Article Text

Download PDFPDF

Low utilisation of bronchoscopy to assess COVID-19 respiratory infection: a multicenter experience
  1. Kamran Mahmood1,
  2. Matt Abbott1,
  3. Keriann Van Nostrand2,
  4. Rabih Bechara3,
  5. Anne V Gonzalez4,
  6. Amanda Brucker5,
  7. Cynthia L Green5 and
  8. Christopher R Polage6
  1. 1Department of Medicine, Division of Pulmonary, Allergy and Critical Care, Duke University, Durham, North Carolina, USA
  2. 2Department of Medicine, Division of Pulmonary and Critical Care, Emory University, Atlanta, Georgia, USA
  3. 3Department of Medicine, Division of Pulmonary and Critical Care, Medical College of Georgia, Augusta, Georgia, USA
  4. 4Department of Medicine, Division of Pulmonary and Critical Care, McGill University, Montreal, Quebec, Canada
  5. 5Department of Biostatistics and Bioinformatics, Duke University School of Medicine, Durham, North Carolina, USA
  6. 6Department of Pathology, Duke University, Durham, North Carolina, USA
  1. Correspondence to Dr Kamran Mahmood; k.mahmood{at}


Objective For the diagnosis of COVID-19, the yield of nasopharyngeal (NP) swabs is unclear, and bronchoalveolar lavage (BAL) is obtained to confirm the diagnosis. We assessed the utilisation of bronchoscopy for COVID-19 diagnosis in a multicenter study and compared the diagnostic yield of BAL versus NP swabs.

Methods This retrospective study included all patients who were admitted with clinical presentation concerning for COVID-19 and underwent BAL from 1 March to 31 July 2020 at four tertiary care centres in North America. We also compared concordance of BAL with NP swabs for diagnosis of COVID-19 infection.

Results Fifty-three patients, with clinical suspicion for COVID-19 and admitted for respiratory failure, underwent bronchoscopy to collect BAL for SARS-CoV-2 testing. During the same period, 2039 bronchoscopies were performed on patients not infected with COVID-19. Of 42 patients with NP swabs and BAL collected within ≤7 days, 1 was NP swab negative but positive by BAL for SARS-CoV-2 (n=1/42 (2.4%)). Across a wide array of testing platforms, the overall agreement between NP swabs and BAL results was 97.6% (95% CI: 93.0% to 100%) with Cohen’s k of 0.90 (95% CI: 0.69 to 1.00). The sensitivity, specificity, positive and negative predictive values of NP swabs compared with BAL were 83.3% (95% CI: 53.5% to 100%), 100%, 100% and 97.3% (95% CI: 92.1% to 100%), respectively.

Conclusions BAL was used infrequently to assess COVID-19 in busy institutions. NP swabs have a high concordance with BAL for COVID-19 testing, but negative NP swabs should be confirmed with BAL when clinical suspicion is high.

  • COVID-19
  • bronchoscopy
  • respiratory infection

Data availability statement

Data are available upon reasonable request.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Key messages

  • Bronchoscopies for bronchoalveolar lavage (BAL) represented 2.5% of total bronchoscopies performed in the participating institutions, and the overall agreement between nasopharyngeal (NP) swab and BAL results was significant at 97.6% (95% CI: 93.0% to 100%).

  • We describe details of bronchoscopy practice and SARS-CoV-2 testing platforms at different institutions, and emphasize that when the suspicion for COVID-19 remains high despite negative NP swab, BAL should be performed to confirm the diagnosis.


The COVID-19 pandemic has led to devastating morbidity and mortality, primarily related to respiratory failure and associated multiorgan dysfunction. Prompt diagnosis of COVID-19 is critical for the treatment of infected individuals and prevention of infection transmission. The diagnosis of COVID-19 is established by clinical presentation, radiologic findings of the chest and testing based on reverse transcription (RT) PCR. The RT-PCR test is performed for the detection of nucleic acid from SARS-CoV-2 on various specimens—nasopharyngeal (NP) swabs being one of the most common source of testing.1 2 However, several studies have reported that the sensitivity of NP swabs is variable.3 4 Since the COVID-19 infection involves the lower respiratory tract, especially in sicker patients, bronchoscopy is often performed in patients with negative NP swabs to obtain bronchoalveolar lavage (BAL), as an early study by Wang et al suggested that lower respiratory samples provide a higher diagnostic yield.4 In this study, the diagnostic yield of BAL was 93%, while the yield of nasal swabs was only 63% from patients suspected of COVID-19.4 However, the bronchoscopy is an aerosol-generating procedure that can potentially expose the healthcare providers to COVID-19 and may strain the limited healthcare resources.5 Several professional societies came out early with guidelines advising against routine use of bronchoscopy for COVID-19 diagnosis, based on the experience from other coronavirus epidemics.6 The objectives of this study were to assess the clinical utilisation of bronchoscopy in COVID-19 diagnosis and compare the diagnostic yield of BAL and NP swabs in patients with suspected or confirmed COVID-19 infection.


Study design and settings

We performed a retrospective review of all patients who were clinically suspected to have COVID-19 and underwent BAL and NP swab testing for COVID-19 testing at four large academic institutions in North America. Medical records were reviewed for all patients who met the criteria from 1 March 2020 to 31 July 2020. Data from participating institutions were entered into a REDCap database maintained by Duke University.

Patient and public involvement statement

It was not appropriate or possible to involve patients or the public in the design, or conduct, or reporting, or dissemination plans of our research, but the study was performed to improve patient care outcomes. We thank our patients and their families for their strength and endurance, as stated in the Acknowledgements.

Specimen collection

NP swabs were collected by trained healthcare personnel, according to Centers for Disease Control and Prevention (CDC) recommendations.7 8 The swabs were made of synthetic fibre attached to a plastic or aluminium shaft. The swabs were inserted in the posterior nasopharynx, parallel to the palate and maintained in place for a few seconds to absorb the secretions. Afterwards, the swabs were removed and placed immediately into a sterile transport tube containing 2–3 mL of viral transport medium or sterile saline. NP swabs for point-of-care nucleic acid amplification (NAA) tests were performed directly on the instrument near the collection location.

BAL was performed via bronchoscopy by trained healthcare providers. The bronchoscope was inserted into the airways and wedged into a distal, segmental bronchus. Then, sterile saline in two to three 50 mL aliquots was instilled into the airway and suctioned back into a trap. From the retrieved BAL fluid, 2–3 mL was sent in a sterile container for SARS-CoV-2 RT-PCR.

COVID testing

SARS-CoV-2 coronavirus testing was performed following the manufacturer’s instructions for use and/or validated protocols on multiple US Food and Drug Administration emergency use authorisation RT-PCR or other NAA test platforms. Tests were independently verified and validated by each performing laboratory before clinical use. NP samples were generally tested on a laboratory RT-PCR assay (Cepheid Xpert SARS-CoV-2, DiaSorin Simplexa COVID-19 Direct, Roche cobas SARS-CoV-2 and Panther Fusion SARS-CoV-2), with a few tested on a rapid point-of-care NAA test (Abbott ID Now COVID-19). BAL samples were only tested by laboratory RT-PCR on platforms validated locally for BAL (CDC SARS-CoV-2, Abbott Alinity m SARS-CoV-2, Cepheid Xpert SARS-CoV-2, Roche cobas SARS-CoV-2 and Panther Fusion SARS-CoV-2).

Statistical analysis

Continuous variables are presented as median with 25th and 75th percentiles (Q1–Q3), and categorical variables as frequency counts with percentages. The concordance of NP swab and BAL testing was assessed using McNemar’s test and Cohen’s k with 95% CI. A two-sided p value of 0.05 or less was considered statistically significant. All analyses were performed with SAS V.9.4 (SAS Institute).


Bronchoscopic BAL was performed to assess for COVID-19 infection on 53 hospitalised patients at four centres during the study period, while 2183 patients were admitted with confirmed COVID-19 diagnosis. Overall, 2039 bronchoscopies were performed on patients not infected with COVID-19 during this time frame. COVID-19 was confirmed in 13 of our cohort of 53 patients who underwent BAL. Confirmed patients with COVID-19 (N=13) underwent a total of 57 non-bronchoscopy procedures and 2 patients had a repeat bronchoscopy performed; bronchoscopies thus accounted for 26.3% of total procedures.

The clinical characteristics of the patients included in the study are summarised in table 1. The patients included in the study were deemed high risk for COVID-19 as they had symptoms and radiographic findings concerning for COVID-19 infection along with respiratory failure or acute respiratory distress syndrome (ARDS). Forty-six patients (86.8%) were admitted in intensive care units (ICU), while others were admitted on medical wards. Diabetes mellitus (32.1%) was the most common comorbid condition. Chest radiographs showed bilateral infiltrates in 83%, unilateral infiltrates in 11.3% and pleural effusions in 15.1% of patients. Chest CT scan showed consolidation in 43.4%, ground-glass infiltrates in 58.5%, interlobular septal thickening in 3.8%, mediastinal lymphadenopathy in 3.8%, and pleural effusions in 26.4% of patients. Forty-six patients (86.8%) were on mechanical ventilation; 31 patients (59.5%) had ARDS and 5 patients (9.4%) required extracorporeal membrane oxygenation.

Table 1

Patient characteristics

Prior to BAL, 41 patients had negative NP swabs, while 12 patients had positive NP swabs for SARS-CoV-2 testing. BAL was done in patients with negative NP swabs to confirm the COVID-19 diagnosis, while the BAL in positive NP swab patients was performed to assess for other coinfections. In the patients (N=13) who were later confirmed to have COVID-19, the BAL was performed at a median of 14 days (Q1–Q3: 8–18 days) from the onset of symptoms. In 46 patients (86.8%), the last NP swab closest to BAL was done when they were on mechanical ventilation, and 7 patients (13.2%) had swabs obtained while spontaneously breathing. A wide array of platforms was used for COVID-19 testing on NP swabs and BAL, as shown in table 2. Some patients had multiple negative NP swabs done before BAL was performed.

Table 2

COVID-19 and other diagnostic testing for patients with NP swabs and BAL performed within 7 days

All bronchoscopies were done in a negative pressure room with powered air purifying respirator or N95 with face shields (table 3). Most bronchoscopies (96.2%) were done with deep sedation or general anaesthesia. Endotracheal tube (92.2%) was the preferred airway device for bronchoscopy. Disposable bronchoscopes were used for 36 (67.9%) procedures. The bronchoscopies were tolerated well and only one patient with septic shock developed transient hypotension requiring temporary escalation of vasopressors.

Table 3

Bronchoscopy data

Across different platforms at the participating institutions, the concordance between NP swabs and BAL was evaluated in 42 patients who had NP swabs and BAL collected within 7 days. The NP swabs closest to BAL were used for concordance assessment, and were collected at a median of 1 day (Q1–Q3: 0–4 days) prior to BAL. One patient had a negative NP swab but subsequent BAL was positive (n=1/42 (2.4%) table 4); the remaining 41 patients had NP swabs and BAL results that were concordant, yielding an overall agreement of 97.6% (95% CI: 93.0% to 100%) and Cohen’s k of 0.90 (95% CI: 0.69 to 1.00). In addition, the p value from McNemar’s test for marginal homogeneity was 0.317; therefore, the null hypothesis that the NP swabs and BAL were concordant could not be rejected. The sensitivity, specificity, positive predictive value and negative predictive value of NP swabs compared with BAL were 83.3% (95% CI: 53.5% to 100%), 100%, 100% and 97.3% (95% CI: 92.1% to 100%), respectively.

Table 4

NP swab* and BAL test results performed within 7 days


This multicenter study evaluated the clinical utilisation of bronchoscopy for COVID-19 diagnosis and showed that bronchoscopy and BAL for diagnosis of COVID-19 made up a small fraction of overall bronchoscopies in busy US and Canadian centres. The high concordance between NP swabs and BAL is reassuring, but BAL should be considered in patients with negative NP swabs who have a high clinical suspicion for COVID-19 infection.

The participating centres in the study were tertiary care, academic medical centres with a high volume of bronchoscopic procedures and patients with COVID-19. However, bronchoscopy and BAL utilisation was low, accounting for only about 2.5% of total bronchoscopies, and NP swabs were primarily used to assess COVID-19. This practice is in accordance with the guidelines issued by American College of Chest Physicians (ACCP) and American Association of Bronchology and Interventional Pulmonology, based on experience from previous coronavirus epidemics, and aimed to prevent the exposure to healthcare workers.6

We found a high concordance rate between NP swabs and BAL. One of the original reports describing the diagnostic yield of respiratory specimens was reported by Wang et al.4 They described the diagnostic yield of 8 nasal swabs, 398 pharyngeal swabs and 15 BAL specimens in patients admitted with COVID-19 infection. The positive test rate from nasal swabs, pharyngeal swabs and BAL was 63%, 32% and 93%, respectively. Gao et al reported their single-centre experience in 123 patients with COVID-19 infection and respiratory failure requiring mechanical ventilation.9 Fourteen patients (11%) had discordant NP swabs and BAL assays. When compared with BAL, they described the sensitivity of the NP swab to be 88.6%, specificity 88.6%, positive predictive value 93.3%, negative predictive value 81.3% and accuracy of 88.6%. In another study, Geri et al evaluated the agreement between negative NP swabs and subsequent BAL in 79 patients admitted with respiratory failure.10 Two patients with negative NP swabs had positive BAL with an accuracy of 97.5% (Cohen’s k=0.487). In another study, in 28 patients with suspected COVID-19, 3 sequential negative NP swabs, and negative IgG and IgM serologies were completely concordant with negative BAL results.11 Barberi et al reported in a cohort of 198 patients with suspected COVID-19 and negative NP swabs, 32 (16%) patients had positive BAL.12 But multiple case reports and studies have reported lower concordance between NP swabs and BAL.13–15 Patrucco et al reported a sensitivity of NP swabs of 23% in a cohort of 43 patients, as 33 patients with negative NP swabs were subsequently diagnosed with COVID-19 on BAL.16 Similarly, Mondoni et al reported a sensitivity of NP swabs of only 44.8% in their series, as 43 out of 78 patients with negative NP swabs tested positive for COVID-19 on BAL.17 Therefore, the published literature about the yield and concordance of NP swabs and BAL is highly variable. We found a high concordance of 97.6% (Cohen’s k=0.90) between NP swabs and BAL with a wide range of testing platforms. Our results are more aligned with published studies of high concordance, which might be related to a better technique of NP swab collection—a critical step stressed by many experts.18 19 The high agreement between NP swabs and BAL and attempt to minimise provider exposure to aerosol-generating bronchoscopy may be the reason why BAL was performed infrequently to diagnose COVID-19 at our institutions.

COVID-19 primarily affects the lower respiratory tract, especially in patients with severe disease, and thus lower respiratory specimens such as BAL are expected to have a higher diagnostic yield.20 There are reports of initial negative upper respiratory RT-PCR tests in patients with clinical or CT scan findings consistent with COVID-19, but subsequent upper respiratory samples were positive on repeat testing.21 22 In addition, NP swabs may be fraught with suboptimal specimen collection technique.18 In a report of four patients with suggestive symptoms and negative NP swab, repeat NP swabs by otolaryngologists within hours were positive.19 All the patients had nasal obstruction, and the initial false-negative test was attributed to inadequate sampling. Another study also found lower human DNA on suspected false-negative NP swabs, suggestive of suboptimal sampling.23 The high yield and concordance of NP swabs with BAL in some published studies and our cohort highlights this important diagnostic consideration.

Radiological findings suggestive of COVID-19 infection might be helpful but have low sensitivity and specificity, as in a study evaluating chest CT based COVID-19 probability scores in negative NP swab patients, 7/46 (15%) patients with atypical or low suspicion CT scans had positive BAL.24 Similarly, in another study evaluating 50 patients suspected of COVID-19 with negative NP swabs, 3 patients with CT scan indeterminate for COVID-19 were found to have positive BAL.15 Therefore, WHO and ACCP guidelines suggest obtaining a lower respiratory specimen if the upper respiratory specimen is negative and clinical suspicion of COVID-19 is high.6 25 Our study findings support the available data and expert opinion, and in patients who remain suspicious for COVID-19 infection despite NP swab testing and radiographic findings, BAL testing should be performed to confirm the diagnosis.

Bronchoscopy may be necessary to establish alternate diagnosis in patients with suspected COVID-19 infection. Torrego et al reported their experience of bronchoscopy in 101 patients with COVID-19 on mechanical ventilation.26 They reported presence of thick secretions and studies positive for other pathogens in 29% of the patients. In our cohort of patients who had NP swab and BAL done concomitantly, the BAL studies isolated other pathogens in 14.2% (6/42) patients. The lower isolation of other pathogens in our study could be related to differences in underlying disease process or ongoing antibiotic coverage.

Strengths and limitations

The strengths of this study include its multicenter design that provides a broader picture of the bronchoscopy practice for diagnosis of COVID-19 in North America compared with the previous single-centre studies and suggests that the negative predictive value of a well-collected NP swab for a subsequent negative BAL may be higher than previous reports. The study also evaluated a broad spectrum of platforms for COVID-19 testing on NP swabs and BAL. The weaknesses of the study are its retrospective design and limited number of patients who had BAL specimens available for comparison with NP swabs, which is in alignment with current guidelines. During the COVID-19 pandemic, bronchoscopies for non-COVID-19 indications continued at our institutions. We instituted a policy to obtain a preprocedure NP swab for COVID-19 testing, and the bronchoscopies were performed only if NP swabs were negative. Since BAL was not performed to assess for COVID-19 during these bronchoscopies, we cannot comment on the value of BAL in asymptomatic patients for COVID-19.


In conclusion, this multicenter experience suggests a high concordance between NP swabs and BAL for diagnosis of COVID-19 in patients with severe acute lower respiratory illness and a high negative predictive value of NP swabs for subsequent negative BAL. This experience supports professional societal recommendations limiting bronchoscopy procedures in persons under investigation for COVID-19 and may have contributed to a decreased utilisation of bronchoscopy for diagnosis of COVID-19 at these centres. However, NP swab results should be interpreted in clinical and epidemiological context, and BAL testing should be considered to definitively rule out COVID-19 in patients with negative NP swabs when clinical suspicion of COVID-19 remains high.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Ethics approval

Institutional review boards (IRB) of all the institutions, including Duke University (IRB no. 00106066), Emory University, Medical College of Georgia and McGill University approved the study.


We acknowledge and thank all the healthcare workers for their dedication, the patients infected with COVID-19 and their families for their strength and endurance. We additionally wish to acknowledge support from the Biostatistics, Epidemiology and Research Design Methods Core funded through Grant Award Number UL1TR002553 from the National Center for Advancing Translational Sciences, a component of the National Institutes of Health (NIH). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.



  • Contributors KM wrote the manuscript which was reviewed, edited and approved by all authors. KM, MA, KMVN, RB, AVG and CRP designed the study and collected the data. AB and CG analysed the data. KM took full responsibility for the integrity of the manuscript.

  • Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.