Anticoagulation in COVID-19 patients – An updated systematic review and meta-analysis

Background: Thromboembolic events are common complications of COVID-19. Clinical study results on safety and efficacy of anticoagulation in COVID-19 are controversial. Material and methods: This report updates our systematic review and random-effects meta-analysis on random- ized controlled trials (RCTs) comparing standard prophylactic anticoagulation and intermediate or therapeutic anticoagulation in COVID-19 patients. We searched eligible studies for the update up to 4 February 2022 by weekly monitoring of RCTs in the Cochrane COVID-19 Study Register. Certainty of evidence was assessed using GRADE (Grading of Recommendations Assessment, Development and Evaluation). Results: For this update we included five new trials; a total of 13 RCTs with 7364 patients. Certainty of evidence was very low to low. We are uncertain whether low-dose prophylactic anticoagulation is favoured over placebo or no anticoagulation in the outpatient- or post-discharge-setting. In hospitalized patients with moderate and severe COVID-19, intermediate-dose anticoagulation may have little or no effect on thrombotic events or death (RR 1.03, 95 % CI 0.86 – 1.24), but may increase severe bleeding non-significantly (RR 1.48, 95 % CI 0.53 – 4.15). Therapeutic-dose anticoagulation may decrease thrombotic events or deaths in hospitalized patients with moderate COVID-19 (RR 0.64, 95 % CI 0.38 – 1.07; fixed-effect model RR 0.72, 95 % CI 0.57 – 0.91), but may have little or no effect in patients with severe disease (RR 0.98, 95 % CI 0.86 – 1.12). With therapeutic-dose anti- coagulation, the risk of major bleeding may increase regardless of COVID-19 severity (RR 1.78, 95 % CI 1.15 – 2.74). Conclusions: Hospitalized, moderately ill COVID-19 patients may benefit from therapeutic-dose anticoagulation, while critically ill patients may not. Risk of major bleeding must be considered.


Introduction
In its severe form, COVID-19, the clinical manifestation of SARS-CoV-2 infection, is characterized by lung failure and high rates of thromboembolic complications [1]. Given the procoagulant status and increased thrombotic risk of COVID-19 patients, modified empirical dosage regimens have been used in many places to treat COVID-19 patients and the different modes of anticoagulation were consecutively subject of several clinical trials. However, the question whether early prophylactic anticoagulation in outpatient settings or intensified prophylactic intermediate-dose or therapeutic anticoagulation in inpatient settings can reduce the risk of disease progression without increasing the risk of adverse events remains unanswered [2]. At the same time, it is still unclear to what extent there is an increased risk of thromboembolism even after hospital discharge, and whether this can be prevented with appropriate anticoagulation [3,4]. We therefore conducted a systematic review with meta-analysis of the available randomized controlled trials (RCTs) to determine the safety and efficacy of anticoagulation at any dosage with standard low-dose prophylactic anticoagulation or no prophylaxis in COVID-19 patients regardless of disease severity and treatment setting. The present meta-analyses informed the German AWMF-S2e guideline for outpatient COVID-19 patients and the AWMF-S3 guideline for the inpatient therapy of COVID-19 patients [5,6].

Materials and methods
This report is an update of a systematic review published in December 2021 [7]. The systematic review protocol was registered with PROSPERO on January 21, 2021 (CRD42021229228). The question is continuously updated in the sense of a 'living systematic review'. A detailed German version of the updated systematic review is accepted for publication at a German journal. This English version addresses the worldwide community of critical care specialists, primary care specialists and other healthcare personnel dealing with COVID-19 patients.

Inclusion criteria for studies in this systematic review
We included RCTs that compared prophylactic anticoagulation at any dosage with anticoagulation at a different dosage or with no anticoagulation in outpatients, hospitalized, or post-discharge patients with confirmed SARS-CoV-2 infection. All studies were considered for inclusion in the analysis, regardless of the severity of the disease, age, gender, and ethnicity of the study participants. COVID-19 severity of study participants was, if possible, classified according to the definition of the WHO 'clinical progression scale' (WHO 0 to 10): outpatient, mildly ill COVID-19 patients (WHO 1-3); inpatient, moderately ill COVID-19 patients (WHO [4][5] and intensive care, seriously ill COVID-19 patients (WHO 6-9) [8]. Separate meta-analyses were performed and reported for patients in outpatient, inpatient and post-discharge settings as well as according to the included severity of COVID-19 disease. There were no restrictions on the type of pharmacological anticoagulation used. All heparinoids, vitamin K antagonists and direct anticoagulants (factor Xa inhibitors and direct thrombin inhibitors) were considered regardless of dosage and administration regimen. Dosage regimens of anticoagulants have been divided into low-dose, intermediate-dose or therapeutic anticoagulation according to the definition of the studies and the general drug recommendations summarized in Table S1 [9]. Standard anticoagulation in the control arm in COVID-19 inpatients included both low-dose anticoagulation with low molecular weight heparin (LMWH) or unfractionated heparin (UFH), as well as intermediate-dose anticoagulation regimens. The latter were amended in the course of the pandemic by adjustments to national therapy guidelines (i.e. in the United Kingdom) and formed part of the standard treatment [10]. We therefore expanded our definition of standard prophylactic anticoagulation to include low-dose and intermediate-dose anticoagulation regimens and created the following comparisons for meta-analyses: • Therapeutic anticoagulation versus standard prophylactic anticoagulation (low-dose or intermediate-dose anticoagulation) • Intermediate-dose anticoagulation versus standard prophylactic anticoagulation (low-dose anticoagulation) • Standard prophylactic anticoagulation (low-dose) versus no prophylaxis or placebo The evaluation of the efficacy of anticoagulation in hospitalized COVID-19 patients was carried out by recording mortality, worsening or improvement in clinical status, thrombotic events with and without death and quality of life (day 28 or longest follow-up). In COVID-19 outpatients, the combined outcome of hospitalization or death was supplemented. For this review update, no new data for the outcomes on clinical worsening or improvement of COVID-19 inpatients were available. The outcomes were therefore not included in this update. The safety of interventions was assessed by recording serious adverse events (SAEs), adverse events (AEs) and severe bleeding (according to ISTH criteria [11]) during the study period. In post-discharge patients, safetyrelevant outcomes were supplemented by clinically relevant but nonsevere bleedings and other bleeding according to the ISTH criteria [11]. Two review authors independently examined titles and abstracts of all entries and the full texts of potentially relevant studies for their inclusion in this review. The study selection process is reported in a flowchart and follows PRISMA guidelines [12].

Data collection and analysis
Two review authors extracted data independently using a custom data extraction sheet according to Cochrane guidelines [13].
The Risk of Bias 2 (RoB 2) tool was used to assess the bias risk of study results that contributed information to the specified outcomes [14]. The RoB 2 ratings per study outcome were independently evaluated by two review authors according to Cochrane's recommendations and for the following domains [15]: selection bias, performance bias, detection bias, attrition bias, and selective reporting bias. For each domain and all domains together (overall bias risk), the bias risk of a study result was classified as low, some concerns or high.
For dichotomous outcomes, the number of events and the total number of participants in both the intervention and control groups were recorded. The relative risk (RR) with 95 % confidence interval (CI) was used as an effect measure.
Meta-analyses were carried out using the Mantel Haenszel method under a random effects model [16]. Random effects meta-analyses were performed with RevMan Web 3.11.1 and R (package "meta", version 5.2-0) [17,18]. Fixed-effects meta-analyses and exclusion of studies with high risk of bias were carried out as sensitivity analyses. For hospitalized individuals with moderate or severe COVID-19, subgroup analyses were performed according to the disease severity at baseline (moderate (WHO 4 to 5) versus severe COVID-19 disease (WHO 6 to 9) as defined by the WHO clinical progression scale [8]). Studies that only provided data for a mixed population of moderately and severely ill participants were included in the subgroup "moderate to severe COVID-19 (WHO 4 to 9)". Statistical heterogeneity was defined as P < 0.1 for the Chi 2 test for heterogeneity or an I 2 statistic ≥50 %.
Quality of the evidence was assessed according to the GRADE methodology (Grading of Recommendations, Assessment, Development and Evaluations) [19]. GRADE assesses the trustworthiness (certainty) of the evidence in four levels: very low, low, moderate, and high. Certainty of evidence was downgraded by one or two levels for risk of bias, imprecision, inconsistency (heterogeneity), indirectness (indirect study results) and publication bias.

Results
The search strategy identified a total of 1153 entries in registers and databases, two more entries were identified from other sources. After removal of duplicates, 1076 titles and abstracts were viewed, of which 942 were deemed irrelevant. The screening of 134 full texts resulted in 20 studies with a total of 32 entries that were excluded and 59 included studies with 102 entries. Of these, 50 studies (66 entries) had not yet been completed at the time of the search. Nine studies (36 entries) were included in this systematic review from the search. Monitoring of the RCT database from 25 September 2021 onwards provided a further 14 entries up to and including 4 February 2022, four of which belonged to studies not yet completed or published in full text (Letter). Four studies (10 entries) were included. The search is summarized in a PRISMA flowchart (Fig. S1).
All studies reported relevant outcomes for this systematic review. The time period for outcomes collection was 28-30 days in the majority of studies. MICHELLE examined a period of 35 days [33] ACTIV-4B 45 days [21] and INSPIRATION reported on the 90-day mortality [27]. No study reported data on quality of life.

Risk of bias
In total, the thirteen studies contributed 48 study results to 23 outcomes reported here, seven outcomes for the comparison "COVID-19 outpatients: standard prophylactic anticoagulation versus placebo", five for the comparison "COVID-19 inpatients: intermediate-dose anticoagulation versus standard prophylactic anticoagulation", five for the comparison "COVID-19 inpatients: therapeutic anticoagulation versus standard prophylactic anticoagulation" and six for the comparison "post-discharge COVID-19 patients: standard prophylactic anticoagulation versus no prophylaxis." One third of the 48 study results (33.3 %) were rated as "overall low risk of bias", 54.2 % as "some concerns about the overall bias risk" and 12.5 % as "overall high risk of bias".
COVID-19 outpatients: standard prophylactic anticoagulation versus placebo Ananworanich-2021 [20] and ACTIV-4B [21] were included in the comparison of standard prophylactic anticoagulation with low-dose anticoagulation versus placebo in the outpatient setting (meta-analyses in Table S2). We are uncertain whether standard prophylactic anticoagulation compared to placebo treatment increases or decreases all-cause mortality (RR 0. 33 (Table S2).

Discussion
In this updated systematic review, the high heterogeneity regarding the anticoagulation schemes continues to make the interpretation of the meta-analyses difficult and reduces the certainty of evidence. The Table 2 Meta-analysis for intermediate-dose anticoagulation versus standard prophylactic anticoagulation in COVID-19 inpatients including absolute effect estimates, risk of bias assessment, and certainty of evidence.  [35] and an assumed increasing thrombotic risk proportional to the disease severity, one could assume that the predominant variant at the time of study has also influenced the presented findings and may explain the heterogeneous results between studies. Unfortunately, no information (e.g. genome sequencing) was available for the included patients to allow further exploration of these assumptions.
To the best of our knowledge, there is currently no comparable and more up-to-date systematic review with meta-analysis which includes more RCTs on this topic available. Despite being published recently, the latest update of the Cochrane Review entitled 'Anticoagulants for people hospitalised with COVID-19' from March 2022 only includes four RCTs based on a search from April 2021 and is therefore lacking important study results [36]. Given the fast pace of the pandemic, we believe it is more than ever important to base clinical practice on up-to-date evidence.
Based on the data available so far, intermediate-dose anticoagulation cannot be expected to have an effect on all-cause mortality after 30 and 90 days or on the risk of thrombotic events or death. However, even intermediate-dose anticoagulation may increase the risk of major bleeding based on a clinically relevant, non-significant effect estimate.
Therapeutic-dose anticoagulation may have a benefit in terms of mortality after 28 days in the subgroup of patients with moderate COVID-19, based on data from the RAPID and HEP-COVID study of 635 participants. While based on the large ATTACC, ACTIV-4a, REMAP-CAP platform study with 2226 moderately ill participants, little or no effect on the risk of in-hospital mortality can be expected. The certainty of evidence for these contradictory results is low and further studies may change the interpretation. Therapeutic-dose anticoagulation may reduce the rate of thrombotic events or deaths in patients with moderate COVID-19 but has had no effect in patients with severe COVID-19. However, this effect was only statistically significant when using the fixed-effect model. Therapeutic-dose anticoagulation can increase the Table 3 Meta-analyses of therapeutic-dose anticoagulation versus standard prophylactic anticoagulation in COVID-19 inpatients including absolute effect estimates, risk of bias assessment, and certainty of evidence. rate of severe bleeding events compared to standard prophylactic anticoagulation. Thus, moderately ill COVID-19 patients without increased bleeding risks may be considered for therapeutic anticoagulation, the increased risk of bleeding should however be taken into account in decision-making and anticoagulated COVID-19 patients should be carefully monitored for bleeding events [5,6]. Severely ill COVID-19 patients may not benefit from therapeutic anticoagulation while at the same time the risk of severe bleeding can be increased. Without further specific indication, therapeutic anticoagulation can therefore not be recommended for critically ill COVID-19 patients [5,6].

Ethics approval and consent to participate
Not applicable.

Consent for publication
Not applicable.

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Funding
The Federal Ministry of Education and Research, Germany, NaFoU-niMedCovid19 (funding number: 01KX2021); part of the project "CEOsys" supported this work (funding ended 31 December 2021). The funders had no role in considering the study design or in the collection, analysis, interpretation of data, writing of the report, or decision to submit the article for publication.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.