Is the international normalised ratio ( INR ) reliable ?

Aim—To determine the reliability of international normalised ratio (INR) measurement in primary care by practice nurses using near patient testing (NPT), in comparison with results obtained within hospital laboratories by varied methods. Methods—As part of an MRC funded study into primary care oral anticoagulation management, INR measurements obtained in general practice were validated against values on the same samples obtained in hospital laboratories. A prospective comparative trial was undertaken between three hospital laboratories and nine general practices. All patients attending general practice based anticoagulant clinics had parallel INR estimations performed in general practice and in a hospital laboratory. Results—405 tests were performed. Comparison between results obtained in the practices and those in the reference hospital laboratory (gold standard), which used the same method of testing for INR, showed a correlation coeYcient of 0.96. Correlation coeYcients comparing the results with the various standard laboratory techniques ranged from 0.86 to 0.92. It was estimated that up to 53% of tests would have resulted in clinically significant diVerences (change in warfarin dose) depending upon the site and method of testing. The practice derived results showed a positive bias ranging from 0.28 to 1.55, depending upon the site and method of testing. Conclusions—No technical problems associated with INR testing within primary care were uncovered. Discrepant INR results are as problematic in hospital settings as they are in primary care. These data highlight the failings of the INR to standardise when diVerent techniques and reagents are used, an issue which needs to be resolved. For primary care to become more involved in therapeutic oral anticoagulation monitoring, close links are needed between hospital laboratories and practices, particularly with regard to training and quality assurance. (J Clin Pathol 1999;52:494–497)