A Measure of Confidence in Bland–Altman Analysis for the Interchangeability of Two Methods of Measurement

Bland-Altman (B-A) analysis has largely replaced the correlation coefficient as the predominant tool for evaluating the interchangeability of two methods of clinical measurement. However, we contend that B-A analysis might lead to erroneous conclusions when the data range is small. We provide an example to illustrate this and explore a possible analysis technique to address this limitation.