Assessing agreement between two methods of measuring blood glucose level without Bland–Altman limits

Abstract

ObjectiveTo describe a new method of assessing agreement between two methods of measuring blood glucose level in view of several problems with the currently used Bland-Altman method. MethodsStudy the deficiencies in the Bland-Altman method and propose a simple method that overcomes these deficiencies. The simple method does not require the calculation of Bland-Altman limits. ResultsClinical tolerance limits, which are needed anyway, can be directly used to obtain the agreement by the percentage of differences within these limits. This method is simple, robust, and nonparametric. It is more flexible also as the clinical tolerance limits can be varied for different values of the measurement such as to have close agreement at critical values and relaxed agreement at other values. Non-symmetric limits can also be set up under the simple method. ConclusionThe method of assessing agreement between two methods of measuring blood glucose level can be greatly improved by directly using clinical tolerance limits without calculating the Bland-Altman limits.