For determination of the international normalized ratio (INR), it has
been suggested that ''highly sensitive'' thromboplastin reagents (Inte
rnational Sensitivity Index [ISI] less than or equal to 1.2) provide t
he most consistent performance and minimize interlaboratory variabilit
y. We compared the INR values obtained from 69 specimens drawn from pa
tients receiving long-term oral anticoagulant therapy, using four thro
mboplastin preparations (manufacturer-assigned ISI range of 0.96-1.10)
and two automated photo-optical analyzers. Multivariate analysis of t
he INR response matrix (552 INR values) indicated that the eight reage
nt-coagulometer combinations did not produce equivalent INR values. Si
milar analysis indicated that INR values were not normalized when unco
rrected prothrombin ratios or INR values, calculated after assignment
of ''local ISI values'' to each thromboplastin reagent, were compared.
The INR differences also seemed to be clinically significant because
17% to 29% of paired thromboplastin values were discordant when all IN
R values were assigned to one of four therapeutic categories used in o
ral anticoagulant therapy (< 2.0; 2.0-3.0; 3.0-4.5; or > 4.5). These d
ifferences in INR values obtained with two photo-optical coagulometers
and four highly sensitive thromboplastin reagents suggest that the ex
isting INR system has not achieved the goal of standardized prothrombi
n time values and does not support the recommendation to use only high
ly sensitive reagents for the regulation of oral anticoagulant therapy
.