Optimal endoscopic hemostasis remains undetermined. This was a systematic review of contemporary methods of endoscopic hemostasis for patients with bleeding ulcers that exhibited high-risk stigmata.

Randomized trials that evaluated injection, thermocoagulation, clips, or combinations of these were evaluated from MEDLINE, EMBASE, and CENTRAL (1990-2006).

A total of 4261 patients were evaluated.

Outcomes were rebleeding (primary), surgery, and mortality (secondary). Summary statistics were determined; publication bias and heterogeneity were sought by using funnel plots or by subgroup analyses and meta-regression.

Forty-one trials assessed 4261 patients. All endoscopic therapies decreased rebleeding versus pharmacotherapy alone, including sole intravenous (IV) proton pump inhibition (PPI) (OR 0.56 [95% CI, 0.34-0.92]); only one trial assessed high-dose IV PPI. Injection alone was inferior compared with other methods, except for thermal hemostasis (OR 1.02 [95% CI, 0.74-1.40]), with a strong trend of increased rebleeding if 1 injectate is used rather than 2 (OR 1.40 [95% CI, 0.95-2.05]). Injection followed by thermal therapy did not decrease rebleeding compared with clips (OR 0.82 [95% CI, 0.28-2.38]) or thermal therapy alone (OR 0.79 [95% CI, 0.24-2.62]). Subgroup analysis, however, suggested that injection followed by thermal therapy was superior to thermal therapy alone. Clips were superior to thermal therapy (OR 0.24 [95% CI, 0.06-0.95]) but, when followed by injection, were not superior to clips alone (OR 1.30 [95% CI, 0.36-4.76]). Surgery or mortality was not altered in most comparisons.

All endoscopic treatments are superior to pharmacotherapy alone; only 1 study assessed high-dose IV PPI. Optimal endoscopic therapies include thermal therapy or clips, either alone or in combination with other methods. Additional data are needed that compare injection followed by thermal therapy to clips alone or clips combined with another method.