Aims

We investigated the incidence, risk factors, clinical characteristics, and outcomes of pulmonary embolism (PE) in patients with COVID-19 attending emergency departments (EDs), before hospitalization.

Methods and Results

We retrospectively reviewed all COVID-19 patients diagnosed with PE in 62 Spanish EDs (20% of Spanish EDs, case group) during the first COVID-19 outbreak. COVID-19 patients without PE and non-COVID-19 patients with PE were included as control groups. Adjusted comparisons for baseline characteristics, acute episode characteristics, and outcomes were made between cases and randomly selected controls (1:1 ratio). We identified 368 PE in 74 814 patients with COVID-19 attending EDs (4.92&). The standardized incidence of PE in the COVID-19 population resulted in 310 per 100 000 person-years, significantly higher than that observed in the non- COVID-19 population [35 per 100 000 person-years; odds ratio (OR) 8.95 for PE in the COVID-19 population, 95% confidence interval (CI) 8.51–9.41]. Several characteristics in COVID-19 patients were independently associated with PE, the strongest being D-dimer >1000 ng/mL, and chest pain (direct association) and chronic heart failure (inverse association). COVID-19 patients with PE differed from non-COVID-19 patients with PE in 16 characteristics, most directly related to COVID-19 infection; remarkably, D-dimer >1000 ng/mL, leg swelling/pain, and PE risk factors were significantly less present. PE in COVID-19 patients affected smaller pulmonary arteries than in non-COVID-19 patients, although right ventricular dysfunction was similar in both groups. In-hospital mortality in cases (16.0%) was similar to COVID-19 patients without PE (16.6%; OR 0.96, 95% CI 0.65–1.42; and 11.4% in a subgroup of COVID-19 patients with PE ruled out by scanner, OR 1.48, 95% CI 0.97–2.27), but higher than in non-COVID-19 patients with PE (6.5%; OR 2.74, 95% CI 1.66–4.51). Adjustment for differences in baseline and acute episode characteristics and sensitivity analysis reported very similar associations.

Conclusions

PE in COVID-19 patients at ED presentation is unusual (about 0.5%), but incidence is approximately ninefold higher than in the general (non-COVID-19) population. Moreover, risk factors and leg symptoms are less frequent, D-dimer increase is lower and emboli involve smaller pulmonary arteries. While PE probably does not increase the mortality of COVID-19 patients, mortality is higher in COVID-19 than in non-COVID-19 patients with PE.

The UMC-19-S9 reports four main findings needing to be highlighted, summarized in Graphical Abstract. First, PE is relatively uncommon as a manifestation of SARS-CoV-2 infection before patient hospitalization (0.5%); however, when we consider standardized incidence per year, this is almost 9-fold greater in the COVID-19 than in the non-COVID-19 population. Second, D-dimer levels >1000 ng/mL and chest pain as a clinical complaint were the strongest risk factors directly associated with PE development in COVID-19 patients. Third, COVID-19 patients with PE differed from non-COVID-19 patients with PE mainly in the clinical characteristics directly related to COVID-19 infection, although, remarkably, they had fewer risk factors for PE, complained less frequently of leg swelling/pain, exhibited a more discrete rise in D-dimers, and thrombi affected smaller pulmonary arteries. Fourth, the mortality was high (one in six patients died during hospitalization), being more than double the number observed in the general population (non-COVID-19) with PE. Conversely, no statistical differences were found in in-hospital mortality of COVID-19 patients with and without PE; therefore, PE should not be considered as a risk factor for death in patients with COVID-19.

Our estimation of the standardized incidence of PE of 35 per 100 000 person-years in the general (non-COVID-19) population (41 considering only the pre-COVID-19 period, when fear of COVID-19 contagion in EDs was not present) is close to that reported in previous studies: 69 in the USA in the period from 1966 to 1990 (with a progressive decline over time being noticed),9 50 in Norway in the period from 1995 to 2001,10 and 38 in Canada in the period from 2002 to 2012 (with a stable incidence over time),11 which makes our estimation quite reliable. Therefore, the incidence of PE of 310 per 100 000 person-years found in COVID-19 patients strongly suggests that SARS-CoV-2 is associated with a marked and significant increment in the risk of PE and could have a direct potential pathogenic role in the development of PE. Nonetheless, our results must be interpreted with caution, as they could be biased by the particular circumstances of the first pandemic wave, including population lockdown, the high number of CTPA ordered in COVID-19 patients (making the diagnosis of PE easier and more frequent) and the fact that non-COVID-19 patients with small PEs could have been less diagnosed as they stayed at home (due to fear of contagion in EDs) or were less frequently explored with CTPA if they complained of non-specific symptoms. However, as difference in standardized incidences in both populations was very large (OR 8.95), and as the OR for PE in COVID-19 patients was also very high when compared with PE diagnosed in non-COVID-19 patients during the pre-COVID-19 period (OR 10.91), and lastly, as nearly half of PE in COVID-19 patients involved the main or lobar pulmonary arteries (only 12% were very small PEs, affecting exclusively subsegmental arteries), we believe that there really is an increase in the incidence of PE in patients with COVID-19.

Our comparison between COVID-19 patients with and without PE was performed before hospitalization and, accordingly, it did not take into account the secondary effects of being bedridden for a long time and/or intubation that can adversely affect hospitalized patients and favour thromboembolic disease. In this scenario of ED, COVID- 19 patients complaining of chest pain and having D-dimer levels over 1000 ng/mL were at increased risk of having PE. D-dimers are a classical biomarker of PE in the general population, although cut-off adjustments must be made in a number of situations, the most relevant being patient age.12 Similarly, it is possible that a specific D-dimer cut-off value for COVID-19 patients is needed, as non-specific mild increments of D-dimer are usually observed in such a population.1 On the other hand, the inverse association of PE with chronic heart failure found in the present study is striking, and contrasts with previously reported data identifying chronic heart failure as a risk factor for PE.13We can hypothesize two potential explanations for our finding: the competing risk of death or the higher presence of atrial fibrillation in chronic heart failure patients (in our series, 28.2% compared to 6.3% in the rest of patients; P< 0.001). Atrial fibrillation usually associates a high use of oral anticoagulants and this could have been, in fact, the factor related to the apparent protective effect of chronic heart failure for development of PE inCOVID-19 patients.

It was of note that leg swelling/pain and risk factors for PE were significantly less present in COVID-19 than in non-COVID-19 patients with PE, despite DVT and proximal DVT being diagnosed in a similar percentage of patients. This suggests that some PE could develop in situ in lungs, favoured by a highly inflammatory involvement and, in fact, in situ immunothrombosis has been proposed to play a role in the pathophysiology of COVID-19-associated PE.14 In this sense, alveolar injury and the inflammatory storm present during COVID-19 pneumonia along with disruption of the thrombo-protective state of the pulmonary vascular endothelial cells might contribute to the formation of deep small vessel thrombi. We made a second review of CTPA for adjudication ofmorphological vascular findings, with special focus on lumen location of thrombi placed in small pulmonary arteries, in order to figure this hypothesis out. Nonetheless, our study does not provide definitive data on this regard: while anatomical involvement of smaller pulmonary arteries would support the hypothesis of a local (lung) process in PE developed in COVID-19 patients, no differences were found in clot location in subsegmental PEs. Finally, it is remarkable that involvement of smaller pulmonary arteries (segmental or subsegmental) does not translate in a different severity of PE with respect to the general population, as we found a similar proportion of patients with PE presenting RVD.

When outcomes were analysed, the need for ICU care was clearly increased in COVID-19 patients with PE compared to COVID-19 patients without PE, in agreement with a previous study by Fauvel et al.3 in 24 French hospitals. Conversely, our results were more inconclusive for in-hospitalmortality. Overall, we did not find significant increments in mortality, in line with the results reported by Fauvel et al.3 However, while the mortality was similar when compared with COVID-19 patients in whom PE was mainly excluded on the basis of clinical findings, it showed a trend to increase when compared with COVID-19 patients in whom PE was excluded using CTPA in all cases, and even achieved statistical significance after adjustment for baseline characteristics. It is difficult to decide which of these two control groups of COVID-19 patients without PE is better for making comparisons since while the former could include some patients having PE in whom PE was not diagnosed during ED management, the latter could include less sick patients in whom lung scans were routinely performed to detect parenchymal lung involvement by COVID-19. Accordingly, caution is recommended when interpreting the impact of PE on mortality in COVID-19 patients, as it is difficult to isolate the adverse effects (death, in this case) of COVID-19 (essentially lung involvement due to inflammation and hyperimmune reaction) from the adverse effects of PE. On the other hand, the mortality was higher in COVID-19 than in non-COVID-19 patients with PE, but the causes of death were not significantly different between the two groups. Although part or all of this increase could be related to severe COVID-19 (with lung parenchymal involvement and secondary respiratory failure), the true role of PE in determining poor outcomes inCOVID-19 patients remains to be elucidated.

Conclusions

The UMC-19-S9 demonstrates that PE in COVID-19 patients is unusual (about 0.5%) at ED presentation, but standardized incidence is about 9-fold higher than expected in the general (non-COVID-19) population. PE by itself probably does not increase in-hospital mortality of the COVID-19 population presenting to the ED. As risk factors and leg symptoms are less frequent, and the increase in D-dimer is lower than in non-COVID-19 patients developing PE, a red flag should be raised by any treating physicians assessing COVID-19 patients in order to detect PE in these patients, and to promptly start specific anticoagulant treatment, as recent data show that this is a safe treatment that should be provided to COVID-19 patients with PE