Screening and the utility of textual analysis software: a retrospective study of RefVizTM

Article type
Authors
Kavanagh J, Thomas J, Brunton J, Rees R, Brunton G
Abstract
Background: Screening of titles and abstracts, and subsequent retrieval of full text papers to screen against inclusion criteria is highly resource intensive. This is particularly the case in systematic reviews that synthesise the findings of studies other than trials and / or synthesise research from the social sciences. Challenges to systematically identifying such literature are often met by developing highly sensitive search strategies. This translates into screening much irrelevant research to find one relevant study. For example in a recent mixed methods systematic review of HIV health promotion and men who have sex with men (MSM) 12,257 titles and/or abstracts were screened, followed by screening of 623 full text papers, 253 were included in a map of the literature 25 of which met the inclusion criteria for in-depth review[1].

RefVizTM is a new interactive text analysis and visualisation software application designed to aid the researcher dealing with large numbers of references and abstracts[2].

Objectives: To explore whether RefVizTM software is a useful tool for the screening stage of a systematic review. Does RefVizTM allow the researcher to interrogate a reference collection to reduce the workload of screening? Can it be used to differentiate between literature on the same topic area but which uses different research methodologies, e.g. between outcome evaluations and research which presents the health related views and experiences of the population targeted by the review (e.g. focus groups, in-depth interviews)? The use of methodological filters are not advisable for comprehensive searches to support systematic reviews of social interventions and it is hoped that RefVizTM will reduce the amount of screening and enable the swifter identification of views studies which are often the most difficult to obtain.

Methods: A retrospective analysis of 253 bibliographic records of those papers included in the mapping stage of the MSM review will be imported into RefVizTM. Analysis will take place to explore the following;

1. How does RefViz organise the records based only on the data available at the import stage? Are they organised into distinct conceptual groups such as those with a similar population, intervention type or study design? 2. How can the researcher use the interactive tools in order to organise the records according to study design or other concepts?

Results: Results will be used and tested prospectively at the screening stage of another mixed methods review.

Conclusions: Screening methods are often overlooked in terms of advancing the methods of systematic reviews. Conclusions from both analyses will be used to inform the advancement of screening methodology.

1. Rees R, Kavanagh J, Burchett H, Shepherd J, Brunton G, Harden A, Thomas J, Oliver S, and Oakley A. HIV Health Promotion and Men who have Sex with Men (MSM): A systematic review of research relevant to the development and implementation of effective and appropriate interventions. (Under review). 2004. London, EPPI-Centre, Social Science Research Unit, Institute of Education, University of London. 2. ISI Software. RefViz. 2004. http:// www.adeptscience.co.uk/go?pg=AA503