Lo et al evaluated the sensitivities of various lupus anticoagulant assays [ I ] , including the system we developed, the rabbit brain neutralization procedure (RBNP) [2], referred to by Lo et al. as the StdiHigh-PC APTT. The RBNP is a kaolin-activated partial thromboplastin time system employing a commercially available rabbit brain phospholipid (Platelin, General Diagnostics) in which clotting time ratios employing onefold standard and fourfold-increased lupus anticoagulant-"neutralizing" phospholipid concentrations are generated, and the difference (onefold minus fourfold) calculated.The data upon which Lo et al. declared that the RBNP is insensitive require reinterpretation. The problem stems from the fact that we had a category of "indeterminate" test results. In our original description, we required of a positive test that the difference between ratios exceed the mean + 2 SD of normal subjects (ratio difference 20.17) und that the fourfold clotting time ratio fall to within the normal range. We have since dropped the latter requirement because it reduces test sensitivity for two reasons: 1 ) we have observed patients with lupus anticoagulants and no other coagulation disturbance whose fourfold ratios do not completely normalize, as Lo ct al. observed, and 2) the fourfold ratio may not normalize when there is a coexisting coagulation disturbance, for example, due to warfarin therapy. We used to apply the term "indeterminate" to describe such results. Lo et al. have suggested that "indeterminate" is not a useful classification, and we agree. Elimination of this category of results does not alter any of the conclusions in our original publication. We now simply express RBNP results as a ratio difference and classify them either as positive (20.17) or negative (<0. 17).In their testing of 18 undiluted APTT inhibitor plasma.\, Lo et al. found the ratio differences in the RBNP to exceed 0.17 in all cases, but because they considered their eight "indeterminate" results to be negative, they calculated a sensitivity of 10118. Had they considered the "indeterminate" results positive, as discussed above, then 181 I8 plasmas were positive.As a separate issue, we are concerned that Lo et al. made specific recommendations concerning which lupus anticoagulant assays should be utilized in routine hospital practice based on analysis of only 18 samples. Calculated from the binomial distribution, the 95% confidence limits for a test scoring 18/18 is fairly broad (81.5-100%), overlapping the confidence limits of a test scoring, for example, 1311 8 (46.5-90.3540). A two-sided Fisher's exact test to determine whether the sensitivities of two tests scoring 13/18 and 18/18 are equal is of borderline significance ( P = 0.045). When account is taken of the many pairwise comparisons possible in analysis of a sizeable group of tests, a P value considerably lower than 0.05 is desirable for declaring the significance of any given pair comparison. Lo et al. found that 9/10 assays (including the RBNP) yielded positive results in 13/18 to...