Main Organiser

Julius Centre University of Malaya

Co-organiser

Department of Social and Preventive Medicine, Faculty of Medicine, University of Malay

Supported by

University of Malaya

TESTING THE AGREEMENT OF MEDICAL INSTRUMENTS: OVERESTIMATION OF BIAS IN THE BLAND-ALTMAN ANALYSIS

Author

Zaki R 1 , Bulgiba A 1 , Ismail N 2

Institution

1. Julius Centre University of Malaya, Department of Social & Preventive Medicine, Faculty of Medicine, University of Malaya.
2. Department of Applied Statistics, Faculty of Economics & Administration, University of Malaya.

Abstract

Objectives: The Bland-Altman method is one of the most popular methods to assess the agreement of medical instruments. It is recommended that the slope of regression line fitted to the Bland-Altman plot should be tested to exclude a proportional bias. The aim of this article is to show that there is an overestimation of bias in the Bland-Altman analysis, even when the proportional bias was excluded.

Methods: A cross-sectional study was conducted to obtain data for this study. Variables collected were blood glucose level, body weight, and systolic blood pressure. The Bland-Altman analysis was performed for the original clinical data set compared with twenty randomly generated data sets with a specific range of bias for all variables.

Results: A total of 300 participants per variable consented for this study. Proportional bias was excluded for all datasets. The range of predicted bias was higher than the simulated bias for all datasets. The overestimation of bias increases as the range of actual bias increases.

Conclusion: Testing the slope of regression line of the Bland-Altman plot does not remove the artifactual bias in the prediction. The Bland- Altman method should be used with caution and should be complemented by other methods.