Objectives: In this workshop the principles of meta-analysis of diagnostic test accuracy studies will be addressed in a practical way. At the end of the workshop, participants will be able to:

- summarise the results of individual diagnostic test accuracy studies graphically and numerically;

- understand the rationale for summarising data using summary ROC curves to account for explicit and implicit differences in threshold between studies;

- conduct basic exploration of sources of heterogeneity.


Summary: This workshop will present a framework for the meta-analysis of diagnostic test accuracy studies as will be implemented within Cochrane reviews of diagnostic test accuracy. We will start with an overview of the various statistics to express the diagnostic performance of an index test, such as sensitivity, specificity, likelihood ratios and the diagnostic odds ratio. Then methods for graphically summarising results from original studies will be presented: forest plots of sensitivity and specificity (including 95% confidence intervals) and a plot of sensitivity-specificity pairs in a ROC-space. Basic methods of estimating a summary ROC-curve will be presented as they are implemented in RevMan 5, and the rationale for using these methods explained. The workshop will conclude with an overview of possible sources of heterogeneity and methods for assessing heterogeneity.

Level of knowledge required to attend: intermediate.