Article type
Year
Abstract
Background: The Cochrane Schizophrenia Group and Enhance Reviews Ltd were funded by the National Institute of Health Research (UK) to perform a series of diagnostic test accuracy (DTA) Cochrane reviews. We describe the challenges faced when screening for DTA schizophrenia papers.
Objectives: We aim to discuss the balance between sensitivity/specificity when performing search strategies and how to screen large number of references. We also discuss what should be considered the reference standard in schizophrenia DTA reviews.
Methods: We have used the methodology recommended by the DTA Working Group to develop the review protocols. We used the protocols inclusion criteria to search MEDLINE, EMBASE, PsychINFO and The Cochrane Library (July 2011) and used a web-based software to handle references that were independently screened by two reviewers.
Results: We started by evaluating the diagnostic accuracy of Schneider’s First Rank Symptoms compared to clinical examination and modern operational criteria. Broad searches were recommended by the DTA Working Group, however, as these criteria appear in most schizophrenia research, the search strategies identified 32874 potentially relevant references. Therefore, we chose to perform literature screening for four reviews altogether. By screening titles and abstracts using a web-based software, two reviewers were able to complete the screen in 3 months, discarding 32 093 as irrelevant and ordering 781 PDFs , of which 296 (0.9%) were considered potentially relevant for at least one review. A fourth level of screen was done for each review separately.
Conclusions: There is a complete lack of standards by which DTA studies are referenced. Broad search strategies are impossible to avoid and content indexing should be improved. Authors are left with large volumes of titles to screen for a relatively few potentially ‘true positive’ studies. Using a web-based software has been invaluable to refine, record, and make accurate the drudgery of this work.
Objectives: We aim to discuss the balance between sensitivity/specificity when performing search strategies and how to screen large number of references. We also discuss what should be considered the reference standard in schizophrenia DTA reviews.
Methods: We have used the methodology recommended by the DTA Working Group to develop the review protocols. We used the protocols inclusion criteria to search MEDLINE, EMBASE, PsychINFO and The Cochrane Library (July 2011) and used a web-based software to handle references that were independently screened by two reviewers.
Results: We started by evaluating the diagnostic accuracy of Schneider’s First Rank Symptoms compared to clinical examination and modern operational criteria. Broad searches were recommended by the DTA Working Group, however, as these criteria appear in most schizophrenia research, the search strategies identified 32874 potentially relevant references. Therefore, we chose to perform literature screening for four reviews altogether. By screening titles and abstracts using a web-based software, two reviewers were able to complete the screen in 3 months, discarding 32 093 as irrelevant and ordering 781 PDFs , of which 296 (0.9%) were considered potentially relevant for at least one review. A fourth level of screen was done for each review separately.
Conclusions: There is a complete lack of standards by which DTA studies are referenced. Broad search strategies are impossible to avoid and content indexing should be improved. Authors are left with large volumes of titles to screen for a relatively few potentially ‘true positive’ studies. Using a web-based software has been invaluable to refine, record, and make accurate the drudgery of this work.