Article type
Year
Abstract
Objective: Randomised clinical trials in duodenal ulcer have been claimed to provide substantially different placebo healing rates, yet few have postulated that part of the variation might have been caused by different placebo regimens. In this systematic review we examine the influence of frequency of placebo administration on duodenal ulcer healing rates. The pooled four-week placebo healing rate of all duodenal ulcer trials that employed a four times a day regimen was compared with that rate obtained from trials with a twice a day regimen.
Methods: Studies were located by various strategies and were eligible for inclusion in this review if a number of methodological criteria were satisfied. Equal, fixed, and random effects models were used to assess the difference in healing rates between the two placebo regimens.
Results: The pooled four week healing rate of the 51 trials with a four times a day regimen was 44.2% (805 of 1821 patients) compared with 36.2% (545 of 1504 patients) in the 28 trials with a twice a day regimen (difference, 8.0% [equal effects model]; 95% confidence interval, 4.6% to 11.3%). Depending on the statistical analysis, the rate difference ranged from 4.0% (multivariate random effects model) to 9.0% (univariate fixed effect model). Subgroup analyses of trials that were carried out in the United States, or that were published from 1980 onwards, or that evaluated a H2-receptor antagonist against placebo, or that had more than 50 patients in the placebo group, showed similar or larger differences.
Discussion: The difference in duodenal healing rate between trials that employed a four and two times a day regimen, is probably not caused by different patient characteristics between trials, type of data analysis, or calender time. We consider the difference is caused by the frequency of placebo administration. We encourage researchers in other fields of medicine to look for comparable examples. If these findings would be confirmed in other models, research into the mechanisms is warranted and the implications for clinical practice and research will have to be discussed.
Methods: Studies were located by various strategies and were eligible for inclusion in this review if a number of methodological criteria were satisfied. Equal, fixed, and random effects models were used to assess the difference in healing rates between the two placebo regimens.
Results: The pooled four week healing rate of the 51 trials with a four times a day regimen was 44.2% (805 of 1821 patients) compared with 36.2% (545 of 1504 patients) in the 28 trials with a twice a day regimen (difference, 8.0% [equal effects model]; 95% confidence interval, 4.6% to 11.3%). Depending on the statistical analysis, the rate difference ranged from 4.0% (multivariate random effects model) to 9.0% (univariate fixed effect model). Subgroup analyses of trials that were carried out in the United States, or that were published from 1980 onwards, or that evaluated a H2-receptor antagonist against placebo, or that had more than 50 patients in the placebo group, showed similar or larger differences.
Discussion: The difference in duodenal healing rate between trials that employed a four and two times a day regimen, is probably not caused by different patient characteristics between trials, type of data analysis, or calender time. We consider the difference is caused by the frequency of placebo administration. We encourage researchers in other fields of medicine to look for comparable examples. If these findings would be confirmed in other models, research into the mechanisms is warranted and the implications for clinical practice and research will have to be discussed.