Article type
Year
Abstract
Background: Recent studies have shown that final reports do not always correspond to what authors originally set out to study in protocols. This phenomenon, which has only been assessed in primary studies, can affect the quality of reporting and can generate selective reporting of results.
Objectives: To assess the agreement between the outcomes stated in protocols and those reported in systematic reviews.
Methods: Original protocols, and the subsequent full reports, of Cochrane systematic reviews (thereafter called 'Cochrane protocolreview pairs' (CPRPs) published in Issues 2,3,4 of 2005 and Issue 1 of 2006 were first identified and then a random sample of 186 CPRPs was drawn from them. For each CPRP agreement between protocols and final reports was assessed. A CPRP was classified as 'in agreement' if the outcomes were present in both protocols and reviews or when the reviewers report that an outcome was not available in primary studies. Agreement between classification of primary/secondary outcomes was also assessed.
Results: A preliminary analysis based on 19 CPRPs has shown that: four (21%) did not provide the protocol, in five (33%) disagreement between the protocol and the final report was observed and in two (13%) primary and secondary outcomes were switched. Results based on a larger sample will be provided at the Colloquium.
Conclusions: One of the stated advantages in Cochrane systematic reviews is the commitment of authors to prepare and publish a protocol before a systematic review is carried out. Therefore, it is important in The Cochrane Collaboration to assess whether this is happening in practice and this study was set out for this.
Objectives: To assess the agreement between the outcomes stated in protocols and those reported in systematic reviews.
Methods: Original protocols, and the subsequent full reports, of Cochrane systematic reviews (thereafter called 'Cochrane protocolreview pairs' (CPRPs) published in Issues 2,3,4 of 2005 and Issue 1 of 2006 were first identified and then a random sample of 186 CPRPs was drawn from them. For each CPRP agreement between protocols and final reports was assessed. A CPRP was classified as 'in agreement' if the outcomes were present in both protocols and reviews or when the reviewers report that an outcome was not available in primary studies. Agreement between classification of primary/secondary outcomes was also assessed.
Results: A preliminary analysis based on 19 CPRPs has shown that: four (21%) did not provide the protocol, in five (33%) disagreement between the protocol and the final report was observed and in two (13%) primary and secondary outcomes were switched. Results based on a larger sample will be provided at the Colloquium.
Conclusions: One of the stated advantages in Cochrane systematic reviews is the commitment of authors to prepare and publish a protocol before a systematic review is carried out. Therefore, it is important in The Cochrane Collaboration to assess whether this is happening in practice and this study was set out for this.