Comparison of risk-scoring systems for heparin-induced thrombocytopenia in cardiac surgery patients

Objectives Several risk-scoring tools have been developed to exclude heparin-induced thrombocytopenia (HIT) in patients with thrombocytopenia, but these scores have not been reproduced or compared in the cardiac surgery population. The objective of this study was to validate and compare the modified 4T's (m4T) and Lillo-Le Louet (LLL) scores for HIT screening in the cardiac surgery population. Methods In this nested case-control study, we retrospectively calculated the m4T and the cardiac surgery-specific score by LLL for 18 cases (HIT-positive) and 54 matched controls (HIT-negative) using characteristics known at the time the HIT assay was ordered post-cardiac surgery and compared their performances by their c-statistic (area under the receiver operating characteristic curve), sensitivity and specificity. Results The median time from surgery to HIT assay order was 9.5 days (IQR 3.75-11.0) in the HIT-positive group and 2 days (IQR 2.0-3.0) in the HIT-negative group (p < 0.0001). The c-statistics for the m4T and the LLL scores were 0.76 (95% CI 0.64-0.85) and 0.63 (95% CI 0.51-0.74), respectively (p = 0.051). Sensitivity and specificity were 61% and 91% for the m4T, and 94% and 32% for the LLL score. Conclusion Performance of the m4T and LLL scores in discriminating HIT-positive from HIT-negative patients was modest among patients post-cardiac surgery. However, differences between the sensitivities of these scores suggest that the LLL score may be a safer tool for ruling out HIT in this population.

Automated summary

The study aimed to validate and compare the performance of the modified 4T (m4T) and Lillo-Le Louet (LLL) scores for HIT screening in cardiac surgery patients. Results showed that these scores had modest performance in discriminating HIT-positive from HIT-negative patients post-cardiac surgery, with differences in sensitivities suggesting that the LLL score may be a safer tool for ruling out HIT in this population.