ASSESSING THE QUALITY OF REPORTS ON PLACEBO-CONTROLLED TRIALS PUBLISHED IN ENGLISH OR GERMAN LANGUAGE
Christoph Junker1 and Matthias Egger2
1Department of Social and Preventive Medicine, University of Berne, Finkenhubelweg 11, CH-3012 Berne, Switzerland; 2Department of Social Medicine, University of Bristol, UK
Objective: To compare the quality of reports on randomized placebo-controlled trials published in German and English language.
Design: Survey of 131 journal articles published by authors from German-speaking Europe between 1985 and 1994. All trials published in 5 German, Swiss,
and Austrian general medical journals and trials published by the same authors in English were identified. Placebo-controlled trials with parallel design were
assessed with 2 tools: a standardized 3-item tool described by Prendiville et al, and an 11-item tool assessing 4 different aspects of reporting quality:
general description of study; reporting of design; reporting of statistics; and reporting of results.
Results: Quality scores reached a mean (SD) of the maximum possible score of 51.7% (17.5%) on the 3-item scale and of 53.3% (22.3%) on the 11-item scale.
The quality of reporting of design features was lowest. There were no differences between German-language and English-language reports on the 3-item scale,
but on the 11-item scale reports in English-language journals scored higher 54.8% (22.3%) than those in German-language journals 49.2% (22.0%).
Among reports in English language, those in journals with low-impact factors showed a somewhat lower quality than those in high-impact journals,
52.4% (21.4%) compared to 57.6% (23.3%). There was also an increase over time in quality measured with the 11-item scale; 1985-1989: 50.1% (22.0%); 1990-1994: 56.7% (22.3%).
Conclusions: The small quality differences do not justify the exclusion of German-language reports of randomized controlled trials in systematic reviews.
Independent of language there is much room for improvement in the conduct and reporting of randomized controlled trials performed in German-speaking Europe.
The 11-item scale appears to be a more sensitive instrument.
Return to Session Information
