In response to Dr. Andriulli and colleagues, the efficacy of high-dose intravenous proton-pump inhibition after successful endoscopic hemostasis has been clearly established by randomized trials in Scandinavia and Asia; recent meta-analyses discussed in the Conference's recommendations; and, in subgroup analysis, another more recent international trial (1). We cited the study by Udd and colleagues (2) but did not consider it further because it included patients with low-risk endoscopic lesions, a subgroup at low risk for rebleeding that was not targeted by that recommendation. We purposefully excluded studies not published in full (except for meta-analyses performed for the Conference) because we felt we could not adequately review their methods and results. We agree, however, that an optimal effective dose threshold remains unknown and may be lower than that used in the randomized trials, as recently suggested by the large Canadian Registry of Patients with Upper Gastrointestinal Bleeding Undergoing Endoscopy (RUGBE) (3). Patients in RUGBE received intravenous pantoprazole, which, combined with the presumed rationale for profound acid suppression and the unavailability of intravenous omeprazole in North America, justified in the panel's judgment the inclusion of both molecules in recommendation 17.

The grading of the level of evidence and recommendations, in response to Dr. Kessler's comments, is a descriptive exercise. It was carried out according to standard methods so that the take-home messages from the different recommendations could be more objectively and uniformly conveyed. We were particularly careful to include clinically relevant topics, even when high-quality evidence was not available. It is then left to individual readers to decide how to integrate and interrelate the grading of the recommendations and consensus that qualify each statement.

We thank Drs. Khuroo and Khuroo for raising an important issue that is often a source of confusion. The increased death rate reported in a single Scandinavian study (4) was not related to excess mortality in the intravenous omeprazole group but rather to an unusually low mortality rate in the placebo group. A recently completed large international randomized trial assessing intravenous pantoprazole failed to show any excess cardiac or neurologic complications compared with ranitidine (1). Individual studies have suggested a trend toward decreased mortality in the intravenous proton-pump inhibitor group (5), and the Conference's recent meta-analysis has even shown a decrease in the mortality rate, as has RUGBE, which included oral and intravenous proton-pump inhibitor use (3). In the face of weak biological plausibility for negative outcomes and an overwhelmingly reassuring safety profile, we believe that there is little evidence to support the concerns raised by Drs. Khuroo and Khuroo. However, we certainly agree that careful postmarketing surveys are always needed to confirm drug safety.