Acute upper gastrointestinal bleed: An audit of the causes and outcomes from a tertiary care center in eastern India.

BACKGROUND/PURPOSE OF THE STUDY: Acute upper gastrointestinal (UGI) bleed is a life-threatening emergency carrying risks of rebleed and mortality despite standard pharmacological and endoscopic management. We aimed to determine etiologies of acute UGI bleed in hospitalized patients and outcomes (rebleed rates, 5-day mortality, in-hospital mortality, 6-week mortality, need for surgery) and to determine predictors of rebleed and mortality. METHODS: Clinical and endoscopic findings were recorded in patients aged > 12 years who presented within 72 h of onset of UGI bleed. Outcomes were recorded during the hospital stay and 6 weeks after discharge. RESULTS: A total of 305 patients were included in this study, mean age being 44 ± 17 years. Most common etiology of UGI bleed was portal hypertension (62.3%) followed by peptic ulcer disease (PUD) (16.7%). Rebleed rate within 6 weeks was 37.4% (portal hypertension 47.9%, PUD 21.6%, malignancy 71.4%). Five-day mortality was 2.3% (malignancy 14.3%, portal hypertension 3.2%); the in-hospital mortality rate was 3.0% (malignancy 14.3%, portal hypertension 3.2%, PUD 0.0%) and 4.9% at 6 weeks (malignancy 28.6%, portal hypertension 5.8%, PUD 0.0%). Surgery was required in 4.59% patients. On multivariate analysis, post-endoscopy Rockall score was significantly predictive of rebleed in both portal hypertension- and PUD-related rebleed. No factors were found predictive of mortality in multivariate analysis. CONCLUSION: Portal hypertension remains the commonest cause of UGI bleed in India and carries a higher risk of rebleed and mortality as compared to PUD-related bleed. Post-endoscopy Rockall score is a useful tool for clinicians to assess risk of rebleed.

Short-Term Outcome of Patients with Cirrhosis and Concurrent Portal Cavernoma Presenting with Acute Variceal Bleeding.

Background and Aim: The outcome of cirrhotic patients with main portal vein occlusion and portal cavernoma after the first episode of acute variceal bleeding (AVB) is unknown. We compared short-term outcomes after AVB in cirrhotic patients with and without portal cavernoma. Methods: Between January 2009 and September 2014, 28 patients with cirrhosis and portal cavernoma presenting with the first occurrence of AVB and 56 age-, sex-, and Child-Pugh score-matched cirrhotic patients without portal cavernoma were included. The primary endpoints were 5-day treatment failure and 6-week mortality. Results: The 5-day treatment failure rate was higher in the cavernoma group than in the control group (32.1% versus 12.5%; p = 0.031). The 6-week mortality rate did not differ between the cavernoma and control group (25% versus 12.5%, p = 0.137). Multivariable Cox proportional hazard regression analyses revealed that 5-day treatment failure (HR = 1.223, 95% CI = 1.082 to 1.384; p = 0.001) independently predicted 6-week mortality. Conclusions: Cirrhotic patients with AVB and portal cavernoma have worse short-term prognosis than patients without portal cavernoma. The 5-day treatment failure was an independent risk factor for 6-week mortality in patients with cirrhosis and portal cavernoma.

Severity and Outcomes of Upper Gastrointestinal Bleeding With Bloody Vs. Coffee-Grounds Hematemesis.

OBJECTIVES: Numerous reviews indicate bloody hematemesis signifies more severe bleeding than coffee-grounds hematemesis. We assessed severity and outcomes related to bleeding symptoms in a prospective study. METHODS: Consecutive patients presenting with hematemesis or melena were categorized as bloody emesis (N=1209), coffee-grounds emesis without bloody emesis (N=701), or melena without hematemesis (N=1069). We assessed bleeding severity (pulse, blood pressure) and predictors of outcome (hemoglobin, risk stratification scores) at presentation, and outcomes of bleeding episodes. The primary outcome was a composite of transfusion, intervention, or mortality. RESULTS: Bloody and coffee-grounds emesis were similar in pulse ≥100 beats/min (35 vs. 37%), systolic blood pressure ≤100 mm Hg (12 vs. 12%), and hemoglobin ≤100 g/l (25 vs. 27%). Risk stratification scores were lower with bloody emesis. The composite end point was 34.7 vs. 38.2% for bloody vs. coffee-grounds emesis; mortality was 6.6 vs. 9.3%. Hemostatic intervention was more common (19.4 vs. 14.4%) with bloody emesis (due to a higher frequency of varices necessitating endoscopic therapy), as was rebleeding (7.8 vs. 4.5%). Outcomes were worse with hematemesis plus melena vs. isolated hematemesis for bloody (composite: 62.4 vs. 25.6%; hemostatic intervention: 36.5 vs. 13.8%) and coffee-grounds emesis (composite: 59.1 vs. 27.1%; hemostatic intervention: 26.4 vs. 8.1%). CONCLUSIONS: Bloody emesis is not associated with more severe bleeding episodes at presentation or higher mortality than coffee-grounds emesis, but is associated with modestly higher rates of hemostatic intervention and rebleeding. Outcomes with hematemesis are worsened with concurrent melena. The presence of bloody emesis plus melena potentially could be considered in decisions regarding timing of endoscopy.

Relationship of time to presentation after onset of upper GI bleeding with patient characteristics and outcomes: a prospective study.

BACKGROUND AND AIMS: We performed a prospective multi-national study of patients presenting to the emergency department with upper GI bleeding (UGIB) and assessed the relationship of time to presentation after onset of UGIB symptoms with patient characteristics and outcomes. METHODS: Consecutive patients presenting with overt UGIB (red-blood emesis, coffee-ground emesis, and/or melena) from March 2014 to March 2015 at 6 hospitals were included. Multiple predefined patient characteristics and outcomes were collected. Rapid presentation was defined as ≤6 hours. RESULTS: Among 2944 patients, 1068 (36%) presented within 6 hours and 576 (20%) beyond 48 hours. Significant independent factors associated with presentation ≤6 hours versus >6 hours on logistic regression included melena (odds ratio [OR], 0.22; 95% CI, 0.18-0.28), hemoglobin ≤80 g/L (OR, 0.47; 95% CI, 0.36-0.61), altered mental status (OR, 2.06; 95% CI, 1.55-2.73), albumin ≤30 g/L (OR, 1.43; 95% CI, 1.14-1.78), and red-blood emesis (OR, 1.29; 95% CI, 1.06-1.59). Patients presenting ≤6 hours versus >6 hours required transfusion less often (286 [27%] vs 791 [42%]; difference, -15%; 95% CI, -19% to -12%) because of a smaller proportion with low hemoglobin levels, but were similar with regard to hemostatic intervention (189 [18%] vs 371 [20%]), 30-day mortality (80 [7%] vs 121 [6%]), and hospital days (5.0 ± 0.2 vs 5.0 ± 0.2). CONCLUSIONS: Patients with melena alone delay their presentation to the hospital. A delayed presentation is associated with a decreased hemoglobin level and increases the likelihood of transfusion. Other outcomes are similar with rapid versus delayed presentation. Time to presentation should not be used as an indicator for poor outcome. Patients with delayed presentation should be managed with the same degree of care as those with rapid presentation.

The weekend effect in patients hospitalized for upper gastrointestinal bleeding: a single-center 10-year experience.

OBJECTIVE: This study was conducted to assess the possible weekend effect in patients with upper gastrointestinal bleeding (UGIB) on the basis of a 10-year single-center experience in Serbia. MATERIALS AND METHODS: A retrospective analysis of hospital records in the University Clinic 'Dr Dragisa Misovic-Dedinje', Belgrade, Serbia, from 2002 to 2012 was conducted. Patients admitted for UGIB were identified, and data on demographic characteristics, symptoms, drug use, alcohol abuse, diagnosis and treatment were collected. Univariate and multivariate logistic regression were used to assess the association between weekend admission and the occurrence of rebleeding and in-hospital mortality. RESULTS: Analyses included 493 patients. Rebleeding occurred significantly more frequently on weekends (45.7 vs. 32.7%, P=0.004). Weekend admission [odds ratio (OR)=1.78; 95% confidence interval (CI): 1.15-2.74], older age (OR=1.02; 95% CI: 1.00-1.03), and the presence of both melaena and hematemesis (OR=2.29; 95% CI: 1.29-4.07) were associated with the occurrence of rebleeding. No difference between weekend and weekday admissions was observed for the in-hospital mortality rate (6.9% vs. 6.0%, P=0.70). Older age (OR=1.14; 95% CI: 1.08-1.20), presentation with melaena and hematemesis (OR=4.12; 95% CI: 1.56-10.90) and need for surgical treatment (OR=5.16; 95% CI: 1.61-16.53) were significant predictors of all-cause mortality. Patients with nonvariceal bleeding had significantly higher rebleeding rates on weekends (44 vs. 32.3%, P=0.013). CONCLUSION: There was no significant weekend effect in the mortality of patients admitted for UGIB, irrespective of the source of bleeding. Increased attention to older patients presenting with a more severe clinical picture is needed to prolong survival and prevent rebleeding.

Causes of bleeding and outcomes in patients hospitalized with upper gastrointestinal bleeding.

GOALS: To evaluate sources of upper gastrointestinal bleeding (UGIB) at an urban US hospital and compare them to sources at the same center 20 years ago, and to assess clinical outcomes related to source of UGIB. BACKGROUND: Recent studies suggest changes in causes and outcomes of UGIB. STUDY: Consecutive patients with hematemesis, melena, and/or hematochezia undergoing upper endoscopy with an identified source at LA County+USC Medical Center from January 2005 to June 2011 were identified retrospectively. RESULTS: Mean age of the 1929 patients was 52 years; 75% were male. A total of 1073 (55%) presented with hematemesis, 809 (42%) with melena alone, and 47 (2%) with hematochezia alone. The most common causes were ulcers in 654 patients (34%), varices in 633 (33%), and erosive esophagitis in 156 (8%), compared with 43%, 33%, and 2% in 1991. During hospitalization, 207 (10.7%) patients required repeat endoscopy for UGIB (10.6% for both ulcers and varices) and 129 (6.7%) died (5.2% for ulcers; 9.2% for varices). On multivariate analysis, hematemesis (OR=1.38; 95% CI, 1.04-1.88) and having insurance (OR=1.44; 95% CI, 1.07-1.94) were associated with repeat endoscopy for UGIB. Varices (OR=1.53; 95% CI, 1.05-2.22) and having insurance (OR=4.53; 95% CI, 2.84-7.24) were associated with mortality. CONCLUSION: Peptic ulcers decreased modestly over 2 decades, whereas varices continue as a common cause of UGIB at an urban hospital serving lower socioeconomic patients. Inpatient mortality, but not rebleeding requiring endoscopy, was higher with variceal than nonvariceal UGIB, indicating patients with variceal UGIB remain at risk of death from decompensation of underlying illness even after successful control of bleeding.

Formic acid poisoning in a tertiary care center in South India: A 2-year retrospective analysis of clinical profile and predictors of mortality.

BACKGROUND: Formic acid (FA), a common industrial compound, is used in the coagulation of rubber latex in Kerala, a state in southwestern India. Easy accessibility to FA in this region makes it available to be used for deliberate self-harm. However, the literature on intentional poisoning with FA is limited. STUDY OBJECTIVES: To determine the patterns of presentation of patients with intentional ingestion of FA and to find the predictors of mortality. A secondary objective was to find the prevalence and predictors of long-term sequelae related to the event. METHODS: We performed a 2-year chart review of patients with acute intentional ingestion of FA. Symptoms, signs, outcomes and complications were recorded, and patients who survived the attempt were followed-up by telephone or personal interview to identify any complications after their discharge from the hospital. RESULTS: A total of 302 patients with acute formic acid ingestion were identified during the study period. The mortality rate was 35.4% (n = 107). Bowel perforation (n = 39), shock (n = 73), and tracheoesophageal fistula (n = 4) were associated with 100% mortality. Quantity of FA consumed (p < 0.001), consuming undiluted FA (p < 0.001), presenting symptoms of hypotension (p < 0.001), respiratory distress (p < 0.001), severe degree of burns (p = 0.020), hematemesis (p = 0.024), complications like metabolic acidosis (p < 0.001) and acute respiratory distress syndrome (p < 0.001) were found to have significant association with mortality. The prevalence of esophageal stricture (n = 98) was 50.2% among survivors and was the most common long-term sequela among the survivors. Stricture was significantly associated with hematemesis (p < 0.001) and melena (p < 0.001). CONCLUSION: This study highlights the magnitude and ill-effects of self-harm caused by a strong corrosive, readily available due to very few restrictions in its distribution. Easy availability of FA needs to be curtailed by enforcing statutory limitations in this part of the world. Patients with hematemesis or melena after FA ingestion may be referred for early dilatation therapy in a setting where emergency endoscopic evaluation of all injured patients is not practical.

Delayed endoscopy increases re-bleeding and mortality in patients with hematemesis and active esophageal variceal bleeding: a cohort study.

BACKGROUND & AIMS: Active bleeding is a poor prognostic indicator in patients with acute esophageal variceal bleeding. This study aimed at determining indicators of 6-week re-bleeding and mortality in patients with "active" esophageal variceal bleeding, particularly emphasizing the presenting symptoms and timing of endoscopy to define the treatment strategy. METHODS: From July 2005 to December 2009, cirrhotic patients with endoscopy-proven active esophageal variceal bleeding were evaluated. Cox proportional hazards regression analysis was used to determine the indicators of 6-week re-bleeding and mortality. Outcome comparisons were performed by Kaplan-Meier method and log rank test. RESULTS: In 101 patients, the overall 6-week and 3-month re-bleeding rates were 25.7% (n=26) and 29.7% (n=30), respectively. The overall 6-week and 3-month mortality was 31.7% (n=32) and 38.6% (n=39), respectively. Door-to-endoscopy time (hr), MELD score, and portal vein thrombosis were indicators of 6-week re-bleeding, while hematemesis upon arrival, MELD score, and hepatocellular carcinoma were indicators of 6-week mortality. Overall mortality was poorer in hematemesis than in non-hematemesis patients (39.7% vs. 10.7%, p=0.007). In hematemesis patients, 6-week re-bleeding rate (18.9% vs. 38.9%, p=0.028) and mortality (27% vs. 52.8%, p=0.031) were lower in those with early (≤ 12 h) than delayed (>12h) endoscopy. In non-hematemesis patients, early and delayed endoscopy had no difference on 6-week re-bleeding rate (17.6% vs. 18.2%, p=0.944) and mortality (11.8% vs. 9.1%, p=0.861). CONCLUSIONS: It is likely that early endoscopy (≤ 12 h) is associated with a better outcome in hematemesis patients, but a randomized trial with larger case numbers is required before making a firm conclusion.

The patient with haematemesis and melaena.

Bleeding from the upper gastrointestinal (GI) tract is a common medical emergency, with an incidence of between 50-150 cases per 100,000 per year.1 A recent audit by the British Society of Gastroenterology showed the mortality rate from upper GI bleeds has fallen from 14%2 in 1993 to 10% in 2007.3 However, despite the use of proton pump inhibitors (PPIs), admission rates for peptic ulcer haemorrhage have increased in older age groups,4 probably related to increased use of antiplatelet agents such as aspirin and clopidogrel and anticoagulants in acute coronary syndromes, stroke and atrial fibrillation. The rising age of the population may also have offset further reductions in mortality and morbidity that may have otherwise come about through improved supportive and endoscopic care.

Independent predictors of in-hospital re-bleeding, need of operation and mortality in acute upper gastrointestinal bleeding.

Prediction of outcome is difficult in patients with acute upper gastrointestinal bleeding (AUGIB). Some factors have been proposed in this regard with varying accuracy. This study aimed to investigate probable predictors of in-hospital outcome in patients with AUGIB. One hundred sixty four patients with AUGIB were studied prospectively in Tabriz Imam Reza Teaching Centre. All these patients were evaluated endoscopically by an expert. Patients' age, gender, presenting complains, transfusion, clinical findings and previous medical history were compared between survived vs. expired, re-bled vs. non re-bled and operated vs. non operated patients. There were 117 males and 47 females with the mean age of 57.12 +/- 17.32 (range: 32-78) years in this study. Hematemesis was the sole independent predictor of in-hospital mortality (82.1 vs. 100%; p < 0.001). In univariate analysis, however, female gender, major hemorrhage and previous neurological disease were associated with higher rate of expiration. Comparing two re-bled and non re-bled groups, hematemesis (76.5 vs. 95.9%; p = 0.003) and need of transfusion > 2U (36.1 vs. 71.4%; p = 0.006) were independent predictors of re-bleeding. In univariate analysis, hematocrit < 30%, major hemorrhage and previous history of hepatic disease or hypertension were predictive of re-bleeding. In comparison between operated and non operated groups no significant predictor was detected. In conclusion, this study showed that presence of hematemesis at the time of admission and need of transfusion > 2U were independent predictors of poor outcome in patients with AUGIB.

Early risk stratification with simple clinical parameters for cirrhotic patients with acute upper gastrointestinal bleeding.

OBJECTIVE: This study aimed to identify pre-endoscopic clinical parameters independently associated with 6-week mortality and to develop a prognostic model in cirrhotic patients with acute upper gastrointestinal (UGI) bleeding. METHODS: A total of 542 consecutive admissions of 389 cirrhotic patients with acute UGI bleeding were retrospectively investigated. Pertinent clinical data obtained at the emergency department were analyzed. Multivariate logistic regression analysis was performed to determine risk factors for 6-week mortality and to develop a predictive model. RESULTS: Forty-four patients (8.12%) died within 6 weeks. The 6-week mortality was independently associated with male sex, hepatocellular carcinoma, non-hepatocellular carcinoma malignancy, hypoxemia with peripheral oxygen saturation less than 95%, serum bilirubin, and prothrombin time. A predictive model consisting of these 6 simple parameters was built. The c statistic of our model was 0.84, significantly superior to that (0.71) of the model for end-stage liver disease score (P = .002). CONCLUSIONS: Simple pre-endoscopic clinical parameters are valuable for early risk stratification in cirrhotic patients with acute UGI bleeding. Our prognostic model warrants prospective validation by further studies.

[Hematemesis in Togo: findings of a 12-month study in an intensive care unit]. / Hématémèses au Togo: expérience d'un an en réanimation.

This prospective study was conducted over a 12-month period on patients who underwent upper digestive endoscopy for hematemesis in the medical intensive care unit (ICU) of the Tokoin University Hospital Center in Lomé, Togo. A total of 44 patients with a mean age of 44 years were included. The sex-ratio was 2.61. Risk factors included use of non-steroid anti-inflammatory (NSAI) in 16 patients (36.4%) and alcohol abuse in 13 (29.6%). At the time of admission to the ICU, 21 patients (47.7%) were in hemodynamic shock and 11 (25%) presented signs of portal hypertension. The underlying etiology was peptic ulcer in 18 cases (40.9%) including 13 cases of duodenal ulcer and 5 cases of stomach ulcer, rupture of esophageal varicosities in 8 (18.2%), gastric tumor in 6 (13.6%), Mallory Weiss syndrome in 5 (11.4%), gastritis in 4 (9,1%), and esophagitis in 3 (6.8%) due to peptic inflammation in 2 and mycotic infection in 1. The mortality rate was 45.5%. The main causes of hematemesis were peptic ulcer and rupture of esophageal varicosities. The death rate was high due to inadequate care facilities.

Correlación clínica- para-clínica en un escolar con hipertensión porta de etiología a esclarecer / Clinical-paraclinical in a school with portal hypertension to determine

La hipertensión porta (HTP) es el resultado del incremento de la presión dentro del sistema venoso porta. Se presenta con poca frecuencia en el paciente pediátrico pero es una de las mayores causas de morbilidad y mortalidad en el niño con enfermedad hepática. La mayoría de los pacientes con http presentan un estado hiperdinámico, lo cual aumenta el flujo venoso porta y mantiene la hipertensión. Puede ser secundaria a obstrucción a nivel prehepático, intrahepático o extrahehepático.

Portal hypertension (PH) is the result of increased pressure within the portal venous system. It occurs infrequently in the pediatric patient but it is a major cause of morbidity and mortality in children with liver disease. Most patients with PH have a hyperdynamic state, which increases venous flow and portal hypertension remains. May be secondary to obstruction at prehepatic, intrahepatic or extrahehepatic.

Delayed endoscopy as a risk factor for in-hospital mortality in cirrhotic patients with acute variceal hemorrhage.

BACKGROUND AND AIMS: Risk factors for mortality in acute variceal hemorrhage remain incompletely understood. Whether endoscopy timing is associated with risk of mortality has not been investigated. We aimed to investigate risk factors for in-hospital mortality in cirrhotic patients with acute variceal hemorrhage, with emphasis on endoscopy timing. METHODS: Three hundred and eleven (73% male and 23% female) consecutive cirrhotic patients presenting with acute variceal hemorrhage from July 2004 to July 2007 were investigated. The univariate association of endoscopy timing as the predictor for in-hospital mortality was examined. Independent risk factors for mortality were determined by multivariate logistic regression analysis consisting of clinical, laboratory and endoscopic parameters. RESULTS: Twenty-five (8.04%) patients died within admission. By plotting the receiver operating curve of endoscopy timing for mortality, we selected 15 h as the optimal cut-off point to define delayed endoscopy. Multivariate regression analysis revealed that independent risk factors predictive for in-hospital mortality included delayed endoscopy performed 15 h after admission (adjusted odds ratio [aOR] = 3.67; 95% confidence interval [CI], 1.27-10.39), every point increment of model for end-stage liver disease (MELD) score (aOR = 1.16; 95% CI, 1.07-1.25), failure of the first endoscopy (aOR = 4.36; 95% CI, 1.54-12.30) and hematemesis as the chief complaint (compared with melena, aOR = 8.66; 95% CI, 1.06-70.94). CONCLUSION: Delayed endoscopy for more than 15 h, high MELD score, failure of the first endoscopy and hematemesis are independent risk factors for in-hospital mortality in cirrhotic patients with acute variceal hemorrhage.

Management of massive peptic ulcer bleeding.

Massive bleeding from a peptic ulcer remains a challenge. A multidisciplinary team of skilled endoscopists, intensive care specialists, experienced upper gastrointestinal surgeons, and intervention radiologists all have a role to play. Endoscopy is the first-line treatment. Even with larger ulcers, endoscopic hemostasis can be achieved in the majority of cases. Surgery is clearly indicated in patients in whom arterial bleeding cannot be controlled at endoscopy. Angiographic embolization is an alternate option, particularly in those unfit for surgery. In selected patients judged to belong to the high-risk group--ulcers 2 cm or greater in size located at the lesser curve and posterior bulbar duodenal, shock on presentation, and elderly with comorbid illnesses--a more aggressive postendoscopy management is warranted. The optimal course of action is unclear. Most would be expectant and offer medical therapy in the form of acid suppression. Surgical series suggest that early elective surgery may improve outcome. Angiography allows the bleeding artery to be characterized, and coil embolization of larger arteries may further add to endoscopic hemostasis. The role of early elective surgery or angiographic embolization in selected high-risk patients to forestall recurrent bleeding remains controversial. Prospective studies are needed to compare different management strategies in these high-risk ulcers.

Lethal outcome of post-tonsillectomy hemorrhage.

Despite the large number of tonsillectomies performed little knowledge exists about post-tonsillectomy hemorrhage (PTH) with lethal outcome. This study was performed to evaluate clinical features in a larger patient population with emphasis on the onset of this complication. A nationwide collection of cases was performed based on personal communication, expert reports to lawsuits and professional boards, and case reports received after a plea published in a professional national journal. Clinical data of 29 patients were collected of whom the 18 were children (64%). With one exception all patients experienced secondary PTH (>24 h) occurring 1-28 days after tonsillectomy. Aspiration contributed to lethal outcome in 13 cases. Fatalities were unavoidable although 21 patients were in the hospital. Massive vomiting of blood was observed in 11 patients. There were 11 patients without (group A) and 18 with (group B) episodes of repeated bleeding. This study suggests that particularly children are endangered by lethal PTH. Inpatient treatment was unable to prevent lethal outcome in this selected patient population. However, it appears wise to re-admit patients with delayed PTH, since excessive PTH may occur. These unexpected and unpredictable situations require an immediate and adequate medical treatment by a skilled staff. The paucity of data currently does not allow calculation of a cut-off point at which the risk of life-threatening PTH significantly decreases. Secondary PTH remains a substantial complication.

Perforation of a gastric tube peptic ulcer into the thoracic aorta.

We present a case of a 52-year-old male patient who died from massive hematemesis as a result of perforation of a benign peptic ulcer into the descending thoracic aorta, 1 year after esophagectomy for esophageal cancer and gastric tube interposition. We also review the literature for mechanisms of ulceration in intrathoracic gastric grafts and for complications of such ulcers.

Predictors of mortality in hospitalized patients with secondary upper gastrointestinal haemorrhage.

OBJECTIVES: To identify clinical, laboratory and endoscopic features of prognostic implication in patients who suffer an upper gastrointestinal haemorrhage whilst hospitalized for other causes. DESIGN/SETTING: A prospective longitudinal study of 125 consecutive cases (89 males, 36 females) seen during 1988-1991 in a tertiary care university hospital. MAIN OUTCOME MEASURES: The crude and adjusted relative risk of mortality associated with each of the various clinical, laboratory and endoscopic variables. RESULTS: The rates of endoscopic haemostasis, persistent or recurrent bleeding and surgery to control bleeding were 48, 37 and 12%, respectively; the overall mortality was 28%. A significantly increased risk of mortality was associated with shock prior to onset of bleeding, sepsis, renal failure, cirrhosis, encephalopathy, presence of red blood in the nasogastric aspirate or per rectum, thrombocytopenia, hypoalbuminaemia, elevation of serum bilirubin, aminotransferases or urea levels, endoscopic evidence of active bleeding, the application of endoscopic haemostasis, rebleeding, transfusion of > or = 6 units of blood, surgical treatment or any subsequent complication. Use of nonsteroidal anti-inflammatory drugs (27%) was associated with a decreased mortality risk (odds ratio 0.2; P = 0.03). A multivariate analysis showed that the features at presentation which were independently associated with an increased risk of mortality were: a history of cirrhosis, sepsis, shock prior to onset of bleeding, hypo-albuminaemia and elevated serum aminotransferases. CONCLUSION: The prognosis of secondary upper gastrointestinal haemorrhage depends on the underlying diseases and on the general condition of the patient, rather than on the actual cause of bleeding.