Time series differences in coagulopathy in mechanically ventilated COVID-19 and bacterial pneumonia patients: a nationwide observational study in Japan

Abstract

Background

See More

Severe acute respiratory syndrome coronavirus 2 infection causes systemic immune overresponse (cytokine storm), which can lead to microthrombi and dysfunction of coagulation such as disseminated intravascular coagulation (DIC) of sepsis. Coronavirus disease 2019 (COVID-19) coagulopathy is known to occur mainly in the pulmonary microcirculation. We aimed to investigate hematological differences in coagulopathy between COVID-19 pneumonia and bacterial pneumonia.

Methods

We performed an observational cohort study using the Japanese REsearch of COVID-19 by assEmbling Real-world data (J-RECOVER) study database for COVID-19 patients and the Japan Medical Data Center (JMDC) database for bacterial pneumonia patients. The J-RECOVER database includes data from patients discharged between January 1 and September 31, 2020. The JMDC database covers patients emergently hospitalized from 2014 to 2022. We analyzed the association between hematological coagulopathy, systematic inflammation, and organ dysfunction in both groups after one-to-one propensity score matching.

Results

We enrolled 572 COVID-19 patients and 2,413 bacterial pneumonia patients who required mechanical ventilation. The COVID-19 group was younger, had higher intensive care unit admission rates, and lower mortality in comparison to the bacterial group (p < 0.05). On day 1, the two groups showed no significant differences in JAAM-2 and sepsis-induced coagulopathy criteria. After matching, platelet counts, antithrombin activity, and prothrombin time-international normalized ratio were consistently maintained within normal ranges in the COVID-19 group. However, trends in D-dimer and fibrin degradation products in the COVID-19 group were similar to those in the bacterial pneumonia group.

Conclusions

COVID-19 coagulopathy differs from bacterial septic DIC by exhibiting lower platelet consumption and minimal vascular hyperpermeability. Consequently, management strategies for COVID-19 coagulopathy should be distinct from those for septic DIC.

Introduction

Coronavirus disease 2019 (COVID-19) is an infectious disease caused by the novel severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), first identified in late 2019 [1]. Since then, COVID-19 has spread worldwide [2]. The disease primarily affects the respiratory system and often requires mechanical ventilation. A key feature of severe COVID-19 is the frequent occurrence of coagulation abnormalities [3,4,5], which have been associated with increased mortality [5,6,7]. Some studies suggested that COVID-19 coagulopathy differs from coagulopathy caused by other conditions, such as bacterial infections [8, 9]. Notably, SARS-CoV-2 infection affects endothelial cells, particularly impairing the pulmonary alveolar endothelium [3, 10]. This leads to a host immune response characterized by the release of inflammatory cytokines and the upregulation of adhesion molecules, which promote immune cell adherence, extravasation, and endothelial dysfunction. This transition from an anticoagulant to a prothrombotic and proinflammatory state contributes to immune and hemostatic dysregulation [10]. The resulting endothelial dysfunction and platelet activation may play a crucial role in respiratory dysfunction and subsequent mortality in COVID-19 [11].

In bacterial sepsis-induced coagulopathy, plasminogen activator inhibitor-1 (PAI-1), a key inhibitor of fibrinolysis, is commonly elevated. However, this pattern of disseminated intravascular coagulation (DIC) with fibrinolysis shutdown is not consistently observed in COVID-19 patients [12,13,14]. Additionally, the fibrinolytic phenotype in COVID-19 showed variability, with the reported incidence of hypofibrinolysis at 1.3% and hyperfibrinolysis at 0.8% (much lower than in typical septic coagulopathy) [15]. In severe COVID-19 cases, hyperfibrinolysis, characterized by elevated D-dimer and fibrin/fibrinogen degradation products (FDP), has been observed [16, 17].

Numerous studies have explored the characteristics of COVID-19-associated coagulopathy and highlighted its differences from septic coagulopathy, emphasizing unique patterns such as elevated fibrinogen and D-dimer levels alongside relatively stable platelet counts and antithrombin activity [8, 18]. However, many of these studies are limited by small sample sizes, single-center designs, or a lack of longitudinal analyses. Furthermore, the temporal changes in coagulation and fibrinolysis markers between COVID-19 pneumonia and non-COVID bacterial pneumonia remain insufficient in the context of large-scale, multicenter datasets.

This study aimed to clarify the temporal differences in coagulation and fibrinolysis markers between COVID-19 pneumonia and bacterial pneumonia in patients requiring invasive mechanical ventilation using a comprehensive multicenter database. These findings may contribute to improved clinical outcomes through targeted therapeutic strategies for managing coagulopathy in COVID-19 patients.

Materials and methods

Study design and setting

In this multicenter observational study, we analyzed data obtained from two databases; the Japanese multicenter REsearch of COVID-19 by assEmbling Real-world data (J-RECOVER) study database for COVID-19 patients and the Japan Medical Data Center (JMDC) database for bacterial pneumonia patients [19]. The J-RECOVER registry consists of data on patients discharged between January 1 and September 31, 2020, with positive SARS-CoV-2 test results, regardless of intensive care unit (ICU) admission status [20]. These data were collected from the Diagnosis Procedure Combination (DPC) and electronic medical records of 64 hospitals in Japan. The JMDC database consists of data on approximately 2.77 million patients discharged between April 2014 and April 2023, and comprises 95 hospitals with laboratory data. In both databases, diagnoses are recorded using International Classification of Disease Tenth Revision (ICD-10) codes.

This study followed the principles of the Declaration of Helsinki, and the study protocol was approved by the Institutional Review Board of Osaka Medical and Pharmaceutical University (approval numbers: 2023–112 and 2020–127-1). Because of the anonymous nature of the data, the requirement for informed consent from patients was waived.

Participants

We included adult COVID-19 pneumonia and bacterial pneumonia patients who required invasive mechanical ventilation. We collected COVID-19 pneumonia patients from the J-RECOVER registry and bacterial pneumonia patients from the JMDC database. Proven/suspected pneumonia was defined as having any of the bacterial respiratory tract infection-related ICD-10 codes previously proposed by the Institute for Health Metrics and Evaluation [21] in the primary diagnosis or the diagnosis that triggered hospitalization. We excluded patients who met the following criteria: age < 18 years, discharge within 24 h of admission, and pregnancy.

Data collection

We collected the following data: age, sex, height, weight, comorbidities, laboratory test results, Charlson Comorbidity Index (CCI) [22], the primary diagnosis on admission, ICU admission, length of hospital stay, treatment information related to oxygen therapy, anticoagulant therapy, complication events, and mortality at discharge. We used predefined ICD-10 coding algorithms to evaluate the CCI [23]. On the basis of the ICD-10 coding, we considered a recorded diagnosis of congestive heart failure, diabetic nephropathy, chronic kidney disease, end-stage renal disease, chronic pulmonary disease, or cerebrovascular disease as complicating diseases. We identified the following mechanical support therapies from claims data: mechanical ventilation, extracorporeal membrane oxygenation, and hemodialysis (including continuous hemodiafiltration). The following drugs were identified for anticoagulant therapy: unfractionated heparin, low molecular weight heparin, antithrombin, and thrombomodulin. Details on the definitions of the ICD-10 codes used in this study are provided in Supplementary Table 1.

Laboratory tests consisted of several hemostatic biomarkers, including platelet count, prothrombin time-international normalized ratio (PT-INR), fibrinogen level, FDP, D-dimer level, and antithrombin activity. In both groups, all biomarkers were collected within 1 week after admission. The incidence of DIC was evaluated at each time point from day 1 to day 7 using existing data. Missing data were excluded. The diagnosis of DIC was based on the Japanese Association for Acute Medicine 2 (JAAM-2) DIC criteria [24] and the sepsis-induced coagulopathy criteria [25].

Statistical analysis

Due to the retrospective nature of the study, it was assumed that an imbalance existed among the covariates in the COVID-19 and bacterial pneumonia groups. We conducted propensity score matching between the groups. Propensity scores were calculated based on 12 variables: patient background characteristics, anticoagulant therapy, and concomitant therapy (Supplementary Table 2). A one-to-one nearest-neighbor matching approach without replacement was applied, using a caliper width of 0.2. Standardized mean difference was calculated to assess the balance between the groups before and after matching, and values < 0.1 we considered to be acceptable. The aim of this study was to elucidate the specific pattern of coagulopathy induced by severe COVID-19 pneumonia by chronologically comparing the hemostatic parameters with those in patients requiring mechanical ventilation due to bacterial pneumonia. Therefore, to examine how the hemostatic parameters changed over time, we fitted multilevel mixed-effects regression models. These models included fixed effects for patient category (COVID-19 or bacterial pneumonia), time points (day 1 to day 7), and their interaction, as well as random effects for individual identification numbers. We additionally conducted multilevel mixed-effects regression analysis to compare time series variations in various parameters during the first seven days after admission, including serum creatinine level, serum bilirubin level, C-reactive protein (CRP), albumin, and white blood cell (WBC) components between the COVID-19 and bacterial pneumonia groups.

In the primary analysis, missing values were not imputed. As a sensitivity analysis, we performed multiple imputations using the mice package in R, generating imputed datasets with predictive mean matching. The same multilevel mixed-effects regression analyses were conducted on each imputed dataset, and the results were averaged across imputations. We also performed subgroup analyses using multilevel mixed-effects regression models to compare temporal changes in coagulation markers between survivors and non-survivors within each disease group (COVID-19 and bacterial pneumonia).

Descriptive statistics were calculated as medians (interquartile range) or proportions (numbers), as appropriate. Univariate differences between groups were assessed using the Mann–Whitney U test or chi-square test, as appropriate. All statistical inferences were two-sided, and p-values of < 0.05 were considered to indicate statistical significance. Because of the underpowered nature of the interaction analysis, we used a two-sided significance level of 20% with statistical inferences for the interaction analyses [26]. Data were analyzed using R (ver. 4.2.3).

Results

Study population

We identified 4,700 patients with COVID-19 pneumonia and 21,764 patients with bacterial pneumonia. Of these, 574 and 2,574, respectively, met the inclusion criteria for invasive mechanical ventilation. In the final analysis dataset, 572 COVID-19 patients and 2,413 bacterial pneumonia patients were included. After matching, 267 propensity score-matched pairs were generated (Fig. 1). Table 1 shows the baseline characteristics, Table 2 shows the laboratory test results on admission, and Table 3 shows the anticoagulant therapies and complication events in this study. Before propensity score matching, the COVID-19 group was younger, had a higher body mass index, a lower mortality risk, and lower rates for several comorbidities excluding diabetes mellitus. Specifically, in the bacterial pneumonia patients, PT-INR values on day 1 were significantly higher, and platelet count, fibrinogen level, and antithrombin activity were significantly lower in comparison to the COVID-19 patients (Table 2). As factors related to coagulation therapy and events, unfractionated heparin was used more frequently and deep vein thrombosis was detected more frequently in the COVID-19 group in comparison to the bacterial pneumonia group (Table 3). After propensity score matching, there were no significant differences in patient characteristics between the two groups.

Fig. 1

Flow chart for patient selection. J-RECOVER, Japanese multicenter REsearch of COVID-19 by assEmbling Real-world data; JMDC, Japan Medical Data Center

Table 1 Baseline characteristics in the COVID-19 and bacterial pneumonia patients

Table 2 Laboratory results on admission in the COVID-19 and bacterial pneumonia patients

Table 3 Anticoagulant therapies and complication events in the COVID-19 and bacterial pneumonia patients

Hemostatic parameters in COVID-19 pneumonia

A multilevel mixed-effects regression model indicated significant differences in the temporal changes in platelet counts during the initial seven days between the two groups (p for interaction < 0.001), with the COVID-19 group maintaining consistently higher platelet counts (Fig. 2). In contrast, the bacterial pneumonia group demonstrated a trend toward prolonged PT-INR, while fibrinogen levels showed minimal differences between the two groups. Antithrombin activity was consistently higher in the COVID-19 group, mostly staying within the normal range. Meanwhile, the bacterial pneumonia group showed higher FDP and D-dimer levels at almost all time points.

Fig. 2

Time-series differences in hemostatic parameters between the COVID-19 and bacterial pneumonia groups after propensity score matching. The regression line with 95% confidence intervals in each group was estimated by a multilevel mixed-effects regression model with a two-way interaction term between the group and time series. Solid black lines represent patients in the COVID-19 group. Dotted red lines represent patients in the bacterial pneumonia group. P indicates the p-value for interaction. CIs, confidence intervals; COVID-19, coronavirus disease 2019; FDP, fibrin/fibrinogen degradation products; PT-INR, prothrombin time-international normalized ratio

The multiple imputation analysis showed a slight decrease in antithrombin activity relative to the primary analysis in the COVID-19 group. However, the overall trends for other hemostatic parameters remained consistent with the main results, supporting the robustness of our findings. These results are shown in Supplementary Fig. 1. Subgroup analyses based on survival status revealed distinct time-course patterns of coagulation markers in both disease groups (Supplementary Fig. 2). In COVID-19 patients, there was no substantial difference in PT-INR between survivors and non-survivors. Antithrombin activity showed a decreasing trend over time, although the reduction in non-survivors was less pronounced than observed in bacterial pneumonia patients. In contrast, D-dimer and FDP levels tended to increase progressively in non-survivors.

Differences in inflammation and organ dysfunction parameters

Timeline differences were also observed in several parameters related to systematic inflammation and liver and kidney dysfunction (Fig. 3). White blood cell counts in the patients with bacterial pneumonia improved with time, while those in the COVID-19 group increased with time, with significant time series differences observed between the two groups (p for interaction < 0.001). The CRP levels in the bacterial pneumonia group initially rose on days 2 and 3 but showed an improving trend thereafter. In contrast, while the CRP levels in the COVID-19 group followed a similar course, their improvement was more gradual. Albumin levels decreased after hospitalization in both groups, with the levels in the bacterial pneumonia group being higher than those in the COVID-19 group after day 2 (p for interaction < 0.001). Serum creatinine levels tended to be lower at all time points in the COVID-19 group. However, serum bilirubin levels in the COVID-19 group increased after hospitalization. The multiple imputation analysis showed trends in inflammation and organ dysfunction parameters that were consistent with the primary analysis. These results are shown in Supplementary Fig. 3.

Fig. 3

Time-series differences in inflammation and organ dysfunction parameters between matched COVID-19 and bacterial pneumonia groups. The regression line with 95% confidence intervals in each group was estimated by a multilevel mixed-effects regression model with a two-way interaction term between the group and time series. Solid black lines represent patients in the COVID-19 group. Dotted red lines represent patients in the bacterial pneumonia group. P indicates the p-value for interaction. CIs, confidence intervals; COVID-19, coronavirus disease 2019; CRP, C-reactive protein

Incidence of DIC

Figure 4 shows the incidence of DIC according to the JAAM-2 and sepsis-induced coagulopathy criteria for each day. According to the JAAM-2 criteria, the incidence of DIC followed a similar trend in both groups, increasing from 10 to 30% after hospitalization. However, the incidence according to sepsis-induced coagulopathy criteria differed from that defined by the JAAM-2 criteria. In comparison to the bacterial pneumonia group, the incidence in the COVID-19 was lower for each day, with both groups maintaining roughly the same incidence after day 2. We found that existing DIC scoring criteria were not sufficient to distinguish between COVID-19 coagulopathy and the pathophysiological conditions of sepsis-induced coagulopathy.

Fig. 4

Seven-day incidence of JAAM-2 DIC and sepsis-induced coagulopathy in matched COVID-19 and bacterial pneumonia groups Squares represent patients in the COVID-19 group. Triangles represent patients in the bacterial pneumonia group. COVID-19, coronavirus disease 2019; JAAM-2, Japanese Association for Acute Medicine version 2; DIC, disseminated intravascular coagulation

Discussion

Principal findings

In recent years, studies have advanced our understanding of COVID-19-related coagulopathy and its pathophysiology, but many were limited by small sample sizes and lack of time-series analyses [8, 27,28,29]. Our multicenter study provided more comprehensive insights into the temporal changes in coagulation markers across a broad patient population. We found that platelet counts in the COVID-19 group were higher in comparison to the bacterial pneumonia group, along with sustained increases in antithrombin activity and fibrinogen. Conversely, the bacterial pneumonia group showed higher levels of FDP and D-dimer. Supplementary Table 3 shows the differences from the known findings related to these parameters.

We also examined markers of inflammation and organ function. In the bacterial pneumonia group, CRP levels rapidly increased and WBC counts gradually decreased, whereas in the COVID-19 group, CRP levels gradually decreased and WBC counts gradually increased. These patterns suggested that bacterial pneumonia triggered a more acute inflammatory response, whereas COVID-19 induced a prolonged, chronic response. Bilirubin levels were elevated in COVID-19 patients, indicating potential hepatic microcirculation issues, possibly due to direct viral effects, hypoxemia, or drug-induced injury [30].

COVID-19 coagulopathy

COVID-19-related coagulopathy has shared some features with septic DIC, such as platelet activation, and endothelial injury [7, 8, 31]. However, a key difference is the elevation of fibrinogen and D-dimer levels, while platelet counts and prothrombin time (PT) remain unchanged [29, 32]. Although typical COVID-19 cases are characterized by elevated D-dimer and fibrinogen levels with normal PT and platelet counts [8, 31], D-dimer and FDP levels typically remain within the normal range during the early stages of septic DIC [33, 34]. Isolated elevation of D-dimer, without accompanying abnormalities in other coagulation markers, often indicates localized thrombus formation, such as pulmonary embolism or deep vein thrombosis [35].

The results for PT-INR and D-dimer presented some divergence from the existing literature [8, 36]. Specifically, while PT-INR remained relatively stable in COVID-19 groups, D-dimer levels showed a progressive increase over time. These findings highlight the potentially distinct progression of coagulation dysfunction between COVID-19 and bacterial pneumonia.

In COVID-19, endothelial injury in the pulmonary microcirculation is a critical finding for coagulopathy. SARS-CoV-2 directly infects and damages endothelial cells, leading to a severe loss of antithrombotic function [3]. Endothelial damage induces procoagulant changes within the deranged glycocalyx, the formation of immunothrombosis, and impaired organ perfusion that contribute to respiratory failure. Therefore, the predominant pathology in COVID-19 involves localized pulmonary vascular endothelial damage and coagulopathy, which differs from septic systemic coagulopathy. High D-dimer levels are associated with disease severity and mortality in COVID-19, and persistent elevation may warrant adjustment from prophylactic to therapeutic-dose anticoagulation. In our study, D-dimer levels remained persistently elevated, and in fatal cases, a sharp rise in D-dimer and FDP was observed on day 4 (Supplementary Fig. 2), suggesting acute thrombus formation. This may indicate a critical time for initiating therapeutic-dose anticoagulation. However, in severe cases, therapeutic anticoagulation carries a potential risk of bleeding [37]. Particularly, when fibrinogen levels are low, plasmin–α2 plasmin inhibitor complex levels are elevated, and fibrinolytic activity is increased [38]. In our study, fibrinogen levels did not decrease on day 4, suggesting a relatively low bleeding risk and a potentially safe opportunity for therapeutic intervention. Time-course monitoring of coagulation and fibrinolysis markers may help identify the appropriate timing for anticoagulation.

The use and dose of prophylactic and therapeutic anticoagulation in accordance with COVID-19 treatment guidelines [39, 40] may have attenuated D-dimer elevation in the COVID-19 group. Additionally, elevated D-dimer levels may have prompted intensified anticoagulant therapy, introducing potential indication bias. In contrast, patients with bacterial pneumonia, particularly older individuals with multiple comorbidities, may not have received equivalent anticoagulant therapy during hospitalization. These discrepancies in anticoagulant use could partially explain the consistently higher D-dimer levels in bacterial pneumonia. Moreover, unmeasured confounders, such as pre-existing thromboembolic conditions or aortic disease, may have contributed to elevated baseline D-dimer levels in our study. Additionally, the lack of sequential organ failure assessment scores limited our ability to adjust for baseline factors in our dataset, potentially affecting the comparison of coagulation profiles between the groups.

Antithrombin trends in patients with severe COVID-19

Antithrombin activity gradually declined in COVID-19 patients, contrasting with the more rapid decrease observed in patients with bacterial pneumonia [41, 42]. The drop in COVID-19 was less pronounced in comparison to bacterial pneumonia, with values rarely falling below 80%. Moreover, the typical correlation between albumin and antithrombin levels was not observed in COVID-19 patients.

The causes of antithrombin reduction are well known to include impaired production due to liver dysfunction, consumption coagulopathy, extravascular leakage, and degradation by neutrophil elastase. In consumption coagulopathy, such as sepsis-induced DIC, antithrombin levels decrease due to ongoing thrombosis [43]. Additionally, in sepsis-induced DIC, inflammation can cause extravasation of serum albumin into the interstitial space due to increased capillary permeability [44], leading to a marked reduction in antithrombin and albumin [32]. Similarly, hypoalbuminemia in COVID-19 is attributed to systemic inflammation [45] and increased pulmonary capillary permeability [46]. Inflammatory processes in COVID-19 may lead to localized changes in pulmonary vascular permeability, contributing to this difference.

Despite these findings, the precise mechanism linking endothelial dysfunction and antithrombin dynamics in COVID-19 remains unclear as our study did not include endothelial biomarkers such as von Willebrand factor, PAI-1, or tissue factor. In sepsis-induced DIC, antithrombin supplementation is sometimes considered when marked depletion occurs. However, in COVID-19, the relatively mild decline in antithrombin levels suggests that routine supplementation may not be necessary. Nevertheless, a more significant decline in antithrombin activity occurs in severe cases [6, 18], which could warrant further investigation into the role of targeted antithrombin supplementation. In line with this, targeted therapies such as antithrombin and recombinant human activated protein C may also merit investigation in COVID-19. Activated protein C may suppress SARS-CoV-2-induced endothelial activation [47, 48], and clinical trials evaluating antithrombin in COVID-19-associated ARDS have also been proposed [49]. These findings underscore the need for further investigation into endothelial injury and endothelial-targeted therapies in COVID-19-associated coagulopathy.

Institutional variability in ICU management and anticoagulant therapies

In our study, ICU admission rates in the COVID-19 group were higher in comparison to the bacterial pneumonia group. This difference can be attributed to the nature of our COVID-19 dataset. During the early pandemic period, limited understanding of the characteristics and potential complications of COVID-19 led to a more cautious approach in clinical management, with ICU admission recommended for moderate and severe cases. Additionally, the hospitals included in the bacterial pneumonia dataset represent both normal and advanced-emergency hospitals, which may have led to a lower proportion of severe cases, potentially contributing to a lower ICU admission rate. Despite performing propensity score matching, differences in ICU admission rates could not be fully adjusted. Furthermore, treatment strategies varied across institutions. Prior studies have reported that the rate of adherence to ventilator care bundles was approximately 60% in mechanically ventilated patients [50], with substantial variation across hospitals. Notably, adherence among COVID-19 patients was as low as 1% [51]. While we adjusted for major treatment variables through propensity score matching, we did not directly account for factors such as mechanical thromboprophylaxis and institutional ICU management protocols.

In addition, pre-hospital use of oral anticoagulants, such as warfarin or direct oral anticoagulants, was not captured in our dataset. This lack of data may have influenced initial coagulation profiles, especially PT-INR and D-dimer values measured upon admission. The inability to adjust for this is a potential confounding factor, and future studies incorporating medication history are warranted to clarify its impact.

Limitations

Our study has several limitations. First, this was a retrospective observational study. The decisions regarding the management strategies were at the discretion of the attending physicians. Therefore, there may have been a potential selection bias in patients with mechanical ventilation. Additionally, treatment strategies were not standardized across institutions, which may have introduced further confounding factors. Second, the accuracy of diagnoses recorded in administrative claims databases is typically lower in comparison to that recorded in prospective studies. Physicians often register diagnosis codes based on clinical judgment, and there are no established accurate criteria. Third, we were unable to obtain data on improvement of oxygenation and weaning outcomes for the patients included in this study. Additionally, detailed information on ventilator settings, such as oxygen concentration and positive end-expiratory pressure, were not accessible. This further limited our ability to assess ventilator-associated outcomes and their relationship to coagulopathy and clinical management. Fourth, genetic data and detailed data on coagulation and fibrinolysis parameters such as PT ratio, thrombomodulin, PAI-1, tissue plasminogen activator, protein-C, protein-S, von Willebrand factor, and tissue factor were not measured. This limited our ability to assess the full spectrum of coagulopathy in COVID-19 and bacterial pneumonia. Although PT-INR is primarily used for monitoring anticoagulant therapy with vitamin K antagonists, it was employed as a substitute for the PT ratio in our study due to the high rate of missing PT ratio data. Finally, we could not fully assess anticoagulant use due to the lack of detailed dosage information and the absence of data on oral anticoagulants used before and during hospitalization. This limitation may have influenced the interpretation of coagulopathy patterns.

Conclusions

Our study using a nationwide inpatient database and a nationwide registry database suggests that there are time series differences in coagulopathy between COVID-19 and bacterial infection in patients requiring mechanical ventilation. Unlike in typical septic DIC, COVID-19-related coagulopathy is characterized by reduced platelet consumption and no increase in vascular permeability. Additionally, certain COVID-19 patients may exhibit increased fibrinolytic activity due to genetic predisposition or other factors, emphasizing the importance of close monitoring to guide individualized treatment. Therefore, in cases of hypercoagulability with significantly decreased antithrombin activity, timely consideration of anticoagulant administration or antithrombin supplementation may be warranted. Thus, the treatment approach for coagulopathy in patients with COVID-19 should be distinct from that for patients with septic DIC.