Correction for lack of coincidence of normal and abnormal calibration slopes in ISI determination

An underlying assumption of the WHO orthogonal regression model for prothrombin time standardisation is that a single line describes the relationship between PT of abnormal and normal plasmas. The aim was to evaluate Tomenson's correction for lack of coincidence using the model of the ECAA human reagent with lyophilized plasma calibrations. Local ISI calibrations using ECAA lyophilized normal and abnormal plasmas were performed on coagulometers with the ECAA human thromboplastin. They gave a high incidence of non-linearity of the two calibration slopes. The mean percentage deviation from assigned values of the target coumarin plasmas coagulometer INR at 45 centres before local ISI correction was 16.6% and 11.9% respectively in two studies. After local ISI calibration, the unsigned deviation was reduced to 5.54% and 6.61% respectively. Tomenson's correction for non-linearity further reduced the mean deviation to 3.08% and 2.96% respectively in the two studies demonstrating the value of the procedure.