Cimetidine and duodenal ulcer: an analysis of methodologic problems in randomized controlled trials

The double-blind placebo-comparison randomized clinical trials (RCTs) of cimetidine in the treatment of duodenal ulcer (DU) are examples of good, well-designed clinical trials. With the publication of the first United States RCT in 1978, which reported superiority for cimetidine for only 2 weeks of treatment, grounds for healthy skepticism arose. We undertook a methodologic review of the 16 RCTs published in English through 1980 testing cimetidine in endoscopically documented DU to ascertain if methodologic problems could explain the disparate results. The lack of a clinical classification of disease severity, the failure to consider the additional effect of iatrotherapy (the healing effect of the investigator), the failure to separate duodenitis from ulceration and to note transitions--change in duodenitis and change in ulcer size, and the failure to calculate beta error statistics for "negative" RCTs are important methodologic problems that could explain the disparate results. Future RCTs of therapy in DU should address these issues.