Dataset Information

Chest radiography versus lung ultrasound for identification of acute respiratory distress syndrome: a retrospective observational study.

ABSTRACT: BACKGROUND:Lung ultrasound may be a reasonable alternative to chest radiography for the identification of acute respiratory distress syndrome (ARDS), but the diagnostic performance of lung ultrasound for ARDS is uncertain. We therefore analyzed the clinical outcomes of ARDS diagnosed according to the Berlin Definition, using either chest radiography (Berlin-CXR) or lung ultrasound (Berlin-LUS) as an alternative imaging method. METHODS:This was a retrospective observational study in a 20-bed medical intensive care unit (ICU). Patients who required noninvasive ventilation or invasive ventilation for hypoxemic respiratory failure on ICU admission from August 2014 to March 2017 were included. Both chest radiography and lung ultrasound were performed routinely upon ICU admission. Comparisons were made using either the Berlin-CXR or Berlin-LUS definitions to diagnose ARDS with respect to the patient characteristics and clinical outcomes for each definition. ICU and hospital mortality were the main outcome measures for both definitions. RESULTS:The first admissions of 456 distinct patients were analyzed. Compared with the 216 patients who met the Berlin-CXR definition (ICU mortality 19.4%, hospital mortality 36.1%), 229 patients who met the Berlin-LUS definition (ICU mortality 22.7%, hospital mortality 34.5%) and 79 patients who met the Berlin-LUS but not the Berlin-CXR definition (ICU mortality 21.5%, hospital mortality 29.1%) had similar outcomes. In contrast, the 295 patients who met either definition had higher mortality than the 161 patients who did not meet either definition (ICU mortality 20.0% versus 12.4%, P?=?0.041; hospital mortality 34.2% versus 24.2%, P?=?0.027). Compared with Berlin-CXR, Berlin-LUS had a positive predictive value of 0.66 (95% confidence interval 0.59-0.72) and a negative predictive value of 0.71 (0.65-0.77). Among the 216 Berlin-CXR ARDS patients, 150 patients (69.4%) also fulfilled Berlin-LUS definition. CONCLUSIONS:For the identification of ARDS using the Berlin definition, both chest radiography and lung ultrasound were equally related to mortality. The Berlin definition using lung ultrasound helped identify patients at higher risk of death, even if these patients did not fulfill the conventional Berlin definition using chest radiography. However, the moderate overlap of patients when chest imaging modalities differed suggests that chest radiography and lung ultrasound should be complementary rather than used interchangeably.

SUBMITTER: See KC

PROVIDER: S-EPMC6098581 | biostudies-literature | 2018 Aug

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Chest radiography versus lung ultrasound for identification of acute respiratory distress syndrome: a retrospective observational study.

See Kay Choong KC Ong Venetia V Tan Yi Lin YL Sahagun Juliet J Taculod Juvel J

Critical care (London, England) 20180818 1

<h4>Background</h4>Lung ultrasound may be a reasonable alternative to chest radiography for the identification of acute respiratory distress syndrome (ARDS), but the diagnostic performance of lung ultrasound for ARDS is uncertain. We therefore analyzed the clinical outcomes of ARDS diagnosed according to the Berlin Definition, using either chest radiography (Berlin-CXR) or lung ultrasound (Berlin-LUS) as an alternative imaging method.<h4>Methods</h4>This was a retrospective observational study i ...[more]

PMID: 30119687

Similar Datasets

Project description:BackgroundAcute respiratory distress syndrome (ARDS) is a common, but under-recognised, critical illness syndrome associated with high mortality. An important factor in its under-recognition is the variability in chest radiograph interpretation for ARDS. We sought to train a deep convolutional neural network (CNN) to detect ARDS findings on chest radiographs.MethodsCNNs were pretrained on 595 506 radiographs from two centres to identify common chest findings (eg, opacity and effusion), and then trained on 8072 radiographs annotated for ARDS by multiple physicians using various transfer learning approaches. The best performing CNN was tested on chest radiographs in an internal and external cohort, including a subset reviewed by six physicians, including a chest radiologist and physicians trained in intensive care medicine. Chest radiograph data were acquired from four US hospitals.FindingsIn an internal test set of 1560 chest radiographs from 455 patients with acute hypoxaemic respiratory failure, a CNN could detect ARDS with an area under the receiver operator characteristics curve (AUROC) of 0·92 (95% CI 0·89-0·94). In the subgroup of 413 images reviewed by at least six physicians, its AUROC was 0·93 (95% CI 0·88-0·96), sensitivity 83·0% (95% CI 74·0-91·1), and specificity 88·3% (95% CI 83·1-92·8). Among images with zero of six ARDS annotations (n=155), the median CNN probability was 11%, with six (4%) assigned a probability above 50%. Among images with six of six ARDS annotations (n=27), the median CNN probability was 91%, with two (7%) assigned a probability below 50%. In an external cohort of 958 chest radiographs from 431 patients with sepsis, the AUROC was 0·88 (95% CI 0·85-0·91). When radiographs annotated as equivocal were excluded, the AUROC was 0·93 (0·92-0·95).InterpretationA CNN can be trained to achieve expert physician-level performance in ARDS detection on chest radiographs. Further research is needed to evaluate the use of these algorithms to support real-time identification of ARDS patients to ensure fidelity with evidence-based care or to support ongoing ARDS research.FundingNational Institutes of Health, Department of Defense, and Department of Veterans Affairs.

Dataset Information

Chest radiography versus lung ultrasound for identification of acute respiratory distress syndrome: a retrospective observational study.

Publications

Chest radiography versus lung ultrasound for identification of acute respiratory distress syndrome: a retrospective observational study.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets