Factors related to biases of randomized controlled trials published from Japan

Article type
Authors
Yoneoka D1, Ota E1, Hisashige A1, Nomura S1, Miyamoto K1, Segawa M1, Kanda M1, Wariki W1, Shibuya K1
1The university of Tokyo, Japan
Abstract
Background: Randomized controlled trials (RCTs) are an important foundation for evidence-based healthcare. Despite an increase in the number of RCTs in Japan, existing international databases fail to capture them, and factors related to biases of Japanese RCTs is lacking.

Objectives: This study assessed the number, characteristics, and risk of bias, and analyzed factors related to biases of RCTs published from Japan in 2010.

Methods: One Japanese database and four international databases were searched for all RCTs published from Japan in 2010. Sixty percent of the included RCTs were randomly sampled and the quality was assessed using the risk of bias tool. Twelve domains of risk of bias was compared with a random sample of 534 trials from 50 Cochrane systematic reviews in the Cochrane Library and the quality of Japanese RCTs classified by the type of journals was compared. Finally, the factors affecting RCT quality were analyzed using a logistic regression model.

Results: Among 2957 studies, 1013 were identified as RCTs. Japanese RCTs had significantly higher odds of risk of bias (p < 0.05) compared with relevant Cochrane systematic reviews in the following domains: sequence generation, allocation concealment, blinding of the outcome assessor, and selective outcome reporting. Non-indexed RCTs in international databases were lower in quality than indexed RCTs in some domains. From the result of regression analysis, factors such as disease and conditions, type of intervention, sample size, trial registration and number of arms were shown to be positively associated with quality, but overall quality was not significantly different.

Conclusions: The CONSORT statement should be endorsed by Japanese researchers, funding bodies and organizations to promote and improve the quality of Japanese RCTs. In addition, since the overall quality of indexed and non-indexed Japanese RCTs was not significantly different, systematic reviewers should consider including Japanese databases.