Chapter - Diagnostic Studies - What the Clinician Needs to Know Regarding Methodology and Statistical Analysis of Assessments of a Diagnostic Test’s Performance

Abstract

Clinicians routinely face the challenge of diagnosing a medical condition, or assessing the severity and prognosis of disease based upon the diagnostic information at their disposal. Using such information to determine a diagnosis (e.g. imagining technology to diagnose glaucoma), is called a diagnostic test. Most diagnostic tests fall comfortably short of perfection in their diagnosis. When assessing a diagnostic test there are two related aspects to be concerned about – how well the test deals with diseased and non-diseased cases. Research is needed to assess the diagnostic performance and more generally the role a diagnostic test should play in clinical care: if, when and how it should be used. In this chapter, measures of diagnostic performance will be described using an example study. The challenges in conducting a robust and clinically relevant diagnostic study will be considered.

Keywords: Sensitivity, Specificity, Likelihood ratio, Predictive value, Diagnostic accuracy, Diagnostic performance, Diagnostic odds ratio, Youden index, Applicability, Prognosis, Diagnostic test.