Association of cumulative fluid balance trajectories with haemodynamics, inflammation, and long-term mortality in critically ill patients with circulatory failure

Abstract

Background: Cumulative fluid balance is usually described using a four-phase concept. This study aimed to identify cumulative fluid balance trajectories in acute care patients with circulatory failure by using growth mixture models, and to analyse their association with haemodynamics, inflammation, and mortality.

Methods: The heterogeneity in cumulative fluid balance trajectories within the first 7 days after ICU admission was analysed using latent process mixed models in a retrospective cohort of patients with acute circulatory failure from the Salzburg Intensive Care database. The optimal number of classes (ranging from 2 to 6) was chosen according to several criteria: Bayesian information criteria, entropy, examination of posterior classification probabilities, and clinical relevance. Haemodynamics, inflammation, and adjusted mortality within 180 days using a Cox model were compared amongst classes.

Results: A total of 1702 patients were included in the analysis. The optimal model identified five distinct classes of cumulative fluid balance trajectories diverging in both slope and shape. Cumulative fluid balance trajectories were associated with patient-centred mechanisms, including haemodynamics and inflammation. A sustained increase in cumulative fluid balance at 2-3 days after ICU admission was associated with higher mortality. Only 567 (33.3%) patients presented a cumulative fluid balance trajectory consistent with the four-phase concept over this period.

Conclusions: There is substantial heterogeneity of cumulative fluid balance trajectories in critically ill patients with circulatory failure. This might inform the design of future interventional trials assessing optimal fluid management strategies.

Keywords: acute inflammation; cumulative fluid balance; de-resuscitation; fluid overload; model; recovery.