key: cord-0864588-zltr39a5
authors: Dalager-Pedersen, Michael; Lund, Lars Christian; Mariager, Theis; Winther, Rannva; Hellfritzsch, Maja; Larsen, Torben Bjerregaard; Thomsen, Reimar Wernich; Johansen, Nanna Borup; Søgaard, Ole Schmeltz; Nielsen, Stig Lønberg; Omland, Lars; Lundbo, Lene Fogt; Israelsen, Simone Bastrup; Harboe, Zitta Barrella; Pottegård, Anton; Nielsen, Henrik; Bodilsen, Jacob
title: Venous thromboembolism and major bleeding in patients with COVID-19: A nationwide population-based cohort study
date: 2021-01-05
journal: Clin Infect Dis
DOI: 10.1093/cid/ciab003
sha: 1ef52bc59b56d6676dde8ac46d48a8613b95c2f9
doc_id: 864588
cord_uid: zltr39a5

BACKGROUND: Venous thromboembolism (VTE) is a potentially fatal complication of SARS-CoV-2 infection and thromboprophylaxis should be balanced against risk of bleeding. This study aimed to examine risks of VTE and major bleeding in hospitalized and community-managed SARS-CoV-2 patients compared with control populations. METHODS: Using nationwide population-based registries, 30-day risks of VTE and major bleeding in SARS-CoV-2 positive patients were compared with those of SARS-CoV-2 test-negative patients and with an external cohort of influenza patients. Medical records of all COVID-19 patients at six departments of infectious diseases in Denmark were reviewed in detail. RESULTS: The overall 30-day risk of VTE was 0.4% (40/9,460) among SARS-CoV-2 patients (16% hospitalized), 0.3% (649/226,510) among SARS-CoV-2 negative subjects (12% hospitalized), and 1.0% (158/16,281) among influenza patients (59% hospitalized). VTE risks were higher and comparable in hospitalized SARS-CoV-2 positive (1.5%), SARS-CoV-2 negative (1.8%), and influenza patients (1.5%). Diagnosis of major bleeding was registered in 0.5% (47/9,460) of all SARS-CoV-2 positive individuals and in 2.3% of those hospitalized. Medical record review of 582 hospitalized SARS-CoV-2 patients observed VTE in 4% (19/450) and major bleeding in 0.4% (2/450) of ward patients, of whom 31% received thromboprophylaxis. Among intensive care patients (100% received thromboprophylaxis), risks were 7% (9/132) for VTE and 11% (15/132) for major bleeding. CONCLUSIONS: Among people with SARS-CoV-2 infection in a population-based setting, VTE risks were low to moderate and were not substantially increased compared with SARS-CoV-2 test-negative and influenza patients. Risk of severe bleeding was low for ward patients, but mirrored VTE risk in the intensive care setting.

Virus Disease 2019 (COVID- 19) , has put healthcare systems worldwide under tremendous pressure.

Venous thromboembolism (VTE) is a potentially fatal complication of hospitalization for surgery or medical conditions including infections. COVID-19 may confer an increased risk of VTE by endothelial dysfunction and inflammatory mediated activation of the coagulation cascade, but also indirectly by immobilization, need for central venous catheter, and mechanical ventilation during admission at an intensive care unit (ICU) [1, 2] . Yet, the magnitude and duration of this risk in COVID-19 is unclear and recent observational studies have suggested that VTE occurred during hospitalization in 1-7% of non-ICU patients and 7-35% of ICU patients [3] [4] [5] [6] [7] [8] [9] [10] . However, these studies consist of small sample sizes of selected patient populations from single or a few hospitals with short and incomplete follow-up and often lack control groups for comparison [3] [4] [5] [6] [7] [8] [9] [10] . In addition, although 3/4 of VTE events are thought to occur outside the hospital setting [11] , data on risks among community-managed patients with COVID-19 remain scarce.

This study combined nationwide registries with manual chart review to examine absolute risks of VTE and major bleeding in non-hospitalized and hospitalized patients with a positive test for SARS-CoV-2. Moreover, risks of VTE and bleeding were compared with non-hospitalized and hospitalized patients with a negative test for SARS-CoV-2 and with influenza patients.

The study was conducted in Denmark where all ~5.800.000 inhabitants are provided with universal, tax-supported health-care, free of charge at the point of delivery. All residents are assigned a unique 10-digit civil registration system (CRS) number, which is used for all health-care contacts, including hospitalizations and prescription medicine, and facilitates individual-level linkage between A c c e p t e d M a n u s c r i p t nationwide Danish registries [12] . The study period was January 27 to June 1, 2020. As of April 21, 2020 , the SARS-CoV-2 test strategy in Denmark changed from examining only symptomatic persons to testing all patients admitted to hospital >24 hours and asymptomatic individuals potentially exposed for SARS-CoV-2 (Supplementary material) [13] . All clinical tests for SARS-CoV-2 during the study period were analyzed at departments of clinical microbiology using reverse-transcriptase polymerase-chain-reaction (PCR). On April 17, 2020 

Data sources for this study included nationwide, population-based administrative health registries combined with electronic medical record (EMR) review in a subgroup of patients ( Figure 1 ).

Established in 1967, the CRS database was used for information on date of birth, sex, and migration and vital status for all study subjects with <0.3% lost to follow-up [12] . The Microbiological Database (MiBa) was used to identify all patients with SARS-CoV-2 confirmed by PCR assays performed on upper or lower respiratory tract specimens [14] . It has received real-time reports from all departments of clinical microbiology in Denmark since 2010. MiBa was also accessed to identify patients with a negative test for SARS-CoV-2. The National Patient Registry was used for information on all hospital admissions, comorbidity, VTE, and bleeding events (Supplementary Table 1 ) [15] . It keeps record of complete WHO International Classification of Diseases (ICD) diagnosis codes on all inpatient hospitalizations since 1977 and outpatient hospital contacts in Denmark since 1995. For each hospitalization, a treating physician assigns one primary discharge diagnosis code for the condition that prompt hospitalization, and mainly affects treatment course, and up to 20 secondary A c c e p t e d M a n u s c r i p t codes. The National Prescription Registry holds data on reimbursed prescriptions at community pharmacies by Anatomical Therapeutic Chemical (ATC) codes since 1995 [16] . It was accessed to obtain data on medication use before hospital admission and after discharge (Supplementary Table   1 ).

In addition, EMRs of all COVID-19 patients treated at departments of infectious disease at Aarhus, Odense, Hvidovre, Rigshospitalet (Copenhagen), Hillerød, and Aalborg university hospitals were reviewed. Clinical data was extracted including data on comorbidities, medication use before and during admission, tobacco and alcohol habits, signs and symptoms at admission, laboratory tests, radiological examinations, and occurrence of VTE or major bleeding events. These data were managed using Research Electronic Data Capture browser-based software (REDCap, Vanderbilt, TN, USA).

First, using health-care registries, a cohort comprising all cases of SARS-CoV-2 infection Denmark, diagnosed until May 1 st , 2020 was assembled. Second, because VTE risk may be related to acute illness in general rather than pathophysiological effects of SARS-CoV-2 per se, a comparison cohort comprising all non-hospitalized and hospitalized patients who tested negative for SARS-CoV-2 (and remained test-negative) until May 1 st , 2020 was identified. Third, as an additional point of comparison, an external cohort of adults born before 1978 with laboratory-confirmed influenza from 2010 through 2018 was used to examine if SARS-CoV-2 confers a greater VTE risk than another serious viral respiratory infection [17] . through May 4 th , 2020, were reviewed.

In the registry-based cohort analyses, VTE and major bleeding events were defined as any discharge diagnosis code made either during index-hospitalization (for patients already hospitalized on date of microbiological testing) or during new hospitalization within 30 days after microbiological testing. In an additional analysis including prescription data, VTE was defined by either a diagnosis code for VTE or a post-discharge reimbursed prescription for new anticoagulant therapy likely due to VTE, as indicated by an algorithm based on relevant ATC codes (Supplementary Table 2 ).

In the EMR cohort analysis, VTE was defined as a deep vein thrombosis (DVT) diagnosed by compression ultrasound or pulmonary embolism (PE) diagnosed by CT pulmonary angiography or lung scintigraphy according to the radiologist's description. Date of VTE was categorized as the day of diagnostic imaging. Major bleeding events were defined as 1) any diagnosis of central nervous system, retroperitoneal or intraocular bleeding, 2) clinical bleeding requiring transfusion of >1 unit of blood, or significant medical or surgical intervention, 3) bleeding significantly contributing to death as judged by the treating physician.

A c c e p t e d M a n u s c r i p t

Categorical variables were presented as numbers and percentages with 95% confidence intervals (CI) and continuous variables as medians with interquartile ranges (IQR). Covariate balances were examined using standardized mean differences. No patients had missing data on exposure or primary outcome and the primary analyses included all study participants.

The date of microbiological testing was defined as cohort entry (index) date. Individuals were excluded if they had either VTE during the year prior to microbiological testing, less than one year of enrollment in the database prior to test date or emigrated out of the country less than 30 days after sample date. All study subjects in registry-based cohorts were followed from the index date until completion of 30 days of follow-up, death or June 1 st , 2020, whichever came first.

For each study cohort, 30-day absolute risks of VTE, major bleeding events, and mortality were computed. Subgroup analyses were performed by age group (0-64, 65+ years of age), sex, presence of risk-factors for VTE (yes/no), and Charlson Comorbidity Index (CCI of 0 and CCI ≥1) [18] . A limited number of primary outcomes among SARS-CoV-2 positive patients precluded planned adjusted analyses of relative risk.

Subjects in the EMR cohort were followed from date of SARS-CoV-2 test until date of VTE, major bleeding, death, emigration out of Denmark, or date of last medical record review (May 4 th ), whichever came first. Next, 30-day risks of VTE, bleeding, and mortality were computed and stratified by ICU admission (yes/no). In non-ICU patients, risk estimates were further examined by 2020-045). Thus, patient consent or approval from an ethical committee was not required for this study in Denmark. 

During the study period, 582 patients were treated for SARS-CoV-2 at six departments of infectious disease in Denmark ( Table 4 ). The median age was 69 years (54-78) and 58% were male (335/582).

At least one pre-existing risk factor for VTE was observed in 22% of patients (130/582) with body mass index >35 kg/m 2 as the most frequent (8%; 49/582). Median duration of COVID-19 symptoms before admission was 7 days (IQR 4-10) and common symptoms included history of fever in 82% (Table 5) .

Overall, VTE occurred in 5% (28/582) of hospitalized patients after exclusion of five patients with a presumptive diagnosis of PE but without radiological confirmation. Major bleeding was observed in 3% (17/582) and 20% (124/582) had a fatal outcome. Risk of VTE was 4% (19/450) among ward patients and 7% (9/132) among patients admitted at the ICU.

For ward patients receiving thromboprophylactic therapy, VTE occurred in 3% (4/140) and major bleeding was observed in 1% (2/140) patients. For ward patients not treated with thromboprophylaxis, VTE was found in 5% (15/310) and major bleeding in 0% (0/310) patients. All patients admitted at the ICU received anticoagulant therapy and major bleeding was observed in 11% (15/132).

In this Danish nationwide, population-based cohort study, 30-day risk of VTE was 0.2% among nonhospitalized and 1.5% in hospitalized SARS-CoV-2 patients (2.3% when adding prescription data). In comparison, major bleeding events occurred in 0.1% of non-hospitalized and 2.3% of hospitalized SARS-CoV-2 positive patients. Risks of VTE and major bleeding were slightly higher by medical record review and were comparable with those of SARS-CoV-2 negative individuals and influenza patients.

A c c e p t e d M a n u s c r i p t Studies addressing VTE risk in SARS-CoV-2 are mostly limited by small or moderate sample sizes from single-or a few centers with selected patient populations and incomplete follow-up [3] [4] [5] [6] [7] [8] [9] [10] . In addition, some studies used VTE screening and included asymptomatic VTEs and subsegmental PE, both of which are of uncertain clinical relevance [19] . Consistent with the current study, several other studies have found VTE risks of 3-7% among ward patients and 7-8% in ICU patients [10, [20] [21] [22] [23] .In contrast, VTE was observed in 3% of non-ICU patients and in 23-35% of ICU patients in two Dutch studies where VTE screening and thromboprophylaxis was standard [4, 8] . Another two ICU studies observed risks of VTE of 17-21% in SARS-CoV-2 patients compared with 8% in influenza patients and 1% (odds ratio 15.2) in a heterogenous group of ARDS patients [6, 9] . In comparison, relatively few cases of VTE were diagnosed in the current study despite more restricted use of thromboprophylaxis. Reasons for the large discrepancy in VTE risk between studies are unclear.

Among hospitalized patients in the current study, VTEs may have gone undetected either through lack of clinical suspicion or due to barriers in performing diagnostic exams such as chest CT's in mechanically ventilated ICU patients. Still, mortality rates among hospitalized COVID-19 patients in our study were comparable with or lower than other studies suggesting that missed fatal VTEs were limited [10, [21] [22] [23] . Moreover, difficulty with diagnostic imaging is unlikely to explain the observed low risk estimates in non-hospitalized patients and non-ICU patients, and compression ultrasound is relatively easy to perform and DVT was rarely diagnosed.

Since most previous studies of VTE in SARS-CoV-2 infection lack comparison cohorts, it remains unclear if risk exceeds that of other infections or medical conditions [24] [25] [26] SARS-CoV-2 may induce thrombosis by immobilization and hospitalization [11] . More directly, the virus may also cause endothelial dysfunction by cell invasion and has been found to invoke hyperinflammation, anti-phospholipid antibody production, and coagulopathy [2] . Some authors suggest that formation of local pulmonary microangiopathic immune-thrombosis may account for a substantial proportion of observed VTEs in SARS-CoV-2 patients rather than embolization from the lower extremities [30, 31] . Although many of these attributes of SARS-CoV-2 have received considerable attention, the very same characteristics have also previously been recognized among a number of other pathogens including viruses [24] [25] [26] .

Thus far, autopsy studies of deceased SARS-CoV-2 patients have yielded varying results ranging from no VTEs [30,32] to findings of DVT in 58% (7/12) and PE in 33% (4/12) of patients [33] . This is consistent with older autopsy studies on more than 5,000 patients where PE were frequent and found in up to 70% of patients who died from infection, primarily pneumonia and sepsis [34, 35] . A c c e p t e d M a n u s c r i p t

In this Danish population-based cohort, SARS-CoV-2 was associated with low risk for VTE among non-hospitalized patients and a moderate risk among hospitalized patients. Importantly, VTE risk was not substantially increased compared with hospitalized SARS-CoV-2 test-negative and influenza patients. Diagnosis of major bleeding in non-hospitalized and hospitalized SARS-CoV-2 patients mirrored that of VTE.

A c c e p t e d M a n u s c r i p t NOTES Acknowledgements:

We thank the Danish Medicines Agency for facilitating the conduct of this study. According to Danish law, data cannot be shared directly by the authors. Data is accessible to authorized researchers after application to the Danish Health Data Authority.

MDP and JB conceived study. MDP, JB, RWT, AP, NBJ, MH, LCL, and HN designed the study. LCL did the registry-based analyses. Medical record review was conducted by JB, RW, TM, OS, SLN, LFL, SBI, LHO and was analyzed by JB. MDP wrote the first draft, which has been critically revised and approved by all authors.

TBL reports personal fees/consulting fees from Boehringer Ingelheim, Bayer, MSD and BMS/Pfizer, during the conduct of the study. AP reports grants from Alcon, grants from Almirall, grants from Astellas, grants from Astra-Zeneca, grants from Boehringer-Ingelheim, grants from Novo Nordisk, grants from Servier, and grants from LEO Pharma, outside the submitted work. RT reports that The A c c e p t e d M a n u s c r i p t M a n u s c r i p t 

Endothelial cell infection and endotheliitis in COVID-19

Coagulopathy and Antiphospholipid Antibodies in Patients with Covid-19

Clinical Characteristics of Covid-19 in New York City

Incidence of venous thromboembolism in hospitalized patients with COVID-19

High incidence of venous thromboembolic events in anticoagulated severe COVID-19 patients

Pulmonary Embolism in Patients With COVID-19

Prevalence of venous thromboembolism in patients with severe novel coronavirus pneumonia

Confirmation of the high cumulative incidence complications in critically ill ICU patients with COVID-19: An updated analysis

High risk of thrombosis in patients with severe SARS-CoV-2 infection: a multicenter prospective cohort study

Venous and arterial thromboembolic complications in COVID-19 patients admitted to an academic hospital in

Risk Factors for Venous Thromboembolism

The Danish Civil Registration System as a tool in epidemiology

Retningslinjer for håndtering af COVID-19 i sundhedsvaesenet

Board of Representatives C. The Danish Microbiology Database (MiBa) 2010 to 2013

The Danish National patient registry: A review of content, data quality, and research potential

Resource Profile: The Danish National Prescription Registry

Association of Nonsteroidal Anti-inflammatory Drug Use and Adverse Outcomes Among Patients Hospitalized With Influenza

A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation

Overdiagnosis of pulmonary embolism: Definition, causes and implications

COVID-19 and coagulation: Bleeding and thrombotic manifestations of SARS-CoV-2 infection

Frequency of venous thromboembolism in 6513 patients with COVID-19: A retrospective study

Thrombosis in Hospitalized Patients with COVID-19 in a New York City Health System

VTE in ICU Patients With COVID-19

Virus Infection of Endothelial Cells

Coagulation in sepsis: All bugs bite equally

Venous Thrombosis and Pulmonary Embolism. A Study of 5039 Autopsies

Achados clínicopatológicos na tromboembolia pulmonar: estudo de 24 anos de autópsias

COVID-19 and Thrombotic or Thromboembolic Disease: Implications for Prevention, Antithrombotic Therapy, and Follow-Up: JACC State-ofthe-Art Review

Vademecum per la cura delle persone con malattia da COVI-19

BTS Guidance on Venous Thromboembolic Disease in patients with COVID-19

Association of Treatment Dose Anticoagulation With In-Hospital Survival Among Hospitalized Patients With COVID-19

Anticoagulant treatment is associated with decreased mortality in severe coronavirus disease 2019 patients with coagulopathy

A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t A c c e p t e d M a n u s c r i p t Figure 1