ABSTRACT
Diagnostic testing is an important biostatistical component of clinical research and requires the understanding of probability concepts related to conditional probability (see Chapter 2: Probability). Chapter 3 covers basic biostatistics concepts related to diagnostic testing and screening procedures, including calculation of sensitivity, specificity, false positive, false negative, positive predictive values, and negative predictive values of a screening test. Concepts are also presented in an epidemiologic framework (incidence, cumulative incidence, prevalence). We discuss the use of receiver operating characteristic (ROC) curves to determine cutoff points for screening tests and to demonstrate how to develop these curves using data in statistical software programs (SAS and Stata; appropriate statistical code is provided for both programs). Readers will learn how to choose cutoff points to optimize sensitivity and specificity of tests based on the ROC curve. Also, in this chapter, Bayes theorem is applied to the calculation of predictive value (positive and negative) of the screening tests.
