Measurement Uncertainty and the Prostate Health Index-A Review.

Abstract

Background: Not all quantity values determined in a medical laboratory are obtained by direct analytical measurement. In many situations, a quantity value is calculated from other measurements through a functional relationship, where the output quantity is derived from one or more input quantities by applying a defined mathematical equation. Even though important for clinical interpretation, the measurement uncertainty (MU) of calculated quantities may not always be considered.

Content: The prostate health index (phi) has been shown to improve the clinical assessment of prostate cancer. Estimates for the MU of phi have been recently provided by a novel approach in which the uncertainty of phi was directly calculated from internal quality control (IQC). However, this method for determining MU generally provides a higher estimate than the procedure recommended by the "Evaluation of Measurement Data-Guide to the Expression of Uncertainty in Measurement" (GUM) and does not allow for the incorporation of a correlation term. A full evaluation of the MU for phi by the GUM procedure is provided.

Summary: The importance of the GUM approach with the inclusion of correlation terms is clearly shown by the phi calculation. By comparison, the relative standard uncertainty obtained by the direct IQC procedure for a mean phi value of 24.48 was given as 7.2%, while by GUM, the relative standard uncertainty for the same mean phi was 3.60% with correlation included and 5.99% without correlation. The influence of correlation terms within the GUM equations is clearly shown for the phi calculation.