Effect of NOACs versus LMWH on COVID-19 mortality
Our findings indicate a significant association between NOAC therapy and a reduced mortality risk in patients with severe COVID-19 compared with LMWH treatment. Specifically, the odds of mortality for patients treated with NOACs are approximately 75% lower than those for patients treated with LMWHs, corresponding to an odds ratio of 0.2455 (95% CI: 0.0816--0.6698), as derived from contingency table analysis (Tables S1, S5). Logistic regression analysis adjusted for additional variables further corroborated these findings, demonstrating that the mortality risk for LMWH patients was approximately 22 times greater than that for NOAC patients, with an odds ratio of 0.045 (95% CI: 0.0072--0.2821, p < 0.001). Please refer to the supplementary material for a detailed breakdown of these calculations.
Additionally, the Cox proportional hazards model highlighted the significant protective effect of NOACs against hospital mortality, with an HR of 0.201 (95% CI: [0.059 to 0.686], p=0.0103) (Figure 5 and Tables S2, S9). Successful PSM, indicated by reduced standardized mean differences, confirmed that the groups were comparable, minimizing potential confounding factors and strengthening the reliability of our findings. This finding was further supported by the Love plot (Figure 7), which demonstrated minimal confounding in the analysis. To ensure robustness, we conducted a variance inflation factor (VIF) analysis to assess covariate interdependencies within our PSM model (Table S6). The SMR for the LMWH group was 60% greater than that predicted by the SAPS3 score, which aligns with the findings of Quintairos and colleagues 20, who reported an SMR of 1.57 among Brazilian ICU patients with COVID-19 (Fig. S4). Conversely, the SMR for the NOAC group was 0.71, indicating a 55% reduction from its predicted mortality, which is 61.4% lower than that of the Brazilian ICU patient data from the same study. Moreover, the SAPS3 and SOFA scores reported by Correia and colleagues 21 were 18% and 25% lower than those reported in our cohort (Table 2), which presented average ICU severity scores higher than the national average.
The interplay between anticoagulation therapy and preexisting conditions
Our study revealed an association between NOAC therapy and decreased mortality via univariate analysis. Hypertension, present in 82.8% of patients treated with NOACs, is typically linked to poorer outcomes in patients with COVID-19. Initially, univariate analysis indicated a significant association between hypertension and reduced mortality; however, this relationship was not confirmed via multivariate analysis (Fig. S2, Table 1). These findings suggest that the protective effect of NOAC therapy may be influenced by multiple factors in a multivariate context, underscoring the complex interaction between preexisting health conditions and treatment outcomes.
Impact on thrombosis and bleeding
Despite concerns about the effectiveness and potential side effects of NOACs, our analysis revealed no significant difference in the incidence of hemorrhagic or thrombotic events between the NOAC and LMWH groups. A recent systematic review suggested that NOACs might be associated with a lower bleeding risk than LMWH in patients who were not critically ill and hospitalized with COVID-19 22. However, the complex interplay between anticoagulation therapy and COVID-19 underscores the need for further investigation to understand its impact on patient survival outcomes 23.
The advantage of NOACs over LMWHs stems from their direct inhibition of Factor Xa, which is crucial in thrombin regulation. This blockade prevents the conversion of prothrombin and directly reduces thrombin concentrations, minimizing clot formation. Unlike LMWH, which requires antithrombin and acts indirectly, NOACs can penetrate thrombi, offering a significant benefit, particularly during severe inflammatory states where antithrombin levels may vary. Additionally, NOACs offer benefits such as oral administration, predictable pharmacokinetics, and no need for routine laboratory monitoring, increasing their clinical utility.
Although COVID-19-induced coagulopathy is similar to severe sepsis, it is characterized predominantly by thrombotic rather than hemorrhagic occurrence 4. Fatal outcomes frequently arise from extensive fibrin deposition and significant thrombin generation, particularly within the lung microvasculature 5. This leads to acute respiratory distress syndrome, coagulopathy, vascular device thrombosis, and occlusive events such as stroke and limb ischemia. Our ICU cohort consisted of patients with severe COVID-19, as reflected by higher-than-average ICU severity scores in Brazil 20.
Bioavailability issues in ICU patients receiving vasopressors
The frequent use of vasopressors, which affects more than half of the patients on their first ICU day (53%, Table 1, Fig. S5), suggests a potential underestimation of their effect on LMWH bioavailability 24. Hemodynamic instability in patients with severe COVID-19 may further contribute to the observed increase in mortality in the LMWH group. Bioavailability is particularly challenging under conditions where hypoxia exacerbates thrombosis through hypoxia-inducible pathways. Moreover, vasopressors, including protein C and thrombomodulin, interact with endothelial and coagulation pathways, complicating the breakdown of prothrombinase and tenase complexes and potentially increasing thrombin levels 4. Additionally, the formation of neutrophil extracellular traps in severe COVID-19 amplifies thrombo-inflammatory responses 25. Consequently, ICU patients receiving vasopressors may require tailored dosing or alternative methods of LMWH administration to optimize therapeutic effectiveness 26.
Prognostic value of D-dimer
The PDDL in patients with severe COVID-19 often exceeds that observed in patients with community-acquired pneumonia and has significant prognostic value 27. Specifically, D-dimer concentrations above twice the ULN have been linked to poor outcomes in advanced COVID-19 patients, frequently indicating increased rates of intubation and mortality. In our subanalysis, we found that the initiation of mechanical ventilation (MV) was strongly correlated with increased mortality, particularly at higher PDLLs (Fig. S6), suggesting that a more significant burden of comorbidities is associated with higher mortality.
Our findings align with a recent Polish randomized study, which demonstrated that anticoagulant therapy, particularly unfractionated heparin, was associated with a lower incidence of MV than LMWH 28. Furthermore, our Cox proportional hazards model—adjusted for severity and comorbidities via propensity score matching (PSM)—revealed a significant increase in mortality risk with increasing PDDL. Specifically, each doubling of the PDDL was associated with a more than twelvefold increase in mortality risk, as shown by our polynomial logistic regression analysis (HR=12.59, 95% CI=4.03–21.15], p=0.00178) (Fig. 6, Table S3). This robust association underscores the critical importance of monitoring the PDDL as a prognostic indicator.
In support of this, the literature has consistently linked high PDDL scores to increased mortality in patients with COVID-19. Ranucci and colleagues 29 reported critical hemostatic disturbances in nonsurvivors, including increased thrombin and fibrin turnover, reduced fibrinolysis, and enhanced baseline fibrinolysis inhibition, as evidenced by elevated plasminogen activator inhibitor-2/plasmin‒antiplasmin ratios. Thrombin generation patterns differ distinctly between survivors and nonsurvivors, contributing to severe thrombo-inflammatory responses and microvascular damage, as further evidenced by elevated PDDL. These findings highlight the prognostic importance of D-dimer levels and support the integration of NOACs into comprehensive strategies for managing coagulopathy in patients with severe COVID-19.
Anticoagulation Strategies Amidst Uncertain Evidence
Our study was conducted during the pandemic's early stages before establishing standardized treatment protocols, such as antivirals, vaccines, corticosteroids, and interleukin-6 inhibitors 30,31. Consequently, our findings provide unique insights into the effectiveness of anticoagulation therapy within an evolving therapeutic context. The absence of these treatments emphasizes the importance of our results and highlights the potential of NOACs when broader treatment options are unavailable. Moreover, as emerging SARS-CoV-2 variants that evade immunity may lead to new waves of severe infections requiring ICU admission, these findings underscore the need for prospective trials within evolving therapeutic standards to validate and refine anticoagulation strategies for severe COVID-19, contributing to global efforts to improve patient outcomes (Table S10).
The need for personalized anticoagulation strategies, informed by the latest research and tailored to individual risk profiles, remains critical for managing conditions characterized by significant inflammatory responses, including severe COVID-19. Billett et al.11 found that early initiation of apixaban or enoxaparin within 48 hours of hospital admission significantly reduced mortality, with apixaban showing effectiveness comparable to that of enoxaparin. Stratification by PDDL indicated that higher levels were associated with increased mortality, highlighting the benefits of anticoagulation, particularly in patients with a PDDL above 10 µg/mL, without increasing the need for transfusion.
Previous research has highlighted the benefits of therapeutic anticoagulation in noncritically ill patients. However, the evidence concerning the efficacy of LMWH or unfractionated heparin in critically ill COVID-19 patients is mixed 32,33. Our cohort study revealed a preference for apixaban (94.3%, 33 out of 35 patients) over rivaroxaban (5.7%, 2 out of 35 patients) among critically ill patients receiving NOACs, despite limited evaluations of NOACs in patients with severe COVID-19 11,12, and established the benefits of LMWH in treating COVID-19-associated coagulopathy. The pronounced use of apixaban suggested a shift in therapeutic strategies, potentially influenced by drug safety profiles 12, individual patient considerations, or emerging evidence not captured in earlier studies. As reported by Gloeck et al. 34, the importance of high-quality randomized controlled trials (RCTs) for comparing different anticoagulation intensities in hospitalized COVID-19 patients has been emphasized, highlighting the critical need for clinicians to carefully consider the risk‒benefit ratio in clinical decision-making.
Our findings align with existing research that links elevated PDDL (scored according to our methodology) with increased mortality risk in patients with COVID-19, demonstrating that this risk is consistent across different anticoagulation intensities (Fig. S2). In contrast to the COVID-19-PREVENT trial 35, which excluded patients requiring ICU admission at randomization, our retrospective study uniquely examined only ICU-admitted patients, a focus rarely observed in COVID-19 research. Similarly, the HEP-COVID trial 36 advocated intensive anticoagulation in patients with PDDL (> 4× ULN) or significant coagulopathy.
Despite the favorable safety profile of apixaban, the CHEST guidelines37 for managing COVID-19 in critically ill patients do not fully endorse its therapeutic use. Apixaban should potentially be considered for reducing end-organ failure and death in noncritically ill patients, but its efficacy in critically ill patients is unclear because of a lack of robust evidence.
DVT incidence analysis
In our study, the incidence of DVT was 9.8% in the LMWH group and 5.7% in the NOAC group among ICU-admitted COVID-19 patients (Figure S3). These rates are significantly lower than the 25% incidence reported by Cui and colleagues and less than the 12.7% incidence (95% CI: 8.7--17.5) noted in a meta-analysis by Kollias and colleagues 38.
Although our study focused on occurrence, the 7.7% incidence of VTE in non-ICU patients reported by Mansory and colleagues 39 highlights the distinct risk profiles and underscores the need for precise VTE management strategies across different care settings.
Chaves and colleagues 40 explored the dual role of apixaban as an anticoagulant and potential antiviral agent because of its noncompetitive inhibition of virus replication. However, the efficacy of therapeutic doses of apixaban for viral inhibition has not been determined 41.
This study highlights the need for standardized anticoagulation protocols to manage severe COVID-19 effectively, as shown by the lower DVT rates in our study (Fig. S3). The challenges of selecting optimal treatment strategies—marked by high rates of undetected DVT and significant autopsy 42 findings—underscore the critical importance of standardizing these practices. Variability in anticoagulation protocols, antithrombin levels, and anti-Factor Xa measurements complicates heparin use, hindering a deeper understanding of therapeutic outcomes in COVID-19 treatment. Despite the high incidence of hypertension (82.8%) in our cohort, mortality rates were lower among patients treated with NOACs. Although univariate analysis suggested an association between hypertension and reduced mortality, this was not confirmed by multivariate analysis (Fig. S2, Table 1), suggesting that the protective effects of NOAC therapy may be influenced by other variables in the multivariate model, illustrating the complex interplay between preexisting conditions and treatment outcomes in patients with COVID-19.
Study limitations
The interpretability of the study is limited by its retrospective design, single-center setting, and short in-hospital follow-up period. Although the moderate sample size was sufficient to detect clinically significant differences, as demonstrated by our post hoc analysis, this may restrict the generalizability of our findings.
A key limitation is the absence of thrombin generation tests and anti-factor Xa measurements in the LMWH group. These findings provide evidence of the pharmacological effect of LMWH and insights into its efficacy in treating severe COVID-19. Without these measurements, we cannot determine whether the poorer outcomes in the LMWH group were due to inadequate anticoagulant effects from antithrombin deficiency (which is crucial for the action of LMWH) or other factors related to COVID-19. This limitation hinders our ability to explain the mechanism behind the observed differences in mortality between the NOAC and LMWH groups.
Variability in clinical practices among physicians and over time may have influenced outcomes, particularly given the early pandemic context with evolving treatment protocols. The unique characteristics of our study population, including disease severity and demographic data, limit the generalizability of our findings to other contexts.
Despite the use of propensity score matching, unmeasured confounding factors may have influenced the results. The short follow-up period limited our ability to assess long-term outcomes and delayed adverse effects. Reliance on electronic health records and manual data extraction could have introduced inaccuracies, particularly regarding comorbidities and treatment adherence.
The lack of randomization hinders causal inferences between NOAC and LMWH use. Potential underreporting or misclassification of adverse events and dosing variability also challenge our findings. The unique nature of our study in anticoagulation research for patients with severe COVID-19 limits direct comparisons with others, emphasizing its novelty and the need for further validation.
Prospects
As COVID-19 evolves with widespread vaccination, future research must address key areas. It is crucial to investigate the efficacy of novel oral anticoagulants (NOACs) in patients with severe COVID-19 who experience vaccine breakthrough infections or compromised immunity, as these patients may remain at high risk for severe outcomes. Understanding the impact of evolving SARS-CoV-2 variants, especially oral direct factor Xa inhibitors, on anticoagulation treatments is essential. Variants such as KP.3 and LB.1, known for increased transmissibility and immune evasion, challenge current protocols (Table S10). Future studies should expand in scale and diversity, incorporating genomic surveillance to track strains. These studies should aim to determine how different variants affect anticoagulation therapy efficacy and whether dosing or drug selection adjustments are necessary.
Lessons from managing coagulopathy in patients with COVID-19 apply to other severe infections with elevated thrombotic risk, such as sepsis. Comparative studies between NOACs and LMWHs could provide insights into broader applications of anticoagulation therapies in critical care. The success of NOACs in critically ill patients with severe thromboinflammatory conditions underscores their potential for wider use 43.
Emerging SARS-CoV-2 variants that evade immunity may lead to new waves of severe infections requiring ICU admission 44. Therefore, prospective trials within evolving therapeutic standards are needed to validate and refine anticoagulation strategies. These trials should consider the efficacy of NOACs against evolving variants and their potential synergistic effects with other treatments 44. Furthermore, developing personalized protocols based on individual risk factors in patients with COVID-19, biomarkers, and genomic data of the infecting strain should be prioritized 45.
Long-term studies on the efficacy and safety of NOACs in preventing severe COVID-19 and their role in preventing and managing post-COVID-19 complications are necessary. Economic evaluations of the expanded use of antibiotics for severe infectious diseases and strategies to address implementation challenges in diverse healthcare settings will be crucial for translating research into clinical practice.
By pursuing these research directions, we can address the dynamic nature of COVID-19 while expanding our understanding of anticoagulation strategies for severe infectious diseases. This approach could revolutionize thrombotic risk management in critical care, improving outcomes for patients with evolving viral threats and adapting to changing therapeutic landscapes.