A novel non-invasive index of oxygenation and prediction of outcomes for patients on high-flow nasal cannula

2021 
Introduction/Aim: Acute respiratory failure (ARF) is a major cause of morbidity and mortality. Predicting patient trajectory has important ramifications. Current tools are limited either by invasiveness or unreliable data points. We explored the utility of a novel non-invasive index of oxygenation (flow x FiO2/SpO2) at predicting outcomes in patients with ARF managed with high flow nasal prongs (HFNP). Methods: This is a retrospective cohort of patients treated with HFNP from July 2018 to June 2019. Patients were treated by a respiratory physician, aged 18-85, and had a diagnosis of ARF. Exclusion criteria included post-operative use of HFNP, or the indication for HFNP was not clear. The index was compared to the ROX index for external validation. The primary outcome was a composite outcome of de-escalation to nasal prongs, room air or discharge, and secondary outcome was the ability of the FOx Index to discern between treatment failure or success. Results: Seventy patients were included in the analysis. Mean age was 65.2± 17.6 (SD) years and 48.6% of the population were female. Exacerbations of COPD was the most common reason for HFNP, with type 2 respiratory failure more common than type 1. Eighty-two percent of patients achieved the primary outcome, with escalation to NIV being the most common reason for failure. The FOx Index, with values of 12.8 or lower, could predict between success and failure at 6 and 12 hours (p = 0.023). The ROX index diverged at 2 hours, with a value above 13.7 predicting success (p = 0.031). The FOx Index performed equally well between type 1 and 2 respiratory failure. Conclusion: The FOx Index was able to discern between success and failure at 6 and 12 hours. This represents an important proof of concept for a prospective study. The outcomes are limited by retrospective design, and susceptibility to success bias. Grant Support: Nil and no conflict of interest to declare by any of the authors.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    0
    References
    0
    Citations
    NaN
    KQI
    []
    Baidu
    map