Predictive performance of ROX index and its variations for NIV failure

Area of Science:

• Critical Care Medicine
• Respiratory Physiology
• Health Informatics

Background:

• Non-invasive ventilation (NIV) is a common ICU treatment.
• Predicting NIV failure is crucial for timely intervention and improved patient outcomes.
• Existing predictive indices require validation in large, diverse patient cohorts.

Purpose of the Study:

• To evaluate the predictive performance of the ROX index and its variations (mROX, ROX-HR, mROX-HR) for NIV failure.
• To assess these indices in large public ICU databases.
• To determine their utility in predicting the risk of intubation in ICU patients receiving NIV.

Main Methods:

• Retrospective observational cohort study utilizing AmsterdamUMCdb and MIMIC-IV databases.
• Included 3344 non-invasively ventilated ICU patients.
• Calculated ROX indices based on SpO2, respiratory rate, FiO2, and PaO2 within the first 12 hours of NIV.

Main Results:

• The ROX index demonstrated the best predictive performance with an AUROC of 0.626 at 10-12 hours.
• NIV failure was associated with higher SOFA scores and CRP levels.
• No significant differences in NIV duration or 28-day mortality were observed between failure and success groups.

Conclusions:

• The ROX index and its variations exhibit only moderate predictive value for NIV failure in large ICU cohorts.
• Current performance levels do not support their use for clinical decision support in predicting intubation.
• Further research may be needed to refine or identify more accurate predictive tools for NIV failure.