Conferences 2018

Evidence synthesis of diagnostic test accuracy for decision making: a 2-day WinBUGS course

  NICE Clinical Guidelines Technical Support Unit & NIHR Complex Reviews Support Unit

 25th January 2018 – 26th January 2018


Course details

CRSU have 5 free places (course fees only – travel and accommodation will not be covered) for NIHR funded and Cochrane UK researchers.

Contact Professor Alex Sutton;


The course will be held at the National Institute for Health and Care Excellence (NICE) London offices:

10 Spring Gardens,



United Kingdom


Closing date - 9th January 2018.


Course Summary


This is a 2-day course on evidence synthesis of diagnostic test accuracy data to inform clinical decision making. It covers:

-         Receiver operating characteristic (ROC) curves

-         Bivariate meta-analysis of diagnostic test accuracy and the hierarchical summary ROC (HSROC) curve model

-         How to include multiple points on the ROC curve from some studies

-         Quantifying summary measures of test accuracy at alternative thresholds for a positive test result

-         Selecting the optimum point on the ROC curve to maximise an objective function, eg the expected net benefit in an economic decision model

-         Exploring reasons for heterogeneity with covariates

-         Evidence synthesis for comparing the accuracy of two or more tests, and combinations of tests

-         Synthesis of test accuracy in the absence of a ‘gold standard’ test


Methods for systematically reviewing the literature will not be covered in this course.




The course material is set at a relatively advanced level.


Participants must have previous experience in WinBUGS, for example having attended an Introduction to WinBUGS course or the tutors’ 3 day ‘Indirect and Mixed Treatment Comparisons’ course at the University of Leicester. We will not cover how to use WinBUGS in this course.


Participants will need to bring a laptop with WinBUGS installed.


What will participants learn?


Participants will learn how to meta-analyse diagnostic test accuracy data and how to go beyond the ‘standard’ methods in order to: include multiple points on the ROC curve from some studies, quantify how test accuracy varies with the threshold for a positive test result, select the optimum point on the ROC curve, explore reasons for heterogeneity, compare the accuracy of multiple tests and account for imperfect reference standards.


Participants will learn how to carry out and interpret results from these analyses in WinBUGS.