[Endoscopic hemostasis for hemorrhagic gastroduodenal ulcer. Meta-analysis of randomized clinical trials].

To demonstrate the advantages of various endoscopic hemostatic methods (laser photocoagulation, electrocautery, injection therapy) for bleeding gastroduodenal ulcer in patients at high risk for continued or recurrent bleeding, a critical review of published randomized clinical trials was made with meta-analytic methods. Only the 15 clinical trials dealing either with patients with visible non bleeding vessels or spurting arterial bleeding were included. Regarding visible non bleeding vessels, the meta-analysis of five trials on electrocautery and two trials on sclerotherapy showed a significant reduction in rebleeding rates in the treatment group compared with untreated controls. The odds ratios were 4 (95 percent confidence levels (CL): 2.4-6.9) (P less than 0.001) and 6.8 (95 percent CL: 2.7-17.2) (P less than 0.001), respectively. As well, the meta-analysis of four trials on electrocautery and the two trials on sclerotherapy showed a significant reduction in the number of emergency surgical operations in the treated groups a compared with the untreated groups. The odds ratios were 5.5 (95 percent confidence levels (CL): 2.7-11.3) (P less than 0.001) and 6.1 (95 percent CL: 2.1-17.8) (P less than 0.001), respectively. Meta-analysis did not show any advantage for laser, electrocautery, or sclerotherapy in terms of mortality. Indirect meta-analysis did not reveal any difference between electrocautery and sclerotherapy. Regarding spurting arterial bleeding, meta-analysis of the two YAG laser trials, the two Argon laser trials, and the two electrocautery trials showed a significant reduction of rebleeding or continued bleeding in the treatment groups as compared with the control groups.(ABSTRACT TRUNCATED AT 250 WORDS)