Evaluation of international normalized ratios by a controlled field survey with 4 different thromboplastin reagents.

A nationwide survey has been performed in Japan involving 75 laboratories to assess the relative reliability of different methods of reporting prothrombin time results in anticoagulant control. The interchangeability of results using prothrombin time, prothrombin activity percentage, prothrombin ratio and international normalized ratios (INR) were compared with four different thromboplastin reagents and a range of coagulometers. A secondary batch of reference thromboplastin of human brain origin (BCT/454) was used to calibrate the local thromboplastins and for comparison of methods of reporting. The study revealed the closest agreement of the results between BCT and the other reagents, and the regression lines of these reagents were almost identical, when the results were reported as INR. Box-Whisker plot analysis showed that the distribution of the results was large with the more deficient plasmas with all methods of reporting. It was found by this analysis that the interchangeability of the results was greatest when the results were expressed by INR, because the mean values obtained of each plasma using different thromboplastin reagents gave the lowest CV and the frequency of the far-out data was least, compared with the other methods of expression. On the other hand, the type of coagulometer had almost as much effect as the thromboplastin reagent on the prothrombin time, even if INR was used. Interchangeability of INR would be further improved by providing ISI values for each reagent/instrument combination.