Postdischarge thrombosis and hemorrhage in patients with COVID-19

Alteration in laboratory coagulation parameters and the occurrence of arterial and venous thrombosis are well recognized in patients with coronavirus disease 19 (COVID-19).^1,2  The immunothrombotic phenomenon is believed to be related to endothelial injury and subsequent activation of the coagulation cascade and thrombin generation.³  In the absence of prospective data, professional societies and individual institutions have published interim recommendations to guide anticoagulation management in COVID-19, including consideration of extended thromboprophylaxis following discharge.^4-6 

In acutely ill medical populations, the risk of venous thromboembolism (VTE) extends beyond the duration of the hospitalization, with up to 80% of events occurring in the post–hospital-discharge period (30-45 days) following index hospitalization.⁷  There is equipoise regarding the net clinical benefit conferred by extended thromboprophylaxis following hospitalization, and the American Society of Hematology (ASH) Guideline Panel recommended only inpatient prophylaxis with low-molecular-weight heparin.^8-13 

In the absence of randomized or prospective data for patients with COVID-19, guidelines suggest postdischarge thromboprophylaxis based on individual bleeding and thrombotic risk.^4-6,14,15  Considering the high rates of in-hospital thrombosis in COVID-19, aggressive thromboprophylaxis is recommended during hospitalization.^4,14,16  Whether the prothrombotic state of COVID-19 persists after hospital discharge is not known.

A retrospective analysis of consecutive patients hospitalized with a diagnosis of COVID-19 at the Beth Israel Deaconess Medical Center (BIDMC) from March to May 2020 was performed. Approval was obtained from the local institutional review board, and the study was conducted in accordance with the Declaration of Helsinki. Inclusion criteria included age of at least 18 years and a positive polymerase chain reaction test for severe acute respiratory syndrome coronavirus 2 from a nasopharyngeal swab (real-time severe acute respiratory syndrome coronavirus 2 assay). A total of 406 patients with COVID-19 were initially admitted during the time period. Patients were excluded if they: remained hospitalized at time of analysis (n = 17); died during hospitalization (n = 102); were discharged without any form of postdischarge contact in the BIDMC electronic medical record by 31 May 2020 (n = 93); or were discharged on therapeutic anticoagulation (n = 27; consort diagram in the supplemental Figure, available on the Blood Web site). Patients discharged on prophylactic-dose anticoagulation were also excluded from primary analysis with thrombosis and hemorrhages reported separately (n = 13). For patients with multiple admissions related to COVID-19, the first admission was considered the index admission.

The primary outcomes of interest were the cumulative incidence of thrombosis or hemorrhage at day 30 after discharge among patients who did not receive postdischarge thromboprophylaxis. Bleeding was assessed according to the International Society on Thrombosis and Haemostasis definitions and included major bleeding and clinically relevant nonmajor bleeding (CRNMB).¹⁷  Baseline data included patient demographics, comorbidities, and use of anticoagulants during hospitalization. Thrombotic and hemorrhagic events were evaluated independently by 2 data abstractors (R.P. and T.B.); disagreements were adjudicated by a third investigator (J.I.Z.).

Estimates of the cumulative incidence of venous and thrombotic arterial events as well as hemorrhage were calculated by Kaplan-Meier methodology along with 95% confidence intervals (CIs). Competing risk analyses were not performed due to low 30-day postdischarge mortality.

A total of 163 patients met eligibility criteria and were included in the primary analysis (characteristics of patients included are available in the supplemental Table). One patient, who was readmitted after the index hospitalization, died of progressive respiratory failure secondary to COVID pneumonia.

Four thrombotic events were recorded in the primary study cohort (4 of 163; 2.5%) including 1 segmental pulmonary embolism (PE), intracardiac thrombus, thrombosed arteriovenous fistula, and ischemic stroke (Table 1). Cumulative incidence of overall (venous and arterial) thrombosis was 2.5% (95% CI, 0.8% to 7.6%) at day 30 after discharge (Figure 1). Median duration to thrombotic event postdischarge was 23 days (interquartile range, 12-33). The cumulative incidence of VTE alone at day 30 postdischarge was 0.6% (95% CI, 0.1-4.6).

Six patients in the cohort experienced hemorrhagic events during the follow-up period (6 of 163; 3.7%). The cumulative incidence of hemorrhage at day 30 after discharge was 3.7% (95% CI, 1.4% to 9.8%) (Figure 1). Four of the hemorrhagic events were classified as CRNMB, and 2 patients developed major bleeds (Table 1). Both major bleeds followed mechanical falls: 1 patient developed a limb hematoma that required multiple transfusions and the other a head strike leading to a subarachnoid hemorrhage. There were no fatal bleeds. Median duration to a bleeding event was 27 days (interquartile range, 16-31 days). The cumulative incidence of major bleeds and CRNMB at day 30 postdischarge was 0.7% (95% CI, 0.1-5.1) and 2.9% (95% CI, 1.0-9.1), respectively.

Among the 13 patients discharged on thromboprophylaxis, there were no observed thrombotic or hemorrhagic complications. Ten patients were discharged on low-molecular-weight heparin, 2 on a direct oral anticoagulant, and 1 patient on unfractionated heparin (all at prophylactic dosages). Three patients were discharged to inpatient facilities where thromboprophylaxis was continued. Seven patients were pregnant when admitted and were discharged on anticoagulation that extended for an additional 2 weeks. Two patients underwent orthopedic procedures during hospitalization and continued thromboprophylaxis per standard of care. One patient was continued on rivaroxaban thromboprophylaxis after discharge due to concern regarding COVID-19 coagulopathy.

Epidemiologic data suggest that up to 80% of all hospitalization-associated VTEs occur after discharge.^7,18,19  Given the high reported rates of thrombosis in hospitalized patients with COVID-19, we anticipated that this increased risk would extend into the outpatient setting. The observed rate of symptomatic thrombotic events in our cohort (0.6%), only 1 of which was a deep vein thrombosis or PE, is low and comparable to those observed in control arms of randomized interventional studies (0.3% to 2.5%) at 28 to 35 days.^9,10,20,21  We believe that the observed rates of thrombosis are reflective of the true symptomatic incidence of thrombosis. For the primary analyses, we excluded 13 patients discharged on thromboprophylaxis. All but 1 of those patients had secondary indications for continued thromboprophylaxis beyond COVID-19 coagulopathy (ie, pregnancy, orthopedic surgery, or continued inpatient rehabilitation). The rate of thrombosis within 30 days postdischarge should be generalizable to the larger population of hospitalized patients with COVID-19 infection in the absence of secondary indications for continued thromboprophylaxis.

There is an emerging signal of increased bleeding during hospitalization for COVID-19 that needs to be balanced against the risk of thrombosis.²²  In randomized studies, the cumulative incidence rates of major and CRNMB at 4 to 5 weeks postdischarge in the placebo arms range between 0.7% and 2%.^9,10,12  In our cohort, the cumulative incidence of major hemorrhage was 0.7% and CRNMB was 2.9%. Given the retrospective study design, true bleeding rates may be lower as patients without follow-up were excluded; these patients may have been less acutely ill and less likely to have experienced bleeding. We also note that 4 bleeding events (including 2 major bleeds) occurred following mechanical falls in debilitated patients who were discharged after lengthy hospitalizations (Table 1), which is of particular relevance when considering the relative risks and benefits of extended prophylaxis following COVID-19 hospitalization.

To the best of our knowledge, this is the first study to systematically evaluate bleeding and thrombosis rates in patients discharged with COVID-19. Thrombosis and bleeding risk is likely modulated by factors including age, coagulation profile, comorbidities, severity of index illness, and degree of immobility. Our study design and event rate did not allow meaningful multivariable analysis of variables predictive of thrombosis or hemorrhage. Inclusion into this analysis required the patients to have at least 1 postdischarge medical evaluation, as we did not want to assume that patients without contact did not experience a thrombotic event. We suspect that the low rate of VTE observed is actually an overestimation of the true event rate as healthier patients with fewer comorbidities are less likely to seek medical care postdischarge (and suffer a thrombotic or bleeding event). We also acknowledge that the relatively limited sample size warrants caution in interpretation of our results; prospective studies are required to definitively establish rates of hemorrhagic and thrombotic events.

Extended-duration pharmacological thromboprophylaxis is recommended for high-risk surgical patients,²³  but the net clinical benefit in acutely ill medical patients is less clear and is thus not universally recommended.²⁴  A meta-analysis of randomized controlled trials assessing in-hospital prophylaxis to “extended-duration” prophylaxis demonstrated that thromboprophylaxis can reduce the risk of VTE, but with an increased risk of hemorrhage.²⁵  Very few patients in our cohort received extended thromboprophylaxis; thus we cannot comment on the relative safety or efficacy of anticoagulation following discharge in COVID-19. However, based on an observed rate of hemorrhage in the nonanticoagulated cohort, we would be cautious in recommending extended-duration thromboprophylaxis without a clear indication (eg, major surgery) outside of a clinical trial.

Contribution: R.P., W.C.A., K.A.B., and J.I.Z. designed the study; R.P., T.B., M.M., A.K., and P.B. collected and analyzed data; J.I.Z. and R.P. wrote the manuscript; and all authors reviewed the final manuscript and contributed to its preparation.

Conflict-of-interest disclosure: J.I.Z. reports research funding from Incyte and Quercegen; consultancy services to Sanofi, CSL, and Parexel; and honoraria from/advisory board participation with Pfizer/Bristol Myers Squibb (BMS), Portola, and Daiichi. K.A.B. reports consultancy with BMS and Takeda. W.C.A. is employed by DynaMed. The remaining authors declare no competing financial interests.

Correspondence: Jeffrey I. Zwicker, Department of Hematology, Beth Israel Deaconess Medical Center, 330 Brookline Ave, Boston, MA 02215; e-mail: jzwicker@bidmc.harvard.edu.