Investigation for randomisation procedure of Chinese original studies

Article type
Authors
Wu T, Chen X, Ni J, Wei J, Zhang J, Chen J, Zhao N, Zhou L
Abstract
Objectives: To investigate randomisation procedures and allocation concealment methods of Chinese original studies those have potential possibility to be included in systematic reviews.

Methods: Study used one of the terms 'random' or 'randomisation' allocation was considered. All of the studies were searched out from Chinese Biomedical Database disc, VIP Chinese Science and Technique Journals Database, CNKI (China National Knowledge Infrastructure).
We contacted authors one by one by telephone directly. The questions were 1. What randomisation procedure was used? 2. How to conceal the random sequence? We sorted the studies according to their source of support, grades of institutes, calculated the rates of adequate, inadequate of both randomisation and concealment.

Result: 384 studies were investigated. Of these studies, 18 were prevention, others were therapy; 5 were supported by government, 11 by both government and World Bank, 5 by pharmaceutical company, others were author selves support; 21 were conducted in or organised by high level institutions include Institutes of National Science Academy, University hospitals, National Medicine Research Bases, Institutes of National CDC, others in low level include county hospitals and/or equal to or lower than this level.

21 studies organised by high level institutions performed randomisation by central office by using random number table or SAS software, none of investigators; 6 used random number table, with a rate of adequate randomisation nearly 7%. 87 used inadequate methods include tossing coin, lots, enrolled number of admission depending on single or even, with a rate of 23%. 265 studies were considered not real randomised controlled trial for their wrong procedure, such as 'randomly sampling', 'according to patient's willing'; random unit were 'ward' or duration such as 'day','month','season', with a rate of 69%. Contacting the authors unsuccessfully in 5 studies. None of the author knows exactly what is the allocation concealment, but 11 preventing studies actually concealed random sequence.

Conclusions: Studies organised by high level institutes always have an adequate randomisation procedure, by low level institutes or self-supporting always have an inadequate approach. These studies investigated just related to a few fields, selection bias should to be considered. Telephone investigation for Chinese original authors is necessary.