Time-varying intensity of mechanical ventilation and mortality in patients with acute respiratory failure: a registry-based, prospective cohort study

Abstract

Background Mortality in acute respiratory failure remains high despite the use of lung-protective ventilation. Recent studies have shown an association between baseline ventilation parameters (driving pressure or mechanical power) and outcomes for patients with acute respiratory distress syndrome. Strategies focused on limiting these parameters have been proposed to further improve outcomes. However, it remains unknown whether driving pressure and mechanical power should be limited over the entire duration of mechanical ventilation and in all patients with acute respiratory failure. We aimed to estimate the association between exposure to different intensities of mechanical ventilation over time and intensive care unit (ICU) mortality in patients with acute respiratory failure.

Methods: In this registry-based, prospective cohort study, we obtained data from the Toronto Intensive Care Observational Registry, which includes all patients receiving mechanical ventilation for 4 h or more in nine ICUs that are affiliated with the University of Toronto (Toronto, ON, Canada). We included all adult (≥18 years) patients who received invasive mechanical ventilation between April 11, 2014, and June 5, 2019. Patients were excluded if they received treatment with extracorporeal life support. The primary outcome was ICU mortality. Bayesian joint models were used to estimate the strength of associations, accounting for informative censoring due to death during follow-up.

Findings: Of 13 939 patients recorded in the registry, 13 408 (96·2%) were eligible for descriptive analysis. The primary analysis comprised 7876 (58·7%) patients with complete baseline characteristics, and a secondary analysis included all 13 408 patients after multiple imputation in the joint model analysis. 2409 (18·0%) of 13 408 patients died in the ICU. After adjustment for baseline characteristics, including age and severity of illness, a significant increase in the hazard of death was found to be associated with each daily increment in driving pressure (hazard ratio 1·064, 95% credible interval 1·057-1·071) or mechanical power (hazard ratio 1·060, 95% credible interval 1·053-1·066). These associations persisted over the duration of mechanical ventilation.

Interpretation: Cumulative exposure to higher intensities of mechanical ventilation was harmful, even for short durations. Limiting exposure to driving pressure or mechanical power should be evaluated in further studies as promising ventilation strategies to reduce mortality in patients with acute respiratory failure.

Funding: Canadian Institutes of Health Research.

Copyright © 2020 Elsevier Ltd. All rights reserved.

Figures

Figure 1

Study profile PaO2=partial pressure of oxygen. FiO2=fraction of inspired oxygen. APACHE=Acute Physiology and Chronic Health Evaluation. *Patients could fulfill multiple exclusion criteria.

Figure 2

Outcomes in relation to gas exchange characteristics at baseline (A) Cumulative probability curves stratified by baseline PaO2/FiO2 ratio. (B) Relative hazard of death in the ICU predicted on the basis of baseline PaO2/FiO2 ratio. (C) Relative hazard of death in the ICU predicted on the basis of baseline ventilatory ratio, calculated as (measured minute ventilation × measured PaCO2) / (expected minute ventilation × ideal PaCO2). Expected minute ventilation is 100 × predicted bodyweight in kg × 100 (mL/min). The unadjusted relationships between the two types of respiratory failure and ICU mortality were estimated using cause-specific Cox proportional hazard models. ICU=intensive care unit. PaO2=partial pressure of oxygen. FiO2=fraction of inspired oxygen. PaCO2=partial pressure of carbon dioxide.

Figure 3

Association between driving pressure and ICU mortality during mechanical ventilation (A) Unadjusted relationship between dynamic driving pressure at baseline and relative hazard of death in the ICU, estimated using a cause-specific Cox proportional hazard model (n=10 591 patients with available dynamic driving pressure at baseline). (B) Differences in the cumulative incidence of death on the basis of static driving pressure at baseline (n=1633 patients with available data on the static measurements). (C) Time-varying HR obtained from a Bayesian joint model estimating the association between dynamic driving pressure and ICU mortality; the association persisted for the entire duration of mechanical ventilation (n=7876 patients with available baseline data on disease severity). (D) A Bayesian joint model including an interaction term with the PaO2/FiO2 ratio at baseline was used to investigate whether the association of time-varying driving pressure and ICU mortality is moderated by the severity of acute respiratory failure (n=7876 patients with available baseline data on disease severity). Shaded areas represent 95% CI in panels A and B, and 95% credible intervals in panel C; lines either side of the dot represent 95% credible intervals in panel D. ICU=intensive care unit. HR=hazard ratio. PaO2=partial pressure of oxygen. FiO2=fraction of inspired oxygen.