Generalized additive mixed model to evaluate the association between ventilatory ratio and mortality in patients: A retrospective cohort study.

Abstract

Previous studies have indicated that a higher ventilatory ratio (VR) is associated with mortality. However, it is unknown whether dynamic changes in VR over time affect the prognosis of critically ill patients. This study aims to investigate the significance of VR during the progression of the disease in critically ill patients. The Medical Information Mart for Intensive Care III database was searched to identify critically ill patients. The primary outcome was 30-day mortality. Multivariable Cox regression was used to elucidate the relationship between the VR and mortality. Finally, we employed a generalized additive mixed model to compare trends in VR over time between survivors and non-survivors. A total of 8024 patients were enrolled. Multivariable Cox regression analysis identified a baseline VR ≥1.89 as an independent risk factor predicting 30-day mortality (hazard ratio: 2.10, 95% confidence interval: 1.89-2.33, P < .001) and 90-day mortality (hazard ratio: 2.18, 95% confidence interval: 1.97-2.41, P < .001) after adjusting for potential confounders. In the subgroup analyses, the observed association between VR and 30-day mortality showed consistent direction across most subgroups. The generalized additive mixed model results highlighted that the difference in VR between survivors and non-survivors increased by an average of 0.01 per day after adjusting for several covariates. In conclusion, VR dynamically mirrors pathophysiological changes in critically ill patients and its escalation is linked to higher mortality rates. Monitoring VR's dynamic shifts might offer more immediate prognostic information, thus aiding in timely interventions and risk stratification.