A Statistician’s Independent Evaluation

Abstract

The European Community Bureau of Reference (BCR) calibration study has established the statistical basis for calibration of thromboplastins (1). Previous approaches to calibration (2,3,4) have been based on an observation by Biggs and Denson (5) that Prothrombin ratios obtained with two different thromboplastins plotted against each other are distributed around a straight line passing through the point (1,1). However, the Biggs-Denson procedure was found to be inadequate in some cases and an alternative calibration model was adopted in the BCR study. A full discussion of the failings of the Biggs-Denson procedure and the arguments for proposing this alternative calibration model are given by Kirkwood (6). The new calibration procedure has now been used succesfully to calibrate three new BCR reference thromboplastins (human, bovine, and rabbit) against the WHO International Reference Preparation (IRP). It is now possible for manufacturers to calibrate their reagents against the BCR reference materials and the results of a Prothrombin test may be reported as an International Normalised Ratio (INR), the theoretical ratio which would have been obtained with the IRP. The procedure has also been used successfully to calibrate the proposed second WHO IRP (BCT/253) against the existing IRP in an international collaborative exercise (7,8). In this latter exercise it was found necessary to modify the procedure in a manner to be described. Furthermore this report describes a detailed retrospective examination of the BCR calibration procedure which reveals a number of other issues which have not yet been resolved satisfactorily and merit further discussion.