Dr. H.(Hans) Reitsma

The number of primary studies evaluating diagnostic tests is rising. Critically summarizing such studies in a systematic review is beneficial for health care professionals seeking the best evidence about the use of diagnostic tests. Reviews of diagnostic studies are more challenging because of more variation in questions & designs, specific sources of bias & variation, and the need to use more complex statistical approaches to meta-analysis. Several advances have recently been made in the methods to perform such reviews. In this course we discuss and practice the methods to assess the methodological quality of primary diagnostic test accuracy studies (QUADAS-2 instrument), the statistical models to meta-analyze the paired measures of test accuracy (bivariate meta-regression model of sensitivity and specificity), and how to critically read and interpret the findings of systematic review of diagnostic studies. The course consists of plenary presentations, small-group discussions, and computer exercises using the statistical software package R.

At the end of the course, the participant is able to:

• Explain the rationale for performing a systematic review of diagnostic studies
• List the key steps of a systematic review of diagnostic studies
• Formulate a focused review question addressing a diagnostic problem
• Understand the key issues when building a proper search strategy to identify relevant  diagnostic studies in electronic bibliographic databases
• Explain the main types of bias in a diagnostic study
• Assess the risk of bias in a diagnostic study using the QUADAS-2 instrument
• Explain in words the assumptions of the bivariate random effects model for meta-analyzing diagnostic accuracy data
• Interpret the results of the bivariate model with and without covariates
• Fit basic bivariate random effects models in statistical package R
• Interpret the main findings of a review in light of the review question(s) 
• List the key advantages of individual participant data meta-analysis compared to traditional aggregate meta-analysis in the diagnostic setting 

In this course we expect participants to have a basic knowledge about the principles of diagnostic research. This includes the basic design of diagnostic accuracy, the various measures of test performance (e.g. sensitivity & specificity, predictive values, odds ratio,  ROC curves). Furthermore, computer exercises will be done using the statistical software R, so an understanding of the basic commands in R is anticipated (see also R tutorials listed under day 1).