Professional Documents
Culture Documents
BACKGROUND: Some studies suggest that lung ultrasonography could be useful for diagnosing
pneumonia; moreover, it has a more favorable safety profile and lower cost than chest
radiography and CT. The aim of this study was to assess the accuracy of bedside lung
ultrasonography for diagnosing pneumonia in adults through a systematic review and
meta-analysis.
METHODS: We searched MEDLINE, Scopus, The Cochrane Library, Web of Science, DARE,
HTA Database, Google Scholar, LILACS, ClinicalTrials.gov, TESEO, and OpenGrey. In
addition, we reviewed the bibliographies of relevant studies. Two researchers independently
selected studies that met the inclusion criteria. Quality of the studies was assessed
in accordance with the Quality Assessment of Diagnostic Accuracy Studies tool. The
summary receiver operating characteristic (SROC) curve and a pooled estimation of
the diagnostic odds ratio (DOR) was estimated using a bivariate random-effects analysis. The
sources of heterogeneity were explored using predefined subgroup analyses and bivariate
meta-regression.
RESULTS: Sixteen studies (2,359 participants) were included. There was significant hetero-
geneity of both sensitivity and specificity according to the Q test, without clear evidence of
threshold effect. The area under the SROC curve was 0.93, with a DOR at the optimal
cutpoint of 50 (95% CI, 21-120). A tendency toward a higher area under the SROC curve
in high-quality studies was detected; however, these differences were not significant after
applying the bivariate meta-regression.
CONCLUSIONS: Lung ultrasonography can help accurately diagnose pneumonia, and it may be
promising as an adjuvant resource to traditional approaches. CHEST 2017; 151(2):374-382
ABBREVIATIONS: DOR = diagnostic odds ratio; PRISMA = Preferred FUNDING/SUPPORT: This work was supported by the Department of
Reporting Items for Systematic Reviews and Meta-Analyses; Clinical Medicine, Miguel Hernández University.
QUADAS-2 = Quality Assessment of Diagnostic Accuracy Studies 2; CORRESPONDENCE TO: Ana Llamas-Álvarez, MD, Department of
ROC = receiver operating characteristic; SROC = summary receiver Intensive Care Medicine, Elche General University Hospital, Camino
operating characteristic de la Almazara 11, 03203, Elche, Spain; e-mail: llamasalvarez@yahoo.es
AFFILIATIONS: From the Intensive Care Unit (Drs Llamas-Álvarez, Copyright Ó 2016 American College of Chest Physicians. Published by
Tenza-Lozano, and Latour-Pérez), Elche General University Hospital, Elsevier Inc. All rights reserved.
Elche, Spain; and the Department of Clinical Medicine (Dr Latour- DOI: http://dx.doi.org/10.1016/j.chest.2016.10.039
Pérez), Miguel Hernández University, Sant Joan d’Alacant, Spain.
Two researchers (A. M. L-A. and E. M. T-L.) conducted an independent Assessment of Risk of Bias
literature search to identify potentially relevant studies in MEDLINE, Two reviewers (A. M. L-A. and E. M. T-L.) independently performed
Scopus, The Cochrane Library, the Web of Science, DARE, HTA the assessment of methodological quality of included articles according
Database, Google Scholar, LILACS, and ClinicalTrials.gov. We also to Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2)
searched for doctoral theses in the TESEO and Tesis Doctorals en criteria,23 resolving discrepancies through consultation with a third
Xarxa databases, for conference papers in the Conference Proceedings expert researcher (J. L-P.).
section of the Web of Science, and for gray literature in OpenGrey. In
addition, we reviewed the bibliographies of relevant studies.
Data Analysis
Our search strategy used the terms pneumonia, ultrasonography, Data were presented in the receiver operating characteristic (ROC)
ultrasonography, ecograph,* and sonograph*, and we did not impose plane and analyzed according to the European Association for
any time restriction, including published studies from database Technology Assessment recommendations.24
journal.publications.chestnet.org 375
Downloaded for Anonymous User (n/a) at University of Guadalajara from ClinicalKey.com by Elsevier on December 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
In the univariate analysis, we constructed a forest plot for We analyzed study heterogeneity graphically and through the
sensitivity, specificity, and diagnostic odds ratio (DOR). The Q test.
correlation between sensitivity and false-positive rate was
explored graphically using the corresponding forest plots and Possible causes of heterogeneity between studies were explored through
statistically by examining the Spearman correlation coefficient prespecified subgroup analyses: study setting (ICU vs other), overall
and 95% CI. We used the Mada application included in the quality of studies (low vs high or unclear risk of bias), and individual
R statistical package (The R Project for Statistical Computing)25 dimensions of quality (QUADAS-2). The existence of interaction
to apply the bivariate Reitsma model,26 which in the absence between these variables and diagnostic accuracy (sensitivity and
of covariates is equivalent to the Rutter-Gatsonis hierarchical specificity) was examined by using a bivariate meta-regression. The
summary receiver operating characteristic (SROC) method.27 figures included were created using Mada and Review Manager.28
CD ¼ clinical data; EC ¼ echocardioraphy; FD ¼ final diagnosis, including diagnostic methods; lab ¼ laboratory data; LUS ¼ lung ultrasonography;
mic ¼ microbiological data.
CAP ¼ community-acquired pneumonia; LVF ¼ left ventricular failure; MV ¼ mechanical ventilation; PS ¼ pneumonia suspected; RF ¼ respiratory failure;
VAP ¼ ventilator-associated pneumonia.
a
Age is expressed according to data extracted from each study as median SD, median (interquartile range), or percentage within a group.
journal.publications.chestnet.org 377
Downloaded for Anonymous User (n/a) at University of Guadalajara from ClinicalKey.com by Elsevier on December 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
Applicability pattern of graphically negative association between
Risk of Bias Concerns sensitivity and specificity. The 95% CI of the overall
Reference Standard
Reference Standard
effect indicates a sensitivity of approximately 80% to
Patient Selection
Patient Selection
Flow and Timing
90% and a specificity of 70% to 90%.
Index Test
Index Test
The overall estimation of the area under the SROC curve
of lung ultrasonography for the diagnosis of pneumonia
is 0.93, and the calculated DOR is 50 (95% CI, 21-120)
Bataille 2014 ? ? + – ? + +
(Fig 5).
Patient Selection
Index Test
Reference Standard
Flow and Timing
Figure 3 – Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2). Each domain is represented in a bar with the proportion of studies
considered high risk, low risk, or unclear. The same applies to applicability concerns.
Figure 4 – Forest plot of sensibility and specificity in each study. FN ¼ false negative; FP ¼ false positive; TN ¼ true negative; TP ¼ true positive.
attributable to the poor quality of the studies, since when of fewer studies (N ¼ 10), 70% of which would merit an
stratifying by the different dimensions of QUADAS-2, test assessment of high risk of bias according to our criteria.
accuracy is numerically better in high-quality studies. Chavez et al also included at least one study in which the
unit of analysis was not the participant.45 Hu et al46
Our findings are consistent with previous reviews
obtained a DOR of 509.99 (95% CI, 128.11-2,014.34)
showing even more favorable data on the diagnostic
accuracy profile of the lung ultrasonography. Chavez
et al15 obtained a weighted sensitivity and specificity of 1.0
94% and 96%, respectively, with an area under the
SROC curve of 0.98. The main differences with our 0.9
review are related to methodology and to the inclusion
Sensitivity
0.8
1.0
0.7
0.8
0.6
Sensitivity
journal.publications.chestnet.org 379
Downloaded for Anonymous User (n/a) at University of Guadalajara from ClinicalKey.com by Elsevier on December 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
and area under the SROC curve of 0.99, although seven ultrasonography in the diagnosis of pneumonia in
of the nine studies analyzed included children and even adults, including a review of gray literature and absence
infants, so the samples are not comparable. In fact, several of language restrictions. Regarding the review’s
pediatric studies suggest superior diagnostic performance limitations, it is important to note the high clinical
for chest ultrasonography in children compared with heterogeneity between included studies and substantial
adults,47,48 which may be related to the fact that children differences in the implementation of the ultrasonographic
usually have a thinner chest wall and a smaller volume of and scanning technique. These issues have been addressed
lung parenchyma. In the review of Ye et al,49 the DOR by Volpicelli et al54 in a consensus document providing
was 151.19 (CI 95%, 38.50-593.77), and the area under the expert recommendations in line with the GRADE55
SROC curve was 0.97, with the handicap of large initiative on the implementation, development, and
imprecision (only five studies included). standardization of lung ultrasonography.
In addition to its diagnostic accuracy, lung ultrasonography Our data suggest that lung ultrasonography may be a
has other advantages over chest radiography and CT valuable resource to consider alongside clinical,
because it can be done quickly at bedside, is easily laboratory, and microbiological data for the diagnosis
reproducible, has a relatively low cost, and avoids any of pneumonia in adults. Regarding imaging techniques,
exposure to ionizing radiation. Conversely, ultrasonography we suggest an adjuvant role for lung ultrasonography
is of limited value in patients with subcutaneous along with chest radiography, since the latter provides
emphysema and in obese people resulting from the important additional information on bone, mediastinal
thickness of the chest wall.50 Moreover, it is not structures, airways, and soft tissues. Nevertheless, it is
possible when access to the patient’s chest is limited by possible that the implementation of lung ultrasonography
large bandages, prosthetic material, or skin disorders. helps minimize the number of radiological examinations,
Another disadvantage of lung ultrasonography is its which could also have a positive impact on patient and
observer-dependent nature, as it implies the need staff safety, optimizing the use of time and economic
for operators with certain skills and experience; resources.
however, the learning curve is relatively fast,51,52 and in
the end the method can obtain high interobserver Conclusions
agreement.53
Lung ultrasonography has high sensitivity and specificity
To our knowledge, this meta-analysis of 16 studies and and appears to be a valuable complement to chest
2,359 participants is the largest one to date assessing radiography for the diagnosis of pneumonia in adults.
journal.publications.chestnet.org 381
Downloaded for Anonymous User (n/a) at University of Guadalajara from ClinicalKey.com by Elsevier on December 28, 2019.
For personal use only. No other uses without permission. Copyright ©2019. Elsevier Inc. All rights reserved.
acute respiratory failure the BLUE diagnosing pneumonia. Am J Emerg Med. radiography for the diagnosis of adult
protocol. Chest. 2008;134(1):117-125. 2013;31(2):401-405. community-acquired pneumonia: review
of the literature and meta-analysis. PLOS
37. Nazemi S, Farokhnia M, Pirouzi M, 43. Macaskill P, Gatsonis C, Deeks JJ, One. 2015;10:6.
Najarion O, Ahmadipour H. Accuracy of Harbord RM, Takwoingi Y. Chapter 10:
ultrasound as a diagnostic tool of Analysing and presenting results. In: 50. Bouhemad B, Zhang M, Lu Q, Rouby JJ.
pneumonia for admitted patients in Cochrane Handbook for Systematic Reviews Clinical review: bedside lung ultrasound
Infectious Disease Section of Afzalipour of diagnostic test accuracy, version 1.0. in critical care practice. Crit Care. 2007;
Hospital. Jam J Sci Tech. 2014;25:48-54. London, UK: The Cochrane Collaboration, 11(1):205.
38. Zagli G, Cozzolino M, Terreni A, 2010. 51. See KC, Ong V, Wong SH, et al. Lung
Biagioli T, Caldini AL, Peris A. Diagnosis 44. Plana MN, Zamora J, Abraira V. Lectura ultrasound training: curriculum
of ventilator-associated pneumonia: a pilot, crítica de revisiones sistemáticas y implementation and learning trajectory
exploratory analysis of a new score based metanálisis en el diagnóstico por imagen. among respiratory therapists. Intensive
on procalcitonin and chest echography. Radiología. 2015;57:23-30. Care Med. 2016;42(1):63-71.
Chest. 2014;146(6):1578-1585.
45. Xirouchaki N, Magkanas E, Vaporidi K, 52. Lichtenstein D, Goldstein I, Mourgeon E,
39. Corradi F, Ball L, Brusasco C, et al. et al. Lung ultrasound in critically ill Cluzel P, Grenier P, Rouby JJ.
Lung ultrasonography fails detection of patients: comparison with bedside chest Comparative diagnostic performances of
non-subpleural community acquired radiography. Intensive Care Med. auscultation, chest radiography, and lung
pneumonia. Intensive Care Med. 2012;38: 2011;37(9):1488-1493. ultrasonography in acute respiratory
S238-S238. distress syndrome. Anesthesiology.
46. Hu Q, Shen Y, Jia L, et al. Diagnostic 2004;100(1):9-15.
40. Fares A, Abo-Naglh A, Moustafa Z. Role performance of lung ultrasound in the
of transthoracic ultrasound in detection of diagnosis of pneumonia: a bivariate meta- 53. Gullett J, Donnelly JP, Sinert R, et al.
pneumonia in ICU patients. Med J Cairo analysis. Int J Clin Exp Med. 2014;7(1): Interobserver agreement in the evaluation
Univ. 2015;83(1):307-314. 115-121. of B-lines using bedside ultrasound. J Crit
Care. 2015;30(6):1395-1399.
41. Nafae R, Eman SR, Mohamad NA, 47. Copetti R, Cattarossi L. Ultrasound
El-Ghamry R, Ragheb AS. Adjuvant role diagnosis of pneumonia in children. 54. Volpicelli G, Elbarbary M, Blaivas M,
of lung ultrasound in the diagnosis of Radiol Med. 2008;113(2):190-198. et al. International evidence-based
pneumonia in intensive care unit-patients. recommendations for point-of-care lung
Egypt J Chest Dis Tuberc. 2013;62(2): 48. Pereda MA, Chavez MA, Hooper- ultrasound. Intensive Care Med. 2012;38(4):
281-285. Miele CC, et al. Lung ultrasound for the 577-591.
diagnosis of pneumonia in children: a
42. Self WH, Courtney DM, McNaughton CD, meta-analysis. Pediatrics. 2015;135(4): 55. Guyatt GH, Oxman AD, Vist GE, et al.
Wunderink RG, Kline JA. High 714-722. GRADE: an emerging consensus on rating
discordance of chest x-ray and computed quality of evidence and strength of
tomography for detection of pulmonary 49. Ye X, Xiao H, Chen B, Zhang S. Accuracy recommendations. BMJ. 2008;336(7650):
opacities in ED patients: Implications for of lung ultrasonography versus chest 924-926.