Impact of ventilatory and laboratory parameter trajectories on short-term survival in acute respiratory distress syndrome patients: a retrospective study using joint models.

backgroundClinical research is based on the parameters at defined time points, such as admission, diagnosis or discharge, for the purpose of risk factor analysis in relation to outcome. However, these parameters are collected with greater frequency in clinical practice. The objective of this study was to demonstrate a correlation between the time course of closely monitored parameters, such as blood gases, ventilatory parameters or routine laboratory values, and the survival of patients with acute respiratory distress syndrome (ARDS) caused by pneumonia.

methodsThis single-center, retrospective study included 274 ARDS patients with primary pneumonia requiring invasive mechanical ventilation. Patients were treated at a German university hospital between January 2014 and April 2021. Ethical approval was obtained from the local ethics committee (BO-EK-374072021). Longitudinal data on ventilatory and inflammatory parameters were collected during ICU stays. The analysis was conducted using descriptive statistics, cox regression and joint models. Joint modelling was used to integrate the progression of these parameters with survival outcomes, with the modelling of longitudinal data performed using quadratic B-splines.

resultsThe cohort included 274 patients, with an ICU mortality rate of 49.6%. Non-survivors were older (67 vs. 62 years, p < 0.001) and had higher SOFA scores at admission (10 vs. 8, p < 0.001). Differences in ventilatory parameters, including driving pressure and the PaO₂/FIO₂ ratio, as well as inflammatory markers such as procalcitonin, were observed between survivors and non-survivors during the ICU stay. The joint model analysis revealed a significant effect of the time course of parameters, such as positive end-expiratory pressure (PEEP), peak airway pressure (Ppeak), driving pressure, minute ventilation, tidal volume, C-reactive protein (CRP) and procalcitonin on mortality. The increase over time (slope-dependent association) for these parameters was strongly associated with mortality. For example, driving pressure was associated with mortality both by its current value (HR 1.16) and by its increase over time (HR 7.10). Similarly, tidal volume (HR 0.72 and 0.07), minute ventilation (HR 0.91 and 0.36), PEEP (HR 1.32 and 13.52), Ppeak (HR 1.20 and 3.28) and CRP (HR 1.14 and 4.25) showed a current value association and a strong slope-dependent association with mortality.

conclusionThis study underscores the importance of analyzing the dynamics of clinical parameters rather than static values for ARDS management. The findings suggest that changes in routine clinical parameters over time provide valuable prognostic information and should be prioritized in risk assessment and therapeutic decision making.