Associations of early restricted S/F oxygenation burden and the plateau pressure–set PEEP pressure proxy with in-hospital mortality in patients receiving invasive mechanical ventilation: a MIMIC-IV landmark study

Single measurements of oxygenation or ventilatory pressure may not capture the sustained physiological burden experienced during invasive mechanical ventilation. We evaluated the individual and joint associations of early restricted peripheral oxygen saturation-to-fraction of inspired oxygen ratio (SpO₂/FiO₂, S/F) oxygenation burden and a plateau pressure–set positive end-expiratory pressure (PEEP) pressure proxy with in-hospital mortality. We performed a retrospective 24-hour landmark cohort study using MIMIC-IV, with supplementary directional-consistency analyses in eICU and MIMIC-III. Adults with early findings compatible with acute hypoxemic respiratory failure who were alive and hospitalized at 24 h were included. During actual invasive mechanical ventilation between 6 and 24 h, we calculated the time-weighted mean restricted S/F ratio and the time-weighted mean difference between plateau pressure and set PEEP. Associations with post-landmark in-hospital mortality were examined using multivariable logistic regression, natural splines, interaction analysis, and marginal standardization. Among 18,285 patients in the landmark cohort, 4,646 had both exposures recorded for at least two distinct ventilation hours; 1,354 (29.1%) died in hospital. Per one-standard-deviation increase in burden, the adjusted odds ratio for mortality was 1.36 (95% confidence interval [CI], 1.24–1.49) for oxygenation burden and 1.25 (95% CI, 1.13–1.39) for the pressure proxy. The standardized mortality risks were 25.6% for neither burden high, 33.6% for high oxygenation burden only, 28.8% for high pressure burden only, and 43.9% for both burdens high. The high-pressure-only group was not clearly different from the neither-high group (risk difference, 3.2% points; 95% CI, − 0.4 to 7.3). Compared with neither burden high, both burdens high was associated with an absolute risk difference of 18.3% points (95% CI, 13.2–23.9). Association directions were consistent across sensitivity analyses, eICU, and MIMIC-III. Among the selected subgroup of patients in whom both measures could be adequately evaluated, greater restricted S/F oxygenation burden and a higher plateau pressure–set PEEP pressure proxy were each associated with post-landmark in-hospital mortality. Prospective studies are needed to assess generalizability and clinical utility. Not applicable.