We congratulate Gil Mosquera et al.1 for their article on the identification of predictors of pulmonary embolism (PE) in patients with SARS-CoV-2 infection (COVID-19).
Numerous studies support the predisposition to the development of PE in patients with COVID-19, conferring a worse prognosis.2 This complication appears to be related to an underlying prothrombotic state in which, among other mechanisms, activation of coagulation and endothelial dysfunction are described. As part of the diagnostic algorithm for PE, clinical probability scores (CPS) are reliable tools that determine pretest clinical probability, helping to avoid unnecessary imaging tests.3 This diagnostic strategy makes more sense in patients with COVID-19, given the risk of nosocomial transmission of infection during patient transfer to the radiology ward and the overlapping signs and symptoms of acute respiratory distress syndrome associated with SARS-CoV-2 infection. Our hypothesis is that, since the pathophysiological mechanism of PE is different in patients with COVID-19, CPS have a low diagnostic yield in this population. The aim of this study was to determine the degree of usefulness of CPS in patients with COVID-19 in our hospital in the event of suspected PE. We conducted an observational, single-center study, based on a retrospective cohort of non-critically ill patients with COVID-19 who underwent chest CT angiography for suspected PE from January 2020 to February 2022. The variables necessary for the construction of the Wells score, revised Geneva score, YEARS criteria and PERC rules at the time of suspected PE were obtained from the electronic medical records. The discriminative capacity of the scores was determined by means of sensitivity, specificity, negative predictive value, positive predictive value, and area under the ROC curve analysis.
A total of 263 suspected PEs were included, of which 129 (49.04%) had PE. 51.3% were male with a mean age of 64.5 (±16.2) years. The predictive ability of the scores is detailed in Table 1. The original Wells score and its dichotomised version adequately discriminated all patients without PD; however, 82.94% (107/129) of patients with PD were classified as low probability. The percentage of PE not identified with the revised Geneva score was 96.89% (125/129), with the score being of low probability in 58.20% (78/134) of patients without PE. Applying the YEARS algorithm, 96.89% (125/129) of the patients with PE would have been correctly detected. On the contrary, this algorithm would have led to the performance of 99 (73.88%) CT angiographies that were not indicated. Using the PERC score a patient with PE would not have been detected at the expense of 127 (94.77%) unnecessary radiological tests.
Predictive ability of the scores in the studied population.
Score | Sensitivity, % | Specificity, % | NPV, % | PPV, % | AUC-ROC (95% CI) |
---|---|---|---|---|---|
Wells ≥ 4 | 4 | 100 | 52 | 100 | 0.519 (0.449−0.589) |
Wells ≥ 2 | 17 | 100 | 56 | 100 | 0.585 (0.516−0.654) |
Geneva ≥ 4 | 83 | 58 | 78 | 66 | 0.706 (0.642−0.769) |
YEARS | 97 | 26 | 90 | 56 | 0.615 (0.547−0.683) |
YEARS + D-dimer ≥ 3,000 ng/mL | 77 | 72 | 76 | 73 | 0.746 (0.685−0.807) |
PERC | 99 | 5 | 88 | 50 | 0.522 (0.452−0.592) |
PERC + D-dimer ≥ 3,000 ng/mL | 75 | 73 | 75 | 73 | 0.742 (0.680−0.803) |
AUC-ROC, area under the ROC curve; 95% CI, 95% confidence interval; NPV, negative predictive value; PPV, positive predictive value.
The CPS analysed did not show predictive ability for the diagnosis of PE in patients with COVID-19. Available evidence has highlighted the role of thromboinflammation in patients with SARS-CoV-2 infection leading to vascular thrombosis in situ.4 Based on this theory, CPS would not have diagnostic yield in patients with COVID-19, mainly because they consider that PE usually originates in the context of a distal embolism and not from a local pulmonary phenomenon. Another limitation of the models lies in considering an alternative diagnosis to PE. Physicians may assume respiratory distress syndrome as the cause of respiratory failure in these patients, except in the absence of pneumonia on chest X-ray.
Patients with severe COVID-19 have significantly elevated D-dimer levels, therefore, it may be necessary to look for levels above 500 ng/mL or age-adjusted as a predictor of thrombosis. The SEMI-COVID-19 registry reported that a cut-off point ≥3,000 ng/mL was useful in predicting venous thromboembolism in these patients.5 In our study, the application of this cut-off point in the YEARS and PERC algorithms would have reduced the number of unnecessary CT angiographies from 99 to 36 and from 127 to 64, respectively, without compromising safety.
Our analysis suffers from the limitations of retrospective studies.
We conclude that new diagnostic prediction scores for PE specifically developed and validated in patients with COVID-19 are needed.
Ethical considerationsThe study was approved by the Clinical Research Ethics Committee of the Hospital Universitario Clínico San Carlos (code 22/282-E). As this was an observational study in which data were analysed retrospectively without clinical intervention on patients, informed consent was not required.
FundingThis research has not received funding.
Conflict of interestsThe authors declare that they have no conflict of interest.