ABSTRACT
Background The role of non-invasive (continuous positive airway pressure (CPAP) or Non-invasive ventilation (NIV)) respiratory support (NIRS) as a primary oxygenation strategy for COVID-19 patients with acute severe hypoxic respiratory failure (AHRF), as opposed to invasive mechanical ventilation (invasive-MV), is uncertain. While NIRS may prevent complications related to invasive MV, prolonged NIRS and delays in intubation may lead to adverse outcomes. This study was conducted to assess the role of NIRS in COVID-19 hypoxemic respiratory failure and to explore the variables associated with NRIS failure. Methods This is a single-center, observational study of two distinct waves of severe COVID-19 patients admitted to the ICU. Patients initially managed with non-invasive respiratory support with laboratory-confirmed SARS-CoV-2 in acute hypoxaemic respiratory failure were included. Demographics, comorbidities, admission laboratory variables, and ICU admission scores were extracted from electronic health records. Univariate and multiple logistic regression was used to identify predictive factors for invasive mechanical ventilation. Kaplan-Meier survival curves were used to summarise survival between the ventilatory and time-to-intubation groups. Results There were 291 patients, of which 232 were managed with NIRS as an initial ventilation strategy. There was a high incidence of failure (48.7%). Admission APACHE II score, SOFA score, HACOR score, ROX index, and PaO2/FiO2 were all predictive of NIRS failure. Daily (days 1-4) HACOR scores and ROX index measurements highly predicted NIRS failure. Late NIRS failure (>24 hours) was independently associated with increased mortality (44%). Conclusion NIRS is effective as first-line therapy for COVID-19 patients with AHRF. However, failure, particularly delayed failure, is associated with significant mortality. Early prediction of NIRS failure may prevent adverse outcomes.