Integrated Lung Ultrasound for the Diagnosis of COVID-19

By Michael Prats


Lung ultrasound for the diagnosis of SARS-CoV-2 pneumonia in the Emergency Department

Annals of Emerg Med Oct 2020 (In Press)

Take Home Points

1. In this high-prevalence population, combining 12-point lung ultrasound with clinical evaluation, can identify cases of COVID-19 pneumonia with a higher sensitivity than initial PCR testing.

2. External validity of these findings are limited due to the highly experienced operators and the subjective determination of a positive test.

3. More evidence is needed to show that incorporating ultrasound into a diagnostic approach is superior to relying on clinical suspicion alone.


Notice: We are still in a pandemic and COVID-19 is still killing people daily. We know that ultrasound is accurate for diagnosing pneumonia caused by SARS-CoV-2 based on several articles. In fact, this is our third post on this disease - see our review of the very early literature and a nice study on how lung ultrasound correlates with severity. By now, most places have a system for diagnosis, usually involving various PCR assays for the virus. None of these tests are perfect, but you work with what you have. Most seem to be very specific, but the lack of sensitivity leads to the problems with false negatives. Any diagnostic strategy that could minimize false negatives would mean less missed cases of COVID-19, which means those patients can be more effectively quarantined or treated. Well - what if ultrasound was actually better than those tests?! It seems well positioned to play such a role. Well these authors dare to investigate that possibility.

Check out more on the prior evidence of POCUS for COVID19 at the POCUS Atlas!


How accurate is an integrated clinical + lung ultrasound exam for the diagnosis of SARS-COV-2 compared to an initial PCR?


Italian study


  • ≥18 years

  • Symptoms: fever, dyspnea, cough, sore throat, diarrhea, ageusia, anosmia, asthenia


  • Known SARS-COV2 infection

  • Urgent psychiatric assessment needed

  • Intubated at time of ED arrival

  • Attending physician with LUS expertise unavailable


Prospective cohort study

Patients arriving to the emergency department (ED) were screened for inclusion.

All potential COVID-19 patients were placed in a separate part of the emergency department (which included everyone in this study).

Patients first received initial evaluation (including history, exam, and ECG).

Treating physician recorded whether or not they thought this was SARS-COV2 pneumonia (dichotomous Yes/No).

Ultrasound was performed prior to labs and chest radiography.

Then same (treating) physician performed LUS and recorded new diagnostic hypothesis (Yes/No) based on incorporating ultrasound into everything else. By this point, many of the patients did have results from an ABG. It is important to note that this is called the “Integrated” Clinical and Lung Ultrasound outcome - because they are using both their initial clinical information (history, exam) and also any ultrasound findings.

After LUS, all patients underwent CXR and/or CT depending on clinical indication

All patients received SARS-COV-2 nasopharyngeal swabs RT-PCR (reverse transcription polymerase chain reaction). Positive tests were considered positive.

In patients with negative initial PCR, a second swab was repeated within 72 hours when clinical, sonographic, laboratory, or imaging evaluations were suspicious for COVID-19. They could also perform tests for other viruses/pathogens at their discretion.

If negative initial PCR but nothing else suspicious - no further SARS-CoV-2 testing required.

All patients had 30 day follow up by telephone (if discharged) or in person (if still in hospital).

Total Ways a patient could be considered COVID-19 positive:

  • Any PCR positive (first or second)


Primary outcome: Diagnostic accuracies of clinical evaluation, clinical-LUS integrated assessment, and first PCR compared to reference standard of determination of COVID-19 based on ultimate diagnosis from repeat PCR or BAL.

Power analysis: needed 228 patients for difference in 8% in sensitivity.

Who did the ultrasounds?

20 attending physicians who has completed a lung ultrasound training course accredited by the Italian Society of Emergency Medicine + at least 40 LUS exams

All physicians received training modules for this study including scanning protocol, LUS findings, and also describing this study (so they were not blinded to the study).

The Scan

Curvilinear probe, abdominal preset from cart-based system or a hand-held device (Butterfly IQ) on lung settings were used.


12 zone protocol (6 per side) - upper and lower anterior, upper and lower lateral, upper and lower posterior on both sides

Presence of focal or diffuse interstitial syndrome with spared areas (“skip lesions”), subpleural consolidations, irregular/thickened pleural line was considered suggestive BUT they used a pragmatic approach - no specific definition, just incorporating findings into the post-test decision.

Learn how to do Ultrasound for Viral Pneumonia on 5 Minute Sono!


Check out COVID-19 Pathology on the POCUS Atlas!



N = 228

  • 46.9% diagnosed with COVID19

  • Median age 57.7

  • Median of 6 days from onset of symptoms

Of ALL patients (not just COVID19)

  • 53.5% discharged home

  • 39.5% admitted

  • 3.9% ICU admission

  • 3.1% died in ED

Of COVID19 patients

  • 25.2% discharged home, 60.8% admitted, 7.5% to ICU, 6.5% died in ED

142 patients tested negative on initial PCR, 51(35.9%) were re-tested, 21 (41.2% of those re-tested) resulted positive

Primary Outcome - Test characteristics of the 3 Diagnostic Strategies

Clinical evaluation alone:

Sensitivity: 81.3% (CI 72.6-88.2)

Specificity: 63.6% (CI 54.4-72.2)

+LR 2.94 (CI 1.7-2.9)

-LR 0.29 (CI 0.19-0.45)

Clinical-LUS integrated assessment:

Sensitivity 94.4% (CI 88.2-97.9)

Specificity 95% (CI 89.5-98.2)

+LR 19 (CI 8.7-41.6)

-LR 0.06 (CI 0.03-0.13)

First PCR:

Sensitivity 80.4% (CI 71.6-87.4)

Specificity would have been 100% because it was part of the reference standard

-LR 0.20 (CI 0.13-0.29)

Other Findings

Net reclassification index (NRI) (this is a controversial statistic to use by the way) for positive COVID-19 infections was 14.0% (CI 8.1-22.1). This means that based on integrated clinical LUS 14% of the people they previously thought did not have it, they would diagnose with it after the test.

LUS was performed within a median of 10 minutes from clinical evaluation. Median duration of scan was 5 minutes.

Interestingly, integrated clinical LUS identified all of the patients with COVID-19 who were negative on initial PCR. The 6 false negatives came from patients that were negative by clinical-LUS but had positive initial PCR.

Of all of the initial false-negative PCRs (n=21), 17 (81%) were identified by Clinical alone. Only an additional 4 were caught by adding ultrasound.

30 day follow up found that all patients diagnosed as SARS-CoV-2 negative in this study did not get a subsequent diagnosis or events consistent with it.



True POCUS - incorporating all of the information with the ultrasound. Ultrasounds performed by same people who were treating patients.


Convenience sample. Highly trained operators.

Incorporation bias - patient deemed negative did not get further testing at times. Confirmation bias - this was in a “COVID-19” part of the ED. If they thought that going into the exam, could have swayed them in interpretation of findings (especially because these were not objectively specified).

Occurred over 20 days. Obviously, given that this study incorporated a subjective element and occurred during a pandemic - these results are unlikely to be extrapolatable to non-pandemic times or environments. The test characteristics could completely change in a time of much lower prevalence, where the providers are not suspecting COVID-19 so often. This could particularly affect the specificity, as other viral pneumonias could eventually become more common as COVID cases downtrend.

Some subjectivity to the test. The integrated clinical-LUS evaluation, although it did incorporate findings of POCUS, did not use objective criteria in a way that it could easily be reproduced. If the clinician had an extremely high pretest probability of disease (e.g. - a roommate with positive test, all of the symptoms, etc), could that have influenced them to be more likely to identify lung pathology on their scans?

A binary yes/no does not often reflect actual decision making. Sometimes if you are forced to make an educated guess you can get it right - doesn’t mean that you would count on that decision in real life. For example, a theoretical “100% sure it’s COVID” versus “51% sure it’s COVID” were both counted as “YES, COVID19” (for both the clinical and the lung US end points). Again, a spectrum of confidence in the clinical diagnosis might affect the subsequent lung ultrasound interpretations to varying degrees.


What do we do with this information? The authors suggest still getting the PCR, but perhaps maintaining a higher level or precaution (isolation) in patients with positive LUS findings. This sounds reasonable, but the utility would likely vary based on practice environment and current practices. Most places with decent resources are probably relying on a combination of rapid testing, PCR, and clinical suspicion in the case of possible false negatives. Since in this study, clinical suspicion did pretty well without ultrasound (more sensitive than PCR in fact!), it is reasonable to weigh the pros and cons of adding ultrasound. Pros - more accurate. Cons - time, decontamination, vector for virus?


In this prospective convenience sample study out of Italy, they enrolled 228 patients with suspected COVID-19 pneumonia. In a population with 46.9% positive COVID-19, they found that a LUS integrated with clinical evaluation had a higher sensitivity than the initial PCR assay (94.4% vs 80.4%).

Take Home Points

1. In this high-prevalence population, combining 12-point lung ultrasound with clinical evaluation, can identify cases of COVID-19 pneumonia with a higher sensitivity than initial PCR testing.

2. External validity of these findings are limited due to the highly experienced operators and the subjective determination of a positive test.

3. More evidence is needed to show that incorporating ultrasound into a diagnostic approach is superior to relying on clinical suspicion alone.

Our score

3 Probes

Expert Reviewer for this Post


Christopher D Thom, MD RDMS @ThomCt9k

Assistant Professor of Emergency Medicine at the University of Virginia and Assistant Director of Emergency Ultrasound

Reviewer's Comments

When the initial PCR was negative, the integrated clinical LUS was 100% sensitive in not missing any additional positives that might have gotten through that initial negative test. It was only in cases of an initial positive PCR test where the integrated clinical LUS had false negative results (n of 6). That makes this integrated protocol look even better than the authors purported. After all, in the real world, we might well be most interested in sensitivity of clinical suspicion, lung US, and the initial RT-PCR test combined. It would be interesting to also have data on a blinded physician performing US scans or have that study design also reflected in future work to avoid the bias of clinical suspicion influencing what we call on US (….at least if the definition of positive or negative US is left completely up in the air). Would have been nice if the authors had included more data on what their US findings were (mostly just B lines, or a lot of irregular pleural findings?), as this may impact our thinking on specificity if we get to downtrending numbers of COVID cases.

I’d be curious if they had more accuracy with one type of probe or another. I doubt they have enough power in the study to detect that, but it’s interesting that they used one curvilinear probe and then a Butterfly probe for the study. They didn’t specify how many patients had one device versus the other. Did one perform better than the other? Was one felt to be easier to decontaminate than the other? In a cart based system, I would often pull out the phased array probe on the lung setting for this evaluation and not the curvilinear, so I found that to be an interesting choice.

I might also opine that their rates of negative LUS (both in high clinical suspicion and low clinical suspicion groups) with positive COVID tests seems low (these were their 6 false negative cases). It’s great for their study results, but the greater body of CT data with COVID gives me pause on thinking that this would apply to other and larger study populations. Not that CT and LUS are equivalent by any means, but other CT studies have shown that a significant portion of patients with positive COVID symptoms and positive RT-PCR tests have entirely normal CT scans of their lungs. Some of that literature seems to suggest roughly 20% to 50% of folks with positive COVID PCR tests will have totally negative CT scans, particularly early in disease course. This would suggest that there should have been more misses with LUS. If the CT scan doesn’t see interstitial fluid or pleural changes, I’m not sure the LUS would be more sensitive or more likely to see things that CT didn’t. But perhaps this is evolving and each of these populations is separate and unique. Or maybe US is just that good at detecting small bits of interstitial fluid. Here are two examples of the Chest CT stuff if interested:

1) Inui S, Fujikawa A, et al. Chest CT Findings in Cases from the Cruise Ship “Diamond Princess” with Coronavirus Disease 2019 (COVID-19). Radiology:Cardiothoracic Imaging. 2020. Doc:10.1148/ryct.2020200110.

2) Bernheim A, Mei X, Huang M, et al. Chest CT Findings in Coronavirus Disease-19 (COVID-19): Relationship to Duration of Infection [published online ahead of print, 2020 Feb 20]. Radiology. 2020;200463. doi:10.1148/radiol.2020200463

Cite this post as

Michael Prats. Integrated Lung Ultrasound for the Diagnosis of COVID-19. Ultrasound G.E.L. Podcast Blog. Published on February 01, 2021. Accessed on April 14, 2021. Available at
Published on 02/01/21 04:00 AM
comments (0)