Anticoagulant use before COVID-19 diagnosis prevent COVID-19 associated acute venous thromboembolism or not: A systematic review and meta-analysis

Abstract

BACKGROUND

Coagulopathy and thromboembolic events are associated with poor outcomes in coronavirus disease 2019 (COVID-19) patients. There is conflicting evidence on the effects of chronic anticoagulation on mortality and severity of COVID-19 disease.

AIM

To summarize the body of evidence on the effects of pre-hospital anticoagulation on outcomes in COVID-19 patients.

METHODS

A Literature search was performed on LitCovid PubMed, WHO, and Scopus databases from inception (December 2019) till June 2023 for original studies reporting an association between prior use of anticoagulants and patient outcomes in adults with COVID-19. The primary outcome was the risk of thromboembolic events in COVID-19 patients taking anticoagulants. Secondary outcomes included COVID-19 disease severity, in terms of intensive care unit admission or invasive mechanical ventilation/intubation requirement in patients hospitalized with COVID-19 infection, and mortality. The random effects models were used to calculate crude and adjusted odds ratios (aORs) with 95% confidence intervals (95%CIs).

RESULTS

Forty-six observational studies met our inclusion criteria. The unadjusted analysis found no association between prior anticoagulation and thromboembolic event risk [n = 43851, 9 studies, odds ratio (OR)= 0.67 (0.22, 2.07); P = 0.49; I² = 95%]. The association between prior anticoagulation and disease severity was non-significant [n = 186782; 22 studies, OR = 1.08 (0.78, 1.49); P = 0.64; I² = 89%]. However, pre-hospital anticoagulation significantly increased all-cause mortality risk [n = 207292; 35 studies, OR = 1.72 (1.37, 2.17); P < 0.00001; I² = 93%]. Pooling adjusted estimates revealed a statistically non-significant association between pre-hospital anticoagulation and thromboembolic event risk [aOR = 0.87 (0.42, 1.80); P = 0.71], mortality [aOR = 0.94 (0.84, 1.05); P = 0.31], and disease severity [aOR = 0.96 (0.72, 1.26); P = 0.76].

CONCLUSION

Prehospital anticoagulation was not significantly associated with reduced risk of thromboembolic events, improved survival, and lower disease severity in COVID-19 patients.

Core Tip: Coronavirus disease 2019 (COVID-19) patients are at an increased risk of developing thromboembolic events and hypercoagulable disorders, which are poor prognostic indicators of COVID-19 disease. The results of different observational studies on the effects of chronic anticoagulation on the mortality and severity of COVID-19 disease in infected patients are inconsistent. This systematic review and meta-analysis provide a comprehensive assessment of the risk of thromboembolic events, mortality, and severity of COVID-19 disease in patients on prehospital anticoagulation treatment.

INTRODUCTION

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) emerged at the end of 2019 and was later termed coronavirus disease 2019 (COVID-19)[1]. While Acute Respiratory Distress Syndrome (ARDS) was the predominant complication associated with a potentially fatal outcome, the infection progression could also be influenced by other risk factors like diabetes, coronary artery disease, chronic kidney disease, metabolic syndrome, psychiatric and neurological complications[2-12]. Due to the complexity of these complications, a standardized treatment approach for COVID-19 has yet to be universally agreed upon to guide healthcare professionals worldwide[10,13-15].

Studies have described that multi-organ involvement in COVID-19 is associated with endothelial dysfunction and thrombosis[16]. Moreover, COVID-19 has been observed to influence clotting system pathways and activate platelets, potentially resulting in vascular inflammation, hypercoagulability, and endothelial dysfunction[17]. As a result, thromboembolic events are found to be a common complication in these individuals, with an incidence rate of around 25% for Venous Thromboembolism (VTE)[16]. SARS-CoV-2-related coagulopathy is distinct from other causes of coagulopathy that result in VTE as it involves a significant element of inducing systemic inflammation and endothelitis, which might not be effectively addressed by conventional anticoagulant treatment methods[18]. Regardless, emerging research evidence suggests that anticoagulation therapy has shown effectiveness in COVID-19 patients, particularly in noncritically ill hospitalized cases[19].

Studies mostly emphasize using anticoagulation therapy during hospitalization; however, the relationship between pre-existing chronic anticoagulation due to non-COVID-19 indication and clinical outcomes in COVID-19-infected patients remains largely unknown. This meta-analysis aims to analyze the relationship between VTE risk in COVID-19 disease and prehospital anticoagulation and to study the impact of prehospital anticoagulation on COVID-19 disease severity and mortality.

MATERIALS AND METHODS

The Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines were followed in this systematic review and meta-analysis[20]. The protocol for this meta-analysis was registered on PROSPERO (CRD42022306893).

Data sources and literature search strategy

Two independent investigators (Iqbal K and Rathore SS) utilized databases, including LitCovid, WHO, and Scopus, to conduct a systemic literature search for relevant articles from December 2019 to June 30, 2023 (Supplementary Table 1). A combination of keywords, such as “anticoagulant”, “direct oral anticoagulants”, “DOAC”, “vitamin K antagonist”, “VKA”, “dabigatran”, “rivaroxaban”, “apixaban”, “edoxaban”, “warfarin”, “heparin”, “prehospital”, “prehospital”, “preadmission”, “pre-admission”, “chronic”, “long-term”, “pre-hospital”, and “prior”, were searched electronically[21].

Study selection and inclusion criteria

Duplicates within the articles retrieved from our systemic search were identified and eliminated using Endnote (Clarivate Analytics, Thomson Reuters Corporation, Philadelphia, Pennsylvania). Two independent researchers (Iqbal K and Rathore SS) screened the articles based on their titles, abstracts, and keywords. The articles were subsequently assessed for relevance through full-text screening. Data review and collection were done by Bhurwal A, Iqbal A, Ahmed J, Iqbal K, Sharma N, Mehdi M, Kumar P, and Rathore SS. References to the shortlisted articles were also screened for additional studies. Any disagreement was resolved by discussing within the group or through an independent reviewer’s (Bansal V) inputs. Articles that failed to fulfill the inclusion criteria were removed. For the meta-analysis, the inclusion criteria were: (1) All retrospective or prospective observational studies enrolling patients ≥ 18 years; (2) Studies that compared outcomes of COVID-19-infected patients who had been on anticoagulation before their COVID-19 diagnosis for any indication vs those that did not receive any prehospital anticoagulation; and (3) Studies including data (raw or adjusted) for at least one of the outcomes of interest. Exclusion criteria consisted of: (1) Studies without a control group; (2) Studies with no outcome of interest (new incidence of thromboembolic events, all-cause mortality, severity of COVID-19 (defined as per WHO Ordinal scale 6-9); and (3) Literature reviews or narrative reviews or case reports.

Our primary outcome of interest was the incidence of VTE in COVID-19-infected adults on chronic anticoagulation prior to the infection. Mortality and disease severity were considered secondary outcomes. COVID-19 disease severity was based on the WHO ordinal scale[22]. A WHO clinical progression scale score of 6-9 indicated severe COVID-19, often necessitating intensive care unit (ICU) admission. In cases where this scale was not used, the highest WHO ordinal scale event was used as a substitute for severe infection. If ICU admission rates were unavailable, intubation or non-invasive mechanical ventilation rates were considered to assess COVID-19 severity. The cases included SARS-CoV-2 positive hospitalized patients with prehospital use of anticoagulants. The controls were COVID-19-positive hospitalized patients with no history of prehospital anticoagulation. Both raw data and their adjusted estimates were derived for the meta-analysis. Estimates from studies that performed propensity-matched scoring were also considered as adjusted estimates.

Statistical analysis

We used Review Manager v.5.4 (The Nordic Cochrane Centre, The Cochrane Collaboration, 2014)[23] for unadjusted and adjusted analysis and Comprehensive Meta-Analysis software package 154 (Biostat, Englewood, NJ, United States)[24] for the meta-regression analysis. For the unadjusted analysis, we calculated the Mantel Haenszel crude odds ratios (ORs) using a random-effects model with a 95% confidence interval (95%CI). We also calculated the adjusted OR (aOR) with 95%CI to measure adjusted estimates. Adjusted estimates were used for the primary analysis to produce more reliable results as it considers the baseline differences between the two groups (prior anticoagulant users and non-users) that may influence the outcomes. In all cases, significance was defined by a P-value < 0.05, and all analyses utilized a random-effects model. Statistical heterogeneity was gauged among the included articles using the Higgins I² statistics, and I² > 50% was regarded as substantial heterogeneity. To reduce the inherent selection bias in observational studies, a subgroup analysis based on study design was performed[25].

We conducted univariate and multivariate meta-regression analyses with random effects (maximum likelihood approach) to examine potential study variations that could impact the effect magnitude. The age and gender distribution of the study sample, the study's country of origin, and the percentage of participants with comorbid conditions like obesity, diabetes, pulmonary illness, dyslipidemia, cardiovascular disease, and hypertension were all considered potential sources of variability. If covariates significantly (P < 0.05) altered the relationship between mortality or severity in COVID-19 hospitalized patients and prehospital anticoagulant usage, they were chosen for additional modeling. Two models were developed: one for mortality and the other for the severity of the condition. Then, using P = 0.05 as the cutoff threshold for elimination, preselected factors were included in a manual backward and stepwise multiple meta-regression analysis. P < 0.05 (P < 0.10 for heterogeneity) was used to define statistical significance. The meta-analysis and meta-regression tests were all two-tailed.

Risk of Bias and Quality assessment

Two investigators (Kumar P, Iqbal A) assessed the included studies’ methodological quality through the Newcastle-Ottawa Scale (Supplementary Table 2)[26]. The following categories were used to evaluate every study: (1) Study group selection; (2) Study group comparability; and (3) Determination of the desired outcomes (Supplementary Table 2). The publishing bias likelihood was tested using Egger regression methods, and the results were visually examined using funnel plots (Supplementary Figure 1). The degree of evidence certainty at the outcome level was evaluated using the GRADE pro profiler (GRADE working group, McMaster University, and Evidence Prime Inc; Supplementary Table 3)[27].

RESULTS

The initial database search produced 2056 articles of potential relevance. After eliminating the duplicates, 956 articles were filtered for appropriateness and relevancy using their titles, abstracts, and keywords. Of them, 86 full-text papers relevant to the manuscript's objectives were examined. The final analysis included 46 studies comprising 41 cohort studies (36 retrospective, 2 ambispective, and 3 prospective), three case-control studies, and one cross-sectional study. The PRISMA flowchart outlining the search process is shown in Figure 1. Table 1 provides an overview of the study characteristics of the included publications, and Table 2 shows the studies reporting adjusted estimates along with the covariates for which the estimates are adjusted.

Figure 1.

PRISMA flow diagram.

Table 1.

Study characteristic table for included studies

Ref.	Country of study	Study design	Setting	Total COVID-19 positive patients	Total patients with pre-admission anticoagulation	Type of anticoagulant	Duration of anticoagulant	Indication for anticoagulant Use	Definition Of Severity	Mean age ± SD (yr)	Female sex proportion (%)	Diabetes proportion (%)	Hypertension proportion (%)	Pulmonary disease proportion (%)	Arrhythmia proportion (%)
Ageno et al[28], 2021	Italy	Retrospective observational study	Inpatient hospitalized	4396	43	DOAC or VKA	-	AF	NA	-	-	-	-	-	56.4
Arachchillage et al[29], 2022	United Kingdom	Ambispective cohort study	Inpatient hospitalized	5883	963	DOAC (rivaroxaban, apixaban, edoxaban, dabigatran, or VKA (warfarin)	-	VTE or heart disease or AF	ICU admission	-	44.81	28.96	47.12	24.55	-
Aslan et al[55], 2021	Turkey	Retrospective cohort study	Inpatient hospitalized	1710	79	DOAC (dabigatran, rivaroxaban, apixaban, edoxaban)	-	AF, venous thrombosis	ICU admission	62 (52-71)	50.5	27	42	6	5
Bauer et al[71], 2021	United States	Case–control study	Hospitalized and non-hospitalized patients	1449	-	-		-	Admitted to the hospital/died	54.7 ± 22.5	63	17	36	22	-
Boari et al[68], 2020	Italy	Retrospective cohort study	Inpatient hospitalized	258	29	-	-	-	NA	71.0 ± 13.8	32.9	26	58.5	14	-
Brouns et al[53], 2020	Netherlands	Retrospective case-series	Inpatient hospitalized	101	18	DOAC or VKA	-	-	-	-	-	-	-	-	-
Buenen et al[40], 2021	Netherlands	Cohort study	Hospitalized and non-hospitalized patients	497	110	DOAC or VKA	-	-	NA	-	36.2	20.5	52.1	26	-
Chocron et al[41], 2021	France	Retrospective cohort study	Inpatient hospitalized	2878	382	DOAC or VKA	-	AF, VTE	ICU admission	-	40.3	30.3	75.3	-	24.8
Corrochano et al[30], 2022	Spain	Retrospective observational study	Inpatient hospitalized	1598	155	VKA (warfarin or acenocoumarol), DOAC (dabigatran, rivaroxaban, apixaban or edoxaban), or heparins	-	-	ICU admission	66.5 (17.1)	47.1	20	50.8	-	-
Covino et al[42], 2021	Italy	Retrospective observational study	Emergency department	184	92	DOAC or VKA	1 month	AF	NA	84 (81-87)	50	18.5	41.8	17.4	-
Denas et al[43], 2021	Italy	Retrospective observational study	Hospitalized and non-hospitalized patients	4697	651	VKA, NOAC (dabigatran, rivaroxaban, apixaban) or edoxaban	6 months	AF	ICU admission	-	45.6	24.1	87.3	-	-
Fauvel et al[54], 2020	France	Retrospective observational study	Inpatient hospitalized	1240	136	VKA 47 (3.8), NOAC 78 (6.3), heparin 11 (0.9)	-	-	ICU admission and mechanical ventilation	-	-	-	-	-	-
Flam et al[59], 2021	Sweden	Prospective cohort study	Inpatient hospitalized	459402	103703	DOAC (dabigatran, apixaban, rivaroxaban or edoxaban)	0–6 months [7394 (7.1%)]; 7–24 months [29012 (28.0%)]; > 24 months [67245 (64.8%)]	AF and atrial flutter	Hospital admission and ICU admission	-	-	-	-	-	-
Fröhlich et al[31], 2021	Germany	Retrospective cohort study	Inpatient hospitalized	6637	731	DOAC or VKA	6 months	AF	NA	-	50	25	93	-	76
Fumagalli et al[44], 2022	Italy	Retrospective cohort study	Inpatient hospitalized	176	91	DOAC or VKA	-	AF	NA	-	48.3	33	71	21	11.9
Gülcü et al[32], 2022	Turkey	Retrospective cohort study	Inpatient hospitalized	5575	451	DOAC (rivaroxaban, apixaban, edoxaban, dabigatran), and VKA (warfarin)	-	-	NA	64 (51–74)	49.8	26.9	49.5	13.9	5.2
Hanif et al[33], 2020	United States	Retrospective cohort study	Inpatient hospitalized	58	33	DOAC (rivaroxaban, apixaban, edoxaban, and dabigatran), VKA (warfarin)	-	-	NA	-	34.5	-	-	-	-
Harrison et al[34], 2021	United States	Retrospective cohort study	Inpatient hospitalized	1027	132	DOAC (rivaroxaban, apixaban, dabigatran), and VKA (warfarin)	-	-	NA	-	52.5	38.7	74.3	-	21.1
Ho et al[45], 2021	United States	Retrospective cohort study	Hospitalized and non-hospitalized patients	28076	304	VKA (warfarin) or DOAC (dabigatran)	Within 3 months prior to SARS-Cov-2 diagnosis	-	ICU admission	-	51.6	8.3	7.2	-	-
Hozayen et al[35], 2021	United States	Prospective cohort study	Outpatient and inpatient	5597	160	Enoxaparin, VKA (warfarin), DOAC	Within 3 months prior to SARS-Cov-2 diagnosis	-	NA	51 ± 22	57.1	15.7	34.7	4.5	12
Iaccarino et al[72], 2021	Italy	Cross-sectional study	Inpatient hospitalized	2377	125	DOAC	6 months	AF, mechanic valvularreplacement, pulmonary thromboembolism prophylaxis	ICU admission	68.2 ± 0.38	37.3	18	59	-	4.7
Klok et al[61], 2020	Denmark	Retrospective cohort study	Intensive care	184	17	-	-	-	NA	-	-	-	-	-	-
Li et al[36], 2020	China	Ambispective cohort study	Inpatient hospitalized	547	16	-	-	-	Severe COVID-19 infection	60 (48-69)	49.1	15.1	30.3	3.1	-
Lodigiani et al[73], 2020	Italy	Retrospective cohort study	Inpatient hospitalized	388	33	DOAC or VKA	-	-	ICU admission	66 (55–75)	32	22.7	47.2	9	-
Ménager et al[60], 2020	France	Retrospective cohort study	Inpatient hospitalized	82	9	VKA (warfarin or acenocoumarol or fluindione)	-	-	NA	88 (85–92)	47.6	23.2	63.4	-	36.6
Middeldorp et al[62], 2020	Netherlands	Retrospective cohort study	Inpatient hospitalized	198	19	-	-	AF	ICU admission	61± 14	34	-	-	-	-
Natali et al[63], 2020	United States	Retrospective case control study	Inpatient hospitalized	400	22	-	-	-	NA	-	-	-	-	--	-
Olcott et al[46], 2021	England.	Retrospective cohort study	Inpatient hospitalized	309	81	DOAC or VKA	-	-	NA	-	47.9	-	-	-	-
Parker et al[47], 2021	United Kingdom	Retrospective cohort study	Inpatient hospitalized	1032	164	DOAC, VKA, LMWH, fondaparinux	Taking the anticoagulant for > 1 month	AF, VTE, metallic heart valve, LV thrombus	ICU admission	71 (56–83)	44.9	29.3	44.3	-	-
Philipose et al[66], 2020	United Kingdom	Retrospective cohort study	Inpatient hospitalized	466	68	-	-	-	NA	-	-	-	50.2	28.1	-
Reilev et al[64], 2020	Denmark	Cohort study	Community-managed and hospitalized	11122	577	-	At least one filled prescription within 6 months prior to the test date	-	ICU admission	48 (33–62)	62	6.4	24	12	4.6
Rieder et al[37], 2022	Multi-Country	Retrospective cohort study	Hospitalized and outpatient	1433	334	VKA or non-VKA DOAC (rivaroxaban, apixaban, edoxaban, dabigatran etexilate)	-	AF	NA	-	39.8	31.2	86.5	-	25.4
Rivera-Caravaca et al[58], 2021	International, HOPE COVID-19 Registry	Retrospective cohort study	Inpatient hospitalized	1002	110	DOAC or VKA	-	AF, VTE, mechanical heart valves	NA	-	40.8	31.2	82.1	18.3	-
Rivera-Caravaca et al[38], 2021	United States, Trinetx	Cohort study	Hospitalized and outpatient	26006	13003	DOAC (dabigatran, apixaban, rivaroxaban or edoxaban)	1 yr	-	NA	-	48.4	37.9	71.9	18.3	48.1
Rodríguez-Molinero et al[67], 2020	Spain	Retrospective cohort study	Inpatient hospitalized	418	34	-	-	-	Need for oxygen therapy through a nonrebreather mask or mechanical ventilation	65.4 ± 16.6	43.1	23.7	52	9.8	10.8
Rossi et al[57], 2020	Italy	Retrospective observational study	Outpatient	70	26	DOAC (eivaroxaban, apixaban, edoxaban, dabigatran)	Regularly taken by the patient for at least 6 months	AF, pulmonary embolism, or DVT	NA	-	50	25.7	61.4	15.7	-
Russo et al[48], 2022	Italy	Retrospective observational study	Inpatient hospitalized	467	87	DOAC (edoxaban, dabigatran, rivaroxaban, apixaban), VKA (warafrin)	-	AF, prosthetic heart valve, venous thromboembolism	NA	-	33.3	25.3	74	18.8	-
Ruzhentsova et al[56], 2021	Italy	Retrospective cohort study	Outpatient	76	26	DOAC (rivaroxaban, apixaban, dabigatran)	-	-	NA	-	56.6	26.3	77.6	-	36.8
Schiavone et al[74], 2021	Italy	Retrospective cohort study	Inpatient hospitalized	844	65	DOAC or VKA	-	-	ICU admission	63.4 ± 16.1	38.3	16.6	45.1	7.4	9.2
Sivaloganathan et al[49], 2020	United Kingdom	Case-control study	Inpatient hospitalized	180	31	VKA (warfarin), DOAC (dabigatran, rivaroxaban, apixaban), or LMWH	-	-	ICU admission	-	-	-	-	-	-
Spiegelenberg et al[50], 2021	Netherlands	Retrospective cohort study	Inpatient hospitalized	1154	190	DOAC or VKA	-	AF, VTE, mechanical valve replacement, cardiac arrest in history, or unknown (5%)	ICU admission	-	32.2	27.3	53.8	26.1	-
Tehrani et al[69], 2021	Sweden	Retrospective cohort study	Inpatient hospitalized	255	49	-	-	-	NA	66 ± 17	41	31	54	13	-
Togano et al[39], 2021	Japan	Retrospective cohort study	Inpatient hospitalized	4026	105	VKA (warfarin), DOAC (dabigatran, rivaroxaban, apixaban, or edoxaban)	-	-	Mechanical ventilation/ supplemental oxygen/SPO2 ≤94% on room air/tachypnea	52.0 (34–69)	40.1	14.1	19	8.1	-
Tremblay et al[70], 2020	United States	Retrospective cohort study	Hospitalized and ambulatory patients	3772	241	-	-	-	Intubation-mechanical ventilation	56.6 (18.2)	45.2	-	-	14.9	-
van Haaps et al[51], 2021	Denmark	Retrospective cohort study	Inpatient hospitalized	3006	445	DOAC, VKA, LMWH	-	-	ICU admission	-	35.1	37.8	59.5	8.3	-
Wargny et al[65], 2021	France	Retrospective cohort study	Inpatient hospitalized	2796	501	-	-	-	NA	69.7 ± 13.2	36.3	100	76.8	9.6	-

Defined according to the 2019 clinical practice guideline from the Infectious Diseases Society of America and the American Thoracic Society for diagnosis and treatment of adults with community acquired pneumonia. DOAC: Direct-acting oral anticoagulants; VKA: Vitamin K antagonists; ARDS: Acute respiratory distress syndrome; NOAC: Non-Vitamin K oral anticoagulants; AF: Atrial fibrillation; VTE: Venous thromboembolism; DVT: Deep venous thrombosis; CAD: Coronary artery disease; NA: Not applicable; IQR: Interquartile range; OAC: Oral anticoagulant.

Table 2.

Studies reporting adjusted estimates and the factors for which they are adjusted

Ref.	Outcome	Adjusted estimates
Ageno et al[28], 2021	Mortality	Age, gender, and heparin use after admission, history of acute MI, T2D, HTN, cancer, COPD, renal function and CRP at hospital entry
Arachchillage et al[29], 2022	90-day mortality, thrombosis, and ICU admission	Age, gender, BMI, antiplatelet treatment prior to admission, autoimmune disease, malignancy, hypercholesterolaemia, heart disease, T2D, smoking status, liver disease, lung disease, existing renal failure and whether renal failure was dialysis dependent
Aslan et al[55], 2021	In-hospital mortality	Age, male gender, T2D, ferritin, d-dimer, neutrophil, lymphocyte, creatinine, CRP, SaO2, procalcitonin, DOAC, HTN, HF, AF, CAD, COPD, systolic blood pressure and hematocrit in univariable logistic regression analysis
Buenen et al[40], 2021	All-cause mortality within 30 d	Age, sex, symptom duration, home medication, and comorbidities
Covino et al[42], 2021	All-cause in-hospital death.	Age, sex, comorbidity (categorized as CCI < 3 or CCI ≥ 3), and illness severity at admission (categorized as NEWS < 6 or NEWS ≥ 6)
Gülcü et al[32], 2022	In-hospital all-cause mortality	Age, gender, HTN, DM, HF, CAD, eGFR, albumin, CRP, D-dimer, hemoglobin, platelet count, LDH, and oxygen saturation variables
Bauer et al[71], 2020	Severity	Age and gender
Ménager et al[60], 2020	7-day mortality	Age, sex, severe undernutrition, T2D, HTN, prior MI, HF, prior stroke and/or TIA, CHA2DS2-VASc score, HAS-BLED score, and eGFR
Philipose et al[66], 2020	Mortality	Age and gender
Rieder et al[37], 2022	COVID-19 related mortality	Age, gender, BMI and smoking status, the phase of disease at diagnosis, solid tumor, AF, CAD, prior MI, peripheral artery disease, HTN, cerebrovascular disease, and T2D
Rivera-Caravaca et al[38], 2021	Mortality and the composite of any thrombotic or thromboembolic event	Age, gender and ethnicity, all the included comorbidities
Rodríguez-Molinero et al[67], 2020	Mortality	Age, sex, obesity, and corticosteroids
Russo et al[48], 2022	Mortality	Age, arterial HTN, T2D, CAD, HF, previous stroke
Schiavone et al[74], 2021	Mortality	Age > 65 years, male gender, CAD, CKD, COPD, HF, OAC, PaO2/FiO2, hydroxychloroquine, tocilizumab, antivirals, heparin
Spiegelenberg et al[50], 2021	All-cause in hospital mortality and ICU admission	Age, sex, body mass index, active malignancy, COPD, T2D, HTN, CAD, MI, HF, non-ischaemic cardiomyopathy, previous heart surgery, electronic heart device, cerebrovascular accident, peripheral artery disease, immunosuppressive medication, no ICU policy
Togano et al[39], 2021	Severity (mechanical ventilation/ supplemental oxygen/SpO2 ≤ 94% on room air/tachypnea)	Age, gender, BMI, smoking, alcohol consumption, myocardial infarction/congestive heart failure, peripheral artery disease, cerebrovascular disease, dementia, paralysis, COPD, liver dysfunction, hypertension, hyperlipidemia, diabetes, obesity, leukemia, lymphoma, immunosuppression
Tremblay et al[70], 2020	All-cause mortality, mechanical ventilation	Age, gender, race, CCI and obesity
van Haaps et al[51], 2021	21-day all-cause mortality and ICU admission	Age, gender, T2D, HTN, CKD, asthma, obesity, time in pandemic, center, chronic cardiac disease, malignancy, liver disease, dementia, organ transplant, autoimmune disorder, and rheumatic disorder
Wargny et al[65], 2021	Death within 28 d	Age
Harrison et al[34], 2021	21-day all-cause in hospital mortality	Age, gender, and confounding variables
Iaccarino et al[72], 2021	Mortality, ICU admission	Age, multimorbidity (combined in the CCI score), and gender
Hozayen et al[35], 2021	Mortality	Age, sex, self-identified race/ethnicity (as a proxy for social, not biological risk factors), Elixhauser comorbidity score, and the presence/absence of any cardiovascular, immunological or hematological comorbidities
Ho et al[45], 2021	ICU admission, VTE, and mortality between date of SARS-CoV-2 diagnosis and 45 d after diagnosis	Age, sex, race/ethnicity, body mass index, CCI, HTN, T2D, and smoking history as well as the week of SARS-CoV-2 diagnosis
Denas et al[43], 2021	ICU admission and all-cause mortality	Age, sex, HF, HTN, cancer, T2D, history of stroke/TIA, previous bleeding, history of MI, peripheral artery disease, abnormal renal function, abnormal hepatic function, use of antiplatelet drugs, NSAIDs and statin use
Corrochano et al[30], 2022	All-cause mortality and ICU admission	Sex, age, CCI, and antithrombotic therapy
Chocron et al[41], 2021	In-hospital mortality and ICU admission	Sex, age, cardiovascular comorbidities (history of HTN, dyslipidemia, BMI, T2D, and current smoking), plasma creatinine level (µmol/L), CRP (mg/L), fraction of inspired oxygen, the degree of pulmonary lesions with ground-glass opacities and areas of consolidation, and the use of in-hospital anticoagulation (preventive low or high dose and therapeutic dose)

T2D: Type 2 diabetes mellitus; HTN: Hypertension; HF: Heart failure; CAD: Coronary artery disease; MI: Myocardial infarction; AF: Atrial fibrillation; TIA: Transient ischemic attack; CKD: Chronic kidney disease; COPD: Chronic obstructive pulmonary disease; VTE: Venous thromboembolism; OAC: Oral anticoagulant; DOAC: Direct oral anticoagulant; CRP: C-reactive protein; LDH: Lactate dehydrogenase; BMI: Body mass index; CCI: Charlson comorbidity index; NEWS: National Early Warning Score; ICU: Intensive care unit; NSAIDs: Non-steroidal anti-inflammatory drugs; eGFR: Estimated glomerular filtration rate; SARS-CoV-2: Severe acute respiratory syndrome coronavirus 2; COVID-19: Coronavirus disease 2019.

Out of the 46 studies included in our meta-analysis, 26 studies[28-54] described using both direct oral anticoagulants (DOACs) and vitamin K antagonists (VKAs). Five studies[55-59] had only DOAC as the prehospital anticoagulant, while a single study by Ménager et al[60] included patients only on VKAs. The anticoagulant type was unspecified in the remaining 13 studies[36,61-72]. In the 32 studies[28-35,37,39-51,53-60,73,74] studying DOACs Apixaban, Dabigatran, Edoxaban, and Rivaroxaban were the most commonly used, while on the other hand, VKAs were represented mostly by Warfarin. However, in two studies[30,60], Acenocoumarin was also taken into consideration. Supplemental prehospital use of other anticoagulants, including LMWH, Fondaparinux, and Enoxaparin, was observed in five studies[30,35,47,49,51].

Quality assessment and publication bias

The included studies’ methodological quality assessment demonstrated that 21 studies[25,28,30,34,36,38,39,43,44,50,54,56,58-60,63,66,68,69,71,72,73] had a high quality with very low risk of bias, while 24 studies[29,31-33,35,37,40-42,46-49,51,55,57,62,64,65,67,70,73,74] were of moderate quality (Supplementary Table 2). Out of the 24 studies with moderate quality, 11 studies[29,33,35,37,41,46-49,52,62] had an unclear risk of bias (not enough information to make a clear judgment), and the remaining 13 studies[31,32,40,46,55,57,64,65,67,70,73,74] had a low risk of bias, but some other potential flaws. Therefore, every study was approved for inclusion in the quantitative analysis. A single study was of poor quality and potentially high risk of bias[53]. Supplementary Figure 1 displays the publishing bias funnel plots. The results showed no discernible publication bias, and Supplementary Figure 1 displays the various P-values from Egger's regression test. We also evaluated the certainty of evidence at the outcome level using GRADE pro profiler (GRADE working group, McMaster University, and Evidence Prime Inc; Supplementary Table 3)[20].

Primary outcome

Thromboembolic event rate: A total of nine studies[29,30,37,40,45,50,54,60,70] evaluated the unadjusted risk, while six studies[29,45,50,54,58,61]calculated the adjusted risk between prehospital anticoagulation and new thromboembolic events in COVID-19 infected patients. Three studies[45,50,62] defined thromboembolic events as VTE, while three others[29,30,37] included both arterial and venous thromboembolic events, and the remaining three[40,70] did not clarify their definition of thromboembolic events.

Overall, anticoagulation prior to COVID-19 diagnosis suggested a reduction in the unadjusted risk of new VTE in adult patients in these nine studies but did not attain statistical significance [n = 43851, OR = 0.67 (0.22, 2.07); P = 0.49; I² = 95%; Figure 2A]. Similarly, in the predefined subgroup analysis, the use of VKA[40,50,54] [2658 participants, OR = 0.32 (0.05, 1.98); P = 0.22; I² = 63%; Supplementary Figure 2A] or DOAC[40,50,54] [2699 participants, OR = 0.36 (0.10, 1.25); P = 0.11; I² = 51%; Supplementary Figure 2B] or any anticoagulants [40960 participants, OR = 1.03 (0.26, 4.08); P = 0.97; I² = 96%; Supplementary Figure 2C] did not show a statistical significant reduction of new VTE event risk associated with COVID-19 infection (P value for subgroup differences = 0.70, I² = 0%). Our prespecified adjusted meta-analysis on six studies[29,45,50,54,61,74] did not achieve statistical significance in the lower odds of thromboembolic events in patients suffering from COVID-19 [aOR = 0.87 (0.42, 1.80); P = 0.71; I² = 99%; Figure 3A].

Figure 2.

Unadjusted meta-analysis for mortality, severity thromboembolic events in prehospital use of anticoagulants vs control cohort in COVID-19. A: Unadjusted thromboembolic events in prehospital use of anticoagulants vs control cohort; B: Unadjusted mortality in prehospital use of anticoagulants vs control cohort; C: Unadjusted severity in prehospital use of anticoagulants vs control cohort.

Figure 3.

Adjusted meta-analysis for mortality, severity thromboembolic events in prehospital use of anticoagulants vs control cohort in COVID-19. A: Adjusted thromboembolic events in prehospital use of anticoagulants vs control cohort; B: Adjusted mortality in prehospital use of Anticoagulants vs control cohort; C: Adjusted severity in prehospital use of Anticoagulants vs control cohort.

Secondary outcomes

Mortality: A total of 35[28-35,37,38,40,41,43,44,46-48,50,51,53,55-57,59,60,63-70,72,74] studies reported mortality in prior anticoagulant users (2388 deaths out of 110975 patients; 2.1%) vs non-users (9457 deaths out of 96317 patients; 9.8%). Prior anticoagulation was found to significantly increase the risk of mortality in COVID-19 patients in the unadjusted analysis [OR = 1.72 (1.37, 2.17); P < 0.00001; I² = 93%; Figure 2B). Subgroup analysis by the type of anticoagulation medication showed that prior VKA use [19747 participants, OR = 1.91 (1.20, 3.06); P = 0.007; I² = 83%; Supplementary Figure 3A] and any anticoagulant use [43643 participants, OR = 1.88 (1.40, 2.52); P < 0.00001; I² = 94%; Supplementary Figure 3B] was associated with an increased mortality risk. However, prior use of DOAC [22374 participants, OR = 1.42 (0.95, 2.12); P = 0.08; I² = 87%; Supplementary Figure 3C] and the adjusted estimates of 28 studies[28,30,32,34,35,37,38,40-45,48,50,51,53,55,57,60,61,65-67,70,72,74] [aOR = 0.94 (0.84, 1.05); P = 0.31; I² = 65%; Figure 3B] revealed no statistically significant association between prehospital anticoagulation and mortality in patients hospitalized for COVID-19.

COVID-19 disease severity: Overall, 22 studies[30,31,33,36,39,41,43,47-51,55,56,59,62,64,67,70,72,73,74] documented the association between prehospital anticoagulation and COVID-19 infection severity. ICU admission was the surrogate marker of COVID-19 severity in 15 studies, while mechanical ventilation was used in five studies, and one study used ARDS development during hospitalization. Six hundred seventy-one patients out of 103,703 who received anticoagulants before COVID-19 diagnosis developed severe COVID-19 infection, while 6155 out of 78890 non-users progressed to severe COVID-19 illness. We derived no statistically significant association between prehospital anticoagulation and COVID-19 disease severity [OR = 1.08 (0.78, 1.49); P = 0.64; I² = 89%; Figure 2C]. On carrying out a subgroup analysis based on the type of anticoagulants, VKAs [6947 participants, OR = 1.26 (0.57,2.77); P = 0.57; I² = 82%; Supplementary Figure 4A], DOACs [149564 participants, OR = 1.12 (0.58, 2.15); P = 0.74; I² = 88%; Supplementary Figure 4B], and any anticoagulant use [36854 participants, OR = 1.07 (0.72, 1.58); P = 0.75; I² = 89%; Supplementary Figure 4C] all reported non-significant association with severe COVID-19 disease. Similarly, prehospital anticoagulation was not significantly associated with COVID-19 disease severity in the adjusted analysis [12 studies[29,30,39,41,43,45,50,51,59,70-72]; aOR = 0.96 (0.72, 1.26); P = 0.76; I² = 80%; Figure 3C].

Multivariate meta-regression model for mortality outcome

To account for variations in the correlation between mortality and prehospital anticoagulant use, a multivariate meta-regression was performed. The findings demonstrated that when considered collectively, the proportion of age, diabetes, female gender, hypertension, and pulmonary diseases was significant. These variables accounted for R² = 90% of the difference in mortality between studies (Figure 4A).

Figure 4.

Meta-regression analyses. A: Mortality meta-regression analysis; B: Severity meta-regression analysis.

Multivariate meta-regression model for severity outcome

Multivariate meta-regression was employed to take into consideration variations in the correlation between COVID-19 severity and prehospital anticoagulation. Together, age, female gender, and the proportion of hypertension, pulmonary disease, and diabetes were found to be significant covariates. Figure 4B demonstrates that these factors, all together, explained R² = 100% for heterogeneity in severity among the included studies.

DISCUSSION

Our analysis offered a thorough examination of the risk of thromboembolic events, mortality, and severity of COVID-19 disease when anticoagulant medication was used prior to hospitalization. The meta-analysis identified no significant association between prehospital anticoagulation and decreased thromboembolic events risk and reduced COVID-19 disease severity. Though the unadjusted analysis exhibited a rise in mortality risk of COVID-19 patients with prior anticoagulation, adjusting the estimates revealed no significant difference in the odds of thromboembolic events, mortality, and severity between COVID-19 infected patients on prehospital anticoagulation and those without antecedent anticoagulation treatment.

Literature review and analysis of existing research

VTE: Coagulopathy and thromboembolic events are described as independent poor prognostic indicators in COVID-19[58,75]. Hospitalized COVID-19 patients tend to have a higher incidence of VTE than individuals with other illnesses[76], yet the exact mechanism of hypercoagulability remains unclear. COVID-19 infection can induce hyperinflammation[77], leading to endothelial dysfunction[78-80], platelet activation, blood stasis[58], and microvascular inflammation[81], all of which can influence consequent respiratory distress and other organ dysfunctioning[78-80]. This acute inflammatory state also increases the arterial and venous TE risk[61].

While ample evidence supports anticoagulation benefits during hospitalization[82-84], the role of prehospital anticoagulation before COVID-19 diagnosis, particularly in thromboembolic event incidence, lacks sufficient data. Despite anticoagulation, people on long-term oral anticoagulation therapy that started prior to COVID-19 infection may be more prone to thrombosis because of the existing SARS-CoV-2 infection[45]. Initial research suggests potential negative effects of VKAs on COVID-19[50]. SARS-CoV-2 infected individuals have reduced extra-hepatic vitamin K stores, which could be further decreased with VKAs[50]. However, studies have produced conflicting results regarding prehospital anticoagulation's impact on COVID-19 patients' thromboembolic risk, with some studies showing increased thromboembolic event risk[58]. In contrast, others observed benefits from prehospital anticoagulation in reducing COVID-related TE risk[29,50]. However, as shown in both unadjusted and adjusted analyses, our study did not find a significant reduction in thromboembolic events among COVID-19 patients receiving prehospital anticoagulation.

Mortality: Diverse study outcomes exist on chronic anticoagulation's impact on COVID-19 mortality. Tremblay et al[70] noted higher mortality among prehospital anticoagulated COVID-19 patients, yet non-significance emerged after age, sex, race, obesity, and Charlson Comorbidity Index adjustment. Rivera-Caravaca et al[58] reported higher all-cause mortality rates in prehospital DOAC-treated patients, remaining significant post-adjustment. Conversely, studies like Fumagillin et al’s[44] observed lower mortality rates in pre-COVID-19 anticoagulated patients[41,43,44,68]. Our research aligns with studies[41-51,69] and an earlier meta-analysis published by Kamel et al[85], revealing no substantial prehospital anticoagulation-associated mortality impact. Unlike Kamel et al[85], our study comprehensively explores pre-admission anticoagulation effects on VTE risk, severity of disease, and mortality in COVID-19.

Severity: Divergent results have been observed regarding the effect of anticoagulation on the severity of COVID-19[47,48,52,64,66,74]. Aslan et al[55] revealed significantly higher ICU admission, ventilation, oxygen therapy, and mortality rates with DOAC use. Yet, Corrochano et al’s[30] study found ICU admission reduction with prehospital anticoagulation after age, sex, and CCI adjustment. Similarly, Iaccarino et al[72] linked oral anticoagulants to ICU admission risk reduction. However, their cross-sectional design for hypothesis generation might influence outcomes[33].

Assessment of potential biases

The existence of selection bias and confounding bias is one possible explanation for inconsistent results among studies[29,30,49,55,58], as some studies include patients with cardiovascular or coagulation issues (poorer premorbid state). Certain studies neglect the effects of other medications during hospitalization[84]. Another reason can be the nature of the study, as observational and retrospective studies limit the causal inference[27-30,55], even though propensity score matching efforts[58], as propensity scoring is dependent on covariates and confounders are not included in the scoring, resulting in a potential bias. Small sample size also introduces bias[28,55]. The focus on hospitalized patients can hamper the generalizability of milder and asymptomatic COVID-19 cases. Similarly, heterogeneous sample sizes impact data collection and yield inconsistent results. Socio-demographics, like age and gender, may also influence drug effects on severity and mortality.

Possible mechanisms underlying the findings

A hypercoagulative state is promoted by COVID-19-induced coagulative abnormalities and alterations in prothrombotic factors, such as enhanced factor VIII, plasma fibrinogen, microparticles, and increased platelet activation[86-89]. This hypercoagulative state, known as COVID-19-associated coagulopathy (CAC), differs from acute disseminated intravascular coagulation (DIC) in presentation, involving thrombosis instead of bleeding[90]. Although D-dimer levels are elevated in both CAC and DIC, additional coagulation factors differentiate CAC (higher levels of Von Willebrand factor (vWF) antigen, Factor VIII activity, and fibrinogen) from DIC (lower levels of fibrinogen, reduced Factor VIII activity, and decreased VWF)[88-91].

Evidence suggests coagulation factor Xa's role in virus entry by cleaving SARS Spike proteins, which certain anticoagulants inhibit, potentially affecting viral fusion with ACE-2 receptors[10]. Ageno et al[28] used this concept to show DOACs' potentially higher antiviral efficacy over VKAs, though with limitations. Meanwhile, in our meta-analysis, DOAC-treated cases showed more severe disease progression than VKA-treated ones.

The underlying pathophysiology of endothelial damage by SARS-COV-2 is thought to be related to the renin-angiotensin-aldosterone system (RAAS). Virus-host cell binding via ACE-2 receptors raises disease severity risk. Thus, ACEI and ARBs (RAAS inhibitors) could heighten disease severity by increasing ACE2 receptor expression[16]. Anticoagulants, not affecting ACE2 receptors, might explain the non-significant enhancement in disease severity, as seen in our meta-analysis with no prehospital anticoagulation-COVID-19 severity correlation.

Clinical implications and future directions

Our study has significant clinical implications. Monitoring coagulation function in ICU patients using repeated platelet count, prothrombin time, and D-dimer measurements is vital. Elevated serum D-dimer predicts VTE and is a prognostic tool for COVID-19 risk stratification[92,93]. High D-dimer or rapid respiratory decline may indicate suspected VTE. Middeldorp et al[62] noted significantly higher D-dimer levels in ICU-admitted COVID-19-infected patients, regardless of chronic prehospital anticoagulation. Our analysis found that chronic anticoagulation does not significantly reduce the risk of new thromboembolic events in COVID-19 patients. This suggests that prior anticoagulation does not protect against COVID-19-related thromboembolic events. Due to the controversial effects of anticoagulant therapy on COVID-19 severity, further studies are needed.

Strengths and limitations

This meta-analysis's strengths include a sizable sample size dispersed across several nations and moderate to high-quality studies included according to the New-Castle Ottawa Scale. However, there are notable limitations. First, the retrospective nature of most included research limits the generalizability of the conclusions, highlighting the need for expansive prospective investigations. Second, while adjusted estimates were prioritized in the original analysis, selection bias and the confounding impact typical of observational research cannot be completely eliminated. Furthermore, inconsistency existed in the definition of severity among the included articles. Lastly, data gaps regarding prehospital anticoagulation specifics—duration, indication, type, and dosage—in some studies impeded comprehensive analysis.

CONCLUSION

The current meta-analysis concludes that prehospital anticoagulation does not significantly correlate with reduced COVID-19-related thromboembolic events, enhanced survival, or lowered disease severity risk. This aligns with recent guidelines advocating prophylactic anticoagulant use in COVID-19 patients, irrespective of VTE risk. To gain deeper insights and robust evidence, we suggest well-designed prospective studies and randomized trials investigating the impact of prior anticoagulant usage on thromboembolic risk, mortality, and disease severity in COVID-19 cases. Furthermore, a thorough exploration of the reasons behind the limited efficacy of chronic anticoagulants in severe infections is warranted. Future research should focus on determining personalized VTE risks for COVID-19 patients, uncovering underlying pathogenic pathways, and identifying optimal anticoagulant interventions for VTE prevention.

ACKNOWLEDGEMENTS

Data from this study was presented as an abstract format for the SCCM 2023 Critical Care Congress in San Francisco, California, United States.