A model for the harmonisation of test results of different quantitative D-dimer methods

The numerical test results of different D-dimer assays vary widely. Because of the complexity of the analyte of target as well as the variability in specificity of different D-dimer assays, only harmonisation of the test results seems to be feasible. The use of a single conversion factor does not take into account for several methods the lack of commutability between test results and consensus values at different D-dimer levels. This is probably related to the mutually different response of methods to high and low levels. We therefore designed a harmonisation model based on the transformation of a method-specific regression line to a reference regression line. We used the data for the measurement of a set of plasma samples with different D-dimer levels by 353 different laboratories using 7 of the most frequently used quantitative D-dimer methods. For each method we calculated the method-specific consensus value for each sample. The overall median value was also estimated. Per method linear regression was applied throughout the method-specific consensus values using the amount of patient pooled plasma added to the different plasma samples as the independent variable. The line through the overall median values of all 7 methods was used as the reference line. Harmonisation between the methods was obtained by transformation of the method-specific regression line to the reference line. This harmonisation resulted in a reduction of the variability between the method-specific consensus values from about 75% to about 5.5%. Clinical validation of this concept had shown significant improvement of the test result comparability. We conclude that this model is a feasible approach in the harmonisation of D-dimer methods. If the harmonisation procedure is included in the calibration procedure by the manufacturers, customers will automatically obtain harmonised test results.