Article type
Year
Abstract
Background: Cochrane methods recommend duplicate data extraction from primary studies to minimise bias. One of the features of Covidence, a new platform for the development of systematic reviews, is its ability to assist in this when extracting data from studies for inclusion in Cochrane Reviews. A similar task is extracting data from Cochrane Reviews. Several groups have searched the Cochrane database to identify implementable evidence or for mapping research gaps. We were faced with the task of extracting data from Cochrane Reviews reporting maternal or neonatal mortality or selected surrogate outcomes.
Objectives: To develop a tool which could assist in the identification and extraction of data from Cochrane Reviews.
Methods: We used Google forms as our basic concept to develop a tool we named the Cochrane Review Screening Tool (CReST). Data were extracted in duplicate by two data extractors working independently from the PDF version of relevant Cochrane Reviews directly into a Google form designed specifically for the purpose. Data included number of studies, total number of participants, effect size and 95% confidence interval and an assessment of quality. CReST was then used to compare data from the two data extractors and consensus was obtained between the two or by involvement of a third person. Finalised data were then automatically transferred to a spreadsheet for analysis.Only authorised data extractors had access to the toolkit and their experience was assessed.
Results: Users described CReST as time saving and convenient. The tool allows two data extractors to work independently at a different place and time. It allowed for real time communication or via comments. Documents were easy to share. It reduced the risk of data error during transfer between documents. Google form is widely accessible without additional cost. It is suitable for team tasks such as reaching consensus after duplicate data extraction.
Conclusions: CReST, although a relatively rudimentary tool uses an available platform. The toolkit could be applied to any project looking at extraction of data across a series of Cochrane Reviews or other large databases.
Objectives: To develop a tool which could assist in the identification and extraction of data from Cochrane Reviews.
Methods: We used Google forms as our basic concept to develop a tool we named the Cochrane Review Screening Tool (CReST). Data were extracted in duplicate by two data extractors working independently from the PDF version of relevant Cochrane Reviews directly into a Google form designed specifically for the purpose. Data included number of studies, total number of participants, effect size and 95% confidence interval and an assessment of quality. CReST was then used to compare data from the two data extractors and consensus was obtained between the two or by involvement of a third person. Finalised data were then automatically transferred to a spreadsheet for analysis.Only authorised data extractors had access to the toolkit and their experience was assessed.
Results: Users described CReST as time saving and convenient. The tool allows two data extractors to work independently at a different place and time. It allowed for real time communication or via comments. Documents were easy to share. It reduced the risk of data error during transfer between documents. Google form is widely accessible without additional cost. It is suitable for team tasks such as reaching consensus after duplicate data extraction.
Conclusions: CReST, although a relatively rudimentary tool uses an available platform. The toolkit could be applied to any project looking at extraction of data across a series of Cochrane Reviews or other large databases.