As one of the statistical reviewers for the Cochrane Infectious Diseases Group www.evidence4health.org, Tonya Esterhuizen, had the opportunity to attend a three day course on statistical methods for diagnostic test accuracy (DTA) reviews run by the Cochrane Collaboration in Birmingham last week. All the international authorities on DTA reviews had converged in Birmingham for a Cochrane DTA review conference, so the teaching faculty read like a “who’s who”. Coordinated by Yemisi Takwoingi, the course featured the likes of Patrick Bossuyt, Petra Macaskill, John Deeks, Richard Riley, and Mariska Leeflang, who all had a part in shaping the development of DTA reviews. The beautiful leafy campus of Birmingham University was the venue, and the uncharacteristically hot and humid weather was an unexpected challenge posed to delegates, as air conditioning is not usually considered necessary in that part of the world.
The thirty five participants on the course were a mixture of statisticians, researchers and clinicians, all with affiliations to either a Cochrane Centre or Review Group. There was a large contingent from the Dutch Cochrane Centre, and representation from other European countries such as Greece and Italy. But most delegates were from in and around the UK, and most were in the process of completing a DTA review themselves.
DTA reviews have been taking far longer than intervention reviews to complete and publish, and this is partly due to the lack of peer review and referee expertise in this area. Hence the need to conduct training sessions like this one, with the aim to recruit trainees as peer reviewers and referees for the DTA review group in the future. Apart from the statistical aspects, study design issues, risk of bias and how to interpret and report the findings were considered. Analysis of DTA reviews has a relatively recent history and is still under development. It is important to ensure that the person doing the analysis considers and understands what they are doing, and prevents the analysis from becoming purely mechanical. For this reason it is key to involve a statistician trained in these methods and with a full understanding of the spectrum of methodological issues in the DTA review.
For more information on DTA reviews: http://srdta.cochrane.org/handbook-dta-reviews
Tonya Esterhuizen, tonyae@sun.ac.za
