How to: evaluate a diagnostic test

Abstract

Background: The development of an in vitro diagnostic test from a good idea to a clinically relevant tool takes several steps, with more stringent requirements at every step.

Objectives: This article aims to summarize the necessary questions to be asked about a test and to illustrate study designs answering these questions. We also aim to relate Regulation (EU) 2017/746 to the needs of evidence-based diagnostic testing, where applicable.

Sources: We used literature on evidence-based diagnostics, a text book on clinical trials in the development and marketing of medical devices and the English version of Regulation 2017/746 of the European Parliament and of the Council on in vitro diagnostic medical devices.

Content: The combination of different test uses and different stages of development determine the required test characteristics and suitability of study designs. In an earlier stage of test development it may be crucial to know whether a test can differentiate diseased persons from healthy controls, although this tells us little about how a test will perform in practice. Later stages focus on the diagnostic accuracy of a test in a clinically relevant situation. However, a test that perfectly distinguishes between patients with and without a certain condition may still have little effect on patient outcomes. Therefore, randomized controlled trials of testing may be needed, as well as post-marketing monitoring.

Implications: Both researchers and users of tests need to be aware of the limitations of diagnostic test accuracy and realize that accuracy is only indirectly linked to people's health status.

Keywords: Clinical epidemiology; Diagnostic research; Diagnostic test accuracy; In vitro diagnostics; Medical test evaluation; Study designs.