Abstract Rationale Noninvasive ventilation (NIV) for preoxygenation before emergency intubation reduces severe hypoxemia and may lower the risk of cardiac arrest. Adoption of new evidence is challenging in high-acuity settings. A learning-health-system framework was applied to measure local NIV use and accelerate implementation. Methods This multisite quality-improvement project was conducted across three hospitals within a tertiary academic system. SMART Aim: increase NIV preoxygenation from 15% to 30% by December 2025. An Ishikawa analysis informed a key driver diagram (standardization, equipment availability, and education). Change concepts: internal NIV guidance, standardized mask stocking (carts/bags), and brief education highlighting PREOXI trial findings and airway-safety goals. Iterative cycles began September 2024. Intubation notes were abstracted quarterly via a natural-language-processing (NLP) workflow and analyzed in Stata/SE 15.1. Process measure: monthly percentage of intubations using NIV for preoxygenation. Outcome measure: proportion with oxygen desaturation 80% within five minutes of the final intubation attempt. Balancing measure: operator reported aspiration. Performance was tracked with a run-chart; centerlines followed Institute for Healthcare Improvement rules. After rates plateaued, a brief operator email was deployed with prespecified barrier categories and later summarized the responses via a Pareto chart. Results From Aug 2023-Aug 2025, 2600 intubations were performed by the critical-care service. Baseline NIV use before PREOXI publication was 15.2% (through June 2024). Following sequential interventions, NIV use increased, reaching 21.3% on average and peaking at 29.3% (May 2025). Run-chart analysis showed a median shift beginning January 2025. Desaturation 80% occurred in 9.1% before and 8.8% after PREOXI publication. Aspiration was 5.6% before and 4.6% after. Post-intubation operator surveys were obtained for 32/46 consecutive non-NIV intubations (69.6% response). Pareto analysis identified patient-related factors (37%; e.g. active emesis, agitation, intolerance), limited time for preoxygenation due to perceived clinical urgency (34%), and equipment access issues (11%) as the most frequent residual barriers, accounting for approximately 82% of responses. Conclusions A structured EMR intubation note plus NLP-based feedback enabled rapid, system-level adoption of NIV preoxygenation. Using a learning-health-system model with run-chart monitoring and targeted barrier analysis, we aligned practice with emerging evidence and improved adoption of evidence based preoxygenation. Further work will focus on addressing residual barriers to use of NIV for preoxygenation. This abstract is funded by: None
Bhakta et al. (Fri,) studied this question.