A comparison of clinical and laboratory manifestations of thrombotic events in patients with COVID-19 and other respiratory viral infections

Chuen Wen Tan (  tan.chuen.wen@sgh.com.sg ) Singapore General Hospital Jing Yuan Tan Singapore General Hospital Wan Hui Wong Singapore General Hospital May Anne Cheong Singapore General Hospital Ian Matthias Ng Singapore General Hospital Edwin Philip Conceicao Singapore General Hospital Guek Hong Jenny Low Singapore General Hospital Heng Joo Ng Singapore General Hospital Lai Heng Lee Singapore General Hospital


Introduction
Towards the end of April 2020, investigators were reporting the association of Coronavirus Disease-19 (COVID- 19) with increased incidence of thrombotic events. Initially con ned to intensive care units (ICU) and critically ill COVID-19 patients, similar ndings were subsequently also noticed in non-critically ill patients 1,2 . Emerging autopsy-derived histopathological evidence have con rmed the presence of pulmonary vasculature microthrombosis in deceased patients 3,4 . Thrombotic complication is however not unique to COVID-19. Other non-CoV-2 respiratory viral (non-CoV-2-RV) infections have also been associated with both arterial 5,6 and venous 7 thrombosis.
The thrombotic manifestations in COVID-19 patients have led to both the American Society of Hematology (ASH) 8 and International Society on Thrombosis and Hemostasis (ISTH) 9 recommending that all hospitalized patients including the non-critically ill, receive venous thromboembolism (VTE) thromboprophylaxis with low molecular weight heparin (LMWH) or fondaparinux. It is uncertain if such recommendations should be universally adopted as there is signi cant heterogeneity in reported thrombotic rates as most studies were carried out on critically ill COVID-19 patients 10 . Additionally, baseline VTE risk is known to differ between geographical regions, race and ethnicity 11,12 . As the absolute risk of bleeding increases by around 0.5% with thromboprophylaxis 13 , stratifying COVID-19 patients with high risk of VTE for thromboprophylaxis is crucial to optimize the risk bene t ratio. Measurable laboratory parameters such as elevated D-dimer and brinogen levels have been reported to be associated with thrombosis in COVID-19. Less accessible global hemostatic assays such as thromboelastogram and clot waveform analysis (CWA) have also demonstrated hypercoagulability, albeit in small groups of critically ill COVID-19 patients. 14,15 In this study, we aimed to determine the thrombotic rates in consecutive COVID-19 patients compared to patients with non-CoV-2-RV infections and to evaluate the dynamic haemostatic changes in both groups of patients using global coagulation assays.

Methods
This single-center study was conducted at Singapore General Hospital, a 1700-bedded tertiary academic center.
For our rst objective, we performed a retrospective cohort study to determine thrombotic event rates in all consecutive COVID-19 patients compared to patients diagnosed with non-CoV-2-RV from January 17 th to April 10 th 2020. During this period, all patients who ful lled our hospital's COVID-19 suspect criteria were admitted regardless of disease severity and tested for both COVID-19 and other respiratory viral pathogens 16 . COVID-19 testing was performed with reverse transcriptase polymerase chain reaction (rt-PCR) on RNA extracted from either oropharyngeal and/or nasopharyngeal swabs specimens. Testing for other respiratory viral pathogens was performed using a 16-target respiratory viral PCR assay (Respiratory syncytial virus A/B, In uenza A/B, Parain uenza viruses 1-4, Metapneumovirus, Rhinovirus A/B/C, Human Coronavirus OC43/229E/NL63, Adenovirus, Human Enterovirus, Human Bocavirus 1-4) on oropharyngeal specimens. Exemption of consent was granted by our institutional review board as this study only involved collection and analysis of routine clinical data (CIRB no. 2020/2535). Study data was obtained from our Hospital's Department of Infection Prevention and Epidemiology database. This database consists of de-identi ed data collected from all con rmed and suspect COVD-19 cases admitted to our hospital during this pandemic period. Data on demographics, co-morbidities, laboratory coagulation pro le parameters, and anti-platelet/anticoagulation medications were obtained. Information on modality of thromboprophylaxis (mechanical vs routine chemical) initiated, thrombotic events during hospitalization, the need for high dependency unit (HDU) and intensive care unit care (ICU), length of stay in ICU and length of hospitalization was also obtained. Need for oxygen supplementation during hospital admission was also collected and was used as a surrogate measure for disease severity in our analysis. We included both arterial and venous thrombotic events. Venous thrombotic events were de ned as any venous thromboembolism including deep vein thrombosis, pulmonary embolism and thrombosis of other sites, which were objectively con rmed on radiological imaging after initial clinical suspicion by attending physicians. Arterial events were de ned as either myocardial infarction or stroke.
For our second objective, laboratory coagulation tests data were pooled from two sources: a) both COVID-19 and non-CoV-2-RV patients as studied above and b) COVID-19 patients from our hospital's infectious disease department novel pathogen study database (CIRB no. 2018/3045) censored at 1 st June 2020. This database is part of a prospective study to characterize emerging infectious diseases and was approved by our institutional review board (CIRB no. 2018/3045). We collected the following coagulation data: prothrombin time (PT), activated partial thromboplastin time (aPTT) and its associated clot waveform analysis (CWA), brinogen and D-dimer.
Coagulation tests taken when patients were on anticoagulants or on citrate-based haemodialysis and patients with active malignancy or pregnancy were excluded from any inter-individual comparison and analysis. APTT-based CWA were generated during the analysis of the standard aPTT assay triggered with Dade Actin FSL reagent (Siemens Healthcare, Marburg, Germany) and retrieved from CS2100i automated coagulation analysers (Sysmex Corporation, Kobe, Japan). The three CWA parameters of interest were min1, min2 and max2 denoting the maximum velocity, maximum acceleration and maximum deceleration of the clot formation process, respectively.

Statistical analysis
Categorical variables were presented as frequencies (percentages) and continuous variable was presented as median (interquartile range) due to its skewed distribution. Univariate analysis was performed using chi square test for categorical variables and Mann Whitney U test for continuous variables. Correlation analysis was performed using Spearman's rank correlation coe cient (R S ). All tests were two sided, with p-values of P<0.05 considered statistically signi cant. Data analysis was performed using SPSS 25.0 (IBM SPSS statistics, IBM Corporation, USA) software.

Results
During the study period, a total of 181 patients testing positive for SARS-CoV-2 and 165 patients positive for non-CoV-2-RV via rt-PCR testing. The respiratory viruses were rhinovirus (n=65), in uenza A and B (n=46), adenovirus (n=13), human coronavirus 229E/NL63/OC43 (n=15), human enterovirus (n=3), metapneumovirus (n=6), parain uenza virus 1 to 4 (n=11) and respiratory syncytial virus (n=6). Among the 186 COVID-19 patients, four patients were co-infected with other respiratory viruses and were excluded from analysis. Age and comorbidities were not signi cantly different between the two groups (Table 1). Compared to the non-CoV-2-RV arm, there were signi cantly higher proportion of male patients and patients of Indian ethnicity in the COVID-19 arm. Majority (94%) of our COVID-19 patients were managed in the general isolation wards and had relatively mild infection as re ected by low proportions requiring any form of oxygen supplementation (11%). COVID-19 patients had longer length of hospitalization and more required ICU support. Mortality rate was low in both groups. The two deaths in our COVID-19 cohort were due to severe COVID-19 pneumonia with multi-organ failure.

Thrombotic outcomes
In both groups, all general ward patients were not given routine thromboprophylaxis. There was no standardized hospital practice with regards to pneumatic calf pumps for general ward patients and thus such data was unavailable. Patients in HDU were only on mechanical thromboprophylaixs (TED stockings). In contrast, patients in ICU were all started on prophylactic subcutaneous enoxaparin 40mg once a day (or 20mg once a day for patients with renal failure) unless contraindicated, together with pneumatic calf pumps. Amongst our COVID-19 ICU patients, two (22.2%) out of nine patients were not on prophylactic thromboprophylaxis because of thrombocytopenia and anemia respectively. There were two arterial thrombotic events in the COVID-19 group, both of which occurred while patients were in ICU. This corresponded to a 22.2% incidence amongst COVID-19 ICU patients and 1.1% incidence amongst all COVID-19 patients. On the other hand, there was only one arterial thrombotic event (0.6% incidence) in the non-CoV-2-RV group and this occurred while the patient was in the general ward. The thrombotic rates in COVID-19 and non-CoV-2-RV patients adjusted for the duration of hospitalization translated to 1 per 1000 patient-days and 1.8 per 1000 patient-days respectively. There were no VTE events in both groups. Only one patient had disseminated intravascular coagulopathy (DIC) by ISTH criteria 17  Both cases of arterial thrombosis in the COVID-19 group were non-ST elevation myocardial infarction (NSTEMI). Both occurred early in the course of disease, at day 9 and day 12 of symptom onset respectively. The arterial thrombosis in our non-CoV-2-RV cohort was also a NSTEMI, developing on day 4 of admission. All of them were on prophylactic enoxaparin at the time of thrombosis. There was only one major bleeding event due to intracranial haemorrhage which occurred in the ICU in our entire cohort.

Hemostatic assays
The coagulation pro les and CWA parameters were analysed. 42 COVID-19 and 5 non-CoV-2-RV patients had only a single data point. 14 COVID-19 patients had data available on two or more time points while only two of the non-CoV2-RV patients had serial coagulation pro le data, both of whom were critically ill.
No biphasic waveform was noted in all the clot waveform curves analysed. The key ndings are summarized here: a) Coagulation functions remained fairly stable in mild COVID-19 and not signi cantly different from baseline (convalescent state) within the same individuals (Table 2) In COVID-19 patients who remained clinically well without a need for supplemental oxygen, intraindividual paired comparison of PT, aPTT and CWA data across various time points were not signi cantly different throughout the course of the disease. Paired comparison between the initial and convalescent coagulation tests also did not show remarkable alteration.
In addition, inter-individual comparison between the presenting coagulation assays taken from patients with mild COVID-19 and coagulation assays taken at convalescent phase also supported this observation -although aPTT was shorter in convalescent phase, CWA pro les and PT were not signi cantly different between the two groups (Supplementary Table 1). b) CWA parameters increased with clinical deterioration of COVID-19 (Table 3 and Figure 1) Comparing the PT, aPTT and CWA data between COVID-19 patients with clinically mild disease and COVID-19 patients who required supplementary oxygen support, a surrogate of more severe disease, aPTT was signi cantly more prolonged and CWA parameters, especially min1, were signi cantly elevated in patients with severe disease. PT and D-dimer results did not differ signi cantly between the two groups.
Serial PT, aPTT and CWA data depicting the various stages, from initial presentation through clinical deterioration to recovery and convalescence were available in one patient and the dynamic changes are shown in Figure 1. Initial clinical deterioration was associated with prolongation of aPTT but aPTT normalised during the subsequent course of ICU stay and remained so at convalescent phase of COVID-19. However, min1 showed marked elevation upon ICU admission and continued to rise throughout the ICU stay despite normalisation of aPTT. Min1 only started to decrease at convalescence and eventual normalised on the 46 th day from symptoms onset. Min2 and max2 demonstrated similar pattern of changes to min1 although the timing of their increment to the levels beyond the respective normal ranges was delayed compared to min1. This patient's PT levels remained normal throughout. c) Critically ill COVID-19 patients CWA showed a trend of having higher CWA parameters as compared to critically ill patients with non-CoV-2-RV. (Table 4) Patients with mild COVID-19 and those with non-CoV-2-RV had similar PT, aPTT and CWA pro les. In contrast, for patients who required ICU admission, critically ill COVID-19 patients CWA showed a trend of having higher CWA parameters compared to similar patients with non-CoV-2-RV, although their PT, aPTT and platelet counts were not different.

d) CWA parameters correlated with D-dimer and brinogen
Based on ISTH recommendations, D-dimer of six folds the upper limit of normal in COVID 19 patients was a predictor of thrombotic events and thus may be used as a surrogate marker for consideration of higher dose of thromboprophylaxis 9 . Eight COVID-19 patients had D-dimer levels of greater than a cut-off of 3.00mg/L FEU (approximately six folds of the upper normal limit). Of these, six patients had severe disease requiring oxygen support, one had mild disease and one was in convalescent phase of COVID-19. Of the two COVID-19 cases with thrombotic complications, only one had a high D-dimer of 4.33mg/L FEU while another had a level of 1.22mg/L FEU at the time when the thrombotic events occurred.
Amongst COVID-19 patients, 35 sets of data were available for correlation evaluation with D-dimer (Supplementary Table 2

Discussion
The latest CHEST guidelines on VTE and its management in COVID -19 patients have provided more mature data on VTE rates 10 . Overall, venous thrombotic events are estimated to be around 8% in all COVID-19 patients and as high as 69% in critically ill patients. Other respiratory viral infections such as in uenza are also known to be associated with thrombotic events 7,19 with a 5.9% prevalence reported during the 2009 in uenza pandemic 20 . Presently, indirect evidence and observations seem to suggest a higher thrombotic rate amongst COVID-19 patients compared to patients infected with other respiratory viruses 21 . However, direct comparison of thrombotic events between concurrent cohorts of COVID-19 and non-CoV-2-RV patients is lacking.
In contrast to other recent reports 2,22,23 , our ndings did not show increased VTE events in either groups. This could be accounted by a number of factors peculiar to our study population. At the time of study, there was a large outbreak involving migrant workers residing in dormitories and all COVID-19 cases regardless of the severity were required to be hospitalized according to national policy and hence the majority of our patients were younger with less comorbidities and had a lower rate of ICU stay. Older age, presence of cardiovascular risk factors and ICU stays are all independent risk factors for VTE. In ICU patients with COVID-19, other studies have reported up to 69% VTE incidence 24,25 . The stark difference in our population may also suggests the in uence of other factors such as racial predilection. Whilst local data show that the incidences of VTE are comparable amongst the major ethnic groups of Chinese, Indian and Malay 26 , Asians in general are purported to have lower risk of VTE compared to Caucasians [27][28][29] although the actual reasons for this difference are still unclear. In our center, VTE screening was not done as a routine for asymptomatic cases. Con rmatory imaging for VTE would only be performed for symptomatic or clinically suspicious cases. As such our VTE detection rates may be lower compared to centers that practice routine screening of their patients. Albeit the variations in threshold or protocols for detecting VTE in the various studies, it is compelling that our thrombotic rates were clearly much lower compared to other centers.
COVID-19 and other acute respiratory viruses have both been reported to be associated with higher risk of arterial thrombotic events as well. Acute coronary syndrome is reported in no greater than 3% of acute in uenza or other viral infections 5, 30 whilst in COVID-19, the rates vary from 3.4% to 11.2% 21,31 . We, however, reported low arterial event rates in our patient cohorts and the potential reasons accounting for these differences are likely to be similar to those discussed above for VTE. In our cohort, the arterial events occurred only in ICU when the patients were sickest while no ischemic events were reported in our patients from the general ward. Nonetheless, early recognition of cardiac complications in both COVID- 19 and other viral infections cannot be over-emphasized to reduce morbidity and mortality.
We also evaluated the coagulation pro les of our COVID-19 patients across the different spectrum of disease severity. Commensurate with the absence of thrombotic events observed, none of our mild COVID-19 cases demonstrated evidence of DIC or SIC and their CWA parameters were not elevated. Marked elevation of D-dimer, de ned as six-times upper limit of normal, was suggested by ISTH as a possible consideration for intensi cation of prophylactic pharmacological anticoagulation in COVID-19 9 . However, this cut-off was neither sensitive nor speci c for thrombotic events in our cohort as none of our patients with markedly raised D-dimers developed venous thromboembolic event and only one of them suffered an arterial event.
To further evaluate the thrombotic potential in COVID-19, we analysed available CWA parameters as a global haemostatic assessment. It has been reported that CWA parameters differ in various types of infections 32 and increased CWA is associated with hypercoagulability 33 . Our ndings suggest that there were no signi cant changes in the overall haemostatic functions in mild COVID-19 but severe COVID-19 was associated with a prothrombotic state and this hypercoagulability gradually normalizes during the convalescent phase. Although CWA parameters correlated with D-dimer and brinogen, the strengths of association were weak and moderate respectively and this suggests that the haemostatic functions measured by these assays do not replicate each other. Hence, composite data from these multiple assays might provide more insight into the overall haemostatic dysfunctions of COVID-19 patients. These assays may therefore complement each other to better risk stratify the thrombotic risk of COVID-19 patients which may guide thromboprophylaxis management. Nonetheless, the coagulation assays in this study only evaluated the plasmatic components of haemostasis. Endothelial functions and cellular components, including platelets and leucocytes, and brinolytic functions are not assessed but may contribute signi cantly to coagulation changes in COVID-19 34,35 .
ISTH has recommended for a universal strategy of routine thromboprophylaxis for all hospitalized COVID-19 patients with standard dose LMWH or unfractionated heparin after assessment of bleeding risk. While existing literature do suggest a signi cantly high rate of thrombotic events in COVID-19 patients, and understandably warrants the recommendation for thromboprophylaxis, our cohort study has not proven the case in our population. As routine thromboprophylaxis is not without risks, the adoption of this strategy for all hospitalised COVID-19 patients may not be universally suitable. Adjustments to these recommendations should be made taking into account the local population pro le and prevalence of thrombotic events. Irrefutably, analysis of hemostatic parameters showed a hypercoagulable state in patients with more severe COVID-19 and thus supporting the use of thromboprophylaxis in critically ill COVID-19 patients.
Although our study has several limitations due to its retrospective nature and the number of thrombotic events in both groups were small thus limiting statistical comparison, we believe our comparison data are su ciently robust as we included all consecutive patients with the same admitting criteria within the same study period and the demographics between both groups were comparable.

Conclusion
Page 10/20 The thrombotic rates in both hospitalised COVID-19 and non-CoV-2-RV patients were low and did not seem to differ signi cantly between the two groups. A more individualized thrombotic risk assessment and management approach, possibly involving the combination of various clinical factors and coagulation markers, is an area of unmet needs and warrants further research works.

Declarations Acknowledgments
We acknowledge the extraordinary work of Singapore General Hospital clinical teams who work tiredlessly in our emergency department, isolation wards and acute respiratory infection wards. Our study could not have been done without their meticulous efforts in history taking and clinical documentation. We would also like to acknowledge Singapore General Hospital Department of Infection Prevention and Epidemiology in providing the clinical data for the manuscript.

Authorship Contributions
CWT and JYT had full access to all the data in the study and takes responsibility for the integrity of the data and accuracy of data analysis. CWT and JYT contributed to statistical analysis, data interpretation and drafting of manuscript. CWT, JYT, LHL, HJN, JGL conceived the study. CWT, JYT, IMN, EPC contributed to acquisition data. LHL and HJN contributed to critical revision of manuscript. All authors agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.  * No D-dimer and brinogen data presented as not more than 2 pairs of these results were available.
Mild COVID-19 de ned as patients who did not require supplementary oxygen support throughout the course of infection.  Table 4: Inter-individual comparison of coagulation data (PT, aPTT and aPTT-based clot waveform analysis pro les) between mild COVID-19 and non-CoV-2 respiratory viral infections^, as well as between critically ill COVID-19 and non-CoV-2 respiratory viral infections # . # Critically ill COVID-19 and non-CoV-2 respiratory viral infections de ned as patients who required intensive care unit support Figures Figure 1 Dynamic changes of the serial haemostatic pro les of a COVID-19 patient from initial presentation to clinical deterioration* to subsequent recovery and convalescent phase.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.