Jonathan Chang
Brigham & Women's Hospital
San Diego, California
Disclosure information not submitted.
Tom Chen, PhD
Instructor
Harvard Pilgrim Health Care Institute, United States
Disclosure information not submitted.
Cara McKenna, MPH
Programmer Analyst
Harvard Pilgrim Health Care Institute, United States
Disclosure information not submitted.
Michael Klompas, MD
Associate Professor
Brigham and Women's Hospital, United States
Disclosure information not submitted.
Chanu Rhee, MD, MPH
Assistant Professor of Medicine
Harvard Medical School, United States
Disclosure information not submitted.
Title: Intubation Versus Ventilator-Sparing Oxygen Support in COVID-19 ARDS: A Multicenter Analysis
Introduction/Hypothesis: Clinical practice has evolved from early intubation for COVID-19 patients with acute respiratory distress syndrome (ARDS) to preferential trialing of ventilator-sparing oxygen support. However, it is unclear whether intubation leads to worse outcomes compared to ventilator-sparing oxygen support when rigorously accounting for patients’ severity-of-illness.
Methods: We conducted a retrospective cohort study using detailed electronic health record data from adults admitted to four hospitals in Eastern Massachusetts between March and December 2020 with PCR-confirmed COVID-19 infection and moderate-severe ARDS (PaO2/FiO2 [P:F] ratio ≤ 200) within 48 hours of hospital arrival. We fit a marginal structural Cox model to estimate the average treatment effect of intubation versus ventilator-sparing oxygen support (i.e., any non-invasive oxygen device) on time to mortality. We used inverse probability weights for both treatment and censoring to adjust for demographics, comorbidities, BMI, calendar time, daily P:F ratio, daily vital signs, and daily labs (including white blood cell count, C-reactive protein).
Results: The cohort included 633 COVID-19 patients with ARDS (median 70 years old, 41% female, 43% with obesity, 35% with diabetes, 32% with chronic lung disease) who contributed 8,974 person-days to the analysis. 235 patients (37%) were intubated a median of 25 hours (IQR 2.8-89) after hospital arrival while 398 patients were managed with non-invasive oxygen support (including 3% with BiPAP, 14% with high flow nasal cannula, 24% with non-rebreather mask). 87 (37%) patients who were intubated and 96 (24%) of those not intubated died in hospital, corresponding to an incidence rate of 16.3 and 25.4 deaths per 1000 person-days, respectively. In the adjusted marginal structural model, however, intubation was associated with a similar risk of mortality as ventilator-sparing oxygen support (hazard ratio 1.21; 95% CI 0.80-1.81).
Conclusions: After rigorous accounting for time-varying confounding by severity-of-illness, intubation was associated with a similar risk of mortality as ventilator-sparing oxygen support in COVID-19 patients with ARDS. These data support trialing non-invasive oxygen support in most COVID-19 patients with ARDS and proceeding to intubation when clinically necessary.