Relevance of surgery in patients with non-variceal upper gastrointestinal bleeding

Upper GI bleeding remains one of the most common emergencies with a substantial overall mortality rate of up to 30%. In severe ill patients, death does not occur due to failure of hemostasis, either medical or surgical, but mainly from comorbidities, treatment complications, and decreased tolerated blood loss. Management strategies have changed dramatically over the last two decades and include primarily endoscopic intervention in combination with acid-suppressive therapy and decrease in surgical intervention. Herein, we present one of the largest patient-based analysis assessing clinical parameters and outcome in patients undergoing endoscopy with an upper GI bleeding. Data were further analyzed to identify potential new risk factors and to investigate the role of surgery. In this retrospective study, we aimed to analyze outcome of patients with an UGIB and data were analyzed to identify potential new risk factors and the role of surgery. Data collection included demographic data, laboratory results, endoscopy reports, and details of management including blood administration, and surgery was carried out. Patient events were grouped and defined as "overall" events and "operated," "non-operated," and "operated and death" as well as "non-operated and death" where appropriate. Blatchford, clinical as well as complete Rockall-score analysis, risk stratification, and disease-related mortality rate were calculated for each group for comparison. Overall, 253 patients were eligible for analysis: endoscopy was carried out in 96% of all patients, 17% needed surgical intervention after endoscopic failure of bleeding control due to persistent bleeding, and the remaining 4% of patients were subjected directly to surgery. The median length of stay to discharge was 26 days. Overall mortality was 22%; out of them, almost 5% were operated and died. Anticoagulation was associated with a high in-hospital mortality risk (23%) and was increased once patients were taken to surgery (43%). Patients taking steroids presented with a risk of death of 26%, once taken to surgery the risk increased to 80%. Patients with liver cirrhosis had a risk of death of 42%; we observed a better outcome for these patients once taken to theater. Clinically, once scored with Blatchford score, statistical correlation was found for initial need for blood transfusion and surgical intervention. Clinical as well as complete Rockall score revealed a correlation between need for blood transfusion as well as surgical intervention in addition with a decreased outcome with increasing Rockall scores. Risk factor analysis including comorbidity, drug administration, and anticoagulation therapy introduced the combination of tumor and non-steroidal antirheumatic medication as independent risk factors for increased disease-related mortality. UGIB remains challenging and endoscopy is the first choice of intervention. Care must be taken once a patient is taking antirheumatic non-steroidal pain medication and suffers from cancer. In patients with presence of liver cirrhosis, an earlier surgical intervention may be considered, in particular for patients with recurrent bleeding. Embolization is not widely available and carries the risk of necrosis of the affected organ and should be restricted to a subgroup of patients not primarily eligible for surgery once endoscopy has failed. Taken together, an interdisciplinary approach including gastroenterologists as well as surgeons should be used once the patient is admitted to the hospital to define the best treatment option.