The LUNG SAFE study: a presentation of the prevalence of ARDS according to the Berlin Definition!

Villar et al. [1] suggest four methodological sources of "bias" that could have led to an overestimation of the acute respiratory distress syndrome (ARDS) incidence in the LUNG SAFE study [2].First, they suggest that an "unvalidated algorithm" was used to classify patients with ARDS.We simply utilized the Berlin Definition for ARDS [3].When the Berlin criteria were met, patients were classified as having ARDS-this is the "computer algorithm".There was no need to "validate" this algorithm, since it simply verifies the presence or absence of Berlin ARDS criteria.The electronic CRF provided additional guidance, including what specifically was meant by "bilateral infiltrates", and investigators were offered web-based training on chest X-ray image interpretation.A second "bias" was the fact that some patients fulfilled the criteria for ARDS for less than 24 hours.The Berlin Definition does not include any time window for ARDS determination.We cannot prove or refute the authors' statement that "Actual ARDS does not resolve in 24 h" [1].In the absence of a gold standard, we cannot know what "actual ARDS" is.A third suggested "bias" was the conduct of the study in winter months, an approach taken to minimize seasonal variation-a possible reason for differing ARDS prevalence estimates in prior reports.Contrary to the authors' assertion, we did not extrapolate to the "year incidence of ARDS" but instead confined our estimates to a 4-week period prevalence [2].