A Dynamic, D-dimer-based Thromboprophylaxis Strategy in Patients with COVID-19

Introduction

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2, COVID) has resulted in significant morbidity and mortality worldwide. One important mechanism for its adverse impact appears to be its promotion of arterial and venous thrombosis, endothelial dysfunction and enhanced angiogenesis, particularly in the lungs.¹ Early data from China found an association between D-dimer and mortality, but the actual incidence of thrombotic events was not reported.² A subsequent study showed that heparin or low molecular weight heparin (LMWH) improved survival in patients with sepsis-induced coagulopathy (SIC) score ≥ 4 or D-dimer value greater than 6 times the upper limit of normal (ULN).³ However, despite standard heparin or LMWH thromboprophylaxis, the cumulative incidence of thrombotic events in COVID-infected patients stay high almost 50% over critically ill patients in an intensive care unit (ICU) in the Netherlands.^4,5 While the recent guidelines support the use of standard thromboprophylaxis in critically ill COVID-19 patients over therapeutic dose anticoagulant in term of survival benefit and days of organ support. However, the use of therapeutic dose heparin in noncritically ill COVID-19 patients reduce the need for mechanical ventilation and improve survival.⁶

The concern for breakthrough thrombosis in hospitalized patients with limitations on radiologic testing access, prompted interest in using laboratory-based guide to anticoagulant dose adjustment at our large metropolitan public hospital. A large retrospective cohort shows a cut-off d-dimer of 1.8 mcg/ml FEU with 70% accuracy in diagnosis of acute symptomatic pulmonary embolism (PE) patients that are associated with worse mortality and morbidity outcomes.⁷ A large retrospective study proposes the D-dimer cut-off of 2495 ng/ml (4.9 mcg/ml FEU) for diagnosis of asymptomatic PE in COVID-19 patients.⁸ A systematic review also shows an association between D-dimer levels between 1000-4800 microgram/L (2-9.6 mcg/ml FEU) and pulmonary embolism in patients with COVID-19.⁹

Based on COVID-related limitation on access to radiologic testing for deep vein thrombosis (DVT) and pulmonary embolism (PE) early in the epidemic, we implemented a D-dimer-stratified anticoagulation protocol for COVID patients as part of a quality improvement protocol for hospitalized patients who need anticoagulant. This protocol is dynamic and adjusted in associated with clinical of the patients and d-dimer level. The primary endpoint was in-hospital survival of who received anticoagulant titration according to the protocol compared with those who received fix anticoagulant dose since the beginning. Secondary endpoints included safety outcome, incidence of breakthrough thrombosis, length of hospitalization and ICU stay, days of mechanical ventilator use, and survival duration.

Methods

This single-center retrospective cohort study is part of an ongoing quality improvement initiative evaluating the use of an anti-factor Xa-driven heparin protocol, which includes a low-dose intravenous (IV) unfractionated heparin (UFH) option. A guideline for the use of anticoagulation in COVID-infected patients was implemented in February 2020 to standardize the management of these patients (Figure 1). The inclusion criteria include all patients admitted with COVID-19 pneumonia confirmed by PCR testing between March 15th and June 15th, 2020. Patients who required therapeutic anticoagulation for other indications at the time of admission, those who presented with radiographically confirmed DVT or PE and those who died within 24 hours of anticoagulant treatment were excluded from this analysis. Baseline D-dimer level was obtained on admission and every 24-48 hours during hospitalization. Levels of D-dimer were measured by means of the STA® - Liatest® D-Di Plus assay (Roche Diagnostics SpA, Milan), an immuno-turbidimetric assay for the quantitative determination of D-dimer in venous plasma. Those who presented with radiographically confirmed DVT or PE were started on treatment dose anticoagulation as per the hospital protocol and were therefore excluded from this analysis. Those without existing thrombosis were stratified into two groups based on their D-dimer: those with a D-dimer below 6 mcg/ml FEU (3.0 ug/mL) were given standard prophylactic anticoagulation and those with a D-dimer above this cutoff received escalated prophylactic dosing (LMWH 0.5 mg/kg subcutaneous every 12 hours or low dose heparin intravenous continuous drip titrated to an anti-Xa level of 0.1-0.3). Subsequent dosing increases were made based on a rise in D-dimer to above 6 mcg/mL FEU, an increase of more than 2 mcg/mL FEU/mL despite >48 hours of prophylactic anticoagulation or inability to dialyze due to clotting in line, filter, or machine. Patients who demonstrated acute respiratory decompensation while receiving prophylactic or escalated prophylactic dosing had further escalation to full therapeutic dosing.

Figure 1. Anticoagulation Treatment Protocol Based on D-Dimer levels.

Based on our hospital’s anticoagulation protocol (Figure 1), we categorized anticoagulant use as follows:

Standard Prophylactic which includes subcutaneous UFH 5,000 units every 8 hours (if creatinine clearance <30 mL/min), or subcutaneous LMWH with enoxaparin 40 mg once daily (40 mg twice daily if BMI>30; 30 mg once daily if Creatinine Clearance <30 mL/min)

Escalated Prophylactic which includes low-dose IV UFH titrated to achieve an anti-factor Xa level of 0.1-0.3 anti-Xa units (if creatinine clearance <30 mL/min) or enoxaparin 0.5 mg per kilogram body weight subcutaneously every 12 hours (if creatinine clearance ≥ 30 mL/min)

Standard Therapeutic dose used, which includes therapeutic dose IV UFH titrated to achieve an anti-factor Xa level of 0.3-0.7 anti-Xa units, or enoxaparin 1 mg per kilogram body weight every 12 hours during the hospitalization.

Patients treated with different anticoagulation regimens during their hospitalization were categorized into 2 categories based on their requirement of dosing up-titration as who received up-titrating according to the protocol and who received fixed anticoagulant dosing (Figure 1).

We collected baseline demographic data including age, sex, body mass index (BMI), ethnicity, comorbidities, baseline inflammatory markers, and clinical parameters associated with disease severity including intensive care unit (ICU) admission, requirement for high flow nasal cannula (HFNC), mechanical ventilation (MV), use of vasopressor for maintenance of hemodynamic stability, and sepsis-induced coagulopathy score (SICS)¹⁰ and also COVID-specific therapies.

The primary endpoints assessed were 30 day-in-hospital survival based on patients who were titrated to either escalated or therapeutic dose compared to fixed dose anticoagulant. We also adjust the potential covariates factors that affect in-hospital survival, such as D-dimer levels, age, ethnicity, BMI, comorbidities, ICU admission, mechanical ventilation, HFNC, SIC score, COVID treatment, thrombosis, and bleeding events. Secondary end points included bleeding outcome according to International Society on Thrombosis and Hemostasis (ISTH) defined major and clinically relevant non-major bleeding events.¹¹ Breakthrough thrombosis was recorded if identified on a radiologic study (venous duplex ultrasonography, CT-pulmonary angiogram, CT with contrast, CT or MRI brain) during the course of the hospital stay. Duration of hospitalization and ICU stay, and days of mechanical ventilator use.

Results

Survival outcome

The mortality rate in this cohort of hospitalized COVID-infected patients was 12.7%. When we compared 30 days in-hospital survival of the patients who received anticoagulant adjustment (up-titrating) according to the protocol to whom that received fixed anticoagulant dosing, the up-titrated group showed statistically significantly higher in-hospital survival until discharge at 30 days of 80.2% compared to 51.3 % in the fixed anticoagulant group (Figure 2).

Figure 2. Kaplan-Meier estimates for in-hospital survival of COVID-infected patients comparing up-titrated anticoagulant group to those who received fixed anticoagulant dose since the beginning.

Based on the multivariable Cox regression, we found factors that independently increased in-hospital mortality that include increasing age (HR 1.05, 95%CI:1.02,1.08), azithromycin treatment (HR 5.80, 95%CI:1.29, 25.97). Lower in-hospital mortality was associated with steroid treatment (HR 0.39, 95%CI:0.16, 0.96). We found that receiving up-titrated heparin was an independent factor associated with lower in-hospital mortality. (HR 0.31, 95%CI:0.13, 0.75). Schoenfeld residuals were tested in the final model, and the results showed that the proportional hazards assumption was valid, allowing us to interpret the independent effect of the up-titrated group separate from the effect of steroids.

Efficacy and safety outcome

Across the cohort, there were 13 patients who experienced bleeding events (5%). Only one major bleeding event (0.38%) occurred in up-titrate anticoagulant group, while 12 patients (4.62%) experienced clinically relevant non-major bleeding (CRNMB) event by ISTH criteria. The prevalence of all type of bleedings were significantly higher in patients who received anticoagulant adjustment compared to who received fixed anticoagulant dosage (12.5% vs 2.13%, p<0.01). There were 7 patients (2.7%) who experienced breakthrough thrombosis. Six patients in the anticoagulant adjustment group developed venous thromboembolism (VTE), while one of the patients in fixed anticoagulant group developed arterial thrombosis. The only arterial thrombotic event that developed in fixed anticoagulant group is an ischemic stroke, and common thrombotic event in anticoagulant adjustment group was deep vein thrombosis (DVT).

D-dimer analysis

Mean profiles over time of D-dimer levels during hospitalization in both groups are shown in Figure 3A. When comparing the mean D-dimer level trends between patients who survived and those who died, we found a higher trend among the deceased, particularly during the first week of hospitalization for patients who received a fixed-dose anticoagulant as shown in Figure 3B. The suppressible level of D-dimer might explain the survival advantage of the up-titrated group in the early course of the disease. However, the timing of D-dimer level progression varied among patients. The progression of D-dimer levels in patients who died while receiving an up-titrated anticoagulant occurred after the first week. Considering the varying timing of D-dimer progression among patients, we also compared and graphically presented the mean profiles over time of D-dimer levels following each escalation step using a random-intercept linear mixed model. This model accounted for the correlation of D-dimer measurements within individuals. The fixed effects included time measuring D-dimer, treatment groups, and the interaction of these variables. The peak effect of escalated and therapeutic heparin on D-dimer levels clearly occurred within the first week of each escalation treatment as shown in Figure 3C. However, a peak D-dimer level above or equal to 6 mcg/ml FEU (3.0 ug/mL) or below 6 FEU/ml (3.0 ug/mL) was not associated with survival outcome nor was it a confounder; therefore, it was not included in the survival model.

Figure 3. A. Mean profiles of D-dimer over hospitalization days. B. Mean profiles over time of D-dimer according to survival outcome and anticoagulant group. C. Mean profiles over time of D-dimer for each escalated step by random-intercept linear mixed model.

Discussion

Many reports and observational studies early in the COVID pandemic era suggest a survival benefit for use of anticoagulation among severely ill COVID patients.^2–4 Considering this, we assessed a quality improvement initiative aimed at implementing a D-dimer-guided anticoagulation strategy for COVID-infected inpatients. We chose a D-dimer-based strategy due to the limited availability of radiologic studies during the COVID-19 pandemic. The most recent guidelines support the use of regular thromboprophylaxis for severely ill COVID-19 patients, as starting with therapeutic-dose heparin as an initial approach did not improve the chances of survival until hospital discharge or having more days without requiring cardiovascular or respiratory organ support.¹² On the other hand, the use of therapeutic-dose heparin in non-severely ill COVID-19 patients reduces the need for mechanical ventilation and improves overall survival.⁶ Our strategy was dynamically adjusted upon the patient’s condition and d-dimer. We stratified patients into three groups, based on D-dimer level on admission and followed a stepwise up-titration of anticoagulation based on D-dimer values, which could occur at any time during the hospitalization. We found that there was no survival benefit for 6 patients who received a fixed therapeutic dose from the beginning of their hospitalization when compared to those who were up-titrated later in their hospital course. This suggests that the dynamic adjustment of anticoagulation treatment yields better survival outcomes compared to a fixed anticoagulant dose, whether it was fixed prophylactic or fixed therapeutic anticoagulation. We observed a suppression of D-dimer values within the first week of using higher doses of anticoagulation, suggesting that up-titrating anticoagulation may mitigate the expected adverse impact on survival. When defining COVID-19 severity according to standard guidelines, the severity is classified as mild for those with mild symptoms. Moderate disease is defined by the presence of clinical or radiographic evidence of lower respiratory tract disease but with a blood oxygen saturation of 94% or higher, and it requires hospitalization. Indicators of severe disease include hypoxemia with oxygen saturation ≤93% or a ratio of partial pressure of arterial oxygen to fraction of inspired oxygen <300. Critical illness is defined by respiratory failure requiring mechanical ventilation or shock requiring vasopressors. According to this definition, we found that 40.9% of patients in the fixed-dose group had severe disease, compared to 15.3% in the up-titrate group. Additionally, there were more critical patients in the up-titrate group, with 72.2% compared to 18.1% in the fixed-dose population. After reclassifying the patients according to standard guidelines, the combination of severe and critical patients accounted for the majority in both groups. The multivariable Cox regression analysis, including severity classified by standard guidelines, shows that severity is an independent factor affecting survival outcomes, while azithromycin treatment has no effect on survival. Therefore, the effect of azithromycin on survival in our data was confounded by patient severity.

An insuppressible trend of D-dimer level among the deceased could explain the worse outcomes in fix dose group. It is possible that deceased patients in the fixed-dose group might have had untreated asymptomatic pulmonary embolisms. According to previous reports, the incidence of VTE among critically ill patients is around 50%,^4,5 compared to 2.7% in our study, where one-third of the population and 84.8% of decrease patients could be classified as critically ill according to standard guidelines. The overall incidence of breakthrough thrombosis in our study is notably low. Furthermore, the mean D-dimer levels over time in deceased patients treated with fixed-dose thromboprophylaxis (Figure 3B) seemed to fall within the D-dimer range associated with pulmonary embolism (2.0-9.6 mcg/ml FEU) from a systematic review.⁹ These patients might require early titration of anticoagulants, additional treatment options, or a combination of both.

The prevalence of bleeding events was higher in the patients who received up-titrate doses of anticoagulation. The one major bleeding event observed was due to esophageal varices in a patient with previously undiagnosed liver cirrhosis, while others are clinically relevant non-major bleeding (CRNMB) that managed with standard supportive care and bleeding was not independently associated with survival outcome. The overall incidence of breakthrough thrombotic events was 2.7%, but this is likely underestimated due to a limited number of diagnostic studies performed in era of new emerging COVID disease.

The strengths of our D-dimer-titrated anticoagulation strategy included the use of a simple diagnostic test to stratify patient risk and allow for a dynamic reassessment over the course of the hospitalization. Using the rise in D- dimer as a trigger to increase anticoagulant allowed us to intervene early in the disease course. Our study is limited by its retrospective nature and lack of uniform diagnostic evaluation for DVT/PE in COVID-infected patients, even in those with suggestive symptoms. Previously, Demelo-Rodriguez et al reported that 15% of patients with COVID and elevated D-dimer levels have asymptomatic DVT.¹³ COVID infection is notably associated with a high risk of thrombotic complications, even in anticoagulated patients; therefore, the number of VTE reported in our cohort is likely underestimated. Furthermore, we did not collect data regarding thrombotic events that might have occurred after patients were discharged, and no autopsies were performed at our site to confirm the causes of death. Finally, we did not include a de-escalation strategy in our guideline, so this was done inconsistently based on physician discretion.

In this quality improvement evaluation of a D-dimer-titrated anticoagulation strategy for hospitalized patients with COVID, we found that D-dimer combined with clinically-driven anticoagulation adjustment was associated with reduced in-hospital mortality with a manageable increase in minor bleeding events.

Consent to participate

Informed consent waiver was obtained by University of Southern California Institutional Review Board. Based on the U.S. Code of Federal Regulations (CFR), the IRB determined that this study was exempt from 45 CFR 46 according to §46.104(d) as category 4 – Exempt Research as follows:

“(4) Secondary research for which consent is not required: Secondary research uses of identifiable private information or identifiable biospecimens, if at least one of the following criteria is met:

(i) The identifiable private information or identifiable biospecimens are publicly available;

(ii) Information, which may include information about biospecimens, is recorded by the investigator in such a manner that the identity of the human subjects cannot readily be ascertained directly or through identifiers linked to the subjects, the investigator does not contact the subjects, and the investigator will not re-identify subjects;

(iii) The research involves only information collection and analysis involving the investigator’s use of identifiable health information when that use is regulated under 45 CFR parts 160 and 164, subparts A and E, for the purposes of “health care operations” or “research” as those terms are defined at 45CFR 164.501 or for “public health activities and purposes” as described under 45 CFR 164.512(b);

(iv) The research is conducted by, or on behalf of, a Federal department or agency using government-generated or government-collected information obtained for non research activities, if the research generates identifiable private information that is or will be maintained on information technology that is subject to and in compliance with section 208(b) of the E-Government Act of2002, 44 U.S.C. 3501 note, if all of the identifiable private information collected, used, or generated as part of the activity will be maintained in systems of records subject to the Privacy Act of 1974, 5U.S.C. 552a, and, if applicable, the information used in the research was collected subject to the Paperwork Reduction Act of 1995, 44 U.S.C. 3501 et seq.”

Ethical approval: Based on the review conducted by the University of Southern California Institutional Review Board (IRB) designee on April 13, 2021, the study has been evaluated and categorized as exempt. The study is authorized to proceed with conducting as approved by the IRB. This exemption indicates that the research poses minimal risk to participants or involves only the collection of existing data, among other criteria outlined in the regulations.

Author contribution statement