Abstract

Background: The Berlin ARDS Definition categorizes lung injury severity based upon the ratio of PaO2/FiO2. In contrast, the Lung Injury Score (LIS) judges the severity of acute lung injury on a 4-point scale encompassing 4 categories (chest radiograph, PaO2/FiO2, PEEP and chest compliance; a score ? 2.75 signifying ARDS). We reasoned that judging ARDS severity by PaO2/ FiO2 alone may result in overlap in severity between the 2 schemes.

Methods: The National Insurance Health(NIH) in Taiwan ARDS quality assurance data base was used; with 2,030 subjects meeting Berlin definition for ARDS between Oct 2020 and May 2021. Subjects were managed with the ARDS Net protocol. LIS severity was categorized pre hoc as: Minimal (< 2), Mild (2-2.5), ARDS (2.8-3) and severe ARDS (> 3) based on ECMO criteria. These distributions were compared across Berlin categories. This strategy was repeated for 3 PEEP categories (< 10, 10-15 and > 15 cmH2O).

Results: Berlin-defined mild ARDS was characterized predominantly by minimal (32%) and mild (55%) lung injury by LIS, whereas 19% of Berlin-defined severe ARDS was designated as mild lung injury by LIS.Considerable overlap occurred between Berlin-defined moderate and severe ARDS with LIS-defined ARDS (30% vs. 37% respectively). Seventy-four percent of Berlin-defined Mild ARDS had PEEP < 10 cmH2O compared to 40 and 41% of Berlin-defined Moderate and Severe ARDS respectively.

Conclusion: Berlin categories of ARDS severity reasonably approximate lung injury severity by LIS. However, considerable overlap in LIS and PEEP distribution occurs between Berlin-defined mild and moderate categories that may lead to underestimation of the true degree of lung injury.