Chest radiography findings of COVID-19 pneumonia: a specific pattern for a confident differential diagnosis

Abstract

Background

Chest radiography (CR) patterns for the diagnosis of COVID-19 have been established. However, they were not ideated comparing CR features with those of other pulmonary diseases.

Purpose

To create the most accurate COVID-19 pneumonia pattern comparing CR findings of COVID-19 and non-COVID-19 pulmonary diseases and to test the model against the British Society of Thoracic Imaging (BSTI) criteria.

Material and Methods

CR of COVID-19 and non-COVID-19 pulmonary diseases, admitted to the emergency department, were evaluated. Assessed features were interstitial opacities, ground glass opacities, and/or consolidations and the predominant lung alteration. We also assessed uni-/bilaterality, location (upper/middle/lower), and distribution (peripheral/perihilar), as well as pleural effusion and perihilar vessels blurring. A binary logistic regression was adopted to obtain the most accurate CR COVID-19 pattern, and sensitivity and specificity were computed. The newly defined pattern was compared to BSTI criteria.

Results

CR of 274 patients were evaluated (146 COVID-19, 128 non-COVID-19). The most accurate COVID-19 pneumonia pattern consisted of four features: bilateral alterations (Expß=2.8, P=0.002), peripheral distribution of the predominant (Expß=2.3, P=0.013), no pleural effusion (Expß=0.4, P=0.009), and perihilar vessels’ contour not blurred (Expß=0.3, P=0.002). The pattern showed 49% sensitivity, 81% specificity, and 64% accuracy, while BSTI criteria showed 51%, 77%, and 63%, respectively.

Conclusion

Bilaterality, peripheral distribution of the predominant lung alteration, no pleural effusion, and perihilar vessels contour not blurred determine the most accurate COVID-19 pneumonia pattern. Lower field involvement, proposed by BSTI criteria, was not a distinctive finding. The BSTI criteria has lower specificity.

Keywords

COVID 19 SARS-CoV-2 virus diagnostic X-ray pneumonitis differential diagnosis

Introduction

The COVID-19 pandemic was declared a world emergency in January 2020 (1). COVID-19 infection may lead to rapidly evolving pneumonia with unfavorable prognosis (2). Nowadays, chest imaging is still used for triage of patients with COVID-19 and assists clinician in both the diagnosis and management of COVID-19 pneumonia (3). Although chest computed tomography (CT) has higher sensitivity compared to chest radiography (CR) in detecting COVID-19 pneumonia (4), the Fleischner statement leaves the choice of imaging modality to the clinical teams, depending on local resources and expertise (3). Because of the lower cost and larger accessibility of CR, this technique remains a valuable resource in an emergency department (ED) setting, also reducing the risk of cross-infection with portable CR systems (5,6). Common features on CR of COVID-19 pneumonia have been described, and CR patterns have been proposed (5,7,8). Among them, the British Society of Thoracic Imaging (BSTI) defined a classic pattern (lower lobe and peripheral predominant multiple opacities that are bilateral >> unilateral), an indeterminate pattern (that does not fit classic or non-COVID-19 descriptors), a non-COVID-19 pattern (pneumothorax/lobar pneumonia/pleural effusion/pulmonary edema), and negative CR, to help ED doctors in diagnosing typical CR alterations of COVID-19 pneumonia (8). However, to define the patterns, the CR features of COVID-19 pneumonia have been not compared to other pulmonary diseases, which could present CR features shared with COVID-19 pneumonia.

Therefore, the aim of the present study was to compare CR features of COVID-19 pneumonia to those of other pulmonary diseases to determine the most specific CR COVID-19 pattern, and then to compare these features to those proposed by the BSTI criteria.

Material and Methods

Study design and patient selection

We retrospectively evaluated patients consecutively admitted to the ED of the Ca' Foncello General Hospital (Treviso, Italy) between March and April 2020 with COVID-19 infection confirmed by real-time reverse transcription polymerase chain reaction (RT-PCR). We then evaluated patients admitted to the ED between March and April 2019, for respiratory symptoms compatible with COVID-19 infection (9), such as cough, dyspnea, sputum production, chest pain, sore throat, or rhinorrhea.

Inclusion criteria were as follows:

available anteroposterior (AP) or posteroanterior (PA) CR, performed at admission;

diagnosis of COVID-19 pneumonia among patients with COVID-19 and of pulmonary disease among non-COVID-19 patients;

good CR image quality.

For both groups, the final diagnosis was reached following local protocols at that time and reported from the ED discharge letters. In particular, following COVID-19 protocol, the diagnosis of COVID-19 pneumonia was verified by CR and lung ultrasound (LUS) and confirmed only when both techniques were concordant. COVID-19 protocol also included a chest CT in case of discrepancy between CR and LUS. LUS was performed by the clinicians at the ED, whereas CR and CT were analyzed by the radiologists.

According to the criteria of the American College of Radiology 2017 statement (10) (Supplementary Material), the quality of CR was assessed by consensus between a thoracic radiologist (NL) with 7 years of experience and a general radiologist (GM) with 25 years of experience.

The following clinical data were recorded in an anonymous database: age, gender, time interval between disease onset and CR, presence of fever, respiratory symptoms, and discharge diagnosis.

The study was approved by the local ethical committee (protocol 837/CE, CESC Treviso and Belluno).

Image acquisition

All CR were performed using a portable CR system (Multix; Siemens, Enlargen, Germany). PA and sitting AP CR had a tube voltage in the range of 110–140 kVp with Automatic Exposure Control. AP bedside CR was performed with 85 kVp and 4.0 mAs.

Image analysis

The most frequent CR findings of COVID-19 pneumonia (11 –14) were identified by literature search and named “main alterations”: interstitial opacities (i.e. interlobular septal thickenings, resembling or not a reticular pattern); ground glass opacities (GGO); or consolidation. In our analysis, the superimposition of interstitial opacities and GGO were considered GGO.

All CRs were fully anonymized and assessed in random order. One chest radiologist with 7 years of experience in chest imaging (NL, reader 1) evaluated all CR. To assess intra-reader concordance, reader 1 repeated the analysis, after one month, on 100 randomly selected images. For the inter-reader concordance, two general radiologists, with 20 years (AD, reader 2) and 10 years (LB, reader 3) of experience, evaluated 50 randomly selected images. Radiologists were blinded to the clinical information of the patients.

All readers were asked to evaluate the following CR features: main alterations presence or absence, their uni-/bilaterality, prevalent location (upper, middle, lower, no prevalence), and distribution (peripheral, perihilar, no prevalence). Moreover, they had to define the predominant finding, as well as its location and distribution. Location was defined by three zones (upper, medium, lower), adapting the division in lung fields as previously described (15) (Fig. 1). Peripheral and perihilar distribution were defined as reported by Wong et al. (16) (Fig. 1). The presence of nodules, bands, cavitation, pneumothorax, pleural effusion, increased heart size (cardiothoracic ratio >0.5), or blurred contour of hilar pulmonary vessels (17,18) were also assessed.

Fig. 1.

Chest radiography anatomical division. Location: Upper-middle-lower zones are identified by two anatomical landmarks: (a) the inferior wall of the aortic arch and (b) halfway between A and the dome of the lower hemi-diaphragm. Distribution: peripheral and perihilar zone are delimited by the halfway between the lateral edge of the lung and hilum (dotted line).

Each radiologist performed a preliminary training on 10 CR of COVID-19 pneumonia and 10 CR of non-COVID-19 pulmonary disease. CR exploited for the training were not included in the final dataset. All lung alterations were defined according to the Fleischner glossary of terms for thoracic imaging (19) (Supplementary Material).

Statistical analysis

Intra- and inter-reader agreements were investigated using intraclass correlation coefficient (ICC). We used a two-way mixed model for the intra-reader agreement and a two-way random model for the inter-reader agreement. ICC were considered as follows: <0.40 = poor; 0.40–0.59 = fair; 0.60–0.74 = good; and 0.75–1.00 = excellent (20). To define a COVID-19 pattern on CR, we considered the findings scored by the reader 1 with a good or very good intra-reader agreement. Moreover, the features should have been described in at least 10 individuals per group. All radiographic features were assessed as dichotomous variables according to presence or absence. We used these variables to calculate odds ratios (OR) (21) between COVID-19 and non-COVID-19 individuals, together with 95% confidence intervals (CI) and Fisher’s exact test (22). P < 0.05 was considered significant. We adopted Bonferroni correction for multiple testing (23).

Finally, considering only the variables with a significant P value, we defined a CR pattern to predict the presence of COVID-19 pneumonia compared to non-COVID-19 patients, using a binary logistic regression model (24): each radiographic feature that had a non-significant P value on the regression analysis was removed. Then, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of the selected CR features were calculated. Lastly, the same analysis was repeated to test the efficacy of the BSTI CR criteria for COVID-19, selecting the following variables: bilateral main alterations; prevalent peripheral distribution; prevalent lower localization; bilateral involvement; absence of pleural effusion; absence of blurred contour of hilar pulmonary vessels; and absence of pneumothorax. The data were analyzed using SPSS version 25.0 (IBM Corp., Armonk, NY, USA).

Results

Patients

A total of 300 patients had an available PA or AP CR. All patients with COVID-19 had confirmed COVID-19 pneumonia, while 21 non-COVID-19 patients were excluded because a diagnosis of pulmonary disease was ruled out at dismission. Five patients (four with COVID-19, one without COVID-19) were discarded because of low image quality.

Thus, 274 patients were finally included in the study (127 PA and 147 AP CR examinations): 146 (53%) patients with COVID-19 and 128 (47%) non COVID-19 patients (Fig. 2).

Fig. 2.

Flow chart of patients’ inclusion. ED, emergency department; RT-PCR, real-time reverse transcription polymerase chain reaction.

Among patients with COVID-19, 21 (14.4%) needed a CT scan at the time of diagnosis to confirm COVID-19 pneumonia. The final diagnoses of non-COVID-19 patients included the following: 64 (50%) cases of pneumonia other than COVID-19; 40 (31%) cases of cardiopulmonary disease (i.e. pulmonary edema or pulmonary embolism); 24 (19%) cases of other diseases, including bronchitis, suspected tumors, and chronic obstructive pulmonary diseases. The mean time interval between disease onset and CR ± standard deviation was 6.9 ± 3.9 days and 6.9 ± 9.1 days in COVID-19 and non-COVID-19 individuals, respectively. Patients’ characteristics are shown in Table 1.

Table 1.

Patient characteristics.

	All	COVID-19	Non-COVID-19	P value
Patients	274 (100)	146 (53)	128 (47)	-
Age (years)	70 ± 17	68 ± 15	73 ± 18	0.001
Male	151 (55)	90 (62)	61 (48)	0.020
Time interval* (days)	6.9 ± 6.5	6.9 ± 3.9	6.9 ± 9.1	>0.05
Symptoms
- Cough	135 (49)	83 (57)	52 (41)	0.007
- Dyspnea	54 (20)	54 (37)	88 (67)	<0.001
- Chest pain	30 (11)	5 (3)	25 (20)	<0.001
Fever	137 (50)	96 (66)	41 (32)	<0.001
Diagnosis
- COVID-19 pneumonia	146 (53)	146 (100)	0 (0)	-
- Non-COVID-19 pneumonia	64 (23)	0	64 (50)	-
- Cardiopulmonary disease	40 (15)	0	40 (31)	-
- Other^†	24 (9)	0	24 (19)	-
Chest radiography
- PA	127 (46)	68 (47)	59 (46)	>0.05
- AP	147 (54)	78 (53)	69 (54)	>0.05

Values are given as n (%) or mean ± SD.

*Time from onset of disease to chest radiography.

^†

Other: bronchitis, suspected tumors, chronic obstructive pulmonary disease.

AP, anteroposterior; PA, posteroanterior.

Intra-reader and inter-reader concordance

Intra-reader concordance was good or excellent for all variables, except for bands and increased heart size (fair). inter-reader concordance was moderate for most variables, except for prevalent localization (good), main alteration location (good, for reader 3), and blurred contour of hilar pulmonary vessels (fair, for reader 2). Intra and inter-reader agreements are shown in Table 2.

Table 2.

ICCs for intra- and inter-reader agreement.

		Inter-reader
CR features	Intra-reader	Reader 2	Reader 3
Presence of main alterations	0.804*	0.561	0.496
Bilaterality	0.825*	0.594	0.573
Prevalent localization	0.868*	0.613^†	0.650^†
Prevalent distribution	0.831*	0.521	0.589
PA	0.872*	0.467	0.572
PA prevalent distribution	0.767*	0.529	0.597
PA prevalent localization	0.801*	0.580	0.642^†
Nodules	0.852*	0.498	0.499
Cavitation	NA	NA	NA
Bands	0.464	0.472	0.452
Pleural effusion	0.886*	0.578	0.521
Pneumothorax	1.000*	NA	NA
Increased heart size	0.566	0.504	0.493
Perihilar vessel contour	0.714^†	0.355	0.521

ICC values of 0.4–0.6, 0.6–0.8, or ≥0.8 are considered to indicate moderate, good, and excellent agreement, respectively.

Excellent agreement

^†

Good agreement

CR, chest radiography; ICC, intraclass correlation coefficient; NA, not available; PA, predominant alteration.

CR findings and odds ratios

CRs were positive for main alterations in 104 (71%) COVID-19 and 69 (54%) non-COVID-19 patients. Main alterations were bilateral in 91 (62%) COVID-19 and 54 (42%) non COVID-19 patients. The most frequent location was the lower zone: 74 (51%) COVID-19 and 54 (43%) in non-COVID-19 patients. The prevalent distribution was usually peripheral in COVID-19 patients (n = 54, 38%), while there was no prevalence in 39 (30%) non-COVID-19 patients. The predominant CR feature was more often represented by interstitial opacities in both groups: 62 (42%) COVID-19 and 45 (35%) non-COVID-19 patients. The predominant CR feature was more often involving lower zones in both groups (79 COVID-19 patients [54%] and 52 non-COVID-19 patients [52%]), the periphery in COVID-19 patients (n = 61, 42%) and had no prevalent distribution in non-COVID-19 patients (n = 42, 33%). The contour of hilar pulmonary vessels was blurred in 26 (18%) COVID-19 and 42 (33%) non-COVID-19 patients. All CR features are reported in Table 3.

Table 3.

Chest radiography features of COVID-19 and non-COVID-19 patients.

CR features		COVID-19	Non-COVID-19	P value
Main alterations	Positive	104 (71)	69 (54)	0.021
Side	Unilateral	13 (9)	15 (12)	0.285
	Bilateral	91 (62)	54 (42)	0.003
Prevalent location	Upper	7 (5)	2 (2	-
	Middle	13 (9)	2 (2)	-
	Lower	74 (51)	55 (43)	0.069
	Diffuse	10 (7)	10 (8)	-
Prevalent distribution	Peripheral	56 (38)	20 (16)	<0.001
	Central	10 (7)	10 (8)	-
	No prevalence	38 (26)	39 (30)	0.248
Predominant alteration	Interstitial opacities	62 (42)	45 (35)	0.251
	GGO	39 (27)	19 (15)	0.016
	Consolidation	3 (2)	5 (4)	-
PA prevalent location	Upper	5 (3)	5 (4)	-
	Middle	13 (9)	1 (1)	-
	Lower	79 (54)	52 (41)	0.206
	No prevalence	7 (5)	11 (9)	-
PA prevalent distribution	Peripheral	61 (42)	20 (16)	<0.001
	Central	8 (5)	7 (5)	-
	No prevalence	35 (24)	42 (33)	0.069
Nodules	Positive	10 (7)	7 (5.5)	-
Pleural effusion	Positive	18 (12)	35 (27)	0.001
Pneumothorax	Positive	0 (0)	1 (1)	-
Increased heart size	Positive	57 (39)	61 (48)	0.094
Blurred contour of perihilar vessels	Positive	26 (18)	42 (33)	0.003

Values are given as n (%).

CR, chest radiography; GGO, ground glass opacities; PA, predominant alteration.

In total, 14 CR features were present with >10 patients per group, with good or very good intra-reader concordance, and were, therefore, further analyzed. After Bonferroni correction, the P value threshold was set to 0.003. Significant OR for patients with COVID-19 were as follows: bilaterality of main alterations (OR = 2.0, 95% CI = 1.2–3.3; P = 0.003); peripheral distribution (OR = 3.3, 95% CI = 1.8–5.8; P < 0.001); prevalent peripheral distribution of the predominant alteration (OR = 3.7, 95% CI = 2.0–6.5; P < 0.001); conversely, for non-COVID-19 patients, ORs were as follows: pleural effusion (OR = 0.4, 95% CI = 0.2–0.7; P = 0.001) and blurred contours of hilar pulmonary vessels (OR = 0.4, 95% CI = 0.3–0.8; P = 0.003).

Best CR COVID-19 pattern and BSTI model performance

By selecting the CR features with a significant P value, we tested them with a logistic regression to define the most specific CR COVID-19 features. The final pattern included four features (Table 4): bilateral main alterations (Expβ = 2.852, P = 0.002); prevalent peripheral distribution of the predominant alteration (Expβ = 2.287, P = 0.013); absence of pleural effusion (Expβ = 0.391, P = 0.009); and no blurred contours of hilar pulmonary vessels (Expβ = 0.341, P= 0.002). This pattern showed a sensibility of 48.6% (95% CI = 40.28–57.03), specificity of 81.3% (95% CI = 73.40–87.60), PPV of 74.7% (95% CI = 66.54–81.49), NPV of 58.1% (95% CI = 53.71–62.37), and accuracy of 63.9% (95% CI = 57.87–68.56). BSTI criteria showed a sensitivity of 50.7% (95% CI = 42.29–59.05), specificity of 76.6% (95% CI = 68.26–83.59), PPV of 71.2% (95% CI = 63.44–77.81), NPV of 57.7% (95% CI = 52.95–62.21), and accuracy of 62.8% (95% CI = 56.75–68.51).

Table 4.

Results of linear regression for CR features in COVID-19 pneumonia.

CR features	Exp-β (95% CI)	P value
Bilaterality	2.852 (1.491–5.458)	0.002
PA peripheral distribution	2.287 (1.187–4.409)	0.013
Pleural effusion	0.391 (0.194–0.788)	0.009
Blurred perihilar vessels	0.341 (0.170–0.684)	0.002

CI, confidence interval; CR, chest radiography; Exp-β, odds ratio for a one-unit increase in CR features; PA, predominant alteration.

Discussion

In the present study, we compared CR features of COVID-19 pneumonia to those of other pulmonary diseases, to determine the most accurate CR COVID-19 pattern for a confident diagnosis at the time of admission to the ED. We then compared these features to the BSTI criteria.

Our analysis showed that the most accurate CR pattern for COVID-19 pneumonia diagnosis consists of four features (Fig. 3): (i) bilateral main alterations (namely interstitial opacities, GGO, or consolidation); (ii) peripheral distribution of the predominant alteration; (iii) absence of pleural effusion; and (iv) contours of perihilar pulmonary vessels not blurred.

Fig. 3.

Application of COVID-19 pneumonia CR models. Sitting in AP CR, COVID-19 pneumonia fits our pattern: bilateral alterations, mostly sustained by GGO with a peripheral predominant distribution (consolidations are also present), absence of pleural effusion, and contour of perihilar vessels not blurred. The alterations mostly involve middle fields: this aspect does not fit the BSTI classic pattern. A nodule is also present in the upper zone of the right lung. Sitting in AP CR, COVID-19 pneumonia fits both our and BSTI classic patterns: bilateral alterations with predominant peripheral and lower involvement distribution (mostly by GGO in this example), absence of pleural effusion, pneumothorax or pulmonary edema, with contour of perihilar vessels not blurred. AP, anteroposterior; BSTI: British Society of Thoracic Imaging; CR, chest radiography; GGO, ground glass opacities.

Our pattern showed a slightly better performance than the BSTI criteria, with higher specificity. Looking at single CR features, the most interesting result is that a lower field distribution of alterations was confirmed as a common feature in patients with COVID-19 but did not result as a distinctive sign, being frequent also in non-COVID-19 patients. Conversely, bilaterality and peripheral distribution were confirmed as useful CR features to diagnose COVID-19 pneumonia. This is in agreement with the findings reported on CT by Shi et al. (25): they observed that in COVID-19 pneumonia all lung segments can be involved, with a prevalent peripheral distribution. Moreover, the peripheral distribution of the predominant alteration is more determinant from a differential diagnostic perspective. On the other hand, other features such as nodules, bands, cavitation, and pneumothorax were rarely encountered in both groups.

In the present study, we also exploited the observed findings to test the efficacy of the BSTI criteria, matching the results of some of the previous works, that directly applied to that model. In particular, the BSTI criteria obtained the same specificity of Tsakok et al. (76%) (26), though they observed higher sensitivity (60%), while Borakaty et al. (27) obtained a slightly higher sensitivity but lower specificity (56% and 69%, respectively). However, the latter considered only GGO among CR opacities, while we considered all most common CR features of COVID-19 pneumonia. Kemp et al. (28) found higher sensitivity and specificity (84 and 83%): nevertheless, statistical analysis was performed on readings by non-radiologists. Authors suggested that clinicians could have been prone to an overdiagnosis, as radiologists more often described CR as negative (data on radiologists’ analysis were not provided). Lastly, Hare et al. (8) described a 44% sensitivity and 100% specificity of the BSTI classic pattern. However, readers were informed on COVID-19 status: we believe that this approach may lead to a significant analysis bias.

Moreover, there are some differences in the aforementioned studies that need to be specified and that, in our opinion, may increase the strength of our work. First, in the other works, with the only exception of Hare et al. (8), patients were chosen in the same time period, splitting COVID-19 and non-COVID-19 by RT-PCR (25 –27). However, RT-PCR can fail in the diagnosis of COVID-19 (4), especially in the early phase of COVID-19 disease, and patients with negative RT-PCR could have been wrongly included among non-COVID-19 patients. For this reason, we decided to enroll non-COVID-19 patients admitted to the ED the year before, when COVID-19 was not yet present. Additionally, we searched our archives during the same range of months (March–April), in order to have similar epidemiological/seasonal conditions. We then selected non-COVID-19 patients among those with respiratory symptoms, while Hare et al. (8) enrolled non-COVID-19 patients with cough and fever. Fever is one of the most common signs in patients with COVID-19 (9) but it was present in less than two-thirds of the total amount of our patients with COVID-19. Thus, we preferred evaluating non-COVID-19 patients with respiratory symptoms, regardless of fever, in order to address patients’ enrollment on chest disease only, but ensuring a large spectrum of possible differential diagnosis. Lastly, all studies (8,26 –28) enrolled COVID-19 patients with positive RT-PCR, while we included COVID-19 patients with both RT-PCR positivity and reported diagnosis of COVID-19 pneumonia. In fact, patients with COVID-19 may be infected without evidence of pneumonia, as demonstrated by CT, the most sensitive imaging tool for COVID-19 pneumonia diagnosis (4). Similarly, we enrolled non-COVID-19 patients with a diagnosis of pulmonary disease.

Regarding inter-reader concordance, the moderate agreement for most of the features was not a surprise: a moderate inter-reader agreement was also observed in BSTI criteria application (8). This highlights the importance of cardiothoracic expertise in imaging evaluation of CR. Bands and heart size analysis had a poor to moderate intra-reader agreement; however, bands were uncommon in both groups and may not influence the resulting pattern. Regarding heart size, moderate intra-reader agreement was expected, since supine AP CR determines a magnification of the heart shadow that would need a correction (17). Following a similar rationale, we preferred to analyze the sharpness of hilar vessels contour (17,18) instead of hilar dimensions (29), as a possible sign of pulmonary edema: in fact, vascular pedicle dimension is also influenced by the same geometric issue (30).

The present study has some limitations. First, the patients were evaluated over a short period and the epidemiological background may not be representative of the whole year. Moreover, to create the CR pattern, we adopted the score of the most experienced reader. This choice aimed to select the radiologist dedicated to chest imaging, and, presumably, with the highest intra-reader concordance for all CR features. Second, protocols for disease diagnosis at the ED were different between COVID-19 and non-COVID-19 patients, as the combination of CR and LUS was specifically defined to manage the COVID-19 wave. Moreover, the ongoing therapy of admitted patients, such as antibiotics and corticosteroids, could have potentially modified imaging manifestation of COVID diseases at time of CR examination. Finally, although the heterogeneity of the non-COVID-19 group ensure a good representation of the general population with pulmonary diseases encountered at the ED, the small sample size per subgroup did not allow for the assessment of the influence of specific disease (i.e. pulmonary edema) in determining the final CT pattern. Further studies on a larger number of patients with a specific non-COVID-19 disease, such as other infective pneumonitis, are warranted.

In conclusion, we proposed a new CR pattern for the identification of COVID-19 pneumonia that should be helpful in differential diagnosis, consisting of four features: (i) bilateral involvement of interstitial opacities, GGO, or consolidations; (ii) prevalent peripheral distribution of the predominant alteration; (iii) absence of pleural effusion; and (iv) perihilar pulmonary vessels not blurred. Bilaterality and a predominant peripheral involvement are confirmed as features able to discriminate between COVID-19 pneumonia and non-COVID-19 pulmonary diseases, but lower field location is not specific for COVID-19 pneumonia. This pattern slightly outperformed the BTSI criteria in terms of specificity and accuracy.

Supplemental Material

sj-doc-1-acr-10.1177_02841851211055163 - Supplemental material for Chest radiography findings of COVID-19 pneumonia: a specific pattern for a confident differential diagnosis

Supplemental material, sj-doc-1-acr-10.1177_02841851211055163 for Chest radiography findings of COVID-19 pneumonia: a specific pattern for a confident differential diagnosis by Nicholas Landini, Giulia Colzani, Pierluigi Ciet, Giovanni Tessarin, Alberto Dorigo, Luca Bertana, Carla Felice, Luca Scaldaferri, Martina Orlandi, Cosimo Nardi, Micaela Romagnoli, Luca Saba, Roberto Rigoli and Giovanni Morana in Acta Radiologica

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship and/or publication of this article.

ORCID iDs

Nicholas Landini

Giovanni Tessarin

Supplemental material

Supplemental material for this article is available online.

References

Garg

Gupta

Maralakunte

, et al. Diagnostic accuracy of CT and radiographic findings for novel coronavirus 2019 pneumonia: systematic review and meta-analysis. Clin Imaging 2021;72:75–82.

Larici

Cicchetti

Marano

, et al. Multimodality imaging of COVID-19 pneumonia: from diagnosis to follow-up. A comprehensive review. Eur J Radiol 2020;131:109217.

Rubin

Ryerson

Haramati

, et al. The role of chest imaging in patient management during the COVID-19 pandemic: a multinational consensus statement from the Fleischner society. Radiology 2020;296:172–180.

Islam

Ebrahimzadeh

Salameh

, et al. Thoracic imaging tests for the diagnosis of COVID-19. Cochrane Database Syst Rev 2021;3:CD013639.

Litmanovich

Chung

Kirkbride

, et al. Review of chest radiograph findings of COVID-19 pneumonia and suggested reporting language. J Thorac Imaging 2020;35:354–360.

Landini N, Orlandi M, Fusaro M, et al. The Role of Imaging in COVID-19 Pneumonia Diagnosis and Management: Main Positions of the Experts, Key Imaging Features and Open Answers. J Cardiovasc Echogr. 2020; 30(Suppl 2):S25–S30.

Dennie

Hague

Lim

, et al. Canadian Society of Thoracic Radiology/Canadian Association of Radiologists consensus statement regarding chest imaging in suspected and confirmed COVID-19. Can Assoc Radiol J 2020;71:470–481.

Hare

Tavare

Dattani

, et al. Validation of the British Society of Thoracic Imaging guidelines for COVID-19 chest radiograph reporting. Clin Radiol 2020;75:710.e9–710.e14.

Rodriguez-Morales

Cardona-Ospina

Gutiérrez-Ocampo

, et al. Clinical, laboratory and imaging features of COVID-19: a systematic review and meta-analysis. Travel Med Infect Dis 2020;34:101623.

10.

ACR–SPR-STR Commettees. ACR–SPR-STR practice parameter for the performance of chest radiography 2017. Available at: https://www.acr.org/-/media/ACR/Files/Practice-Parameters/ChestRad.pdf (accessed March 2021).

11.

Cozzi

Albanesi

Cavigli

, et al. Chest X-ray in new coronavirus disease 2019 (COVID-19) infection: findings and correlation with clinical outcome. Radiol Med (Torino) 2020;125:730–737.

12.

Ippolito

Maino

Pecorelli

, et al. Chest X-ray features of SARS-CoV-2 in the emergency department: a multicenter experience from northern Italian hospitals. Respir Med 2020;170:106036.

13.

Vancheri

Savietto

Ballati

, et al. Radiographic findings in 240 patients with COVID-19 pneumonia: time-dependence after the onset of symptoms. Eur Radiol 2020;30:6161–6169.

14.

Vespro

Andrisani

Fusco

, et al. Chest X-ray findings in a large cohort of 1117 patients with SARS-CoV-2 infection: a multicenter study during COVID-19 outbreak in Italy. Intern Emerg Med 2021;16:1173–1181.

15.

Borghesi

Maroldi

. COVID-19 outbreak in Italy: experimental chest X-ray scoring system for quantifying and monitoring disease progression. Radiol Med 2020;125:509–513.

16.

Wong

HYF

Lam

HYS

Fong

, et al. Frequency and distribution of chest radiographic findings in patients positive for COVID-19. Radiology 2020;296:E72–E78.

17.

Milne

Pistolesi

Miniati

, et al. The radiologic distinction of cardiogenic and noncardiogenic edema. Am J Roentgenol 1985;144:879–894.

18.

McHugh

Forrester

Adler

, et al. Pulmonary vascular congestion in acute myocardial infarction: hemodynamic and radiologic correlations. Ann Intern Med 1972;76:29–33.

19.

Hansell

Bankier

MacMahon

, et al. Fleischner society: glossary of terms for thoracic imaging. Radiology 2008;246:697–722.

20.

Koo

. A guideline of selecting and reporting intraclass correlation coefficients for reliability research. J Chiropr Med 2016;15:155–163.

21.

Andrade

. Understanding relative risk, odds ratio, and related terms: as simple as It Can Get. J Clin Psychiatry 2015;76:e857–e861.

22.

Ernst

Basten

Ilsen

, et al. Pulmonary disease in cystic fibrosis: assessment with chest CT at chest radiography dose levels. Radiology 2014;273:597–605.

23.

Armstrong

. When to use the Bonferroni correction. Ophthalmic Physiol Opt 2014;34:502–508.

24.

Austin

Merlo

. Intermediate and advanced topics in multilevel logistic regression analysis: multilevel logistic regression. Stat Med 2017;36:3257–3277.

25.

Shi

Han

Jiang

, et al. Radiological findings from 81 patients with COVID-19 pneumonia in Wuhan, China: a descriptive study. Lancet Infect Dis 2020;20:425–434.

26.

Tsakok

Shaw

Murchison

, et al. Diagnostic accuracy of initial chest radiograph compared to SARS-CoV-2 PCR in patients with suspected COVID-19. BJR Open 2020;2:20200034.

27.

Borakati

Perera

Johnson

, et al. Diagnostic accuracy of X-ray versus CT in COVID-19: a propensity-matched database study. BJR Open 2020;10:e042946.

28.

Kemp

Watson

Swanson-Low

, et al. Comparison of chest X-ray interpretation by emergency department clinicians and radiologists in suspected COVID-19 infection: a retrospective cohort study. BJR Open 2020;2:20200020.

29.

Pistolesi

Milne

Miniati

, et al. The vascular pedicle of the heart and the vena azygos. Part II: acquired heart disease. Radiology 1984;152:9–17.

30.

Milne

Pistolesi

Miniati

, et al. The vascular pedicle of the heart and the vena azygos. Part I: the normal subject. Radiology 1984;152:1–8.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.02 MB