How high can a correlation coefficient be? Effects of limited reproducibility of common cardiological measures

In clinical studies the linear correlation coefficient is commonly used to quantify the strength of the association between two variables, such as hei ght and weight: the value of r indicates whether the relationship is a stro ng one. However, actual clinical data includes an underlying physical varia ble plus an inevitable measurement error component that represents the repr oducibility of the test used. If test reproducibility is poor, then even if the underlying physical variables are perfectly correlated, the actual obs erved correlation coefficient cannot be one but must be somewhat less. We p resent a method for calculating the reduction in correlation coefficient du e to limited reproducibility, and discuss its implications with respect to experimental design and interpretation. (C) 1999 Elsevier Science Ireland L td. All rights reserved.