Why have controlled trials failed to demonstrate a benefit of esophagogastroduodenoscopy in acute upper gastrointestinal bleeding? A probability model analysis.

Numerous prospective randomized trials have failed to demonstrate a benefit attributable to early diagnostic esophagogastroduodenoscopy (EGD) in acute upper gastrointestinal bleeding (UGIB). The clinical implications of these studies have received extensive editorial comment and analysis. We have employed a probability model to further analyze the reasons why these studies have failed to demonstrate an impact of EGD on UGIB. The clinical course of each bleeding lesion can be predicted from the literature. For each lesion, the mortality associated with early specific intervention afforded by an early specific diagnosis can be compared with the mortality of intervention delayed by applying EGD only to those patients who have a complicated course marked by continued bleeding or rebleeding. Using optimistic assumptions that would tend to overstate the impact of EGD, this analysis estimates the maximum decrease in overall mortality in any of these trials afforded by early diagnostic EGD to be 1.2% which would require randomization of over 5000 patients to demonstrate this benefit in a prospective trial.