Anticoagulation variability between centres: implications for comparative prosthetic valve assessment.

One of the major determinants in the choice of a mechanical prosthetic valve is that valve's thromboembolic record but the thromboembolic (TE) rates may be substantially influenced by the levels of anticoagulation achieved. A detailed study of anticoagulation variability was undertaken in 834 patients who received one or more of a particular prosthesis (Medtronic-Hall) in one centre during a 7-year period from 1979 to 1987, but who attended 27 different anticoagulant clinics spread over a wide area. In addition, a questionnaire was sent to all 89 practising cardiac surgeons in the UK asking for their preferred range of International Normalised Ratio (INR) for patients with mechanical prosthetic valves. Both the local study (with 16,866 INR observations) and the national questionnaire (with a 53% response) revealed an enormous amount of variability. Median INR values (semi-interquartile range) varied from 2.2 to 3.9 (0.8-2.5) according to the anticoagulant clinic attended. The range of INR preferred by UK cardiac surgeons, but presumably not necessarily achieved, varied from 1.8-2.2 to 3.0-4.8, with 64% of surgeons preferring an INR less than 3.0. In comparison, standard US practice is to maintain prothrombin times equivalent to INR values of 4.0-5.0. Unless anticoagulant practice can be standardised internationally, comparison of TE complications between centres is meaningless, and casts doubt on the validity of TE rates quoted for particular prostheses, unless accompanied by a detailed analysis of anticoagulant control.