Article Text

Download PDFPDF

1116 Diagnostic accuracy of point-of-care lung ultrasound for COVID-19: a meta-analysis of 19 studies including 3954 patients
  1. Ashley Matthies1,
  2. Michael Trauer2,
  3. Karl Chopra3,
  4. Robert Jarman4
  1. 1Homerton University Hospital
  2. 2Guy’s and St Thomas’ NHS Foundation Trust
  3. 3Homerton University Hospital NHS Foundation Trust
  4. 4Royal Victoria Infirmary


Aims/Objectives/Background There is a growing evidence-base concerning the role of Point-of-Care (POC) lung ultrasound (LUS) for the diagnosis of COVID-19. LUS is well-established for many respiratory illnesses and may convey several advantages over conventional imaging modalities and single initial reverse-transcriptase polymerase chain reaction (RT-PCR) testing for COVID-19.

The purpose of this study was to perform a systematic review and meta-analysis of the diagnostic accuracy of LUS for the diagnosis of COVID-19. This is the first meta-analysis in this evolving area of research and combines data from almost 4,000 patients.

Methods/Design This work was synthesised according to the PRISMA guidelines. A research question was formulated according to the PICOTS framework and a search strategy was developed in accordance with existing guidance. A screening and selection tool was used to identify studies from the initial search. All prospective and retrospective trials of adult patients comparing LUS to either RT-PCR testing, chest CT or a final clinical diagnosis were included.

Data analysis was performed in R. Random effects bivariate binomial models were used to estimate overall sensitivity, specificity, and the hierarchical summary receiver operating characteristic (HSROC) curve for LUS. The QUADAS-2 tool was used to assess the quality of included studies.

Abstract 1116 Figure 1

Forest plot of LUS sensitivity

Abstract 1116 Figure 2

Forest plot of LUS specificity

Abstract 1116 Figure 3

Random effects meta-analysis

Results/Conclusions 19 studies were included in the meta-analysis, providing data for 3,954 patients in total. LUS demonstrated an overall (pooled) sensitivity of 87.3% (95% CI 83.4–90.5) and specificity of 70.0% (61.6–76.4). The overall positive and negative likelihood ratios were 3.04 (2.27–4.06) and 0.16 (0.12–0.22) respectively. Heterogeneity between studies was found to be high, with I2 values of 72.6% (59.2–91.7) and 87% (83.5–95.9) for sensitivity and specificity respectively.

LUS was found to be highly sensitive in a high prevalence population and may improve detection of COVID-19 pneumonia compared to CXR. In patients requiring hospital admission a normal LUS should prompt consideration of alternative diagnoses.

Statistics from

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.