Chest imaging
Infiltrates
The committee agreed that chest imaging criteria should include bilateral radiologic (chest radiograph or computed tomography) or ultrasound findings suggestive of loss of lung aeration that are not fully explained by effusions, atelectasis, or nodules/masses.
Although the identification of bilateral opacities by chest radiograph has poor inter-rater reliability, the chest radiograph is the most common imaging modality in critically ill patients, which contributed to the recommendation to retain it in the definition despite its limitations.
In the LUNG SAFE study, the mortality of patients with unilateral opacities in two quadrants was similar to those with bilateral opacities in two quadrants. After considerable discussion, the consensus was that patients with unilateral opacities in two quadrants should not meet criteria for ARDS. Accepting unilateral opacities to diagnose ARDS was thought to have less face validity, conflicting with the conceptual model of a diffuse pulmonary process and increasing the likelihood of diagnosing patients with lobar pneumonia or atelectasis as having ARDS. Further, mortality alone does not reflect ARDS pathobiology. Thus, although some data support the similarity between patients with unilateral opacities in two quadrants and those with bilateral opacities, more investigation is required. Furthermore, it is unclear if bilateral opacities identified by other imaging modalities such as computed tomography are associated with the same outcomes. There was agreement that further research is needed in this area.
Ultrasound
In addition, the committee recommended that ultrasound be accepted as a modality for identifying signs of loss of lung aeration consistent with (non-cardiogenic) pulmonary edema or lung consolidation, especially when a chest radiograph or computed tomography are not available. There is evidence that ultrasound can be reliable if the operator is trained to detect bilateral consolidations and non-cardiogenic pulmonary edema, an approach that should have value especially in resource-limited areas.
For ultrasound, B-lines alone have been found to be very sensitive but not specific enough to differentiate different respiratory conditions. This is particularly true for ARDS diagnosis with lung ultrasound, where the criteria established in the Kigali modification of the Berlin definition have been shown in a study conducted in a resource-rich setting and in mechanically ventilated patients, to lead to high sensitivity but only moderate specificity. In that study, the false positive cases could all be explained by non-congruent chest radiographs, suggesting that lung ultrasound could overdiagnose ARDS. For this reason, more recent studies, including a recently published study, explored the role of other lung ultrasound findings, such as pleural line abnormalities in addition to B-lines and consolidations for increasing specificity. To confirm and test this approach, further studies should involve multiple operators across a range of clinical settings.
The committee concluded that appropriate training in the use of lung ultrasound should be emphasized, and therefore some recent recommendations for training have been included.
The consensus recommendation is that ultrasound should be used when a chest radiograph is not readily available. The committee anticipates prospective studies that will evaluate the potential complementary or independent value of ultrasound in any setting, including where chest radiographs and CTs are available. High quality data on inter-rater agreement for lung ultrasound interpretation are currently missing and most of the literature suggesting high interrater agreement of lung ultrasound was produced by extremely skilled lung sonographers and therefore not likely to reflect real world performances.