What’s wrong with pragmatic trials?

26 January 2017 - 17:00-18:30

LectureNancy Cartwright, University of Durham and University of California, San Diego (Work with Sarah Wieten)

Video Recording

(See this event also on Facebook)


In principle nothing is wrong with pragmatic trials: trials that ‘seek to determine the effectiveness of an intervention in a real-world setting to inform clinical decision making’ (Roland and Torgerson, 1998). In aid of this, pragmatic trials eliminate some of the exclusion conditions that are usual in what are labelled ‘ideal’ or ‘explanatory’ trials. The trouble comes with the concepts of ‘effectiveness’ and ‘external validity’. Positive results in well-conducted trials, whether ideal or pragmatic, show only that the treatment has worked for some members of the population enrolled in the trial. They cannot establish that it works in general nor what other factors help or hinder. Similarly, a pragmatic trial can establish that the treatment worked in the particular ‘real world’ setting – the one in which it was conducted, not that it works in ‘real world clinical practice’. Nor can they tell us what, if anything, is causally relevant in those settings. What can they do then? We shall argue that pragmatic trial results can be used in just the same way as ‘explanatory’ trial results: in tandem with a great deal of other evidence and theory, especial ‘midrange’ theory, to build credible local claims about how specific populations – or possibly even a specific individual – in specific places and circumstances may respond.