Keywords
thrombosis - ambulatory - outpatients - postdischarge - COVID-19
Introduction
Novel coronavirus disease 2019 (COVID-19) caused by the severe acute respiratory syndrome-coronavirus-2
(SARS-CoV-2) leads to endothelial and coagulation dysfunction which puts patients
at an increased risk of both venous and arterial thrombotic events and also increases
the morbidity and mortality associated with the disease.[1] Hospitalized COVID-19 patients are at a particularly high risk of thrombosis with
estimates of venous thromboembolism (VTE) risk up to 8% in hospitalized patients and
24% in patients admitted to intensive care, according to two large systematic reviews.[2]
[3] Arterial events are estimated to occur in 1% of ward patients and 5% of critically
ill patients.[4] However, the majority of published studies have focused on the incidence of thromboembolism
during hospitalization, and it is unclear what the thrombotic risk is in outpatients
who have a milder illness and do not require hospitalization and among patients who
have been discharged after hospitalization for COVID-19 infection. This information
could be helpful in deciding the need for thromboprophylaxis in these patient populations.
Methods
This review was performed in accordance with the Preferred Reporting Items for Systematic
Reviews and Meta-analysis (PRISMA) guidelines ([Fig. 1]) and is registered in PROSPERO (
https://www.crd.york.ac.uk/PROSPERO
; registration number: CRD42021292010).
Fig. 1 Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) flow
diagram for study selection. CTPA, computed tomography pulmonary angiography.
The literature search was performed using EMBASE and MEDLINE using medical subject
headings and free text words through Boolean operators. The retrieved papers were
also scanned for additional references in addition to searching preprint databases
(preprints.org, biorxiv.org) for accepted papers not yet published. Full details on the search strategies of
the main databases can be found in the [Supplementary Material S1].
We included studies that looked at venous or arterial events in adult patients with
COVID-19 either postdischarge, following hospitalization or ambulatory with no history
of hospitalization. Full-text articles, abstracts, letters, brief reports, editorials,
and correspondence were eligible for inclusion if they reported on randomized control
trials (RCTs), observational cohort studies (prospective or retrospective), or case-control
studies. We excluded studies with no original data, studies with less than 100 patients
and studies that would not reflect the general risk of thrombosis in COVID-19 patients
(e.g., studies that only included patients with computed tomography pulmonary angiogram
(CTPA)). There was no language restriction.
We conducted an initial broad screening according to title and reviewed all abstracts
judged to be relevant. Potentially relevant papers were reviewed in total length.
Articles were independently assessed by two reviewers (E.M.M. and M.A.-F) to verify
eligibility and decide on inclusion vs exclusion. Disagreements were resolved by consensus
or in conjunction with a third reviewer (A. L.-L.). Translation of included papers
from Dutch to English was conducted using Google Chrome's built-in translation tool.
We extracted data on study characteristics, study methodology (including sample size,
study design, health care setting, and ultrasound screening strategy), patients and
diseases (hospitalization, thrombosis risk factors, and thromboprophylaxis strategy),
and outcomes including (venous thromboembolism or arterial thrombosis.
The study aimed to look at the proportion of venous thromboembolism (i.e., deep venous
thrombosis and pulmonary embolism) or arterial thrombosis in outpatients or postdischarge
patients with COVID-19 infection as the primary outcome. We used the proportion of
patients diagnosed with a thrombotic event in the included studies (prevalence) to
compensate for the differences in event reporting between studies. Some studies used
prevalence, while others used incidence rates depending on the study type. If needed,
authors were contacted to obtain missing data. Additionally, we also estimated incidence
rates if possible.
The quality and risk of bias of included observational studies were assessed using
the Newcastle–Ottawa Scale (NOS) to evaluate the quality of nonrandomized studies
in meta-analyses.[5] For randomized trials, we planned to use the Jadad scale.[6]
Statistical Analysis
A meta-analysis of proportions was done for the frequency of VTE and arterial thrombosis.
We estimated pooled proportions through a Freeman–Tukey transformation, using fixed
and random effects models. We also estimated incidence rates (IR) from imperfect data
assuming no losses to follow-up for the reported observation periods. We applied standard
methods to obtain 95% confidence intervals (CIs) for the IR that we used to estimate
standard errors. We calculated pooled estimates for the IR using the generic inverse
variance method.[7]
The reported pooled proportions were obtained by a random-effects model given statistical
heterogeneity. Sensitivity analyses were performed according to setting (outpatients
vs. postdischarge), thrombosis type (VTE vs. arterial), and study type (retrospective
vs. prospective). Additionally, we conducted a meta-analysis of the relative risk
for VTE for the studies that included both postdischarge and outpatients. Exploratory
analyses were also conducted according to percentage of use of thromboprophylaxis
(<20 vs. >20%) and duration of follow-up in the study (<42 vs. > 42 days). Heterogeneity
between studies was evaluated using Cochrane Q and Higgins I
2 tests. Publication bias was assessed using Egger's tests and funnel plots. The analysis
was done using Microsoft Excel 365 (Microsoft Corp. Redmond, Washington, United States)
and MedCalc Statistical Software version 19.2.6 (MedCalc Software Ltd., Ostend, Belgium).
Results
Search Strategy and Included Studies
The database search included papers published up to July 21, 2021. The search produced
543 studies through Medline and Embase and an additional two articles from other sources.
A total of 368 studies were excluded after title and abstract screening, and a total
of 26 articles were reviewed in full text. Of those, 16 studies met our eligibility
criteria and were included. The reasons for excluding studies are summarized in [Fig. 1]. All the included studies were cohort studies (11 retrospective and 5 prospective).
All studies were in English except one study published in Dutch.[8]
The studies were conducted in North America (seven in the United States and one in
Canada), Europe (three from the United Kingdom and one each from Norway, the Netherlands,
Denmark, and Belgium), and there was one multicenter study. Ten studies reported events
in postdischarge patients, three reported on outpatients, and three reported on both
outpatients, and postdischarge. The follow-up duration in the studies ranged from
30 days up to 6 months.
Only one study performed regular VTE screening at 6-week postdischarge.[9] With regard to thromboprophylaxis, many studies used some form of thromboprophylaxis
on an individual basis or according to risk scores.[10]
Characteristics of included studies are shown in [Table 1]. Details of the included papers and a detailed reference list can be found in the
[Supplementary Tables S1] and [S2].
Table 1
Characteristics of included studies
|
Characteristics
|
No of studies
|
|
Country
|
|
|
North America
|
8
|
|
Europe
|
7
|
|
Multicenter study
|
1
|
|
Study design
|
|
|
Prospective cohort
|
5
|
|
Retrospective cohort
|
11
|
|
Setting
|
|
|
Outpatients only
|
3
|
|
Postdischarge only
|
10
|
|
Both
|
3
|
|
Events reported
|
|
|
VTE events only
|
8
|
|
Both Venous and arterial
|
8
|
|
Thromboprophylaxis strategy
|
|
|
Prophylaxis used
|
7
|
|
No prophylaxis used
|
9
|
Abbreviation: VTE, venous thromboembolism.
All estimates showed high statistical heterogeneity; therefore, we focused our analysis
on the results of random effects estimates, although the fixed effects estimates are
shown for completeness when appropriate. Funnel plots are included in the [Supplementary Figs. S1] to [S8].
Venous Thromboembolic Events
The results are summarized in [Table 2]. The 16 included studies reported on 102,779 patients diagnosed with COVID-19 infection
treated as either outpatients (85,051) or posthospital discharge (17,728). In total,
388 patients had a VTE event with a pooled prevalence estimate of 0.80% (95% CI:0.44–1.28;
Q-test: p < 0.0001, I
2 = 97.33%).
Table 2
Estimates of the proportion of venous and arterial thrombotic events in ambulatory
and postdischarge patients with COVID-19 infection
|
Patient population
|
No. of total patients
|
Percentage of patients with VTE (%)
|
95% confidence interval
|
|
Venous thromboembolic events
|
|
All patients
|
102,779
|
0.80
|
0.44–1.28
|
|
Postdischarge
|
17,728
|
1.16
|
0.69–1.74
|
|
Outpatients
|
85,051
|
0.28
|
0.07–0.64
|
|
Subanalysis according to study design
|
|
Prospective studies
|
14,904
|
1.37
|
0.48–2.69
|
|
Retrospective studies
|
87,875
|
0.62
|
0.25–1.14
|
|
Arterial events
|
|
All patients
|
20,187
|
0.75
|
0.27–1.47
|
|
Postdischarge
|
7,816
|
1.46
|
1.10–1.86
|
|
Outpatients
|
12,371
|
0.23
|
0.02–0.66
|
|
Subanalysis according to study design
|
|
Prospective studies
|
13,889
|
0.71
|
0.01–3.09
|
|
Retrospective studies
|
6,298
|
0.78
|
0.34–1.39
|
Abbreviations: COVID-19, novel coronavirus disease 2019; VTE, venous thromboembolism.
In the outpatient population, 141 patients developed a VTE event with a pooled prevalence
estimate of 0.28% (95% CI: 0.07–0.64, p < 0.0001, I
2 = 97.45%). Among the postdischarge patients, 247 developed a VTE event with a pooled
prevalence estimate of 1.16% (95% CI: 0.69–1.73; Q-test: p < 0.001, I
2 = 87.21%).
Prospective studies included 14,904 patients with a pooled VTE prevalence estimate
of 1.37% (95% CI: 0.48–2.69; Q-test: p < 0.001, I
2 = 94.75%) and retrospective studies included 87,875 patients with a pooled VTE prevalence
estimate of 0.62% (95% CI: 0.25–1.14; Q-test: p < 0.001, I
2 = 97.63%).
Estimates for the pooled IR of VTE and arterial events are shown in [Table 3]. Funnel and forest plots are included in the [Supplementary Figs. S9] to [S16]. The estimated pooled IR for VTE events in outpatients was 0.06 (95%CI: 0.03–0.08)
events per 1,000 patient-days of observation and the estimate for postdischarge patients
was 0.12 (0.07–0.19).
Table 3
Pooled estimates for incidence rate of venous and arterial thrombotic events in ambulatory
and postdischarge patients with COVID-19 infection
|
IR (events per 1000 patient-days)
|
Heterogeneity
|
|
IR
|
95% CI
|
Q-test (p)
|
Higgins I
2 (%)
|
|
Venous thromboembolic events (outpatients)
|
|
Fixed effects
|
0.02
|
0.02–0.03
|
<0.001
|
95.5
|
|
Random effects
|
0.06
|
0.03–0.08
|
|
|
|
Venous thromboembolic events (postdischarge)
|
|
Fixed effects
|
0.06
|
0.04–0.07
|
<0.001
|
82.7
|
|
Random effects
|
0.12
|
0.07–0.19
|
|
|
|
Arterial events (outpatients)
|
|
Fixed effects
|
0.01
|
0–0.01
|
0.001
|
93.2
|
|
Random effects
|
0.10
|
0–0.30
|
|
|
|
Arterial events (postdischarge)
|
|
Fixed effects
|
0.21
|
0.17–0.25
|
0.182
|
35.8
|
|
Random effects
|
0.26
|
0.16–0.37
|
|
|
Abbreviations: CI, confidence interval; COVID-19, novel coronavirus disease 2019;
IR, incidence rates.
A meta-analysis conducted for studies including both discharged patients and outpatients
showed that, compared with the latter, the pooled relative risk of VTE among discharged
patients was 3.87 (95% CI: 0.38–39.18; p = 0.252) with a high heterogeneity (Cochrane's Q = 128.34 (p < 0.001); Higgins' I
2 = 96.88 (95% CI: 94.82–98.13; [Fig. 2]). Additional comparisons according to study design showed that in prospective studies
the relative risk for discharged patients compared with outpatients was 0.46 (95%
CI: 0.12–1.72; p = 0.246; Q-test: p = 0.498; I
2 = 0%; n = 890). In comparison, the relative risk reported in retrospective studies was 11.31
(95% CI: 0.57–224.85; p = 0.112; Q-test: p < 0.001, I
2 = 98.23; n = 82,666). Estimates of VTE events according to the reported percent of utilization
of thromboprophylaxis in studies, as well as duration of follow-up showed similar
results among groups ([Supplementary Table S3]).
Fig. 2 Forest plot for the risk of venous thromboembolism in discharged patients compared
with outpatients. Forest plot showing the pooled relative risk estimate of VTE for
COVID-19 patients discharged from hospital compared with outpatients. The pooled relative
risk was 3.87 (95% CI: 0.38–39.18; p = 0.252) with a high heterogeneity (Cochrane's Q = 128.34 [p < 0.001]; Higgins' I
2 = 96.88 (95% CI: 94.82–98.13). CI, confidence interval; COVID-19, novel coronavirus
disease 2019; VTE, venous thromboembolism.
Arterial Events
Studies that evaluated arterial events included a total of 20,187 patients (including
both outpatients and postdischarge) with an overall proportion of arterial events
of 0.75 (95% CI: 0.27–1.47; Q-test: p < 0.001, I
2 = 94.52%). In the outpatient population (12,371 patients) arterial thromboses occurred
in 0.23% (95% CI: 0.019–0.66; Q-test: p < 0.001, I
2 = 88.85%). Among 7,816 postdischarge patients, arterial events occurred in 1.46%
(95% CI: 1.10–1.86; Q-test: p = 0.242, I
2 = 25.57%).
Retrospective studies included 6,298 patients with a pooled proportion of arterial
events of 0.77% (95% CI: 0.34–1.39; Q-test: p = 0.001, I
2 = 77.81%), while prospective studies included 13,889 patients with a pooled proportion
of arterial events of 0.71% (95% CI: 0.01–3.09; Q-test: p < 0.0001, I
2 = 99.11%).
Estimates for the pooled IR of arterial events per 1,000 patient-days of observation
were 0.10 (95%CI: 0–0.303) in outpatients and 0.26 (95% CI: 0.16–0.37) in postdischarge
patients, respectively. Given a lack of data, we were unable to conduct a meta-analysis
comparing discharged patients to outpatients.
Discussion
Hospitalization for COVID-19 infection is associated with an increased risk of thrombosis,[2]
[3] but whether the COVID-19 infection increases the risk of VTE outside of the hospital
setting is not well established. In this systematic review, we found that patients
with COVID-19 infection not severe enough to require hospitalization or those, who
are discharged and not in the acute phase anymore, do not have a markedly increased
risk of thrombosis, both venous and arterial and in the case of postdischarge patients,
their risk might be similar to other medically ill outpatients after hospital discharge.
The MARINER trial looked at the benefit of using Rivaroxaban 10 mg. daily for thromboprophylaxis
after hospital discharge in medical patients at high risk of VTE based on a score
of ≥4 in the modified IMPROVE-VTE score or ≥2 plus D-Dimer level > 2 ULN (upper limit
of normal) at discharge. In this study, the occurrence of symptomatic VTE in the placebo
and intervention groups were 1.10 and 0.83%, respectively.[11] For comparison, the occurrence of VTE events is more than double (up to 2.8%) in
patients undergoing major orthopedic surgery which is an accepted indication for anticoagulant
prophylaxis.[12] More recently, the benefit of anticoagulant prophylaxis to reduce mortality in populations
at risk for VTE has been questioned as the use of anticoagulants is associated to
a higher risk of bleeding counterbalancing the relevance of VTE prevention.[13] Furthermore, our results showed that compared with fully ambulatory patients, the
relative risk for VTE among discharged patients was not statistically significant,
despite being numerically higher. Additionally, exploratory analyses showed similar
proportion of VTE in patients according to the reported proportion of use of thromboprophylaxis
in individual studies. These and the previously mentioned findings are important,
since the use of anticoagulant prophylaxis in all these patients would be of doubtful
benefit with the reported rates of events and uncertainty about the increased risk
of VTE, and it would likely result in a higher risk of bleeding complications.
The findings of our review are consistent with other reports and expand on the information
reported by another systematic review that also looked at the risk of VTE in postdischarge
COVID-19 patients and reported a VTE risk of 1.8% (95% CI: 0.8–4.1%, I
2 = 96.0%),[14] in agreement with our findings. We believe that the difference in calculated risk
stems from including two recent large cohorts in our systematic review that were not
included in the previous one.[15]
[16] Additionally, our study reports on pooled estimates of incidence rates which also
demonstrate a low IR of venous and arterial events.
Although some studies comparing COVID-19-positive with COVID-19-negative patients
have reported a higher risk of thrombosis in COVID-19 patients compared with other
medical patients, the results were not statistically significant. One study assessed
the occurrence of postdischarge VTE in patients with COVID-19 compared with a 2019
control cohort and reported an odds ratio (OR) of 1.6 (95% CI: 0.8–3.1).[17] Another study that looked at the incidence of complications in the post–acute phase
of SARS-CoV-2 infection in outpatients, comparing SARS-CoV-2-positive with SARS-CoV-2-negative
individuals, found the VTE risk to be slightly higher in COVID-19 infection with 0.2
versus 0.1% (risk difference = +0.1% [95% CI: 0.0–0.2]; risk ratio = 1.77 [95% CI:
1.09–2.86]).[18] Pasha et al showed that the incidence of VTE increased within the first week following
positive COVID-19 testing and the rate then decreased and returned to baseline by
the sixth week; however, in this study, only 3% of VTE events occurred in patients
who were never hospitalized.[15] In ambulatory patients, a randomized trial including 657 symptomatic outpatients
with COVID-19 evaluated the use of different thromboprophylaxis strategies compared
with placebo.[19] The study was terminated early because of a very low event rate. Among randomized
participants who were on aspirin (81-mg once daily), apixaban (2.5-mg twice daily),
apixaban (5.0-mg twice daily), or placebo, the rates of composite outcomes (including
symptomatic venous or arterial thromboembolism, or hospitalization for cardiac or
pulmonary cause) after 45 days were 0.0, 0.7, 1.4, and 0.0%, respectively, with no
significant differences between the active groups and the placebo group. Interestingly,
a study on outpatients with COVID-19 who were on outpatient anticoagulation at the
time of diagnosis reported a reduction in the risk of hospitalization but not mortality.[20]
It is estimated that arterial events complicate 3.9% of COVID-19 admissions, although
the amount of literature concerning arterial events is more limited.[21] Similarly, the burden in non-COVID-19 patients is unknown. A study investigating
the rates of myocardial infarctions and stroke following hospital admission for pneumonia
found an incidence of 2.5% for myocardial infarction and 0.2% for stroke.[22] The estimate for the risk of cardiovascular events in general medically ill outpatients
was unclear.
It has been suggested that the administration of low molecular weight heparin during
the earlier phases of COVID-19 infection may be beneficial not only in the prevention
of thrombosis risk but also in reducing systematic and pulmonary inflammation and
limiting viral invasion.[23]
[24] A randomized controlled trial, the OVID trial (ClinicalTrials.gov Identifier: NCT04400799),
is currently being conducted. This study aims to evaluate whether prophylactic-dose
enoxaparin (vs. no treatment) can decrease early all-cause mortality and subsequent
hospitalizations in adult symptomatic ambulatory COVID-19 patients with no other indications
to receive anticoagulation.[25]
The use of thromboprophylaxis in postdischarge patients has been explored previously
in patients with medical illness by the MARINER trial which showed no reduction in
the risk of symptomatic venous thromboembolism and death due to venous thromboembolism
compared with placebo.[11] A recent randomized control study applied the IMPROVE-VTE score on discharged patients
posthospitalization with COVID-19 and reported that high-risk patients with a score
of ≥4, or 2 to 3 with a D-dimer >500 ng/mL, had improved outcomes when on thromboprophylaxis
with rivaroxaban 10 mg/day for 35 days.[26] This study used a composite outcome including symptomatic, asymptomatic, and fatal
VTE (all detected by mandatory screening), as well as arterial thromboembolism including
myocardial infarction, stroke, adverse limb events, and cardiovascular death. The
generalizability of these findings is questionable for several reasons. The study
accrued only 30% of potentially eligible patients, and more than half of the patients
in this study had been admitted to the intensive care unit suggesting that the population
included a larger than usual proportion of patients with severe disease. Additionally,
the reported proportion of events in this study amounted to near 10% in the control
arm which is not in agreement with large observational studies suggesting that the
included population in the trial was a higher risk one. However, the findings of the
trial raise the question about the need to identify specific subpopulations of patients
at a higher risk of postdischarge VTE and that might benefit from thromboprophylaxis.
Current guidelines suggest against (1) using thromboprophylaxis in medical outpatients
with no major provoking risk factors for VTE, and (2) using extended-duration outpatient
VTE prophylaxis in the case of discharged patients.[27] Most importantly, the American College of Chest Physicians (ACCP) and the American
Society of Hematology recommendations suggest that routine extended thromboprophylaxis
after hospital discharge of COVID-19 patients may not have a net clinical benefit.[28]
[29]
Our findings support these recommendations as the proportion and incidence of venous
and arterial thrombosis observed in our study are similar to the reported risk of
bleeding associated with the use of thromboprophylaxis, in addition to the added burden
and cost. This aligns with the findings of a study by Giannis et al who reported that
patients discharged on anticoagulation had a major bleeding rate of 2.45% compared
with 1.63% in those who were discharged without any anticoagulation.[10] Therefore, the decision to use thromboprophylaxis should not be influenced by COVID-19
status but should be based on an overall assessment of patients' thrombosis risk factors
balanced with their bleeding risk, although a precise and proven algorithm to assess
these risks is currently not available.
Limitations
A limitation to this meta-analysis is that the included studies for postdischarge
patients had different rates of patients on prophylactic anticoagulation ranging from
0 to 97%, but in general, there was only a small percentage of high-risk discharged
patients started on prophylactic anticoagulation, but no clear criteria were used.
Given the heterogeneity in approaches, subgroup analyses might be misleading and and
need to be interpreted with much caution. Additional limitations to our study are
those inherent to the included reports, mainly the fact that the majority of patients
were included in retrospective studies. However, the proportion of events was almost
double in prospective studies compared with retrospective ones, suggesting the possibility
of recall bias, although the overall incidence is still very low even considering
the possibility of events underreporting. Additionally, the estimate of IR needs to
be interpreted with caution, as it was conducted under the assumption of no losses
to follow-up which is unlikely to be the case. However, given the short observation
period (less than 42 days in 11 of 16 studies) losses to follow-up are likely low
and would result in marginal modifications to the estimates. In any case, the proportions
of events were similar in studies with shorter or longer follow-up. Another issue
is the potential for COVID-19 vaccines to influence the risk of VTE. In this regard,
most recent studies have reported no influence of vaccines on the incidence of VTE,[30]
[31] except perhaps for the ChAdOx1 vaccine,[32] but overall it seems that vaccines are safe in this regard. This information was
not available in the studies included in this review. Finally, although statistical
heterogenicity was found between studies, all the proportion estimates were very low,
and there was no evidence of publication bias for most outcomes ([Supplementary Table S4]).
Conclusion
In conclusion, our findings suggest that patients with COVID-19 infection are not
severe enough to require hospitalization and those who are postdischarge and past
the acute phase are not at a particular high rate of thrombosis, and their thrombosis
risk is similar to other patients with medical illness. We believe that these findings
support the current recommendations, suggesting that prophylactic anticoagulation
should not be routinely prescribed in these groups of patients. However, to better
inform clinical decisions, further research is needed to define subgroups that could
potentially benefit from anticoagulant prophylaxis by identifying those patients at
a high risk of postdischarge thrombosis and a low bleeding risk that would derive
the most benefit from this intervention.
