key: cord-0773304-3gjs0kct authors: Trauer, M.; Matthies, A.; Mani, N.; McDermott, C. B.; Jarman, R. title: Utility of Lung Ultrasound in COVID-19: A Systematic Scoping Review date: 2020-06-16 journal: nan DOI: 10.1101/2020.06.15.20130344 sha: 7f3053ff0642bae7669113480ce9bbc3d623f27a doc_id: 773304 cord_uid: 3gjs0kct Lung ultrasound (LUS) has an established evidence base and has proven useful in previous viral epidemics. An understanding of the utility of LUS in COVID-19 is crucial to determine its most suitable role based on local circumstances. A scoping review was thus undertaken to explore the utility of LUS in COVID-19 and guide future research. 33 studies were identified which represent a rapidly expanding evidence base for LUS in COVID-19 however the quality of the included studies was relatively low. LUS certainly appears to be a highly sensitive and fairly specific test for COVID-19 in all ages and in pregnancy, however its accuracy may be influenced by various factors including disease severity, pre-existing lung disease, scanning protocol, operator experience, disease prevalence and the reference standard. High quality research is needed in various fields including: diagnostic accuracy in undifferentiated patients; triage and prognostication; monitoring progression and guiding interventions; persistence of residual LUS findings; inter-observer agreement; and the role of contrast-enhanced LUS. Coronavirus 2 (SARS-CoV-2) and was declared a global pandemic on the 11 th March 2020 by the World Health Organisation. As of the 10 th of June, there have been over seven million confirmed cases and over 400,000 deaths [ 1 ] . The evidence base for lung ultrasound (LUS) is well established. In 2008, LUS was found to have an accuracy of greater than 90% for some of the most common causes of dyspnoea [ 2 ] . In 2011, an international, evidence-based, consensus statement recommended its use in pneumothorax, interstitial syndrome, consolidation and effusion [ 3 ] . In 2015, a prospective study of over a thousand patients found incorporation of LUS into clinical assessment significantly improved sensitivity (97%) and specificity (97.4%) for acute heart failure [ 4 ] . And in 2018, a meta-analysis of over five thousand patients found LUS to be 92% sensitive and 93% specific for community-acquired pneumonia [ 5 ] . LUS has also proven useful during recent viral epidemics. In the 2009 influenza (H1N1) epidemic, LUS was found to be accurate in differentiating viral and bacterial pneumonia [ 6 ], and during the avian influenza (H7N9) epidemics LUS was found to be superior to CXR (sensitivity 94%, specificity of 89%) [ 7, 8 ]. In admitted patients with COVID-19, CXR has a reported sensitivity of between 59% and 69% [ 9,10 ]. In ambulatory patients with symptomatic COVID-19, CXR sensitivity has been reported at 42% [ 11 ]. The sensitivity of PCR for COVID-19 has been estimated at 70% [ 12 ] and depends upon factors including the quality of sampling and stage of illness. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.15.20130344 doi: medRxiv preprint CT of the thorax is highly sensitive for COVID-19 [ 13 ] however LUS has several logistical advantages. The capacity to perform routine CT in suspected COVID-19 may become overwhelmed if large numbers of patients attend hospital. LUS has also been shown to reduce healthcare worker exposure to COVID-19 by reducing the intra-hospital transfers associated with conventional imaging [ 14 ] . Other advantages of LUS over CT include reduced cost, repeatability, lack of radiation exposure and rapid image acquisition time [ 15 ] . LUS has been shown to improve diagnostic accuracy in patients who present with acute respiratory symptoms [ 16 ] and is increasingly used by frontline clinicians who assess these patients. Ultrasound machines continue to improve in quality, affordability and portability [ 17 ] and new technologies such remote teleguidance have the potential to further extend the accessibility of point-of-care ultrasound. The LUS findings in COVID-19 are well described and include B lines, pleural line abnormalities and consolidation [ 18 ] . However the most suitable role for LUS in COVID-19 is still unclear. Various roles have been proposed including triage, diagnosis, prognostication, severity scoring, monitoring progression, and guiding interventions [ 19 ]. An understanding of the utility of LUS is crucial to determine its most suitable role in COVID-19 based on local circumstances. To review the evidence of the utility of LUS in COVID-19 and guide future research Patients with suspected or confirmed COVID-19 • Concept: The utility of LUS • Context: Clinical management . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The protocol was drafted in line with PRISMA [ 20 ] and registered on https://figshare.com/ on 13/6/2020 (10.6084/m9.figshare.12478820) Inclusion criteria: • Patients of any age with suspected or confirmed COVID-19 • Explores the utility of LUS in COVID-19 • Trials or case series (prospective or retrospective) Exclusion criteria: • Case reports and recommendations • Non-English language Traditional online databases were searched including: Medline, Embase, SCOPUS, The Cochrane Library, The TRIP database, Google Scholar and www.clinicaltrials.gov. Given the dynamic nature of the pandemic, other less traditional sources were also searched including point-of-care ultrasound (PoCUS) websites, specialty college websites, pre-publication websites and social media platforms (see Appendix I). An initial search strategy was formulated by MT and reviewed by AM using the PRESS checklist [ 21 ]. This initial search was performed on two databases (Medline and Embase) (See Appendix II). . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.15.20130344 doi: medRxiv preprint Keywords were identified from the above abstracts and another search was performed of all relevant databases (See Appendix III). A screening and selection tool was applied to the identified studies by two independent reviewers (MT & AM) with a third reviewer (NM) available to resolve disagreements (See Appendix IV). The reference lists from these included studies were then reviewed for further relevant studies. The authors of the included studies were contacted regarding relevant unpublished or recently published evidence. Data was extracted on study design, numbers of participants, population, and data relating to the utility of LUS in COVID-19. Given the heterogeneity of the data, findings are described in a narrative style. A flow diagram in line with PRISMA [ 20 ] is presented in Figure 1 and displays the number of studies screened, excluded and assessed. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.15.20130344 doi: medRxiv preprint Records identified through database searching (n=227) A total of 33 studies were identified from countries including China, Italy, Spain, France and the USA. 17 were single-centre, three were multi-centre and the numbers of participants in each study ranged from three to 107. The topics explored in each study are summarised in Table 1 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2020. and found B lines were the most common and consistent finding but other LUS findings had a high degree of heterogeneity. In non-peer reviewed data from is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.15.20130344 doi: medRxiv preprint had a sensitivity and specificity of 100% and 80% respectively when performed by the treating clinician however this fell to 92% and 37% when the images were reviewed in isolation by an independent reviewer. Compared to CT, the sensitivity of CXR and crackles on auscultation was 25% and 8% respectively. In Yang et al. [ 29 ], 29 adult patients with confirmed COVID-19 received simultaneous LUS and CT and the lung fields were divided into 12 regions. 63% of regions displayed abnormal findings on LUS (3 or more B lines, consolidation or pleural effusion) compared to 39% of regions on CT (ground-glass opacity, consolidation or pleural effusion). The authors concluded that LUS was more sensitive at identifying the above findings than CT. In non-peer reviewed data from Benchoufi et al. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2020. There were various issues regarding the methodology of the included studies including convenience sampling, unrepresentative populations (often only admitted patients), lack of power calculations, variability of index test (operator experience, scanning protocol), variability of reference standard (CT, single PCR test, multiple PCR tests) and . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. There may be particular LUS findings and patterns that are more specific for COVID-19. In argued that relative specificity can be attributed to the classic bilateral, patchy distribution with spared areas and multifocal confluent B lines ('white lung'), especially in relatively young patients without a history of lung disease. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.15.20130344 doi: medRxiv preprint There is a spectrum of LUS findings in COVID-19 ranging from subtle to highly suggestive. Many studies have focused on LUS severity scores however in terms of diagnostic utility an assessment of likelihood rather than severity may be more useful. The differential diagnosis of each specific patient will also influence which LUS findings are most specific to COVID-19. If the alternative is pulmonary oedema, the presence of pleural thickening and irregularity is relatively specific for COVID-19. However if the alternative is pulmonary fibrosis this finding would not help to discriminate between these processes. In Hankins et al. [ 28 ], diagnostic accuracy was higher when LUS was interpreted by the treating clinician as opposed to being reviewed in isolation. This highlights the importance of integrating LUS findings with clinical findings. This Bayesian approach of combining a pre-test probability with point-of-care ultrasound findings is well described [ 48 ]. It should be noted that all of the included studies were conducted during a period of high disease prevalence and it is likely that measures of diagnostic accuracy will be affected by fluctuations in disease prevalence over time [ 49 ] . In Xing et al. [ 50 ] , 20 adult patients with confirmed COVID-19 underwent 36 scans at various time intervals after onset of symptoms. The authors found that the extent of LUS findings reached a peak at the second week and then there was gradual improvement (but not complete resolution) until the fourth week. In Shkoohi et al. [ 51 ] , three physicians with confirmed COVID-19 monitored themselves at home and in all cases the LUS findings had resolved by day 14. More information is urgently needed regarding the persistence of LUS findings as clinicians will be increasingly encountering patients who may have recently recovered from COVID-19. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. Further studies relating to inter-observer agreement are warranted however it appears the element of practical training is important. Novel technologies such as remote teleguidance could help to achieve this. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. , three patients with confirmed COVID-19 received contrast-enhanced ultrasound (CEUS). Perfusion defects were noted within these lesions and the authors concluded that this was at least in part caused by ischaemic or necrotic changes rather than inflammation or atelectasis. This is consistent with the findings of Huang et al. The issue of ionizing radiation is of great concern in children and pregnant women. Several small studies were identified that examined the utility of LUS in COVID-19 in these patient groups and were suggestive that LUS is as useful as it is in non-pregnant adults. The recent emergence and dynamic nature of the COVID-19 pandemic has led to the rapid publication of research and it is inevitable that new studies will continue to be released before this review is published. A thorough and systematic literature search was performed including non-traditional sources (see Appendix I) however all relevant evidence may not have been identified due to publication bias and non-English language publications being excluded. The evidence base for LUS in COVID-19 is rapidly expanding but the methodological quality of the identified studies was relatively low. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.15.20130344 doi: medRxiv preprint It is difficult to make a precise estimate of diagnostic accuracy of LUS in COVID-19 as both sensitivity and specificity may be influenced by various factors including disease severity, pre-existing lung disease, scanning protocol, operator experience, disease prevalence and the reference standard. However, LUS appears to be a highly sensitive and fairly specific test for COVID-19 in all ages and in pregnancy. LUS is probably more sensitive than CXR and possibly more sensitive than CT for COVID-19. High quality research is needed to better define the utility of LUS in COVID-19 and thus inform clinicians of its most suitable role in a local context. Although the LUS findings in COVID-19 are now well described, further research is needed regarding the relative specificity of the various LUS findings and patterns. High quality, prospective studies assessing diagnostic accuracy in undifferentiated patients in an era of lower prevalence would also be of great value. The role of LUS in triage, prognostication, severity scoring, monitoring progression and guiding interventions has not yet been adequately explored. An understanding of the persistence of residual LUS findings post infection will be increasingly important going forwards. Larger studies assessing inter-observer agreement would both estimate reproducibility but may also help inform necessary training standards for novices. Further research into contrast-enhanced LUS and colour Doppler is warranted as this may significantly augment traditional LUS and contribute to a broader understanding of the disease process. International consensus is required regarding training standards, scanning protocols and an appropriate reference standard. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted June 16, 2020. LUS findings began at symptom onset and resolved within 14 days . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2020. . https://doi.org/10.1101/2020.06.15.20130344 doi: medRxiv preprint World Health Organisation. Coronavirus Disease (COVID-19) Dashboard Relevance of Lung Ultrasound in the Diagnosis of Acute Respiratory Failure: The BLUE Protocol International evidence-based recommendations for point-of-care lung ultrasound Lung-Ultrasound Implemented diagnosis of acute decompensated heart failure in the ED Lung Ultrasound in Diagnosing Pneumonia in the Emergency Department: A Systematic Review and Meta-Analysis the emergency department: a randomized controlled trial Variation of a test's sensitivity and specificity with disease prevalence Lung ultrasound findings in patients with COVID-19 pneumonia Lung ultrasound monitoring in patients with COVID-19 on home isolation Training for lung ultrasound score measurement in critically ill patients The role of imaging in the detection and management of COVID-19: a review Deep learning for classification and localization of COVID-19 markers in point-of-care lung ultrasound Sonography of the Lungs: Diagnosis and Surveillance of