Abstract
IMPORTANCE: Most studies on acute respiratory distress syndrome (ARDS) group patients by severity based on their initial degree of hypoxemia. However, this grouping has limitations, including inconsistent hypoxemia trajectories and outcomes. OBJECTIVES: This study explores the benefits of grouping patients by resolver status based on their hypoxemia progression over the first 7 days. DESIGN SETTING AND PARTICIPANTS: This is an observational study from a large single-center database. Medical Information Mart for Intensive Care (MIMIC)-IV and MIMIC Chest X-ray JPEG databases were used. Mechanically ventilated patients that met the Berlin ARDS criteria were included. MAIN OUTCOMES AND MEASURES: The primary outcome was the proportion of hypoxemia resolvers vs. nonresolvers in non-COVID-19 ARDS patients. Nonresolvers were defined as those whose hypoxemia worsened or remained moderate or severe over the first 7 days. Secondary outcomes included baseline admission characteristics, initial blood gases and ventilation settings, length of invasive mechanical ventilation, length of ICU stay, and ICU survival rates across resolver groups. RESULTS: A total of 894 ICU admissions were included in the study. Of these, 33.9% were hypoxemia nonresolvers. The resolver groups showed no significant difference in age, body mass index, comorbidities, or Charlson score. There was no significant difference in the percentage of those with initial severe hypoxemia between the two groups (8.1% vs. 9.2%; p = 0.126). The initial Pao(2)/Fio(2) ratio did not significantly increase the odds ratio (OR) of being a nonresolver (OR, 0.84; 95% CI, 0.65-1.10). Nonresolver mortality was 61.4%, comparable to the survival rates seen in nonresolvers in a previous large COVID-19 ARDS study. CONCLUSIONS AND RELEVANCE: Our study shows that resolver status is a valuable grouping in ARDS. It has significant advantages over grouping by initial degree of hypoxemia, including better mapping of trajectory and comparable outcomes across other studies. While it may offer insights into disease-specific associations, future studies should include resolver status analysis for more definitive conclusions.