OVERVIEW

Many researchers who wish to carry out diagnostic accuracy meta-analysis are not familiar with rjags. To address this we developed a Shiny app that is available on the net here. The programs behind this are the ones described in our previous blogs articles illustrating Rjags programs for diagnostic meta-analysis model (Bivariate model and Bivariate latent class model).

This blog article is intended to serve as a guide for the usage of the R Shiny app that implement the models discussed in the blog articles mentioned above. The main goal will be to describe the various options in the app via applied examples. The interested reader can learn more about the statistical theory in Dendukuri et al(2012).

The R Shiny app provides options to perform both Bayesian Bivariate meta-analysis and latent class meta-analysis.

MOTIVATING EXAMPLE AND DATA

For our motivating example we will use the data from a systematic review of studies evaluating the accuracy of GeneXpertTM (Xpert) test for tuberculosis (TB) meningitis (index test) and culture (reference test) for 29 studies (Kohli et al(2018)). Each study provides a two-by-two table of the form

reference_test_positive Reference_test_negative
Index test positive tp fp
Index test negative fn tn

where:

  • tp is the number of individuals who tested positive on both Xpert and culture tests
  • fp is the number of individuals who tested positive on Xpert and negative on culture test
  • fn is the number of individuals who tested negative on Xpert and positive on culture test
  • tn is the number of individuals who tested negative on both Xpert and culture tests

We will explain later how to correctly supply the data into the app. Let’s first look at the 4 tab menus of the Shiny app.