Evaluating the impact of ESICM 2023 guidelines and the new global definition of ARDS on clinical outcomes: insights from MIMIC-IV cohort data.

Summary

Using MIMIC-IV, the new ARDS definition enabled earlier diagnosis and included lower-mortality patients compared with Berlin. Patients meeting the new but not Berlin criteria had better responses to non-invasive ventilation (p=0.009). An XGBoost classifier achieved AUC 0.88, and simple measures (respiratory rate, BUN) aided diagnosis in resource-limited settings.

Key Findings

• The new ARDS definition diagnosed patients earlier and captured a lower-mortality cohort compared with the Berlin definition.
• Patients meeting the new but not Berlin criteria showed favorable responses to non-invasive ventilation (p=0.009).
• An XGBoost classifier predicted ARDS subphenotypes with AUC 0.88±0.02; RR and BUN were practical diagnostic aids in resource-limited settings.

Clinical Implications

Clinicians may diagnose ARDS earlier under the new definition and consider NIV in patients meeting the new but not Berlin criteria; simple variables (RR, BUN) can support triage where diagnostics are limited.

Limitations

• Retrospective, single-database analysis without external validation; potential center and practice pattern biases.
• Treatment effects (e.g., NIV benefit) are observational and prone to confounding by indication.

Future Directions

Prospective, multicenter validation of the new definition’s performance and ML classifier, with randomized evaluation of NIV strategies in newly defined subgroups.