In the process of establishing a diagnosis, the clinician calculates the disease probabilities, which keep changing as data is gathered from clinical history, physical examination and laboratory and imaging data. This paper sets out to explain, in a simple and practical way, how the characteristics of a test can be applied in order to determine post-test probability, that is, the probability of a specific disease being present. ⋯ To calculate the post-test probabilities, the sensitivity and specificity of the test are used to calculate the positive and negative predictive values for a determined pretest probability, as well as likelihood ratios (through graphics or nomograms, approximations or equations). Among the test characteristics, the most useful for calculating post-test probabilities are likelihood ratios, which have several advantages over sensitivity and specificity.
Sérgio Paulo, Susana Mendes, Ricardo Vizinho, and António Vaz Carneiro.
Centro de Estudos de Medicina Baseada na Evidência, Faculdade de Medicina de Lisboa, Lisboa, Portugal.
Rev Port Cardiol. 2004 Sep 1;23(9):1187-98.
AbstractIn the process of establishing a diagnosis, the clinician calculates the disease probabilities, which keep changing as data is gathered from clinical history, physical examination and laboratory and imaging data. This paper sets out to explain, in a simple and practical way, how the characteristics of a test can be applied in order to determine post-test probability, that is, the probability of a specific disease being present. Before applying the test, we start with a pretest probability that corresponds to the prevalence of the disease. With each additional test this probability changes, until a threshold is reached that includes or excludes the clinical diagnosis. To calculate the post-test probabilities, the sensitivity and specificity of the test are used to calculate the positive and negative predictive values for a determined pretest probability, as well as likelihood ratios (through graphics or nomograms, approximations or equations). Among the test characteristics, the most useful for calculating post-test probabilities are likelihood ratios, which have several advantages over sensitivity and specificity.