Authenticity investigation and quality assessment of claimed randomized controlled trials on upper respiratory infection

Tags: Poster
Zhang J, Wu T, Xie L, Yan Q

Background: Lots of 'claimed' randomized controlled trials (RCTs) on upper respiratory infection have been published in China and always with positive results. However, their quality is unknown.

Objectives: To identify the quantity of 'true' RCTs in 'claimed' RCTs and assess their quality.

Methods: A research protocol was developed. Three investigators were trained for relative knowledge, communication skills and quality control. Literature was searched for in the electronic database China National Knowledge Infrastructure (CNKI) (from 1994 to June, 2005). Investigators interviewed by telephone the authors of 'claimed' RCTs about their randomization methods and related quality features used to judge the authenticity, and further assessed the quality of 'true' RCTs.

Results: Six hundred and seventy-four 'claimed' RCTs were identified. Of which 26 (4%) were duplicated reports. Authors of 131 (20%) research reports failed to be contacted. Authors refused to provide information in 16 (2.5%) trials. Only 42 (6.5%) 'true' RCTs were identified, the remaining were quasi- or pseudo- RCTs. Of the 'true' RCTs, the randomization methods were respectively computer randomization (14), randomization by number table (26), and simple randomization (2). Single- and double-blind methods were used in 10 and 15 RCTs respectively. Allocation concealment was used in 11 RCTs. Baseline data were reported as comparable in 26 RCTs. Of 12 trials in which participants dropped out or withdrew, intention-to-treat analysis was not used.

Conclusions: Most non-RCTs were reported as RCTs in China. The term 'randomization' was incorrectly used by many authors for lack of knowledge of randomization approaches and by a few authors with adequate knowledge but an absence of a serious scientific attitude. Critical registration and an auditing system can result in well-designed RCTs. The methodological quality of RCTs was improved.