Article type
Year
Abstract
Background: In some Cochrane Reviews the authors claim to exclude quasi-randomised trials (qRCTs). This is contrary to guidance in the handbook and is futile as many reports of trials just state participants were randomised without enough information about how this was done.
Objectives: To determine how many authors specify that they exclude quasi-randomised trials from Cochrane Reviews, and what review group instructions say.
Methods: A random sample of 300 Cochrane Reviews was selected and the methods scrutinised to see if qRCTs were excluded. Details of all review groups were read to determine policies regarding qRCTs.
Results: Forty-seven of the 300 stated that qRCTs were excluded (15.7%, 95% CI 11.7–20.3%). In addition, of the remaining 253 trials, 138 simply stated that RCTs were included (54.5%, 95% CI 48.2–60.8%). In these it was unclear whether this referred to securely randomised trials only. One Cochrane Review group states in their suggested methods that qRCTs should be excluded, and six suggest that they should only be included if there are only a few or no RCTs, or only for adverse effects. Seven review groups say that RCTs should be included with no further elucidation so it is unclear whether they would like qRCTs to be excluded.
Conclusions: Contrary to the advice in the handbook many Cochrane Reviews attempt to exclude qRCTs and some review groups support this. Many reviews and groups are unclear with respect to qRCTs. While it is a good idea to use only trials at low risk of bias this is unlikely to be achieved by excluding qRCTs that are the better reported ones.
Objectives: To determine how many authors specify that they exclude quasi-randomised trials from Cochrane Reviews, and what review group instructions say.
Methods: A random sample of 300 Cochrane Reviews was selected and the methods scrutinised to see if qRCTs were excluded. Details of all review groups were read to determine policies regarding qRCTs.
Results: Forty-seven of the 300 stated that qRCTs were excluded (15.7%, 95% CI 11.7–20.3%). In addition, of the remaining 253 trials, 138 simply stated that RCTs were included (54.5%, 95% CI 48.2–60.8%). In these it was unclear whether this referred to securely randomised trials only. One Cochrane Review group states in their suggested methods that qRCTs should be excluded, and six suggest that they should only be included if there are only a few or no RCTs, or only for adverse effects. Seven review groups say that RCTs should be included with no further elucidation so it is unclear whether they would like qRCTs to be excluded.
Conclusions: Contrary to the advice in the handbook many Cochrane Reviews attempt to exclude qRCTs and some review groups support this. Many reviews and groups are unclear with respect to qRCTs. While it is a good idea to use only trials at low risk of bias this is unlikely to be achieved by excluding qRCTs that are the better reported ones.