key: cord-0864805-0wvt01y8
authors: Karimian, M.; Jamshidbeigi, A.; Badfar, G.; Azami, M.
title: Laboratory findings in coronavirus disease 2019 (COVID-19) patients: a comprehensive systematic review and meta-analysis
date: 2020-06-08
journal: nan
DOI: 10.1101/2020.06.07.20124602
sha: c661703a9daa4d4b8748e239f30262eef82ba4e4
doc_id: 864805
cord_uid: 0wvt01y8

Background: In early December 2019, the first patient with COVID-19 pneumonia was found in Wuhan, Hubei Province, China. Recent studies have suggested the role of primary laboratory tests in addition to clinical symptoms for suspected patients, which play a significant role in the diagnosis of COVID-19. Therefore, the present study was conducted to evaluate laboratory findings in COVID-19 patients. Material and methods: The present meta-analysis was reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines. This protocol is registered with the code CRD42019145410 in PROSPERO International Database. Results: Finally, 52 studies involving 5490 patients with COVID-19 entered the meta-analysis process. The prevalence of leukopenia, lymphopenia, elevated c-reactive protein (CRP), elevated erythrocyte sedimentation rate (ESR), elevated serum amyloid A, elevated ferritin was estimated to be 20.9% (95%CI: 17.9-24.3), 51.6% (95%CI: 44.0-59.1), 63.6% (95%CI: 57.0-69.8), 62.5% (95%CI: 50.1-73.5), 63.6% (95%CI: 57.0-69.8), 62.5% (95%CI: 50.1-73.5), 74.7% (95%CI: 50.0-89.7), and 72.6% (95%CI: 58.1-83.5), respectively. The prevalence of elevated interleukin-6 was 59.9% (95%CI: 48.2-70.5), CD3 was 68.3% (95%CI: 50.1-82.2), reduced CD4 was 62.0% (95%CI: 51.1-71.6), reduced CD8 was 42.7% (95%CI: 32.2-53.9). The prevalence of elevated troponin-I was 20.6% (95%CI: 9.0-40.5), elevated creatine kinase-MB (CKMB) was 14.7% (95%CI: 7.1-28.0), elevated brain natriuretic peptide (BNP) was 48.9% (95%CI: 30.4-67.7), elevated blood urea nitrogen was 13.1% (95%CI: 6.6-24.4),, elevated creatinine was 7.2% (95%CI: 4.4-11.8), elevated lactate dehydrogenase (LDH) was 53.1% (95%CI: 43.6-62.4), hyperglycemia was 41.1% (95% CI: 28.2-55.5), elevated total bilirubin was 48.9% (95%CI: 30.4-67.7), reduced albumin was 54.7% (95%CI: 38.1-70.2), reduced pre-albumin was 49.0% (95%CI: 26.6-71.8), and reduced PT was 53.1% (95% CI: 43.6-62.4), and D-dimer was 44.9% (95%CI: 31.0-59.6). Conclusion This study provides a comprehensive description of laboratory characteristics in patients with COVID-19. The results show that lymphopenia, elevated CRP, elevated ESR, elevated ferritin, elevated serum amyloid A, elevated BNP, reduced albumin, reduced pre-albumin, reduced CD3, reduced CD4, reduced CD8, elevated D-dimer, reduced PT, elevated interleukin-2, elevated interleukin-6, elevated LDH and hyperglycemia are the common findings at the time of admission.

The present meta-analysis was reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines (63). This protocol is registered with the code CRD42019145410 in PROSPERO International Database. All steps were performed by two authors (M.A and A.J.) and disagreements between the two authors regarding the quality of studies or extracted data were resolved after discussion with the senior author (M.K.).

We did a comprehensive literature review in databases PubMed/Medline, Scopus, EMBASE, Web of Science, Cochrane Library (Cochrane Database of Systematic Reviews -CDSR), Ovid, Science Direct, CINAHL and EBSCO to find citations from the beginning of January 2019 to the beginning of April 2020 without any restrictions. A search for gray literature was conducted at https://www.medrxiv.org/. To ensure a thorough systematic literature review, we reviewed the reference lists of eligible studies and systematic eligible surveys identified from the above sources.

The search terms included "Novel coronavirus", "COVID-19", "2019 nCoV", "Novel coronavirus 2019", "acute respiratory infection", "Wuhan pneumonia", "Wuhan coronavirus", "SARS-CoV-2", "Laboratory", "Laboratories", "clinical features", and "Clinical Characteristics".

The following is a sample of the search in PubMed: (Novel coronavirus OR COVID-19 OR 2 0 1 9 nCoV OR Novel coronavirus 2019 OR acute respiratory infection OR COVID-19 OR Wuhan pneumonia OR Wuhan coronavirus OR SARS-CoV-2) AND (Laboratory OR Laboratories OR clinical features OR Clinical Characteristics).

Studies with the following criteria were included: 1) All cross-sectional studies examining laboratory findings and 2) the studies with full test. Studies with the following criteria were excluded: 1) the sampling method was non-random, 2) duplicate studies, 3) lack of relevance to the subject, 4) diagnostic intervention for COVID-19 other than laboratory confirmation, 5) sample size of less than 10 participants, 6) low quality in the qualitative evaluation and 8) case studies, review articles, letter to the editor without quantitative data.

. CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020 . . https://doi.org/10.1101 /2020 Extracted data include: author's name and year of publication, country and province, study source, study design, total sample size, number of men and women, mean age and standard deviation, patient description, number of patients referred to the intensive care unit (ICU), COVID-19 diagnostic method, test sample for COVID-19 diagnosis (respiratory secretions, blood, etc.), sample location (nasal, pharyngeal), and available laboratory findings (including CBC components, inflammatory markers, cellular immunity tests, cytokines, and cardiac, renal, hepatic, muscular, and coagulopathic tests). In this study, only the number of cases that increased or decreased in each test was extracted based on normal range, and the mean and standard deviation of the experiments were not extracted.

If two articles were published by the same authors or by the same institute, one of those studies was selected with a bigger sample size or the information of both articles was used.

Critical evaluation of studies was performed using the Newcastle-Ottawa Scale (NOS) (64).

Studies with six stars or more (based on a scale of 0 to 9) were considered moderate to good quality studies. In terms of study quality, studies with scores 6-7 were considered as average quality and studies with scores 8-9 were considered as good quality.

In the present meta-analysis, the total number and positive cases for each test were used. We tested the statistical heterogeneity between the studies using I 2 and Q statistic; I 2 >50% and P<0.10 were statistically significant (65). In the case of significant heterogeneity, we proposed meta-analysis based on random effects model, because this model can estimate the distribution of true effect size (ES) (66). Sensitivity analysis was also performed by omitting one study at a time and combining the results of the remaining studies. We did the meta-analysis in Comprehensive Meta-Analysis Software (CMA) version 2. The publication bias was assessed using the Begg's test and Egger's test for most variables with the highest number of studies (67, 68). Figure 1 shows the flowchart for the selection of studies. In the electronic search, 2652 related topics were retrieved. Articles were reviewed based on their title and abstract, and 821 duplicate and 1682 unrelated articles were deleted. Ninety-seven studies were removed after a full review . CC-BY-NC-ND 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 June 8, 2020 . . https://doi.org/10.1101 /2020 of the text because they did not meet the criteria for entering the study (Figure 1) . Finally, 52 studies entered the meta-analysis process after qualitative evaluation. All studies were conducted in China and all of them were of good quality (Table 1 ). The average age of the participants was estimated to be 52.09 years (95% CI: 48.80-55.38).

The prevalence of leukocytosis, lymphocytosis, neutrophilia, and monocytosis in patients with COVID-19 was estimated to be 9.9% (95% CI: 7.7-12.7), 4.0% (95% CI: 1.3-11.4), 18.9% (95% CI: 13.1-26.6), and 17.1% (95% CI: 127-22.6), respectively (Figure 2A -D).

The prevalence of leukopenia, lymphopenia, neutropenia and monocytopenia in patients with COVID-19 was estimated to be 20.9% (95% CI: 17.9-24.3), 51.6% (95% CI: 44.0-59.1), 11.0% (95% CI: 8.0-15.1), 1.7% (95% CI: 0.9-3.3), respectively ( Figure 2E -I).

The prevalence of thrombocytosis was 4.7% (95% CI: 2.3-9.1) and thrombocytopenia was 17.2% (95% CI: 13.7-21.5) (SF 2A-B). The prevalence of polycythemia was 26.5% (95% CI: 15.7-41.2) and anemia 3.3% (95% CI: 0.3-27.4) (SF 2 C-D).

The prevalence of elevated c-reactive protein (CRP), elevated erythrocyte sedimentation rate (ESR), elevated TNFα, elevated serum amyloid A, elevated ferritin and elevated procalcitonin (PCT), in patients with COVID-19 was estimated to be 63.6% (95% CI: 57.0-69.8), 

In patients with COVID-19, the prevalence of elevated interleukin-2 was 32.0% (95% CI: 8.0-71.8), elevated interleukin-6 was 59.9% (95% CI: 48.2-70.5) and elevated interleukin-10 was 12.7% (95% CI: 2.9-41.8) ( Figure 3A -C).

The prevalence of reduced CD3, reduced CD4, reduced CD8, reduced CD4/CD8 and elevated CD4/CD8 in patients with COVID-19 was estimated to be 68.3% (95% CI: 50.1-82.2), 62.0% (95% CI: 51.1-71.6) and 42.7% (95% CI: 32.2-53.9), 9.3% (95% CI: 5.6-15.2), and 15.3% (95% CI: 4.6-39.9), respectively (SF3 A-E).

. CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. .

In patients with COVID-19, the prevalence of elevated troponin-I (cTnI) was 20.6% (95% CI: 9.0-40.5), elevated creatine kinase-MB (CKMB) was 14.7% (95% CI: 7.1-28.0), elevated brain natriuretic peptide (BNP) was 48.9% (95% CI: 30.4-67.7) (SF 4).

The prevalence of elevated blood urea nitrogen (BUN), elevated creatinine, hyperkalemia and hypokalemia in patients with COVID-19 was estimated to be 13.1% (95% CI: 6.6-24.4), 7.2% (95% CI: 4.4-11.8), 15.1% (95% CI: 8.1-26.4), and 15.6% (95% CI: 12.8-19.0), respectively (SF 5).

In patients with COVID-19, the prevalence of elevated alanine aminotransferase (ALT) was 20.6% (95% CI: 9.0-40.5), elevated aspartate aminotransferase (AST) was 14.7% (95% CI: 7.1-28.0), elevated lactate dehydrogenase (LDH) was 53.1% (95% CI: 43.6-62.4), hyperglycemia was 41.1% (95% CI: 28.2-55.5), elevated total bilirubin was 48.9% (95% CI: 30.4-67.7), elevated direct bilirubin was 14.7% (95% CI: 7.1-28.0) and elevated globulin was 5.3% (95% CI:

1.8-14.5) (SF 6-7). The prevalence of reduced albumin and reduced pre-albumin in patients with COVID-19 was estimated to be 54.7% (95% CI: 38.1-70.2) and 49.0% (95% CI: 26.6-71.8), respectively (SF 7).

In patients with COVID-19, the prevalence of elevated of myoglobin was 18.3% (95% CI: 9.9-31.5) and elevated creatine phosphokinase (CPK) was 13.7% (95% CI: 11.0-16.9) (SF 8).

The prevalence of elevated D-dimer, elevated prothrombin time (PT), elevated partial thromboplastin time (PTT) in patients with COVID-19 was estimated to be 44.9% (95% CI: 31.0-59.6), 15.2% (95% CI: 5.9-33.7) and 20.6% (95% CI: 11.3-34.6), respectively ( Figure 5 )

In patients with COVID-19, the prevalence of reduced PT was 53.1% (95% CI: 43.6-62.4) and reduced PTT was 4.1% (95% CI: 1.1-14.5) (SF 9 C-D).

Sensitivity analysis for all meta-analyses showed that the overall result is robust even after the elimination of one individual study (SF 10 A-GG).

. CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.07.20124602 doi: medRxiv preprint

Publication bias for performed meta-analyses showed that publication bias had no effect on studies. Begg's test and Egger's test results were as follows: Egger = 0.115 and Begg = 0.840 for lymphocytosis, Egger = 0.307 and Begg = 0.199 for CRP increase, Egger = 0.155 and Begg = 0.220 for procalcitonin increase, and Egger = 0.779 and Begg = 0.186 for AST increase (SF 11).

Our systematic review and meta-analysis included 52 studies and 5490 patients, which provides the most comprehensive overview of the laboratory findings about patients with COVID-19. However, all of these laboratory markers are non-specific, which limits their clinical use. When assessing suspected cases, physicians cannot rely on these laboratory abnormalities to reject or confirm the diagnosis of COVID-19, and they only play a supporting role. These abnormalities are similar to those previously seen in patients with .

In this meta-analysis, changes in lymphocytes predominantly included a decrease in lymphocytes. Pathological results in patients with COVID-19 showed that overactive T lymphocytes, characterized by increased Th17 cells and high toxicity of CD8 + T cells, caused some severe immune damage. This may be the main reason for the disappearance of lymphocytes in these patients (74, 75) .

In the present meta-analysis, a decrease in CD3, CD4, and CD8 cells was observed in most COVID-19 patients. CD4 + T cells play an important role in regulating immune responses, cell removal and strengthening of immune cells, especially CD8 + T cells (76). CD4 + T cells facilitate the production of specific antibodies to eradicate the virus by activating T-dependent B cells. On the other hand, CD8 + T cells exert their effects mainly through two mechanisms, . CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.07.20124602 doi: medRxiv preprint including cytolytic activity against target cells or secretion of cytokines, including IFg, TNFα, and IL-2, as well as many chemokines (77). However, continuous stimulation by the virus may cause T cell exhaustion, leading to loss of cytokine production capacity and decreased function (78, 79) . Other studies have shown that the number of T cells in patients with COVID-19 is significantly reduced, and the surviving T cells appear to be worn out. Studies also showed that in non-ICU patients, complete T cells, CD8 + T cells and CD4 + T cells were less than 800 microliters, 300 microliters and 400 microliters, respectively. More invasive interventions may be required even if severe symptoms are not observed. That's because they are at high risk for worse conditions (80).

Since there is no proven antiviral treatment yet, there may be ways to boost the immune system.

On the other hand, the most common hematological changes in COVID-19 patients are lymphopenia and immune system defects. We assume that hematopoietic growth factors such as G-CSF, which mobilize endogenous blood stem cells and endogenous cytokines, may be a possible hematological treatment for COVID-19 patients (81).

In the present study, inflammatory markers such as CRP, ESR, ferritin, PCT, and serum amyloid A were increased in most patients. CRP is a non-specific acute-phase reactive protein derived from IL-6 in the liver and is a biomarker sensitive to inflammation, infection, and tissue damage In a recent meta-analysis on patients with COVID-19-induced ARDS, physicians were advised to closely monitor the number of WBCs, the number of lymphocytes, the number of platelets, IL-6, and serum ferritin as a sign of potential progression to critical illness. PCT should also be measured regularly to act as an indicator of secondary bacterial infection, which is frequently . CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.07.20124602 doi: medRxiv preprint seen in patients who do not survive. Finally, these findings should be continuously reviewed in the upcoming months (59).

In the present meta-analysis, abnormalities in liver tests such as AST, ALT, LDH, bilirubin, globulin, albumin and pre-albumin were observed in COVID-19 patients. The liver has bile receptors, and these abnormalities can be the result of bile damage (33, 90). This could be an explanation for the liver's abnormal laboratory findings in the early stages of infection. Hypoxia is a serious event in COVID-19 and is one of the leading causes of sudden death in patients, so an increase in CPK may be the result of hypoxia and should be causally interpreted (59).

The similarity between DNA sequencing of the COVID-19 and SARS-COV suggests that the mechanism of action may be similar (91) In one meta-analysis, cTnI analysis in four studies showed that the levels of cTnI in patients with severe SARS-CoV-2 infection increased compared to those with milder forms of the disease. It is therefore reasonable to assume that the initial measurement of cardiac injury markers in SARS-CoV-2 patients immediately after hospitalization, as well as long-term monitoring during hospitalization, may help identify patients with possible cardiac injury. This way, it is possible to predict the progression of COVID-19 towards worse clinical conditions (97).

In the present study, elevated levels of creatinine, hyperkalemia and hypokalemia were observed in patients with COVID-19. The exact mechanism of kidney involvement is unknown. The hypothetical mechanism includes sepsis, which leads to cytokine storm syndrome or direct cell damage caused by the virus. Angiotensin converting enzyme (ACE) and dipeptidyl peptidase IV (DPP-IV) have both been expressed in renal tubular epithelial cells, and have been identified as a connecting partner for SARS-CoV and MERS-CoV, respectively (98). Viral RNA has been identified in kidney tissue and urine in both infections (72). Recently, Zhong's lab in Guangzhou . CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020 . . https://doi.org/10.1101 /2020 successfully isolated SARS-CoV-2 from the urine sample of an infected patient and identified the kidney as the target of this novel virus. Previous reports of SARS and MERS-CoV infections have shown acute kidney injury (AKI) in 5% to 15% of cases and high mortality rates (60-90%).

Preliminary reports indicate a lower prevalence (9%) of AKI in people with COVID-19 infection (1, 17, 55) . AKI was an independent risk factor for hospital mortality (99).

Recent literature shows that D-dimer values are often increased in patients with  which was reported to be 44.9% in the present meta-analysis. What is clear is that D-dimer values are severe in patients with COVID-19, even in patients with milder forms, and therefore, D-dimer measurement may be associated with a worsening clinical picture. Although an increase in D-dimer is known to be a multifactorial cause, Lippi et al. stated that an increase in D-dimer and disseminated coagulopathy might be a trivial finding in patients with severe COVID-19 forms compared to other infectious diseases such as human immunodeficiency virus (HIV), Ebola, Zica virus, and Chikungunya virus. Therefore, immediate studies should be planned to determine whether antithrombotic adjunctive treatments (e.g., anticoagulants, antithrombin, or thrombomodulin) may be helpful in patients with severe COVID-19 (100).

In fact, it has been reported that coagulation is activated and accelerated in response to several infections because this mechanism may enhance the physiological response (101-103).

Coagulation also has immune function, which can be another line of defense against severe infection (104). Although it is reasonable to assume that homeostasis is deviated in patients with COVID-19, overuse of coagulation factors increases the risk of disseminated intravascular coagulation (DIC), which has a clear negative effect on prognosis (105). A study by Han et al.

found that coagulation in COVID-19 patients was clearly deranged compared to a healthy control population (106).

The limitations of this study include the following: 1) All studies were conducted in China and laboratory manifestations may be influenced by ethnic factors. 2) Most patients are hospitalized and patients who have milder symptoms or patients who are not hospitalized may cause bias in the results. 3) In most early studies, laboratory findings were not considered separately for patients admitted to the intensive care unit or the isolated ward. Our results are based on laboratory findings at the time of admission, but patients may have experienced symptoms before admission (because laboratory findings are influenced by the clinical course of the disease), and during this time, the patient may receive antiviral or antibacterial drugs. 4) Since all studies were . CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.07.20124602 doi: medRxiv preprint performed in China in 2020 using the similar diagnostic method, we were unable to determine the cause of the heterogeneity.

This study provides a comprehensive description of laboratory characteristics in patients with COVID-19. The results show that lymphopenia, elevated CRP, elevated ESR, elevated ferritin, elevated serum amyloid A, elevated BNP, reduced albumin, reduced pre-albumin, reduced CD3, reduced CD4, reduced CD8, elevated D-dimer, reduced PT, elevated interleukin-2, elevated interleukin-6, elevated LDH and hyperglycemia are the common findings at the time of admission.

Coronavirus disease . CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. 

Not applicable.

Not applicable.

We declare no competing interests.

This study was funded by the Ilam University of Medical sciences. Funder role was only financial support.

MA, ARJ, GhB, and MK acquired the data. MA analyzed and interpreted the data. MA, ARJ, GhB, and MK drafted the manuscript; MA, ARJ, GhB, and MK critically revised the manuscript for important intellectual content. MK supervised the study. All authors have read and approved the manuscript.

We sincerely thank Ilam University of Medical Sciences for helping us with this research.

. CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.07.20124602 doi: medRxiv preprint Team EE. Note from the editors: World Health Organization declares novel coronavirus (2019-nCoV) sixth public health emergency of international concern. Eurosurveillance. 2020;25 (5) Cui P, Chen Z, Wang T, Dai J, Zhang J, Ding T, et al. Clinical features and sexual transmission potential of SARS-CoV-2 infected female patients: a descriptive study in Wuhan, China. medRxiv. 2020. . CC-BY-NC-ND 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 June 8, 2020. . 

The prevalence of thrombocytosis (A), thrombocytopenia (B), polycythemia (C), and anemia (D) in patients with COVID-19

The prevalence of reduced CD3 (A), reduced CD4 (B), reduced CD8 (C)

SF 4: The prevalence of elevated troponin-I (A), elevated Creatine kinase-MB (B), elevated brain natriuretic peptide (C) in patients with COVID-19

The prevalence of elevated blood urea nitrogen (A), elevated creatinine (B), hyperkalemia (C) and hypokalemia (D) in patients with COVID-19

elevated aspartate aminotransferase (B), elevated lactate dehydrogenase (C), hyperglycemia (D), elevated total bilirubin (E), elevated direct bilirubin (F) and elevated globulin (G) in patients with COVID-19

The prevalence of elevated of myoglobin (A) and elevated creatine phosphokinase (B) in patients with COVID-19

elevated partial thromboplastin time (PTT) (B), reduced PT (C) and reduced PTT (D) in patients with COVID-19

 in the Fever Clinic of a teaching hospital in Beijing: a single-center, retrospective study . medRxiv. 2020. Huang Y, Zhou H, Yang R, Xu Y, Feng X, Gong P. Clinical characteristics of 36 nonsurvivors with COVID-19 in Wuhan, China. medRxiv. 2020. Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics. 1994:1088-101.COVID-19: A systematic review and meta-analysis. Travel medicine and infectious disease. 2020:101623.Cao Y, Liu X, Xiong L, Cai K. Imaging and clinical features of patients with 2 0 1 9 novel coronavirus SARS-CoV-2: a systematic review and meta-analysis. Journal of medical virology. 2020. Ling W. C-reactive protein levels in the early stage of COVID-19. Medecine et Maladies Infectieuses. 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.07.20124602 doi: medRxiv preprint . CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.07.20124602 doi: medRxiv preprint . CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.07.20124602 doi: medRxiv preprint . CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.07.20124602 doi: medRxiv preprint . CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.07.20124602 doi: medRxiv preprint . CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.07.20124602 doi: medRxiv preprint . CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.07.20124602 doi: medRxiv preprint . CC-BY-NC-ND 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.

The copyright holder for this preprint this version posted June 8, 2020. . https://doi.org/10.1101/2020.06.07.20124602 doi: medRxiv preprint