For citation purposes: Thomas JM. Diffusion of innovation in systematic review methodology: Why is study selection not yet assisted by automation? OA Evidence-Based Medicine 2013 Oct 21;1(2):12.

Review

 
Systematic Reviews

Diffusion of innovation in systematic review methodology: Why is study selection not yet assisted by automation?

J Thomas
 

Authors affiliations

(1) EPPI-Centre, Social Science Research Unit, Institute of Education, London

* Corresponding author Email: j.thomas@ioe.ac.uk

Abstract

Introduction

Systematic reviews, the foundation of much Evidence-Based medicine, are suffering from increasing ‘data deluge’: reviewers often need to manually assess many thousands of titles and abstracts to determine their relevance. Automation has been advanced as a potential solution; but given that its efficacy was first demonstrated in 2006, why is it not yet widely used? The Diffusion of Innovations framework by EM Rogers is used to structure an exploration of why this might be the case.

Discussion

According to Rogers, five characteristics affect the rate of adoption of innovations: those perceived as having greater relative advantage, compatibility, trialability and observability, and less complexity, will be adopted more rapidly than others. The relative advantage of automation has been demonstrated empirically, though usually in narrowly focused reviews in clinical areas, rather than more challenging areas for automation, such as public health. Detailed methods and procedures for their use have yet to be established, addressing transparency, replicability and reporting practices. Although issues concerning the compatibility of new technology with existing infrastructure are probably surmountable, the use of automation may challenge contemporary notions of what constitutes a systematic search and how publication bias is addressed using sensitive search techniques.

The remaining factors are interrelated. The technologies are complex, both to understand and to deploy. This affects the trialability of automation: technical expertise is required and there are thus few opportunities for reviewers to observe others using these technologies.

Conclusion

Further technical and empirical work is needed where systematic reviewers work with information and computer scientists to develop solutions which have a demonstrative relative advantage and which are clearly compatible with the needs of systematic reviewers and their users. Such work may have a significant role to play in addressing the deluge of new research publications which threaten to overwhelm systematic review processes.

Licensee OA Publishing London 2013. Creative Commons Attribution License (CC-BY)
Keywords