Low Pre-Test Probability Scores Are Not Sufficient to Exclude Heparin-Induced Thrombocytopenia in Critically Ill Patients

Background: Heparin-induced thrombocytopenia (HIT) is a serious complication of heparin administration. Diagnosis of HIT is a clinical challenge. The 4Ts scoring model (0-3 low, 4-5 intermediate, 6-8 high probability of HIT) has been validated in several studies including the recent meta-analysis (Blood 2012;120:4160-7), which demonstrated that a low probability 4Ts score conferred a high negative predictive value (0.998; 95% CI, 0.970-1.000) for excluding HIT. Some experts propose a diagnostic approach for HIT based on the 4Ts scoring model and exclude HIT in patients with a low probability score without serologic testing for HIT. Another recently constructed model 'HIT expert probability' (HEP) score (< 2 unlikely, ≥ 2 likely) demonstrated better diagnostic performance in mainly surgical patients. However, in critically ill patients who receive heparin, other concomitant causes of thrombocytopenia are common and may interfere with clinically diagnosis of HIT. In this study, we aimed to determine the diagnostic accuracy of the 4Ts and the HEP score for excluding HIT in a population of critically ill patients.

Methods: Consecutive patients admitted in critical care units during 2006-2015 were included in this study. Clinical and laboratory data of individuals were retrospectively reviewed from medical records. The 4Ts and the HEP score were blindly computed by two independent reviewers (NU and RV). The rapid particle gel immunoassay (platelet factor 4/heparin-PaGIA) was used for HIT screening. Subjects yielding positive PaGIA were sent for the confirmatory testing using the in-house platelet aggregometry measuring heparin-induced platelet aggregation (HPA). However, during the shortage of PaGIA, HPA was performed in all cases. HPA using platelet-rich plasma from healthy donors with known reactive platelets was performed as previously described with a few modification. Aggregation values of at least 20% above negative controls in the presence of 0.5 or 1.0 U/ml of heparin, which were subsequently inhibited by the addition of 100 U/ml of heparin, were defined as positive results.

Results: There were 92 critically ill patients with suspected for HIT. Among them, 56 (60.9%), 33 (35.9%) and 3 (3.3%) yielded low, intermediate and high probability 4Ts score, respectively, while 33 (35.9%) and 59 (64.1%) yielded unlikely and likely high probability HEP score, respectively. Of 78 with obtainable PaGIA, 25 cases (37.2%; 6/6 HPA+ and 19/72 HPA-) yielded positive results. Eleven patients (12.0%) yielded positive results for HPA were diagnosed as HIT. There were 6 (54.5%) developing thrombosis (4 new proven and 2 progressive). Clinical data of all documented HIT were summarized in the table 1. Documented HIT was diagnosed in 5.4%, 18.2% and 66.7% of low, intermediate and high probability 4Ts score, respectively, whereas HIT was demonstrated in 9.4% and14.3% of unlikely and likely probability HEP score, respectively. The receiver operating characteristic curve analysis demonstrated that the 4Ts score was tended to display better diagnostic performance than the HEP score with the area under curve of 0.740 and 0.587 (P = 0.053), respectively. The HIT cases with low pre-test probability scores were due to concomitant causes of thrombocytopenia resulting in earlier onset, lower nadir of platelet counts and/or more minus scores from alternative etiologies of thrombocytopenia.

Conclusions: The diagnostic performance of the 4Ts and the HEP score is limited in complicated and/or critically ill patients due to multiple etiologies affecting onset and severity of thrombocytopenia. Both low probability 4Ts score and unlikely HEP score are unsafe for excluding HIT in this patient group.