Challenges in harmonizing immunoassays: The use of the Bland-Altman based harmonization algorithm

Area of Science:

• Clinical chemistry
• Biomedical engineering
• Data harmonization

Background:

• Harmonization of clinical laboratory measurements is crucial for data comparability.
• Existing methods like weighted Deming regression-based harmonization (WD-BHA) have limitations.
• A novel Bland-Altman plot-based harmonization algorithm (BA-BHA) has been proposed.

Purpose of the Study:

• To compare the effectiveness of BA-BHA against WD-BHA.
• To evaluate the applicability of both harmonization algorithms for clinical measurands.

Main Methods:

• BA-BHA and WD-BHA were applied to 19 measurands using LiCA medical devices.
• Data from 80 patient sera per measurand were analyzed.
• Evaluation utilized Bland-Altman plots, curve estimation, and Passing-Bablok regression.

Main Results:

• BA-BHA resulted in mean differences close to zero and acceptable limits of agreement for all 19 measurands.
• Passing-Bablok regression confirmed acceptable harmonization effects for all measurands with BA-BHA.
• WD-BHA achieved acceptable harmonization effects for only a subset of the measurands.

Conclusions:

• BA-BHA demonstrates effective harmonization by adjusting both the mean and distribution of percent differences.
• BA-BHA exhibits more extensive applicability compared to WD-BHA.
• The study supports BA-BHA as a robust method for harmonizing clinical laboratory data.