MetaDTA and MetaBayesDTA: Interactive web applications to conduct meta-analysis of diagnostic test accuracy studies
Alex Sutton, University of Leicester
Background Diagnostic tests form an essential part of current medical practices aiming to distinguish between patients with the disease and healthy individuals. They are used across a diverse range of healthcare settings and are often a pre-requisite to identifying treatment options and enabling access to services. Recommended statistical methods for meta-analysis of diagnostic test accuracy (DTA) studies require the fitting of complex non-standard statistical models, which can be a barrier to their application. MetaDTA (https://crsu.shinyapps.io/dta_ma/) is a free interactive online application which, for meta-analyses DTA studies using the bivariate model, plots the summary receiver operating characteristic curve, encourages sensitivity analysis and incorporates quality assessment results from the QUADAS-2 tool. MetaBayesDTA (https://crsu.shinyapps.io/MetaBayesDTA/) is a version that extends the functionality to allow for imperfect gold standards, subgroup analysis, meta-regression and comparative test accuracy evaluation. Both apps produce visualisations which facilitate the communication of results to all stakeholders, including patients and healthcare professionals. Due to the rich feature-set and user-friendly “point and click” interface of these apps, this workshop should appeal to a wide audience, including non-statistical experts. Objectives To illustrate how MetaDTA and MetaBayesDTA can be used to conduct DTA meta-analyses (using methods recommended in Version 2 of Cochrane Handbook) and for participants to get hands-on experience of using the apps in a structured and supportive environment. Description This workshop will provide an introduction to MetaDTA and MetaBayesDTA and offer participants experience using these web-based applications. During this workshop, we will: i. Demonstrate how MetaDTA and MetaBayesDTA can be used to conduct a Cochrane DTA meta-analysis using an example from a Cochrane reviews; ii. Demonstrate the wide array of interactive functions available within MetaDTA and MetaBayesDTA; iii. Provide participants with a dataset and worksheet designed to guide them, at their own pace, through the stages of conducting an analysis in MetaDTA and MetaBayesDTA; and iv. Provide time for questions, discussion and feedback. Topics covered will include loading data, customising summary receiver operating characteristics (SROC) plots, obtaining statistics, visualising study quality results and covariate effects, conducting sensitivity analyses and exporting tables and figures. All participants must bring a laptop or tablet or may share with another participant.