Pulmonary embolism at the emergency department during the COVID-19 pandemic. A comparative cohort study from a tertiary level hospital in southern Spain

1. Introduction

Acute pulmonary embolism (PE) is a potentially life-threatening condition which mandates urgent diagnosis and treatment [1]. It is the third leading cause of cardiovascular mortality in emergency departments (EDs) worldwide, following acute myocardial infarction and stroke [2]. Its incidence in the population ranges from 0.5 to 1 per 1000 individuals per year, with no significant sex differences, and increases with age [3]. Several risk factors for PE have been identified, including confinement in a hospital or nursing home, active cancer, presence of a central venous catheter or pacemaker, and recent pregnancy [4]. In the context of the global Coronavirus disease 2019 (COVID-19) pandemic, an increased risk of venous thromboembolic (VTE) disease, including PE, has been reported since the first published studies [5, 6] particularly in hospitalized and Intensive Care Unit (ICU) patients [7, 8, 9, 10], with associated increased mortality risk [11, 12]. The pathophysiological mechanism of PE in COVID-19 patients has been suggested to involve multiple factors related to inflammation, prothrombotic state, hypoxia, immobilization, diffuse intravascular coagulation and endothelial dysfunction, among others [13]. However, subsequent studies reported discrepant results regarding the incidence, clinical significance, and characteristics of PE during the pandemic. Gallastegui et al. [14] were among the first authors who challenged the generalized assumption that PE was unusually highly incident in hospitalized patients. In line with other authors, some of the main factors underlying discrepancies among studies lie in the heterogeneity regarding different methodological approaches [15], including but not limited to patient selection, study design, and assessment tools. Currently, unanswered questions remain in spite of several systematic reviews and meta-analyses on this topic [16, 17, 18]. It is still unclear whether PE is more frequent in COVID versus non-COVID cohorts with the same degree of disease severity.

The clinical presentation of acute PE encompasses a wide spectrum of signs and symptoms, and routine exams at the ED have low specificity for its detection [19]. Currently, the ‘old standard’ technique for the diagnosis of PE is computed tomography pulmonary angiography (CTPA), which has high sensitivity and specificity, allowing to detect even small (i.e., subsegmental) embolisms. In addition, this technique may show radiological signs associated with right ventricle strain and pulmonary hypertension (RVS/PH), which have been associated with a worse prognosis [2]. For these reasons, CTPA has been suggested to have a fundamental role in the diagnostic work-up of COVID-19 patients for the diagnosis of acute PE [20]. The early-recognized, alleged association between COVID-19 and higher risk of PE in the setting of a hypercoagulability state together with the characteristic pulmonary involvement of this virus, could have motivated a greater number of CTPA orders to rule out PE during the pandemic. Accordingly, some institutions carried out universal CTPA screening to rule out PE, which has been associated with an increased incidence of PE [21]. This increase in the number of CTPAs performed may have magnified the actual incidence of PE by increasing the detection of subsegmental PEs, which are frequently asymptomatic and whose clinical impact is still unclear [2]. Conversely, the number of CTPAs could have been very infrequent in some settings, particularly during the first waves of the pandemic, due to factors such as scarcity of resources, fear to concerns pertaining to the contamination of computed tomography (CT) scanners and in-hospital transmission of disease [22], the decrease in patient visits to the ED, or differences in respiratory severity of the disease. This would imply an underestimation of PE incidence [6]. In sum, whether the number of CTPA orders increased or decreased during the pandemic, and whether such trend was homogeneous throughout the pandemic remains unknown. The subsequent impact on the incidence of PE, as well as other related factors associated with prognosis and mortality (e.g., distribution of PE, associated signs of RVS/PH) is also unclear.

The aim of this work is to analyze the number of CTPAs ordered to rule out PE at our ED during the first year of the pandemic, as well as the incidence, distribution and radiological features associated with PE.

2. Methods

3. Results

3.1 Flow diagram of the study and description of the cohorts

Of the total of 2719 CTPAs registered in the PACS of our institution, 1905 CTPAs were included after applying the exclusion criteria. Of these, 530 (27.8%) belonged in the pre-COVID cohort and 1375 (72.2%) corresponded to the COVID cohort. Fig. 1 shows the flow diagram followed for the selection of both cohorts.

Fig. 1.Flow diagram of the selection of cohorts in our study. CTPA, computed tomography pulmonary angiography. PACS, picture archiving and communication system. ED, emergency department. COVID, coronavirus disease 2019 sub-group. PE, pulmonary embolism. *Indicates the categories of the reason for exclusion of CTPAs.

The mean age of the entire sample included in the study was 68.3 years (standard deviation (SD), 16.5), and sex distribution included 981 (51.5%) women. Comparative analyses showed no significant differences between groups regarding sex or incidence of PE, although the mean age was significantly different (p = 0.009). In the group with confirmed PE, comparative analyses showed no significant differences regarding age or sex, but a significantly higher proportion of peripheral PE was observed in the COVID cohort. Table 1 summarizes the characteristics and contrastive analysis of the cohorts included in the study. Table 2 shows the adjusted analyses of the multivariate logistic regression models. PE was defined as the outcome (dependent variable), and the covariates included were sex, age, and cohort (i.e., preCOVID or COVID). The pre-pandemic cohort showed higher odds of presenting PE, although no significant association was found (odds ratio [OR]: 1.23, 95% confidence interval [CI]: 0.97–1.57).

Table 1.Characteristics and bivariate analyses of the two cohorts included in the study.

		Total n (%), x (s)	Pre-COVID cohort n (%), x (s)	COVID cohort n (%), x (s)	p value†
Total	1905 (100.0)	530 (27.8)	1375 (72.2)	-
Age*	68.3 (16.5)	69.9 (16.7)	67.7 (16.4)	0.009
Sex*
	Female	981 (51.5)	278 (52.5)	703 (51.1)	0.604
	Male	924 (48.5)	252 (47.5)	672 (48.9)
PE frequency	392 (20.6)	123 (23.2)	269 (19.6)	0.078
Age**	69.0 (16.3)	70.7 (17.0)	68.2 (15.9)	0.160
Sex**
Female**	208 (53.1)	65 (52.8)	143 (53.2)	0.954
Male**	184 (46.9)	58 (47.2)	126 (46.8)
PE location
Central**	269	115 (93.5)	156 (58.0)	<0.001
Peripheral**	122	8 (6.5)	113 (42.0)
*Data relative to the entire sample. **Data relative to the PE+ group. †p value of the chi-squared test for qualitative variables and Student’s t test for the continuous variable “age”. PE, pulmonary embolism.

Table 2.Multivariate analysis of the two cohorts included in the study.

		OR*	95% CI
Sex
	Female	0.93	0.75–1.17
	Male	Ref	-
Age	1.00	0.97–1.01
Cohort
Pre-pandemic	1.23	0.97–1.57
Pandemic	Ref	-
*Odds ratio (OR) for presenting pulmonary embolism for each category. CI, confidence interval.

4. Discussion

A number of thrombotic events including PE have been reported to be more frequent in COVID-19 patients as a consequence of a hyper-inflammatory response [26]. However, the degree of severity has been observed to be variable depending on the patient profile. For instance, patients with mild COVID-19 disease may show scarce abnormalities in laboratory parameters, while critically ill patients exhibit a hyperinflammatory and procoagulant phenotype with significantly higher levels of ferritin, C-reactive protein, fibrinogen, D-dimer, and lactic acid [27]. Considering the dramatic impact of this disease in health care systems, important efforts have been made to facilitate identification, risk stratification and optimization of treatment in COVID-19 patients. This includes consensus guidelines and recommendations that are now available to facilitate clinical decision-making [28]. For example, the need for specific prediction rules has been advocated for estimating the risk of PE in hospitalized COVID-19 patients, differentiating ICU from non-ICU patients, and taking into account anticoagulation prophylaxis, comorbidities, and the time from COVID-19 diagnosis [29]. However, it is not clear whether mild forms of COVID-19 infection lead to a significant increase in the risk of developing thrombotic complications in comparison with other diseases of similar nature and severity, particularly in the context of general wards.

Due to the rapid emergence of this pandemic, many of the studies exploring the incidence of PE in COVID-19 patients show heterogeneity, selection biases or inadequate designs. This has prompted systematic reviews and meta-analyses to shed light on the actual incidence of PE in COVID-19 patients. However, discrepancies have also been found in these pooled studies. Early meta-analyses observed an increased incidence of PE in both ICU and non-ICU patients [30], but figures were highly variable ranges (e.g., 2–35%) [31]. Posterior, meta-analyses such as the one published by Mai et al. [16] including more than 40,000 patients found an increased risk of VTE occurrence among COVID-19 patients hospitalized in the ICU, but no overall difference in risk in COVID-19 cohorts compared to non-COVID-19 cohorts. Conversely, a more recent review by Tufano et al. [17] including more than 1,000,000 patients reported a risk difference (RD) for PE between COVID-19 and non-COVID-19 patients of 0.03, while the RD for PE between COVID-19 and non-COVID-19 patients was 0.021 in retrospective studies and 0.11 in ICU studies.

In Spain, few studies have been published on the incidence of PE during the COVID-19 pandemic. A study by Martínez-Chamorro et al. [32] comparing the number of CTPAs performed to rule out PE and the incidence of PE at the beginning of the pandemic found a greater number of CTPA orders (69.9% vs. 30.1%), but no significant differences in the incidence of PE. When adjusted for COVID infection, significant differences were found, with PE being more frequent in patients with COVID-19 infection. These results are in agreement with our findings; our ratio of CTPAs was very similar (72.2% vs. 27.8%), but no significant differences were observed in the incidence of PE (in fact, it was lower in the COVID cohort). In addition, the analysis of the COVID cohort showed that the frequency of PE was significantly higher in patients with laboratory-confirmed COVID-19 disease compared to non-COVID-19 patients (77% vs. 23%). Moreover, an added value of our study is the longer follow up (first year of the pandemic), with stratified analyses based on health care pressure intervals which showed interesting results. First, the incidence of PE was not significantly different across waves, although it was lower during the second and third waves compared to the first wave. Interestingly, the ratio of peripheral PE was also significantly lower in the first wave compared to the rest of the pandemic, and these were in turn significantly higher compared to the pre-COVID cohort. These findings could be explained by a lower number of CTPAs performed during the first pandemic wave. Of note, we observed statistically significant differences regarding age both in the comparative analysis between the pre-COVID and COVID cohorts and between the different pandemic waves, but this seems to be secondary to large numbers rather than actual relevant differences (less than 3 years between cohorts).

One of the most interesting results of our study relates to the frequency of peripheral and central PE, both before and during the pandemic, and across the waves of the pandemic. We observed that, whilst the ratio of peripheral PE was anecdotal prior to the pandemic (6.5%) —in agreement with previous studies—, this proportion was significantly higher during the pandemic (42.0%). These findings parallel those reported by a retrospective study conducted in COVID-19 patients in France by Grillet et al. [33] who found a PE incidence of 23%, with a more frequent peripheral distribution (51%), and also are in line with the results of the meta-analysis conducted by Jiménez et al. [34] in hospitalized patients. In Spain, Mestre-Gómez et al. [35] conducted a retrospective study with 91 patients who underwent a CT scan to rule out PE in a tertiary hospital from Madrid. The authors found a 31.9% PE incidence, with a significantly more predominant distribution of peripheral vs. central location (69% vs. 31%). Similarly, Benito et al. [36] reported a higher incidence of PE in hospitalized COVID-19 patients, with a predominance of peripheral (78.1%) vs. central (21.9%) distribution. These findings could be explained by two main hypotheses. On the one hand, this could be due to a higher number of incidental PEs resulting from the 3-fold increase in the number of CTPAs performed during the pandemic. On the other hand, COVID-19 disease could be specifically associated with a higher risk of small size pulmonary emboli rather than with central/massive PE forms. Interestingly, some authors have raised the hypothesis that pulmonary in situ clot formation is an added mechanism of PE in COVID-19 patients, since the prevalence of deep vein thrombosis (DVT) has been found to be low in COVID-19 patients as compared with PE [37, 38, 39]. A combination of both hypotheses (i.e., increased incidental detection of subsegmental PEs and increased formation of peripheral PE in COVID-19 patients) also seems reasonable.

A particularly innovative aspect of our study is the analysis of signs of RHS/PH. Early reports such as the study by Argulian et al. [40] suggested right ventricular dilation in patients hospitalized due to COVID-19, yet potential biases secondary to small sample size and retrospective nature limited its validity. Potential causes for such finding included increased susceptibility of PE, PH and acute respiratory syndrome [41]. It seems that disease severity plays a role in the development or aggravation of PH. Accordingly, patients with severe COVID-19 were reported to have a higher proportion of pulmonary hypertension as compared to mild COVID-19 disease (22% vs. 2%) [42]. A study by van den Heuvel et al. [43] including 51 COVID-19 patients admitted to hospital. One of the subgroup analyses carried out in this study compared COVID-19 patients with PE (n = 9) vs. those with no PE (n = 42), and found that none of the patients with PE had right ventricular dysfunction, and right ventricular dimensions were normal, although a higher tricuspid annular plane systolic excursion and increased right ventricular systolic excursion velocity were found in PE patients, presumably due to small sample size and limited severity of embolism. A retrospective study conducted by Tilliridou et al. [44] including a total of 1286 CTPAs performed to rule out PE in three hospitals from Scotland. They compared CTPA frequency and PE severity in April and May 2020 versus 2019. The authors assessed PE severity with the Modified Miller score and also assessed the presence of right heart strain (RHS). The authors found a 17% reduction in the number of CTPA performed and an increase in the proportion of PEs detected (26% vs. 15%). Although there was no difference in PE severity in 2020 compared to 2019, the authors observed an increased frequency of RHS in May 2020 (29 vs. 12%, p = 0.029). Of note, the authors only used a RV/LV >1.2 as radiological indicator of RHS. In our study, the comparison between the pre-COVID and COVID cohorts showed no significant differences in the RV/LV ratio, but significant differences were observed in septal modification (with the presence of signs of RHS/PH being more frequent in the pre-COVID cohort). In addition, stratified analyses based on pandemic waves showed that RHS/PH signs were less frequent in the second and third waves, paralleling the pattern of distribution of peripheral PE, and suggesting a causal association (i.e., more peripheral PEs with less RHS/PH signs were present throughout the pandemic).

The main limitations of our study lie in its retrospective observational design and single-centered nature, as well as the relatively short follow-up period. In addition, it should be noted that the radiological signs of RHS/PH are indirect signs that do not always correlate with clinical symptoms. For example, contrast reflux is associated with rapid contrast injections or other pathologies such as mitral regurgitation [45]. On the other hand, the quantification of reflux is variable in the literature, ranging from classifications with 4 categories [46] to 6 categories [47, 48]. These limitations should be overcome in future studies, which shall address the incidence and characteristics of PE considering present challenges, such as the influence of vaccination and changes in virus dynamics with new variants such as Omicron.

5. Conclusions

Despite a significantly higher number of CTPAs were ordered in our institution during the first year of the pandemic, the incidence of PE was similar to that of the pre-pandemic era. A higher number of peripheral PE and less radiological signs of RVS/PH were observed during the pandemic. Although PE was more frequent in patients with laboratory-confirmed COVID-19, our overall findings suggest that a high number of incidental PEs were detected during the pandemic.

Availability of Data and Materials

The data presented in this study are available on reasonable request from the corresponding author.

Author contributions

AJLRB, BMC and FGS—designed the research study and wrote the first draft of the manuscript; AJLRB, BMC and FGS—conducted the research and advised on methodology; ERC, MCB, FJPG and CJM—extracted the data; MRI—mentored the work. All authors read and approved the final manuscript.

Ethics approval and consent to participate

This study complies with the principles of the Declaration of Helsinki. A completely anonymized database was used for the analyses. No identification data were used. The need for informed consent has been waived owing to its retrospective nature. The study was approved by the Provincial Research Ethics Committee of Granada (ref. TEPA18).

Acknowledgment

The authors wish to thank Benítez Sánchez of the University of Granada for his statistical assistance.

Funding

This work was supported by the ‘Artificial Intelligence for the diagnosis and prognosis of COVID-19’ project (CV20-29480), funded by the Consejería de Transformación Económica, Industria, Conocimiento y Universidades, Junta de Andalucía, and the FEDER funds.

Conflict of interest

The authors declare no conflict of interest. Antonio Jesús Láinez-Ramos-Bossini is serving as one of the Guest editors of this journal. We declare that Antonio Jesús Láinez-Ramos-Bossini had no involvement in the peer review of this article and has no access to information regarding its peer review. Full responsibility for the editorial process for this article was delegated to MC.