Publication Characteristics Tied to Treatment Effects
Larger tx effects for published versus unpublished trials; larger effects in non-English publications
THURSDAY, Sept. 6, 2018 (HealthDay News) -- Published trials have larger treatment effects than unpublished trials, while trials published in a language other than English have larger treatment effects versus those published in English, according to research published online Aug. 21 in the Annals of Internal Medicine.
Agnes Dechartres, M.D., Ph.D., from the Université Paris Decartes, and colleagues compared treatment effects between published and unpublished randomized controlled trials (RCTs) and between trials published in English and other languages.
Of the 5,659 RCTs included in 698 meta-analyses, 93.7 and 6.3 percent were published as journal articles and unpublished, respectively. Of the 5,303 journal articles, 92.6 and 7.4 percent were published in English or another language, respectively. The researchers found that compared with unpublished trials, published trials had larger treatment effects (combined ratio of odds ratios for 174 meta-analyses, 0.90). Larger treatment effects were seen in trials published in a language other than English versus in English (combined ratio of odds ratios for 147 meta-analyses, 0.86).
"Our study has important implications for the planning and conduct of systematic reviews. Our results confirm that restricting a search to published trials may lead to overestimation of treatment effects, possibly affecting meta-analysis results and conclusions," the authors write. "Trials published in a language other than English showed larger treatment effect estimates than those published in English. Our results suggest that these studies have lower methodological quality."