The devil in the detail: data extraction forms for systematic reviews of complex interventions

Stephanie Taylor, Sonja Hood

Objectives:

To gain a better understanding of the challenges of extracting data from reports of trials, particularly from trials of complex interventions;

To share strategies and good practice around the approach to data extraction and the design of data extraction forms for trials, particularly for trials of complex interventions;

To collate the issues raised and the conclusions of the workshop into a brief report offering practical guidance to those working in this area (and which we hope to submit to the editors of the Cochrane Handbook for Systematic Reviews of Interventions for consideration for inclusion in the section on data extraction).

Summary: Complex interventions in health care are non-pharmacological interventions "made up of various interconnecting parts"1. Trials of complex interventions are increasingly becoming the subject of systematic reviews. Data extraction is a key step in systematic reviewing and the content and structure of the data extraction form (either electronic or paper) shape the completed systematic review. We have found that data extraction may be particularly difficult where reports of complex interventions are concerned since published accounts of trials of complex interventions are often very dense and the multi-faceted nature of the interventions predisposes the reporting of a wide variety of endpoints and outcomes (not to mention the diverse elements of the intervention itself). In our experience, the design of the data extraction form is intimately linked to the overall quality of the completed review. A well-designed data extraction form can promote rigour, validity and efficiency. In this workshop we will discuss practical issues around data extraction and share some of our ideas about the way data extraction might be approached. We will also examine how other people's data extraction tools have shaped their completed systematic reviews. The session will be very participatory and a real 'work shop' with people bringing examples of their own data extraction tools and sharing examples of good data extraction tools and practice. Findings of the workshop will be compiled into a set of simple, practical guidance to help others working in the field. References 1. Campbell M, Fitzpatrick R, Haines A, Kinmonth AL, Sandercock P, Spiegelhalter D, Tyrer P. Framework for design and evaluation of complex interventions to improve health. BMJ 2000; 321:694-6. Level of knowledge required to attend: basic (but experience of data extraction and an understanding of complex interventions would be helpful).

Dublin 2006 -- W29