Bias and uncertainty in comparative diagnostic accuracy research

Healthcare professionals frequently ask how the accuracy of one diagnostic test compares with that of another. To compare the accuracy of tests, studies that directly compare two or more tests are performed, called comparative accuracy studies. These studies can be designed with shortcomings that render the results at risk of being biased. Guidance on how to identify these shortcomings and assess risk of bias can help those performing systematic reviews and those designing comparative accuracy studies. This thesis describes the development a tool for assessing the risk of bias in comparative accuracy studies, named QUADAS-C (Quality Assessment of Diagnostic Accuracy Studies – Comparative). The thesis also describes preparatory work that informed the development of this tool, a guidance on how to assess the overall certainty of a body of evidence about comparative accuracy, and an application of the developed methods in a systematic review about point-of-care tests for tuberculosis disease.