Chronic Anti-Coagulation Therapy Reduced Mortality In Patients With High Cardiovascular Risk Early In COVID-19 Pandemic

Background: Coronavirus disease 2019 (COVID-19) is associated with provoked thrombo-inflammatory responses. Early in the COVID-19 pandemic this was thought to contribute to hypercoagulability and multi-organ system complications in infected patients. Limited studies have evaluated the impact of therapeutic anti-coagulation therapy (AC) in alleviate these risks in COVID-19 positive patients. Our study aimed to investigate whether long-term therapeutic AC can decrease the risk of multi-organ system complications (MOSC) including stroke, limb ischemia, gastrointestinal (GI) bleeding, in-hospital and intensive care unit death in COVID-19 positive patients during the early phase of the pandemic in the United States. Methods: A retrospective analysis was conducted of all COVID-19 positive United States Veterans between March 2020 and October 2020. Patients receiving continuous therapeutic AC for a least 30 days prior to or after their initial COVID-19 positive test were assigned to the AC group. Patients who did not receive AC were included in a control group. We analyzed the primary study outcome of MOSC between the AC and control groups using binary logistic regression analysis (Odd-Ratio; OR). Results: We identified 48,066 COVID-19 patients, of them 879 (1.8%) were receiving continuous therapeutic AC. The AC cohort had significantly worse comorbidities than the control group. On the adjusted binary logistic regression model, therapeutic AC significantly decreased in-hospital mortality rate (OR; 0.67, p = 0.04), despite a higher incidence of GI bleeding (OR; 4.00, p = 0.02). However, therapeutic AC did not significantly reduce other adverse events. Conclusion: AC therapy reduced in-hospital death early in the COVID-19 pandemic among patients who were hospitalized with the infection. However, it did not decrease the risk of MOSC. Additional trials are needed to determine the effectiveness of AC in preventing complications associated with ongoing emerging strains of the COVID-19 virus.

arrhythmias, myocarditis and acute heart failure (6). Early in the pandemic, cardiac injury was a common occurrence and conferred a higher risk of in-hospital mortality (7). In addition to cardiac ischemia noted in autopsy studies, pathological micro-thrombosis was a common observation among patients with severe COVID-19 illness, suggesting a pro-thrombotic state may be contributing to these cardiovascular complications and overall poor recovery outcomes (8,9).
Since the early phase of the pandemic there has been an abundance of evidence to suggest that thrombosis alongside in ammation contributed to the poor outcomes observed with COVID-19 illness.
The impact of anticoagulation in improving outcomes in already infected COVID-19 patients was therefore previously explored. In one multiplatform randomized controlled clinical trial, administration of therapeutic anticoagulation appeared to reduce thrombotic complications but not improve cardiopulmonary outcomes in already hospitalized critically-ill patients with COVID-19 (10)(11)(12). It is argued that hypercoagulable states associated with COVID-19 infection are often present during the early phase of the disease process, and this could in part account relatively modest bene ts observed with therapeutic anticoagulation.
However, the data regarding the role of chronic anticoagulation in outcomes of COVID-19 has since remained con icting. Parker et al. reported reduced rates of mechanical ventilation (8.5% vs 17.4%) in chronically anticoagulated patients with COVID-19 but a higher (40.2% vs 30%) mortality rate (13). On the other hand, Tremblay et al. demonstrated no mortality bene t in a propensity matched cohort of 241 patients receiving anticoagulation prior to COVID-19 infection (14). As such, we sought to evaluate the incidence of multisystem organ failure in individuals with high cardiovascular risks during the early phase of the COVID-19 pandemic. We hypothesize that patient who contracted COVID-19 during this time period, while already receiving therapeutic anticoagulation would have potentially bene ted from reduced COVID-19-related complications.

Patient Population
We evaluated 48,066 COVID-19 positive U.S. Veterans who received care at the Department of Veteran Affairs (VA) during the early phase of the COVID 19 pandemic in the U.S. between March 1, 2020 and October 29, 2020. Patients were assigned to the AC group if they received a continuous prescription of therapeutically dosed anticoagulants at least 30 days prior to and 30 days after their initial COVID-19 positive test. The VA collects data from routine medical visits and compiles and stores them in a secure Corporate Data Warehouse (CDW). Patient information was de-identi ed before use in this study. This study was conducted retrospectively, and was approved with a waiver of informed consent by the institutional review board of the Department of Veterans Affairs St. Louis Health Care System.

Study Variables
Patients with a COVID-19 positive test were identi ed from the VA COVID-19 Shared Data Resource, and con rmed from a combination of laboratory results and physician notes. Determination of therapeutic anticoagulant prescription dosing for assignment to the AC group was ascertained from the CDW outpatient pharmacy domain. Incidence of MSOC, between 5-30 days after initial COVID-19 positive status of stroke, limb ischemia, gastrointestinal (GI) bleeding, intensive care unit (ICU) death, ICU death, as well as composite MSOC outcomes were identi ed from International Statistical Classi cation of Diseases and Related Health Problems, Tenth Revision (ICD-10) diagnosis codes in CDW outpatient and inpatient encounters domains. Admission outcomes between 1-30 days after initial COVID-19 positive status related to hospitalization, ICU admission, death while in the hospital, death while in the ICU, hospital maximum length of stay, and ICU maximum length of stay were also extracted from the CDW inpatient encounters domain and the VA vital status database (date of death).
Study covariates included patient demographic information, including age, race, sex, and smoking status;

Propensity matching
To reduce confounding caused by unbalanced covariates in anticoagulant group and control group, we also performed propensity score analyses. The patients in control group are about 53 times of the patients in anticoagulant group. Since there is little to gain in terms of statistical e cacy with over 4 controls to 1 case (15, 16), we studied 1:1, 1:2, 1:3 and 1:4 matching. We used greedy nearest neighbor matching without replacement within 0.25 caliper method to form 1:1, 1:2, 1:3, 1:4 matching with propensity score, then chose the best matching ratio for entire cohort. Similarly, we performed the same propensity score analysis separately in hospitalized patients, as well as in ICU and non-ICU patients. Standardized mean difference pre-and post-1:1, 1:2, 1:3, 1:4 matching was performed to assess the quality of the propensity methods. From this, 1:2 and 1:4 matching appeared to be the best quality. We applied the same methods to hospitalized and ICU cohorts, and observed that 1:4 matching provided the best quality data for these cohorts. Therefore, to be consistence, we utilized 1:4 matching for all three cohort assessments, and then conducted conditional logistic regression analyses for all the study outcomes.

Statistical Analysis
Characteristics of the overall study population, are reported as mean (95% con dence interval) or counts (percentage). T-tests for continuous covariates and chi-square tests for categorical covariates were conducted between the anticoagulant group and control group for each cohort. Multivariate binary logistic regression models, adjusted for covariates ascertained prior to the initial COVID-19 positive test, were used to compare various outcomes between those in the anticoagulation group and those in the control group. The unadjusted logistic regression model is summarized in eTable 1. Odds ratios with 95% con dence intervals are reported for those in the overall cohort and those in the hospitalized cohort. Effect modi cation for the composite MSOC outcomes was conducted for various covariates as well. All statistical tests were two sided, where a p < 0.05 or a 95% con dence interval that did not cross 1.0 was considered statistically. No imputation was conducted. All analyses were done using SAS Enterprise Guide, version 7.1 (SAS Institute Inc).

In-hospital outcomes
During the 30-day follow-up period, between the AC and control groups in patients who were hospitalized with COVID-19 (

Discussion
Our study investigated whether continuous therapeutic AC decreased the risk of MOSC, ICU admission and in-hospital mortality in patients infected with COVID-19 during the early phase of the pandemic in the US. In the study cohort, the rate of in-hospital mortality was lower in the AC group compared to the control group, despite a higher incidence of GI bleeding. However, the rate of MOSC and ICU admission were similar between the two groups.
Several studies reported an increased risk of hypercoagulable state in COVID-19 positive patients (17)(18)(19)(20)(21)(22)(23). In a meta-analysis of 91 studies evaluating 35,017 COVID-19 patients, Mansory et al. found the overall incidence of VTE events in all hospitalized and ICU patients was 12.8% and 24.1% respectively (23). Reports showed hyper-thrombotic laboratory changes (e.g., d-dimer) in COVID-19 patients, and these changes are observed to be negative independent predictors of end-organ-damage and death (1,(24)(25)(26). In addition, critically-ill COVID-19 patients manifest with a wide range of clinical symptoms that include interstitial pulmonary edema, ST elevation myocardial infarction, brainstem infarcts, gastro-intestinal mucosal bleeding, acute kidney failure, liver dysfunction, and skin petechial rashes (7,(27)(28)(29)(30)(31). The mechanism by which COVID-19 viremia and cytokines induce thrombosis and in ammation is still not fully understood. However, some have postulated that hypercoagulability and MSOC in patient with COVID-19 results from a sequence of events that rst starts with endothelial cell dysfunction and excess thrombin generation and brinolysis shutdown (32,33). Next hypoxia associated with severe COVID-19 can lead to further thrombosis via activation of hypoxia-inducible transcription factor-dependent signaling pathway (34). Lastly, immobilization during critically-ill states, can lead to blood ow stasis and higher incidence of in situ thrombosis.
Limited studies to date have assessed outcomes of therapeutic AC in patients with COVID-19, and the few that have provide contradictory ndings. Early observational studies and a meta-analysis reported no difference in mortality in patients with COVID-19 who received AC therapy relative, but these studies suffered from small sample sizes, heterogenous patient populations, and unclear inclusion and exclusion criteria (14,35,36 (39). On the other hand, AC was not shown to reduce the incidence or severity of MSOC such as acute kidney injury (13).
In the present study, patients who were on a continuous prescription of therapeutically-dosed AC therapy for at least 30 days prior to, and 30 days after their initial COVID-19 positive test, demonstrated reduced risk of in-hospital mortality when compared to control patients (aOR 0.67, 95% CI: 0.46 -0.99, p = 0.04). Interestingly, our study observed that COVID-19 positive patients who received AC and had higher cardiovascular co-morbidities, actually had an equivalent incidence of MOSC. However, the potential bene ts of AC in this higher risk population needs to be weighed against the risk of bleeding. Our study observed that AC therapy was associated with a higher incidence of GI bleeding, which is consistent with prior literature (25). As treatment protocols continue to evolve for patients a icted with COVID-19, personalized assessments of the risks and bene ts of AC therapy should be tailored to meet the most favorable anticipated outcomes.
The ndings of this study should be interpreted with certain limitations. First, the associations explored were retrospective in nature and as such can include inherent biases associated with the study design, and method of data extraction and analysis. Second, there are factors that were not included in our propensity score matching that could also potentially impact mortality and morbidity. One such risk factor is smoking status, which was not discernable in the CDW. Third, within our study cohort we were unable to determine the primary cause of death for patients with COVID-19 during the early phase of the pandemic, which we found was often not clearly de ned in the patient's medical record.

Conclusion
In summary, our study demonstrates that among U.S. veterans a icted by COVID-19, chronic AC therapy reduced the risk of in-hospital death. However, AC therapy did not ameliorate the risk of MOSC. These ndings suggest that MOSC in patients with COVID-19 are unlikely to be mainly determined by the patient's pro-thrombotic state. Randomized control trials are needed to further evaluate the effectiveness of AC in preventing COVID-19-related MOSC, particularly in later and ongoing stages of this evolving viral illness. De-identi ed datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests