Information on clinical setting was available for 37 patients (ie, all but 9 patients from 1 of the studies³⁰ ): post–cardiac surgery/post–vascular surgery/post–percutaneous coronary intervention (n = 10); treatment of venous thromboembolism (n = 9); post–orthopedic surgery (n = 8); hemodialysis (n = 4); heparin flushes for catheters (n = 3); medical thromboprophylaxis (n = 2); and general surgery thromboprophylaxis (n = 1). One patient¹⁷  had limb amputation (frequency, 1 [2.2%] of 46); this patient reportedly had limb ischemia secondary to inoperable arterial thrombosis before starting rivaroxaban.

Thrombus that occurred in association with HIT, not thrombosis present before HIT.

Bleed that occurred 9 days after stopping rivaroxaban is not included here.

The articles by Ng et al and Ong et al are combined because the 3 patients first reported by Ng et al are also included among the 9 patients reported by Ong et al; the median platelet count data were provided by Ng and Ong (Heng Joo Ng and Shin-Yeu Ong, Singapore General Hospital, Singapore, Singapore, e-mail, 4 January 2017).

Aggregate data only, rather than data for individual patients (per Sharifi et al); the estimated mean platelet count at the start of DOAC was ∼90 × 10⁹/L; in addition, 2 of the 11 patients who received rivaroxaban tested negative for HIT antibodies, and thus are excluded in this table.

Most of the HIT-associated thrombotic events were venous; however, 3 patients had 1 or more arterial thrombi (bilateral lower limb,¹⁷  radial artery,³¹  or carotid artery/saphenous vein graft to coronary artery³² ) of whom 2 patients were shown to have clinical improvement along with partial resolution of arterial thrombi (the third patient did not have repeat imaging performed).