Abstract

The challenges of evidence-based policy making in areas such as public health, social care, education and criminal justice are located in the complexity of the problems addressed and the prominence of methods derived from evidence-based medicine which were originally developed for assessing well-defined clinical interventions. The retrieval of evidence on complex topics is identified as being particularly problematic. The application of established systematic review search methods is difficult due to the interdisciplinary nature of many of the topics under review, the absence of clearly defined questions, the dynamic process of sense-making involved in reaching an understanding of complex phenomena and the wide range of direct and indirect evidence required to address complex decision problems. Difficulties with searching have typically been addressed using pragmatic coping strategies to make the search process more manageable. These may be judged acceptable on a project by project basis but taken as a whole they could be judged as making the search methods available to us methodologically vulnerable. Looking beyond search methods as a set of procedures it is possible to identify a methodological gap between the theoretical information retrieval model that underpins established search methods and the tasks involved in reviewing complex topics. This talk explores this methodological gap and seeks solutions from the wider discipline of information science.

Suzy Paisley is a Senior Research Fellow and is Head of the Information Resources Group at ScHARR, University of Sheffield. She has extensive experience of searching for evidence to inform policy decision-making in a wide range of health related areas. Previously she has been the Managing Director of two of ScHARR’s NICE assessment programmes to inform Technology Appraisal and Public Health guidance. Her research interests focus on the development of search methods for decision-analytic models of cost-effectiveness and systematic reviews of complex topics.