Respiratory failure & ventilation
Oxygenation thresholds for invasive ventilation in hypoxemic respiratory failure: a target trial emulation in two cohorts
Authors' conclusion
Initiating invasive ventilation at lower hypoxemia severity will increase the rate of invasive ventilation, but this can either increase or decrease the expected mortality, with the direction of effect likely depending on baseline mortality risk and clinical context.
Study context
- Population
- ICU patients with hypoxemic respiratory failure receiving FiO2 ≥ 0.4 via non-rebreather mask, NIV or HFNC (MIMIC-IV 2008–2019; AmsterdamUMCdb 2003–2016)primary analysis
- Participants
- 3,357 patients
- Exposure
- Invasive ventilation initiation thresholds SF < 110 vs < 98 vs < 88
- Outcome
- Mortality
- Follow-up
- 28 days
Methodological detailsnonparametric Bayesian G-computation, MIMIC-IV primary analysis, AmsterdamUMCdb secondary analysis
- Nonparametric Bayesian G-computation
- MIMIC-IV primary analysis
- AmsterdamUMCdb secondary analysis
Main finding
As reported in the abstract
SF < 110
22.2%
posterior mean · CrI 19.2–25.0
SF < 98
47%
predicted 28-day probability of invasive ventilation
SF < 88
19%
predicted 28-day probability of invasive ventilation
Observational data: associations, not effects.
Further findings
Predicted 28-day mortality (AmsterdamUMCdb secondary analysis, 1,279 patients) · 28 days — SF < 98 vs SF < 110 · CrI SF < 110 posterior mean 14.6% (CrI 7.7–22.3) · absolute risk decrease 0.5% (0.0–0.9)
Predicted 28-day probability of invasive ventilation
| Outcome | Absolute risk decrease (95% CrI) |
|---|---|
| Predicted 28-day mortality (AmsterdamUMCdb secondary analysis, 1,279 patients) · 28 daysSF < 88 vs SF < 110 · CrI | 1.9% (0.9–2.8) |
- Predicted 28-day mortality (AmsterdamUMCdb secondary analysis, 1,279 patients): 95% CrI
- Predicted 28-day mortality (AmsterdamUMCdb secondary analysis, 1,279 patients): mortality highest with SF < 110, in contrast with the primary analysis
Original publication
Composed by AI from the published abstract and checked against it. How Briefs are made →