Abstract
The administration of intravenous fluid to critically ill patients is one of the most common, but also one of the most fiercely debated, interventions in intensive care medicine. Even though many thousands of patients have been enrolled in large trials of alternative fluid strategies, consensus remains elusive and practice is widely variable. Critically ill patients are significantly heterogeneous, making a one size fits all approach unlikely to be successful.New data from basic, animal, and clinical research suggest that fluid resuscitation could be associated with significant harm. There are several important limitations and concerns regarding fluid bolus therapy as it is currently being used in clinical practice. These include, but are not limited to: the lack of an agreed definition; limited and short-lived physiological effects; no evidence of an effect on relevant patient outcomes; and the potential to contribute to fluid overload, specifically when fluid responsiveness is not assessed and when targets and safety limits are not used.Fluid administration in critically ill patients requires clinicians to integrate abnormal physiological parameters into a clinical decision-making model that also incorporates the likely diagnosis and the likely risk or benefit in the specific patient's context. Personalised fluid resuscitation requires careful attention to the mnemonic CIT TAIT: context, indication, targets, timing, amount of fluid, infusion strategy, and type of fluid.The research agenda should focus on experimental and clinical studies to: improve our understanding of the physiological effects of fluid infusion, e.g. on the glycocalyx; evaluate new types of fluids; evaluate novel fluid minimisation protocols; study the effects of a no-fluid strategy for selected patients and scenarios; and compare fluid therapy with other interventions. The adaptive platform trial design may provide us with the tools to evaluate these types of interventions in the intrinsically heterogeneous intensive care unit population, accounting for the explicit assumption that treatment effects may be heterogeneous.