[Multicenter prospective study of the current status of treatment for bleeding ulcer in Germany]

The current state of ulcer treatment in Germany was analysed in a prospective multi-centre study. It was based on 1139 consecutive patients admitted to the participating hospitals because of upper gastrointestinal bleeding. The source of the bleeding was identified by diagnostic endoscopy in 1075 patients (94%), from a gastric and/or duodenal ulcer in 546 of them (mean age 62 +/- 18 years). Using Forrest's classification, 4% of patients were in bleeding stage Ia, 17% in stage Ib, 16% stage IIa, 30% stage IIb and 33% stage III. An attempt to arrest bleeding through the endoscope was made in 233 patients (43%): more often with tissue-preserving substances (epinephrine +/- NaCl in 36%, fibrin glue +/- epinephrine in 24%) than with tissue-damaging procedures (epinephrine + polidocanol +/- NaCl in 26%, epinephrine + thermocoagulation in 7%). Primary haemostasis was achieved in 219 patients (94%). There was a total of 66 recurrences of bleeding (12%), but the rate was 18% after endoscopic haemostasis. 64 patients (12%) required operative intervention, including initial emergency operations. Severe complications (infections, organic failure) occurred in 82 patients (16%). 114 of the 546 patients were in the high risk group (older than 60 years; high amount of bleeding). Their bleeding recurrence and mortality rates (27 and 22%, respectively) were significantly higher (P < 0.01) than those of the total group. Overall mortality rate was 11% (58 patients). The mortality rate depended on the severity of initial bleeding (26% for Forrest group Ia). After recurrent bleeding the mortality rate was 34% with conservative and 33% with operative treatment. 7% of all deaths were the direct result of bleeding. The following factors prognostically closely correlated with mortality rate: age of patient (P < 0.01); haemoglobin < 8 g/dl on admission (P < 0.05); initial severity of bleeding (Forrest group I; P < 0.05); and recurrence of bleeding (P < 0.001).