complication or failure to reduce the size of all internal hemorrhoids with three or more treatments. RESULTS: With similar background variables and no difference in treatment times, rectal bleeding and other symptoms were controlled in a shorter time with the heater probe than with the bipolar probe (77 versus 121 days). Five complications (fissures, bleeding, or rectal spasm) occurred with the bipolar probe, and two occurred with the heater probe. The heater probe caused more pain during treatments but had significantly fewer failures and crossovers. CONCLUSIONS: For patients who had failed medical management of chronically bleeding internal hemorrhoids, the techniques and complications of heater and bipolar probes were similar, but pain was more common, failures and crossovers were less frequent, and the time to symptom relief was shorter with the heater probe than with the bipolar probe.

The economic impact of esophageal variceal hemorrhage: cost-effectiveness implications of endoscopic therapy.

Esophageal variceal hemorrhage (EVH) is a serious and expensive sequela of chronic liver disease, leading to increased utilization of resources. Today, endoscopic sclerotherapy (ES) and endoscopic ligation (EL) are the accepted, community standards of endoscopic treatment of patients with EVH. However, there are no published studies comparing the economic costs of treating EVH using these interventions. As part of a prospective, randomized trial comparing ES and EL for the treatment of EVH, we estimated the direct costs of health care utilization and cost-effectiveness for the prevention of variceal rebleeding and patient survival at 1-year follow-up. Treatment groups were similar in incidence of variceal rebleeding (41.9% vs. 42.9%), variceal obliteration (41.9% vs. 40.0%), hospital days, blood transfusions, shunt requirements, and survival (71.0% vs. 60.0%). There were significantly more treatment failures for active bleeding using EL (42% vs. 0%; P =.027) and esophageal stricture formation in the ES-treated patients (19.4% vs. 2.9%; P = 0.03). Median total direct cost outcomes were similar between groups (EL = $9,696 and ES = $13,197; P =.46). EL and ES had similar cost/variceal rebleeding prevented ($28,678 vs. $29,093) and cost/survival ($27,313 vs. $23,804). In the subgroup of active bleeders, ES had a substantially lower cost/survival ($28,523 vs. $51,696). We conclude that resource utilization was similar between treatment groups and that the choice of endoscopic therapy for EVH must still rely on clinical grounds. Further studies comparing costs and resource utilization in this patient population are needed.

An economic analysis of patients with active arterial peptic ulcer hemorrhage treated with endoscopic heater probe, injection sclerosis, or surgery in a prospective, randomized trial.

BACKGROUND: There are no published, detailed assessments of the direct costs of endoscopic hemostasis for actively bleeding peptic ulcers. We compared the direct costs of care for patients with active ulcer hemorrhage treated with endoscopic or medical-surgical therapies and correlated these costs with patient outcomes. METHODS: In a prospective, randomized, controlled trial, 31 patients with active ulcer hemorrhage at emergency endoscopy were randomly assigned to heater probe, injection, or medical-surgical treatment. For further ulcer bleeding, heater probe and injection patients were re-treated endoscopically and medical-surgical patients were referred for surgery. Direct costs were estimated using fixed and variable costs for resources consumed and Medicare reimbursement rates for physician fees. RESULTS: Compared to medical-surgical treatment, the heater probe and injection groups had significantly higher primary hemostasis rates (100% and 90% vs 8%) and lower rates of emergency surgery (0% and 10% vs 75%), blood transfusions, and median direct costs per patient ($4153 and $5247 vs $11,149). Furthermore, compared to medical-surgical treatment, the heater probe group had a significantly lower incidence of severe ulcer rebleeding (11% vs 75%). CONCLUSIONS: Heater probe and injection sclerosis are similarly efficacious treatments for active ulcer hemorrhage, and both treatments yield significantly lower direct costs of medical care and cost savings.

Clinical and economic outcomes of individuals with severe peptic ulcer hemorrhage and nonbleeding visible vessel: an analysis of two prospective clinical trials.

OBJECTIVE: We report the clinical outcomes and direct medical costs of 155 patients with severe peptic ulcer hemorrhage and a nonbleeding visible vessel at emergency endoscopy treated with endoscopic hemostasis or medical-surgical therapy. METHODS: In two consecutive, prospective, randomized, controlled trials, patients were randomly assigned to endoscopic hemostasis (heater probe, bipolar electrocoagulation, or injection sclerosis) or medical-surgical treatment. Study endpoints included the incidence of severe ulcer rebleeding and emergency surgery, length of hospital stay, blood transfusion requirements, mortality rate, and direct costs of utilized health care. Direct medical costs were estimated using combined fixed and variable institutional costs for consumed resources and Medicare reimbursement rates. RESULTS: Compared with medical-surgical treatment, endoscopically treated patients had significantly lower rates of severe ulcer rebleeding (p = 0.004), emergency surgery (p = 0.002 and p = 0.019, 0.024), and blood transfusions (p = 0.025). Observed inter-trial differences in ulcer rebleeding rates may be partially explained in a multivariate model by covariates of comorbid disease and inpatient ulcer bleeding. In both trials, length of hospital stay, mortality rates, and treatment-related complications were similar. Estimated median direct costs per patient differed: The first trial had lower costs with endoscopic hemostasis ($4254, vs $4620 for electrocoagulation and $5909 for medical-surgical treatment), yet the second trial yielded lower costs with medical-surgical treatment ($3169, vs $3477 for injection sclerosis and $4098 for heater probe). CONCLUSIONS: Compared with medical-surgical therapy, endoscopic hemostasis for severe ulcer hemorrhage and a nonbleeding visible vessel yielded significantly better patient outcomes and was safe. This procedure may or may not yield lower direct medical costs and cost savings.