[Treatment of gastroesophageal reflux: analysis of randomized double-blind trials]

The aim of this study was to analyze the results and the quality of methodology of 51 controlled double blind trials in the medical treatment of gastroesophageal reflux. The results of H2 receptor antagonist treatment were evaluated by the pooling method. Evaluation of methodology was carried out by using a special form filled in by two independent observers. The major criticisms in methodology were: small sample size, unblind evaluation of end-points, inappropriate statistical tests for small samples, and inaccurate handling of the withdrawals. There were only two trials concerning antacids versus placebo: one showed that Novaluzid improved symptoms and another that Maalox did not differ from placebo. The effectiveness of alginic acid and domperidone on either symptoms or endoscopic lesions was not demonstrated. Metoclopramide and bethanechol produced significant relief of reflux symptoms. Sucralfate and bethanechol were better than placebo in improvement of esophagitis endoscopic lesions. The H2-inhibitors efficiently relieved symptoms and esophagitis. Pooling analysis showed that H2-inhibitors were superior to placebo in the healing of esophagitis; the odds ratios were 2.5 for cimetidine and 3.3 for ranitidine, without significant difference. Omeprazole was better than ranitidine in relief of symptoms and esophagitis. The comparison of cimetidine alone with cimetidine plus metoclopramide showed that combined therapy was better in one trial out of two. New controlled trials are necessary to compare these different drugs and their association.