Comparison of activation and antigen assays for HIT by receiver operating characteristics (ROC) curve analysis.

For the orthopedic patients, the activation and antigen assays are compared at various cutoff points between negative and positive results. Activation assay (SRA, serotonin release assay, thick solid line). Although the sensitivity-specificity trade-off analysis was performed at 5% increments of serotonin release, the data points for cutoff 20% or more, 50% or more, and 90% or more serotonin release are shown. Antigen assay (EIA, enzyme immunoassay, thin solid line indicating standard dilution of blood sample $150$⁠; thin broken lines indicating higher dilutions [1/100, 1/250, 1/500, 1/750], as indicated). Although the sensitivity–specificity trade-off analysis was performed at 0.1-OD increments, the data points for cutoff of 0.5 or more, 1.0 or more, and 2.0 or more are shown. The ROC curve analysis suggests that both the activation and the antigen assays are highly informative for diagnosing HIT in this patient population. For comparison, the inset shows the full-scale ROC curve analysis for both activation and (at conventional 1/50 dilution) antigen assays together with a (theoretical) noninformative assay (shown as dashed line). The data indicate that the operating characteristics of the antigen assay are not improved by performing the assay using a greater dilution of patient serum/plasma.