The authors reply.

We thank Imamura (1) for their interest in our recently published study (2) in Critical Care Medicine.

Patient-related outcome measures are crucial in clinical studies. We looked at all thrombotic events that would require therapeutic anticoagulation as a standard outcome metric. As for severe or fatal thrombosis as the primary endpoint and causation, this was a retrospective study with its recognized limitations. Those with thrombosis did not have a causal relationship with admission d-dimer, WBC, lactate dehydrogenase, or ferritin. We did not look at severe thrombosis, a clinico-physiologic determination subject to definitions. We did not look at physiologic parameters, but echocardiographic evidence of right ventricular dysfunction (RVD) in our coronavirus disease 2019 (COVID-19) acute respiratory distress syndrome (ARDS) cohort has been reported separately (3). Of our cohort, 31 patients had moderate or severe RVD at admission by echocardiography, of whom 24/31 had macrothrombosis on CT angiography. Of the 24 with thrombosis and RVD, 13 were on vasoactive agents for hemodynamic support; one patient died. Vasopressor use was not necessarily for RVD. One patient with RVD without thrombosis on noradrenaline died. Thus, in the study by Mirsadraee et al (2), analysis of outcomes by degree of severity did not differentiate survivors from nonsurvivors. Inhaled nitric oxide was used in a few patients, for rescue oxygenation, rather than RVD (4). Moderate or severe RVD was too infrequent for valid subgroup analysis.

Cause of death is an important determinant. Of 72 patients, 14 with and three without thrombosis died. Indeed, we did report that the presence of thrombosis portended a higher chance of death than absence. However, it is difficult to prove direct causality. A meta-analysis of studies demonstrated increased thrombotic events in those on ICU with high admission d-dimer levels. However, a cutoff has not been suggested due to uncertainty of its validity (5).

In regard to enhanced treatment interventions, a prospective propensity matched cohort study found a priori therapeutic heparin versus thromboprophylaxis to reduce the prevalence of thromboses without an increase in bleeding complications. However, this is not reproduced or recommended by international guidelines. Thus, data from several clinical trials show benefit of thromboprophylaxis in moderate or severe COVID-19 disease, but uncertainty about treatment versus enhanced or extended thromboprophylaxis without confirmed thrombosis (6). In our cohort, only a few patients had a repeat scan within a few days, indicated by a clinical deterioration. Therapeutic anticoagulation was maintained on confirmation. Systemic thrombolysis was used occasionally in selected cases of arteriovenous pulmonary thromboses, and if thrombotic burden, or its impact were life threatening.

The burden of thromboses in patients with COVID-19 ARDS requiring ICU was higher than pre-COVID-19 ARDS (2). Furthermore, the prevalence of thrombosis in our cohort was higher than other published reports. Either due to the severity of disease of these patients, many of whom were on extracorporeal membrane oxygenation, or the fact that all patients systematically underwent contrast-enhanced CT at admission to ICU. The ideal investigational algorithm for screening patients admitted to ICU with COVID-19 ARDS remains elusive, as do unified anticoagulation treatment recommendations in the absence of confirmed thrombosis (6).