Heparin-induced thrombocytopenia (HIT) is a thrombotic disorder, occurring in 3 percent or more of individuals exposed to heparin, in which antibodies to platelet factor 4-glycosaminoglycan complexes form on platelet surfaces, leading to thrombi, which can be life-threatening. Diagnosis of this disorder can be difficult because the clinical picture may be confusing and assays are limited by poor specificity and lack of availability in real time.
To tackle this problem, a multi-institution team of investigators constructed a HIT-test clinical scoring model based on the expert opinions of 26 physicians who regularly diagnose and manage HIT. The experts, who are participants in the NHLBI-funded Transfusion Medicine Hemostasis (TMH) Clinical Trials Network, were asked to rate the diagnostic importance of each of eight HIT clinical findings selected from the literature, ranging from -3 (strongly against a diagnosis of HIT) to +3 (strongly for a diagnosis of HIT). The median ratings for these variables were used to derive the pre-test probability model, the HIT Expert Probability (HEP) score. The model was then validated in 50 consecutive patients referred for HIT testing at a single reference laboratory at the University of Pennsylvania, with tests including a polyspecific HIT antibody ELISA and an in-house serotonin release assay. Clinical information on the 50 patients was obtained at baseline testing and then by phone interview at 30 days post-testing. A panel of three experts determined whether HIT was likely or unlikely in the patients, and two hematology fellows rated each patient independently using the HEP score, the 4 Ts,1 (a diagnostic method using four criteria to determine a HIT probability score), and a third system, the Lillo-Le Louët model,2 (a diagnostic score using three criteria to determine a HIT probability score following cardiopulmonary bypass) to determine inter-observer agreement.
| . | HEP Score . | 4 Ts . | p-value . |
|---|---|---|---|
| Clinical Features |
|
| |
| Inter-Observer Agreement (Correlation Coefficient) | 0.88 | 0.71 | |
| Concordance with Lab, Panel (AUROC) | 0.91 | 0.74 | p=0.017 |
| Mean Score in ITP | 1.98 | 4.09 | |
| Sensitivity | 1.00 | 1.00 | |
| Specificity | 0.60 | 0.44 | |
| PPV | 0.29 | 0.23 | |
| NPV | 1.00 | 1.00 | |
| Score in SRA(+) | 5.4 | 1.3 | p=0.0003 |
| Score in SRA(-) | 4.5 | 4.0 | p=0.33 |
| . | HEP Score . | 4 Ts . | p-value . |
|---|---|---|---|
| Clinical Features |
|
| |
| Inter-Observer Agreement (Correlation Coefficient) | 0.88 | 0.71 | |
| Concordance with Lab, Panel (AUROC) | 0.91 | 0.74 | p=0.017 |
| Mean Score in ITP | 1.98 | 4.09 | |
| Sensitivity | 1.00 | 1.00 | |
| Specificity | 0.60 | 0.44 | |
| PPV | 0.29 | 0.23 | |
| NPV | 1.00 | 1.00 | |
| Score in SRA(+) | 5.4 | 1.3 | p=0.0003 |
| Score in SRA(-) | 4.5 | 4.0 | p=0.33 |
AUROC is area under the receiver operating curve; SRA is serotonin release assay
As compared with the standard 4 Ts pre-test system, the eight-criteria HEP score showed better inter-observer agreement, better correlation with HIT laboratory tests, and better agreement with a HIT diagnosis by the expert panel. Further, the HEP score showed better specificity and was better able to distinguish patients with a positive serotonin release assay than the 4 Ts system. The authors conclude, based on these findings, that the HEP score reduced the number of false positive diagnoses by 40 percent, thereby reducing the unnecessary use of direct thrombin inhibitors in parallel.
In Brief
The new pre-test probability score for HIT appears to be innovative and impressive, considering the fact that it is based on the expert opinion of 26 busy hematologists who agreed to complete the questionnaire. The score is straightforward, as it is based on standard clinical and laboratory features of HIT, and will likely be test-driven by hematologists in the coming year. Before the scoring system can be implemented, however, a few critical issues need to be considered. For example, it will be important to determine how the HEP score model works when not all data are available, which was the case in 10 percent of the validation group. It will also be critical to validate the scoring system in surgical, medical, and critical-care HIT patients, who constituted fewer than half of the cases in the validation group. Finally, as the score was tested with patients from a single institution, it will be important to assess the HEP score at other institutions and with different serotonin release assays and cutoffs. Notwithstanding these caveats, as the HEP score is tested in the field, it should be noted that hematology fellows with no prior experience were able to apply the score and closely reproduce the findings of the expert panel.
References
Competing Interests
Dr. Ragni indicated no relevant conflicts of interest.
