SESSION TYPE: Hematologic Problems in the ICU
PRESENTED ON: Monday, October 22, 2012 at 11:15 AM - 12:30 PM
PURPOSE: Both over and under-diagnosis of Heparin Induced Thrombocytopenia (HIT) can lead to significant clinical sequelae. The recommended diagnostic approach suggests initial risk stratification using the 4Ts score followed by a screening immunoassay (EIA). A confirmatory serotonin release assay (SRA) is reserved for positive EIAs. We hypothesized that this strategy could lead to a high rate of misclassification in high prevalence populations.
METHODS: We created a decision model reflecting alternative approaches for the diagnosis of HIT. We relied on published estimates for the sensitivity and specificity for tests for HIT. The sensitivity and specificity used for the 4Ts score, EIA, and SRA were 80% and 69%, 95% and 50%, and 98% and 97%, respectively. We varied the prevalence of HIT in our theoretical population of 10000 subjects from 1%-5% to reflect the reported rates of HIT in the medical and cardiothoracic surgical populations, respectively. Our base case model consisted of sequential testing with the 4Ts, EIA, and confirmatory SRA. We compared this to simply utilizing SRA alone in all subjects. The false negatives (FN) resulting from each approach served as the primary outcome. Positive and negative predictive values (PPVs and NPVs) served as secondary endpoints.
RESULTS: In the base case where the prevalence of HIT is 1%, 23 FNs result and the PPV and NPV are 20.0% and 99.9%, respectively. In contrast, if all subjects receive SRA testing alone, only 2 FNs arise and the PPV rises to 24.8% with NPV unchanged. Increasing the prevalence to 5% causes the number of FNs to climb (116 FN) for sequential testing and does not alter the NPV while doubling PPV. If only SRA testing is used, few FNs result (n=10) and the PPV is similar to sequential testing.
CONCLUSIONS: Estimated prevalence of disease should be considered to optimize the diagnostic approach to HIT. As prevalence increases, direct testing with an SRA may be optimal. In lower prevalence situations, sequential testing is likely preferred.
CLINICAL IMPLICATIONS: Sequential testing should be applied in populations where the prevalence for HIT is low. When the prevalence is higher (eg, cardiothoracic surgery), the evaluation for HIT should automatically include SRA.
DISCLOSURE: The following authors have nothing to disclose: Chee Chan, Christian Woods, A. Shorr
No Product/Research Disclosure InformationWashington Hospital Center, Washington, DC