COVID‐19 and coagulation dysfunction in adults: A systematic review and meta‐analysis

Abstract The outbreak of 2019 novel coronavirus disease (COVID‐19) has posed a grave threat to the global public health. The COVID‐19‐induced infection is closely related to coagulation dysfunction in the affected patients. This paper attempts to conduct a meta‐analysis and systematically review the blood coagulation indicators in patients with severe COVID‐19. A meta‐analysis of eligible studies was performed to compare the blood coagulation indicators in patients with severe and nonsevere COVID‐19. PubMed, Embase, Web of Science, and the Cochrane Library were searched for studies published between 1 December 2019 and 7 May 2020. A total of 13 studies with 1341 adult patients were enrolled in this analysis. Platelet (weighted mean difference [WMD] = −24.83, 95% confidence interval [CI]: −34.12 to −15.54; P < .001), d‐dimer (WMD = 0.19, 95% CI: 0.09‐0.29; P < .001), and fibrinogen (WMD = 1.02, 95% CI: 0.50‐1.54; P < .001) were significantly associated with the severity in patients with COVID‐19. The meta‐analysis revealed that no correlation was evident between an increased severity risk of COVID‐19 and activated partial thromboplastin time (WMD = −1.56, 95% CI: −5.77 to 2.64; P = .468) or prothrombin time (WMD = 0.19, 95% CI: −0.13 to 0.51; P = .243). The single arm meta‐analysis showed that compared with the nonsevere group, the severe group had a lower pooled platelet (165.12 [95% CI: 157.38‐172.85] vs 190.09 [95% CI: 179.45‐200.74]), higher d‐dimer (0.49 [95% CI: 0.33‐0.64] vs 0.27 [95% CI: 0.20‐0.34]), and higher fibrinogen (4.34 [95% CI: 1.98‐6.70] vs 3.19 [95% CI: 1.13‐5.24]). Coagulation dysfunction is closely related to the severity of patients with COVID‐19, in which low platelet, high d‐dimer, and fibrinogen upon admission may serve as risk indicators for increased aggression of the disease. These findings are of great clinical value for timely and effective treatment of the COVID‐19 cases.


| INTRODUCTION
The global outbreak of 2019 novel coronavirus disease  has posed enormous impacts on the public health, with severe and critically ill patients accounting for 20% of all COVID-19 patients 1 and the fatality rate of critically ill patients amounting to 49%. 2 Severe patients are more likely to develop acute respiratory distress syndrome (ARDS) and multiple organ dysfunction syndrome, which may affect the prognosis of patients with COVID-19. 3 Therefore, an early screening of severe and nonsevere patients is critical to reduce the mortality rate of patients with COVID-19.
Of note, abnormal coagulation function has been prevalent in 20% of the patients with COVID-19. 4 Moreover, the prevalence of disseminated intravascular coagulation (DIC) in COVID-19 cases is higher than that of severe acute respiratory syndrome (SARS) patients. 5 A recent study reports that the mortality of COVID-19induced DIC is 71.4%. 6 Hence, coagulation dysfunction is closely related to the severity of COVID-19 cases and can endanger patients' lives. 7 However, a close examination indicates that uncertainties and controversies still persist and await further research efforts to shed new light on them.
This meta-analysis first followed a strict definition of "severity" and focused on the coagulation blood indicators, including platelet, ddimer, prothrombin time (PT), activated partial thromboplastin time (APTT), and fibrinogen. It attempted to explore the difference in coagulation dysfunction between severe and nonsevere adult COVID-19 patients, so as to screen the severe patients and to timely adjust the therapeutic regimen to improve the prognosis.

| Data sources and search strategy
This meta-analysis followed the PRISMA recommendations and was registered with PROSPERO-The International Prospective Register of Systematic Reviews (registration No. CRD42020186941). PubMed, Embase, Cochrane Central Register of Controlled Trials, and Web of Science were systematically searched for papers published between 1 December 2019 and 7 May 2020. The language restriction was English and the search terms were "COVID-19" or "2019-nCoV" or "SARS-CoV-2" or "Novel Coronavirus-Infected Pneumonia" or "2019 novel coronavirus" or "coronavirus 2019" and "severe" or "severity." The search strategy has been provided in the Supporting Information Material (File S1). The classification criteria for severity observed the "The American Thoracic Society (ATS) guidelines for community-acquired pneumonia," 8 "WHO COVID-19 Clinical Guidelines," 9 or "COVID-19 Diagnosis and Treatment Protocol" of China. 10 ATS guidelines for community-acquired pneumonia, 8 in which the severe type is defined according to either one major criterion (including septic shock in need of vasopressors and respiratory failure requiring mechanical ventilation) or three or more minor criteria (including respiration rate [RR] > 30/min, PaO 2 /FiO 2 < 250 mm Hg, multilobar infiltrates, confusion/disorientation, uremia, leukopenia, thrombocytopenia, hypothermia, and hypotension requiring aggressive fluid resuscitation). WHO's interim guidelines for COVID- 19,9 in which the severe type includes severe pneumonia (for adolescents or adults: fever or suspected respiratory infection, plus one of the following: RR > 30/min; severe respiratory distress; or SpO 2 ≤ 93% on room air), ARDS (PaO 2 /FiO 2 ≤ 300mm Hg with positive end expiratory pressure or continuous positive airway pressure ≥5 cm H 2 O, or nonventilated; SpO 2 / FiO 2 ≤ 315 or else), sepsis, and septic shock. The diagnostic and treatment guidelines for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) issued by Chinese National Health Committee, 10 in which the severe type is designated as one meeting any of the following indices:

| Data extraction and quality assessment
Two reviewers (YH and HHM) independently extracted prespecified data elements (first author, year of publication, country, sample size, age, sex, value of coagulation indicators) from each study. Median and range were used to estimate mean and standard deviation (SD). The percentiles were converted to SD according to the following formula: SD ≈ Norm interquartile range = (P75 − P25) × 0.7413 (P75: 75th percentile and P25: 25th percentile). 11 Study quality was assessed using the Newcastle-Ottawa Scale (NOS) checklist. Disagreements were settled by consensus.
A high-quality study was defined as a study with a score ≥7. 12

| Data synthesis and analysis
All the statistical analyses were performed with the STATA software (version 12.0; STATA Corporation, College Station, TX). Weighted mean difference (WMD) with 95% confidence interval (CI) was calculated for each blood coagulation indicator. I 2 was employed to evaluate statistical heterogeneity with I 2 values of <25%, 25%-75%, LIN ET AL.
| 935 and >75%, respectively, indicating low, moderate, and high heterogeneity. The choice of the proper-effect models was based on the analysis results: the fixed-effect model was used for I 2 ≤ 50% and the random-effect model for I 2 > 50%. The single-arm meta-analysis of proportions with 95% CIs was conducted for the value of blood coagulation indicators. Subgroup analysis was performed according to sample size, if adapting the random-effect model. In addition, a further sensitivity analysis was performed to test the stability of the results. The Begg test was performed to assess publication bias if a coagulation indicator was retrieved from 10 or more studies. The statistical significance was set at P < .05.

| Study selection
The review progress is summarized in Figure 1. A total of 7192 articles were identified basing on the search strategy, and 1453 duplications were removed. Then, 173 were potentially relevant to the review question after screening of titles and abstracts and 160 studies were further excluded according to the aforementioned criteria. Finally, 13 studies 13-25 with 1341 patients and 371 severe COVID-19 adults were enrolled in the meta-analysis.

| Study characteristics and quality assessment
All these studies were from China and published in 2020, with different sample sizes ranging from 21 to 204 patients, with a clear severity distinction (all the studies used the Chinese definition). The features of the 13 enrolled studies are summarized in Table 1. The quality assessment was based on NOS, with quality score ranging from 6 to 8. The quality results are shown in Table 1 and the assessment of each literature in the NOS scale is depicted in Table S1.
Despite the high heterogeneity, the result should be interpreted with caution, for the 95% CI of WMD ranged from −5.77 to 2.64.

| PT
For PT, five studies with 302 nonsevere and 151 severe COVID-19 patients were eligible for meta-analysis. In four studies, PT was higher in the severe group than in the nonsevere one, but the difference was not significant (WMD = 0.19, 95% CI: −0.13 to 0.51; P = .243, I 2 = 65.2%; Figure 2).

| Sensitivity analysis and publication bias
As shown in Figure S1, from the results of the sensitivity analysis, the combined results did not change with the exclusion of any of the studies. Thus, sensitivity analysis suggested that these metaanalyses (d-dimer, APTT, PT, and fibrinogen) were steady. Because only d-dimer was retrieved from 10 studies (≥10), a funnel plot regarding the d-dimer showed that the P-value of the Begg test was .858. The Begg test of "d-dimer" suggested that no stable evidence of publication bias was present in the meta-analysis ( Figure 5). Previous studies suggest that influenza-associated pneumonia is associated with thrombotic events. 26 Similarly, COVID-19 can induce thromboembolic complications, especially in severe patients. 27 In severe cases, COVID-19 triggers a cytokine storm, which activates the coagulation cascade, resulting in the thrombotic phenomena. 28 Nevertheless, the mechanism of COVID-19-related coagulation disorder is far more complicated to be explained by inflammation. 29,30 A previous study proposed that the immunity-related pulmonary Thrombocytopenia is considered as a dysregulated host response in critical sepsis patients, 32 which usually occurs after a viral infection (such as influenza and HIV). 33 In SARS-related diseases, the virus has been suspected to cause platelet consumption in the lungs by damaging epithelial cells 34 and infect hematopoietic stem cells and megakaryocytes. 35 Similarly, COVID-19 is homologous to SARS. A recent study by Yang et al 36 showed that 20.7% of patients with COVID-19 had a low platelet (<125 × 10 9 /L) and 5% of them had a very low platelet (range, 0-50 × 10 9 /L) with 92.1% mortality. 36 However, the incidence of thrombocytopenia was much lower in a study by Zheng et al 37 39 ; the other reported that thrombocytopenia was associated with the severity of COVID-19, but with a high heterogeneity (I 2 = 92%). 40 Compared with the two meta-analysis studies, the current meta-analysis has a clearer definition of "severity," focuses on "adult patients," and enrolls more studies. Therefore, the heterogeneity is low and the results are more accurate. Fibrinogen is an essential part of the blood coagulation cascade.
In sepsis, decreased fibrinogen is associated with an increased mortality, 49 which is related to consumptive coagulopathy. The elevated fibrinogen has been documented in infectious diseases, acute stroke, and myocardial infarction. 50 The current study is the first metaanalysis to summarize increased fibrinogen levels in severe COVID-19 adults. In addition, fibrinogen is an acute-phase protein, which is induced by interleukin 6 and associated with inflammatory responses. 51 Once infection occurs, hepatic synthesis of fibrinogen increases 2 to 10 times. 52 Possibly different from septic coagulation disorder, early severe COVID-19 patients present hypercoagulability rather than consumptive coagulopathy. 53,54 Prolonged PT and APTT are linked to anticoagulant, coagulation factor deficiency, and fibrinolysis, which have been used as laboratory tools to predict bleeding. 55,56 The performance of PT and APTT was contradictory in the initial research of severe COVID-19 cases.
Some studies showed shortened PT and APTT in severe COVID-19 patients 14 while another study reported prolonged PT and APTT. 22 Our meta-analysis found no difference in PT and APTT between the severe and nonsevere groups upon admission. Probably, PT and APTT may fail in an early recognition and be influenced by many factors (eg, anticoagulant). However, this finding should be interpreted with caution because of the high heterogeneity and much less included literature. More clinical trials are urgently needed to investigate the relation between PT/APTT and COVID-19 severity.

| LIMITATIONS
The following limitations should be mentioned: first, patients from the studies were all from China. There may exist racial differences in blood coagulation. 57 Further studies on coagulation from other races are highly awaited. Second, blood indicators were included upon admission, without considering previous use of antiplatelet or anticoagulant drugs. Finally, not all five indicators were included in some studies, so we hope that more studies related to blood coagulation will expand the sample size in the future.

DATA AVAILABILITY STATEMENT
All data are fully available online without restriction.