Abstract
Despite the best medical and endoscopic efforts, some patients with nonvariceal upper gastrointestinal bleeding suffer recurrences. Because high risk stigmata (visible vessels, active bleeders and adherent clots) often persist despite apparently successful initial hemostasis and have a variable natural history, it would seem reasonable to at least consider a routine second look endoscopy. However, a review of the literature revealed six randomized trials that, in aggregate, do not support such a strategy. In fact, a second look does not appear to be effective and is associated with an increased number of procedures, treatment sessions and possibly retreatment-related complications. In addition, the cointerventions in these trials are already out of date and the potential absolute risk reductions are low when a second look is used with intravenous proton pump inhibitors and/or the application of endoscopic hemoclips or combination endoscopic therapy. Finally, the Forrest classification may provide dangerously misleading estimates of prognosis because it is being used out of context. This review critically analyzes routine second look endoscopy.