Gc. Critchfield et St. Bennett, THE INTERNATIONAL NORMALIZED RATIO AND UNCERTAINTY - VALIDATION OF A PROBABILISTIC MODEL, American journal of clinical pathology, 102(1), 1994, pp. 115-122
The motivation behind the creation of the International Normalized Rat
io (INR) was to improve interlaboratory comparison for patients on ant
icoagulation therapy. In principle, a laboratory that reports the prot
hrombin time (PT) as an INR can standardize its PT measurements to an
international reference thromboplastin. Using probability theory, the
authors derived the equation for the probability distribution of the I
NR based on the PT, the International Sensitivity Index (ISI), and the
geometric mean PT of the reference population. With Monte Carte and n
umeric integration techniques, the model is validated on data from thr
ee different laboratories. The model allows computation of confidence
intervals for the INR as a function of PT, ISI, and reference mean. Th
e probabilistic model illustrates that confidence in INR measurements
degrades for higher INR values. This occurs primarily as a result of a
mplification of between-run measurement errors in the PT, which is inh
erent in the mathematical transformation from the PT to the INR. The p
robabilistic model can be used by any laboratory to study the reliabil
ity of its own INR for any measured PT. This framework provides better
insight into the problems of monitoring oral anticoagulation.