Clinical value of laboratory measurements

Abstract

Clinical accuracy of a laboratory test refers its ability to differentiate alternative states of healths, i.e, to classify patients into clinicaly relevant groups. Basic parameters and methods used for the assessment of clinical accuracy are the following: specificity, sensitivity, ROC-curve, predictive values and diagnostic efficiency. Sensitivity and specificity are inherent characteristics of laboratory tests, which are not influenced by the prevalence of target disease. ROC-curve, which is a sensitivity/specificity pair for different cut-off limits in the whole range of the obtained values, offers the best insight into the clinical performance of a laboratory test. The use of ROC-curve is the best way to compare clinical accuracies of different tests and to identify the cutoff limits that yield optimal test sensitivity and specificity. Predictive values are influenced by the disease prevalence, and are used mostly for interpretation of results or in building sequential diagnostic strategies. Diagnostic efficiency is a combined parameter which is easy to obtain, but is not very useful for the assessment of clinical accuracy.