Multi-organ point-of-care ultrasound for detection of pulmonary embolism in critically ill COVID-19 patients – A diagnostic accuracy study

Purpose Critically ill COVID-19 patients have an increased risk of developing pulmonary embolism (PE). Diagnosis of PE by point-of-care ultrasound (POCUS) might reduce the need for computed tomography pulmonary angiography (CTPA), while decreasing time-to-diagnosis. Materials & methods This prospective, observational study included adult ICU patients with COVID-19. Multi-organ (lungs, deep vein, cardiac) POCUS was performed within 24 h of CTPA, looking for subpleural consolidations, deep venous thrombosis (DVT), and right ventricular strain (RVS). We reported the scan time, and calculated diagnostic accuracy measures for these signs separately and in combination. Results 70 consecutive patients were included. 23 patients (32.8%) had a PE. Median scan time was 14 min (IQR 11–17). Subpleural consolidations' diagnostic accuracy was: 42.9% (95%CI [34.1–52.0]). DVT's and RVS' diagnostic accuracy was: 75.6% (95%CI [67.1–82.9]) and 74.4% (95%CI [65.8–81.8]). Their sensitivity was: 24.0% (95%CI [9.4–45.1]), and 40.0% (95%CI [21.3–61.3]), while their specificity was: 88.8% (95%CI [80.8–94.3]), and: 83.0% (95%CI [74.2–89.8]), respectively. Multi-organ POCUS sensitivity was: 87.5% (95%CI [67.6–97.3]), and specificity was: 25% (95%CI [16.9–34.7]). Conclusions Multi-organ rather than single-organ POCUS can be of aid in ruling out PE in critically ill COVID-19 and help select patients for CTPA. In addition, finding RVS can make PE more likely, while a DVT would preclude the need for a CTPA. Registration: www.trialregister.nl: NL8540.


Introduction
Clinical manifestations of Coronavirus Disease 2019 (COVID-19) extend beyond that of a pneumonia [1]. Critically ill COVID-19 patients have an increased risk of developing pulmonary embolism (PE), compared to hospitalized and intensive care unit (ICU) patients with other (respiratory) infections [2][3][4][5]. Computed tomography pulmonary angiography (CTPA) is the gold standard to diagnose PE. However, transport outside of the controlled ICU environment carries substantial risks for patients, while necessitating post-transport decontamination [6,7] It also requires more personnel and reduces time that can be spent on patient care.
Point-of-care ultrasound (POCUS) might reduce the need for CTPAs in these patients in already overwhelmed acute and critical health care systems worldwide, particularly in low-and middle-income [8][9][10]. Recent consensus statements recommend the use of POCUS as an initial test in critically ill COVID-19 patients [3,5,11,12]. However literature on POCUS in the detection of PE in these patients is still limited.
Compression ultrasound of the deep veins of the lower extremities (deep vein POCUS) has demonstrated excellent accuracy in detecting deep venous thrombosis (DVT)a proxy for PEacross a myriad of settings, including this pandemic [19][20][21][22]. Furthermore, cardiac POCUS may help identify pulmonary embolism (PE) indirectly by the detection of new right ventricular strain (RVS), which has high specificity for PE. Cardiac POCUS has been reported to be as reliable as formal echocardiography in detecting RVS [23]. However, RVS is found in 22-50% of patients with ARDS, which could limit its use in COVID-19 [19,24].
Combining different POCUS modalities (lung, deep veins, and cardiac) to detect pulmonary embolism has been advocated as well [3,5,11,12,25]. In the emergency department (ED) setting there are promising results of multi-organ POCUS in the detection of PE [7,10,26], while others found no benefit [27]. However, there have been no studies in critically ill and/or COVID-19 patients.
Therefore, our aim was to investigate the diagnostic accuracy of multi-organ POCUSlung, deep vein and cardiacseparately, and in combination in the detection of PE in critically ill COVID-19 patients.

Study design and setting
This is a prospective, observational study conducted at the academic adult ICU of the Amsterdam UMC, location VUmc, Amsterdam, The Netherlands, between October 20th 2020 and February 20th, 2021. Patients were followed up until May 20th, 2021. The study, and use of data gathered during routine ultrasound and CTPA, was approved by the Medical Ethical Committee of the VUmc, along with a waiver of informed consent (2020.011). The trial was registered in the Dutch trial registry (NL8540).

Patients
Consecutive adult patients (≥18 years) with a laboratory confirmed diagnosis of COVID-19 were eligible for inclusion upon admission to ICU. Patients were included when the multi-organ POCUS examination (index test) was performed within 24 h of a CTPA (reference test). They could be included again if a CTPA was repeated. Patients were excluded if they were already on therapeutic anticoagulation, transferred to another hospital within 24 h of CTPA, or died within 24 h of CTPA. Baseline characteristics along with ventilator settings and laboratory values were collected from the electronic patient record closest to the time of CTPA. Please see the Supplemental Material for the diagnostic and treatment protocol on our ICU.

Medical work up
As part of the departmental protocol, patients received a CT(PA) when their clinical condition did not improve or deteriorated approximately every 7 days to diagnose PE, to determine the extent of pulmonary involvement, and possible superinfection. If a PE was diagnosed, a CTPA was typically not repeated. A normal CT could still be repeated though to monitor the degree of pulmonary involvement or detect a possible superinfection. As commonly used clinical decision rules (i.e., WELLS, GENEVA or YEARS-criteria) are not validated in the ICU nor in the COVID-19 setting [28][29][30][31], the decision to perform a CTPA was made by the treating ICU-consultant in consultation with a daily multidisciplinary team consisting of a consultant microbiologist, a consultant pulmonologist and at least three other ICU consultants. Please see the Supplemental Material for the full diagnostic and treatment protocol on our ICU.

Index test: Point-of-care ultrasound (POCUS)
POCUS is part of the standard care on our ICU. All examinations were performed or supervised by experienced ultrasound physicians using a COVID-19 unit-restricted SonoSite-Edge II ultrasound machine [32]. They were blinded for CTPA result, but not for the clinical data. Lastly, the scan time was recorded for all investigations. Pulmonary embolism has been associated with subpleural consolidations on lung ultrasound [13,33]. A 10-5 MHz linear transducer or 5-3 MHz curvilinear transducer with lung examination setting was used. The lung ultrasound protocol consisted of a structured assessment of 6 zones of each hemithorax [34]. The recently modified criteria from Mathis et al. were used to diagnose PE in COVID-19 patients [14,18]. Likelihood of PE was determined to be: • High: when two or more subpleural consolidations (≥1 cm) were detected ( Fig. 1, Video 1); • Probable: when only one subpleural consolidation (≥1 cm) was detected; • Possible: when two (or more) subpleural consolidations (<1 cm) were detected (Fig. 2, Video 2); • Low: when no consolidations were detected.

Deep vein POCUS: Deep venous thrombosis (DVT)
A two-point compression ultrasound of the common femoral and popliteal vein was performed, using a 10-5 MHz linear transducer on venous setting following a previously described protocol (Video 3-4) [35,36]. If an indwelling central venous catheter (CVC) was present, it was also examined as far as possible after skin insertion. Deep venous thrombosis was defined as a non-compressible vein (Video 5-7).

Computed tomography pulmonary angiography (CTPA)
CTPAs were evaluated by our local radiologists with varying degrees of experience, who did have access to the clinical informationbut not the POCUS examinations. Pulmonary embolism was defined as a constant intravascular filling defect on CTPA (see Supplemental Material for the scan protocol). Location of a filling defect was registered for each lobe until most distal subsegmental levels. Right ventricular strain (RV/LV ratio ≥ 1) was also registered.

Statistical analysis
Baseline characteristics and outcome variables were presented as means ± standard deviations (±SD), medians and interquartile range [IQR], or numbers (percentages %) as appropriate.
We determined diagnostic accuracy measures; sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), positive predictive value (PPV), negative predictive value (NPV), and diagnostic accuracy for subpleural consolidations, right ventricular strain, and deep venous thrombosis separately and in different combinations. In addition, we determined a receiver operating characteristic (ROC) curve with corresponding area under the curve (AUC) for subpleural consolidations ≥1 cm. We established the optimal cut-off with the Youden index. Missing data were handled by pairwise deletion.
McNemar's tests were used to compare sensitivities and specificities of each single-organ modality with multi-organ POCUS. Missing data were handled by pairwise deletion. Statistical analyses were performed using SPSS IBM version 24.0 (SPSS Inc., Chicago, IL, USA).

Results
140 CTPAs were made in 70 consecutive patients, with a median of 2 scans per patient [IQR 1-3]. In 11 instances multi-organ POCUS examinations data was missing because the scan was not conducted withing 24 h of the CT scan. In 2 patients all of the POCUS scans were incomplete or could not performed, this was also the case for 1 additional lung POCUS, 4 deep vein POCUS and 2 cardiac POCUS (Fig. 3). This was mainly due to the positioning of the patient (i.e., prone or lateral position). Patient baseline characteristics and outcomes are shown in Table 1

Discussion
The main findings of this prospective, observational study in adult critically ill COVID-19 patients are that: 1) Single-organ POCUS modalities alone have low sensitivity and therefore are not helpful in ruling out PE; 2) Subpleural consolidations are hardly useful in the detection of PE; 3) DVT has good specificity for PE and its detection alone obviates the need for CTPA; 4) RVS has similar diagnostic accuracy to other settings, despite concomitant ARDS or mechanical ventilation; 5) Multi-organ POCUS however has good sensitivity for PE and may have a role in ruling out PE.    We found rather poor diagnostic accuracy for subpleural consolidations in the detection of PE in critically ill COVID-19 patients, regardless of the criterium used or the number of subpleural consolidations. In contrast, the latest systematic reviews in the non-COVID-19 setting showed diagnostic accuracy of lung ultrasound was good in studies where the diagnostic threshold was at least one subpleural consolidation; with a sensitivity of 81.4-87%, and a specificity of 81.8-87.4% [13,33]. Studies using a diagnostic threshold of at least two subpleural consolidations had a worse sensitivity 44.2%, but better specificity 96.5% [33]. Most included studies were conducted in ambulatory patients [14,15,33,39,40], and the authors noted a high potential of selection bias, limiting generalization of those results to critically ill COVID-19 patients.
Our results might be less remarkable when we revisit the pathophysiology of subpleural consolidations. Subpleural consolidations originate from the depletion of air in the lung. This may be caused by infarction, atelectasis or inflammation, and as such can be found in PE, COVID-19 pneumonia and ARDS [15,16,41,42]. This overlap is probably what caused the poor diagnostic accuracy in this particular setting. In addition, 52.2% of PEs found in our study were subsegmental. Outside the critically ill COVID-19 setting sensitivity of lung POCUS also drops from 89%in central, lobar, and segmental PEto 21% in subsegmental PE [13].
Contrarily, in early reports, one group have argued that since thrombi are often located distally, and obscured by consolidation they often cannot be reliably visualized by CTPA, while lung POCUS actually does pick-up those PEs [18,43]. However, there is debate about what the clinical significance of such small emboli is, and if they should warrant detection or treatment [44,45]. PEfound by CTPAhas been associated with a higher mortality in COVID-19 [4]. The REMAP-CAP trial showed no benefit of therapeutic anticoagulation in critically ill patients with COVID-19, while the majority of those patients will probably have subpleural consolidations [46]. 4.1.1.2. DVT. The prevalence of PE (32.8%) and DVT (21.4%) found in our study was comparable with the prevalence found in other screening studies in COVID-19 patients [4,47,48].The pathophysiology of PE in COVID-19 is still not entirely clear. In addition to 'classic' PE, it is probably multifactorial as COVID-19 patients exhibit all components of Virchow's triad: an immune-mediated hypercoagulable state, endothelial injury, and stasis/turbulence of blood flow (due to immobilization, hypoxemia induced vasoconstriction, and further exacerbated by the insertion of CVCs) [3,4,22,[47][48][49]. Indeed, PEs are associated with lower limb DVT in only 10-13.6% of COVID-19 patients, while this is 56-61% in undifferentiated cohorts, explaining the lower sensitivity [7,50,51]. Specificity is good though, indicating that finding a DVT can make the diagnosis of PE a little more likely, but its absence cannot rule it out.
On the other hand, detecting a DVT by POCUS would preclude the necessity of a CTPA, since therapeutic anticoagulation should be started regardless according to international guidelines [33,52]. Systematic DVT scanning by POCUS could save costs, time, and potentially lives given the increased mortality risk with concomitant venous thromboembolism [4]. Meanwhile it would circumvent the downsides of having to take a potentially contagious and unstable patient to and from radiology for a CTPA or formal DVT scan. In line with current guidelines, we therefore advocate its use, as a DVT scan takes less than 2 min to perform, and results are just as reliable as a formal radiology exam [3,4,19,21].

RVS.
As is the case with subpleural consolidations, RVS is not a direct marker of PE. It is a proxy for clot burden and cardio-pulmonary reserve [19,23]. The sensitivity and specificity found for RVS in our study are comparable to what is found for different features of RVS (i.e., RV: LV ratio ≥ 1, 'D-sign', and McConnell's sign) in the non-COVID-19 literature; 40% versus 24-54%, and 83% versus 87-98.6% respectively [19,23,33]. This supports the premise that RVS might help rule PE in, but cannot rule it out [19,23,33]. This could be explained by the fact that (COVID-19 related) ARDS itself can cause RVS [19,24,53]. Mechanical ventilation can also be of influence, but this was deemed less likely as we adhered to stringent lung protective mechanical ventilation protocols [24,54]. In order to discriminate between different causes of RVS like PE or (COVID-19 related) ARDSit might be prudent to perform scans at baseline and regular intervals. We hypothesize that a sudden increase in RV:LV ratio without changes in respiratory mechanics might be associated with PE, but further research is required before drawing definitive conclusions.

Multi-organ POCUS
The sensitivity found for multi-organ POCUS: 87.5% is similar to that in the non-COVID ED setting: 90% [7,33,55], and cardiac-deep vein POCUS alone: 91% [10]. Our results suggest that multi-organ POCUS has a role in ruling out PE, or at least have the potential to reduce the amount of CTPA's. The main driver of the good sensitivity seems to be subpleural consolidations, which as mentioned before seem to be ubiquitous in critically ill COVID-19 patients. The high likelihood PE criterium (CON2) was found in almost two-thirds of patients and falsely positive in >80% of them. This causes specificity to drop well below the values found in a non-COVID-19 ED setting: 25% versus 86.2%. The specificity for solely cardiac-deep vein POCUS was comparable though 73.5% and 81% [7,10].

Global perspective
COVID-19 has laid bare health disparities along socio-economic, racial, cultural, and ethnic lines within and across regions and nations. CTPA is not always readily available in low-and middle-income countries, especially when resources are overwhelmed. POCUS is an affordable and easy-to-use tool, which could reduce barriers to timely adequate care [10]. Two descriptive studies suggest that POCUS could reduce >50% of CTPAs [33]. The emergence of handheld devices has made POCUS even more accessible. A complete multi-organ POCUS scan takes less than 15 min. Moreover, it is easy to learn for (para)medical personnel [56,57]. In our study, all POCUS scans were performed by ICU residents. Considering that non-specialist physicians see and treat the bulk of COVID-19 patients, this underscores the applicability of POCUS in a real-life setting.

Limitations and strengths
First, this was a single center study with a relatively limited sample size. Still, this is the first and largest prospective study of consecutive COVID-19 patients to investigate the use of multi-organ POCUS in the detection of PE. Furthermore, our sample size was similar to those of most included studies in the aforementioned systematic reviews on VTE [4,47,48]. Second, although we are a tertiary center, our case mix was reflective of the Dutch COVID-19 ICU population since ICU patients were divided across the country according to a fair share principle. As such selection bias was minimal. Third, our way of assessing RVS combines the features most commonly reported in the literature [33]. We decided to use parameters that interpreters would be able to 'eye-ball' as we believed this is simpler, quicker and requires less skill than performing exact measurements (i.e., tricuspid annular plane systolic excursion [TAPSE], or measuring RV:LV ratio), while having similar diagnostic traits [19,23,33]. We thought our results would be applicable to a larger group of operators this way. Fourth, diagnostic accuracy studies on multi-organ POCUS in PE often suffer from suboptimal reporting [33]. Mainly information on the blinding of POCUS operators and the time interval between POCUS and the reference test are rarely reported. In our study, POCUS operators were always blinded from the CTPA results, which minimizes information bias. Time between POCUS and CT was always below 24 h, so we believe any imprecision of the diagnostic accuracy measures was therefore curtailed. Additionally, our reference standard is the current gold standard in diagnosing PE. Our strict diagnostic protocol minimized the chance of missing any PEs, thereby reducing the risk of negatively influencing the diagnostic accuracy measures.

Future research
Future prospective investigations or randomized controlled trials should be conducted to analyze the impact of multi-organ POCUS in cost-effectiveness, reduced radiation exposure, and time to definitive diagnosis of PE. Further research should also focus on aspects that can help distinguish the different etiologies of consolidations like doppler and contrast enhanced ultrasonography [18,[58][59][60][61]. Additionally, it would be interesting to see if POCUS signs could be used in concert with other parameters (i.e. et-CO2 gap or rapid D-dimer rise), or clinical decision rules [62,63].

Conclusion
Multi-organ rather than single-organ POCUS can be of aid in ruling out PE in critically ill COVID-19 and help select patients for CTPA. In addition, finding RVS can make PE more likely in this setting, while a DVT would preclude the need for a CTPA.

Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of Competing Interest
The authors have no conflict of interest nor any financial disclosures.