Elevated D-Dimer Levels are Associated with Increased Risk of Mortality in COVID-19: A Systematic Review and Meta-Analysis

Introduction. The 2019 novel Coronavirus (2019 nCoV), now declared a pandemic has an overall case fatality of 2 to 3% but it is as high as 50% in critically ill patients. D-dimer is an important prognostic tool, often elevated in patients with severe COVID-19 infection and in those who suffered death. In this systematic review, we aimed to investigate the prognostic role of D-dimer in COVID-19 infected patients. Methods. We searched PubMed, Medline, Embase, Ovid, and Cochrane for studies reporting admission D-dimer levels in COVID-19 patients and its effect on mortality. Results. 18 studies (16 retrospective and 2 prospective) with a total of 3,682 patients met the inclusion criteria. The pooled mean difference (MD) suggested significantly elevated D-dimer levels in patients who died versus those survived (MD 6.13 mg/L, 95% CI 4.16 to 8.11, p <0.001). Similarly, the pooled mean D-dimer levels were significantly elevated in patients with severe COVID-19 infection (MD 0.54 mg/L, 95% CI 0.28 to 0.8, p< 0.001). In addition, the risk of mortality was four-fold higher in patients with positive D-dimer vs negative D-dimer (RR 4.11, 95% CI 2.48 to 6.84, p< 0.001) and the risk of developing the severe disease was two-fold higher in patients with positive D-dimer levels vs negative D-dimer (RR 2.04, 95% CI 1.34 to 3.11, p < 0.001). Conclusion. Our meta-analysis demonstrates that patients with COVID-19 presenting with elevated D-dimer levels have an increased risk of severe disease and mortality.


Introduction
The 2019 novel Coronavirus (2019-nCoV) or Severe Acute Respiratory Syndrome Corona Virus 2 (SARS-CoV-2), now declared a pandemic, first originated in December 2019 in Wuhan city of Hubei province, China and has since caused a significant impact on mankind 1 .
As of April 24th, 2020, 2. 6  The overall case fatality rate (CFR) for COVID-19 was reported at about 2% in China.
Still, it was noted to be higher at 7.2% in Italy, which was felt secondary to the higher mean age of the overall population 5 . The CFR is significantly high in patients with severe COVID-19 infection with CFR as high as >50% in patients admitted to the Intensive Care Unit (ICU) 6 . Due to high mortality in critically ill COVID-19 patients, the detection of biomarkers which may help identify them earlier in their course of illness can be crucial. D-dimer is one such biomarker that has emerged as an important prognostic tool, with elevated levels in critically ill patients and those deceased. In this systematic review, we aimed to investigate the prognostic role of admission D-dimer levels in patients hospitalized with COVID-19.

Search strategy
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020. . https://doi.org/10.1101/2020.04.29.20085407 doi: medRxiv preprint The reporting of this systematic review and meta-analysis complies with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) guidelines (Supplement Table 1) 7 .
The initial search strategy was developed by two authors (SS and SP). We performed a systematic search, without language restriction, using PubMed, EMBASE, SCOPUS, Google Scholar, and two preprint servers (https://www.medrxiv.org/ and https://www.ssrn.com/index.cfm/en/coronavirus/) from inception to April 16th, 2020, for studies that reported D-dimer levels in COVID-19 patients. We utilized the "related articles" function in PubMed to find relevant articles that were missed by the initial search. In addition, reference lists of the included studies were hand-searched to further locate relevant articles that were missed in the primary search. We used the following keywords and medical subject heading: "COVID-19", "SARS-CoV-2", "Wuhan coronavirus", "Coronavirus 2019", "2019 n-CoV", "Ddimer", "laboratory".

Study Selection and data extraction
To be included in our systematic review and meta-analysis the study had to fulfill the The data from included studies were extracted using a standardized protocol and a data extraction form. Any discrepancies between the two investigators were resolved with a . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020. . https://doi.org/10.1101/2020.04.29.20085407 doi: medRxiv preprint consultation with the senior investigator (JG). Two independent reviewers (SS and SP) extracted the following data from the eligible studies: author name, study design, publication year, followup duration, number of patients, age, gender, diabetes mellitus (DM), hypertension (HTN), coronary artery disease (CAD), acute cardiac injury, arrhythmias, shock, and outcomes. The Newcastle Ottawa Risk bias assessment tool was used to appraise the quality of the included studies (Supplement Table 2).

Clinical outcomes
The primary outcome of interest in our study was all-cause mortality and severity of COVID-19.

Statistical Analysis
Mantel-Haenszel risk ratio (RR) random-effects model (DerSimonian and Laird method) was used to summarize data between the groups 8 . The D-dimer levels in the studies were reported as median and Interquartile Range (IQR). We used the Wan method to estimate the mean and standard deviations 9 . We then calculated the pooled difference in means (MD) to evaluate the association of levels of D-dimer between the groups. Higgins I-squared (I 2 ) statistic was used to assess the test of heterogeneity. A value of I 2 of 0-25% represented insignificant heterogeneity, 26-50% represented low heterogeneity, 51-75% represented moderate heterogeneity, and more than 75% represented high heterogeneity 10 . A pre-specified randomeffects meta-regression analysis was conducted for the primary outcome in relation to the baseline demographics, comorbid condition, biomarkers to test the relationship between D-dimer and disease severity, and all-cause mortality. Publication bias was formally assessed using . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 5, 2020.

Search results
A total of 920 citations were identified during the initial search (Figure 1). Nine hundred and two records were excluded. After a detailed evaluation of these studies, 12 studies met the inclusion criteria. We included six manuscripts from 2 pre-print servers (https://www.medrxiv.org/ and https://www.ssrn.com/index.cfm/en/coronavirus/), to accommodate the rapidly evolving nature of information for COVID. We acknowledge that the manuscripts from these two sources are not peer-reviewed. Eighteen articles including 3,682 patients were included in the final analysis.

Study characteristics
This systematic review and meta-analysis of 18 studies incorporated a total of 3,682 is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 5, 2020. . https://doi.org/10.1101/2020.04.29.20085407 doi: medRxiv preprint ≤ 93% on room air or an arterial blood oxygen partial pressure (PaO2)/oxygen concentration (FiO2) ≤ 300 mm Hg and was consistent across all studies. The severe group included patients with severe COVID-19 and those needing ICU care for acute respiratory failure requiring mechanical ventilation, or for shock, or multiorgan failure. The acute cardiac injury was defined as an elevation in cardiac troponin or new changes of ischemia on electrocardiography (ECG) or new wall motion abnormalities on an echocardiogram. Table 1 summarizes the baseline characteristics of 6 studies which compared dead versus survived patients and 1 study which compared patients with elevated D-dimer vs normal Ddimer level. Among the 6 studies which compared dead versus survived patients, the mean age of the study population in this group was 62.5±14.8 years and 56.3% were males. Overall, hypertension (HTN) was the most common comorbidity (36.6%), followed by diabetes (DM) (16.8%) and coronary artery disease (CAD) (11.7%). The acute cardiac injury was present in 19.3% of patients while shock was observed in 8.9% of patients. New-onset arrhythmias of some form were observed in 12% of patients. Table 2 summarizes the baseline characteristics of 11 studies that compared severe versus non-severe COVID-19 patients. The mean age of the study population in this group was 49.9±17.2 years and 54.6% were males. Overall, HTN was the most common comorbidity (18.8%), followed by DM (9.2%) and CAD (3.9%). The acute cardiac injury was present in 11% of patients while shock was observed in 3.6% of patients, of which, 2% of patients have septic shock while it was undefined in other patients.

All-cause mortality
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 5, 2020. . https://doi.org/10.1101/2020.04.29.20085407 doi: medRxiv preprint The data for D-dimer levels were available in 5 studies 11, 13-16 . The pooled mean Ddimer levels were significantly elevated in patients who died versus those who survived (MD 6.13 mg/L, 95% CI 4.16 -8.11, p ≤ 0.001, I 2 = 81.41%) (Figure 2). No publication bias was observed (Egger's P = 0.39, Supplement Figure 1). Meta-regression analysis demonstrated no significant associations were found between age, male sex, hypertension, diabetes, coronary artery disease, C-Reactive Protein, and troponins in COVID-19 infected patients who died versus who survived ( Table 3).

Severity of COVID-19
The data for D-dimer levels were available in 9 studies 19-25, 27, 28 . The pooled mean Ddimer levels were significantly elevated in patients with severe COVID-19 infection (MD 0.54 mg/L, 95% CI 0.28 -0.8, p≤ 0.001, I 2 = 90.74%) ( Figure 4A). No publication bias was observed (Egger's P = 0.13, Supplement Figure 3). Meta-regression analysis showed a significant association between CAD, C-Reactive Protein, and severe COVID-19 disease, but the results were not significant for age, male sex, comorbidities (hypertension, diabetes, troponin levels) (  CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020. . https://doi.org/10.1101/2020.04.29.20085407 doi: medRxiv preprint Supplement Figure 4). A sensitivity analysis was performed by removing one study at a time (n-1 analysis) to investigate the significant heterogeneity. No significant change in the findings was observed with the sensitivity analysis.

Discussion
Elevated D-dimer is one of the abnormal laboratory parameters in patients with COVID-19 infection. D-dimer is the fibrin degradation products released upon cleavage of crosslinked fibrin by plasmin 29 . Historically, the role of D-dimer is limited due to its nonspecificity, with elevated levels are often seen with advanced age, African American race, female sex, active malignancy, surgery, pregnancy, immobility, cocaine use, connective tissue disorders, end-stage renal disease and prior thromboembolic disease 30  is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020. . largest to date to the best of our knowledge. The key findings of our pooled analysis are: 1) the D-dimer levels were higher in patients with severe COVID-19 infection and those who succumbed to death compared to non-severe disease and those who survived, respectively after adjusting for age, comorbid condition, CRP levels; 2) patients with elevated D-dimer levels were at increased risk of developing severe COVID-19 infection and increased all-cause mortality compared to those with normal D-dimer levels.
Zhou et al. 13 reported that D-dimer level >1 mg/L on admission was independently associated with increased odds of mortality, findings that echoes in our pooled analysis as well.
Also, patients with advanced age, higher Sequential Organ Failure Assessment (SOFA) score, elevated troponin, and B-type natriuretic peptide (BNP) have been associated with poor outcomes and mortality in COVID-19 infection 13, 33, 34 . Furthermore, using a higher cutoff value of D-dimer (levels > 2mg/L) predicted in-hospital mortality even better as noted by Zhang at el. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020. . https://doi.org/10.1101/2020.04.29.20085407 doi: medRxiv preprint microvascular thrombosis 14, 46 might be beneficial (provided a thorough risk-benefit assessment given these patients are also at risk of spontaneous bleeding) (however, our study was not designed to assess this difference). Besides that, it remains unclear at this time regarding the optimal dosing and duration in this patient population and hence needs to be explored further.
Although, extended DVT prophylaxis with oral anticoagulation at discharge (for up to 45 days) may be reasonable in patients at higher risk for the thromboembolic event (i.e. active malignancy, immobility and elevated D-dimer level > two times the upper limits of normal) and lower bleeding risk 47, 48 . Thus, using D-dimer levels as a surrogate marker for disease severity and underlying thromboembolic disease, especially, in COVID-19 patients who cannot get dedicated imaging might be beneficial.
Our study has a few important limitations. First, all studies included in our meta-analysis were from China, while the United States and Europe have the majority of COVID-19 cases currently. However, the preliminary reports from the United States and Europe have shown similar trends in COVID-19 infection in terms of clinical presentation and outcomes 5,49 . Our pooled analysis provides the best available data regarding trends of D-dimer levels in patients with COVID-19 infection and the likelihood of developing severe infection or mortality in patients with elevated D-dimer levels. Secondly, all studies included in our analysis were either prospective or retrospective reports, which is currently the best available evidence; and, therefore, subject to potential confounding and publication bias. Third, significant heterogeneity was observed between studies in our pooled analysis. Fourth, details on anticoagulation or trends of D-dimer over the course of hospitalization were not available. Finally, patient-level data to perform additional. detailed analyses are not available.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Conclusion
Our meta-analysis demonstrates that patients with COVID-19 presenting with elevated D-dimer levels have an increased risk of severe disease and mortality.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020. . https://doi.org/10.1101/2020.04.29.20085407 doi: medRxiv preprint Table 1: Baseline characteristics of studies included in the meta-analysis comparing COVID-19 infected patients who died versus who survived Table 2: Baseline characteristics of studies included in the meta-analysis comparing severe versus non-severe COVID-19 infected patients Table 3: Meta regression of baseline characteristics with mean difference in D-dimer levels in COVID-19 patients -dead versus survived Table 4. Meta regression of baseline characteristics with mean difference in D-dimer levels in severe versus non-severe COVID-19 infected patients Supplement  Table 2: Risk of bias assessment of studies included in our meta-analysis using Newcastle-Ottawa Scale    The Forest plot for pooled difference in mean D-Dimer levels in severe versus non-severe COVID-19 patients, followed by random-effects meta-regression analysis plots depicting the relationship between mean differences in D-Dimer levels (on y-axis) and (B) Coronary Artery disease (CAD) and (C) C-Reactive Protein (CRP). Each included study is represented by a circle, the size of which is proportional to its respective weight in the analysis. The line indicates the predicted effects (regression line). There was significant association between CAD (β = 0.8, P = 0.02), and CRP levels (β = 0.02, P = 0.03) and mean differences in D-Dimer levels. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Studies included in final analysis: 18
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020.   . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020.    The Forest plot for pooled difference in mean D-Dimer levels in severe versus non-severe COVID-19 patients, followed by random-effects meta-regression analysis plots depicting the relationship between mean differences in D-Dimer levels (on y-axis) and (B) Coronary Artery disease (CAD) and (C) C-Reactive Protein (CRP). Each included study is represented by a circle, the size of which is proportional to its respective weight in the analysis. The line indicates the predicted effects (regression line). There was significant association between CAD (β = 0.8, P = 0.02), and CRP levels (β = 0.02, P = 0.03) and mean differences in D-Dimer levels.    . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 5, 2020. . https://doi.org/10.1101/2020.04.29.20085407 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020.   is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020.  is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 5, 2020.