Fluid management in the intensive care unit: bioelectrical impedance vector analysis as a tool to assess hydration status and optimal fluid balance in critically ill patients.

BACKGROUND: Fluid balance disorders are a relevant risk factor for morbidity and mortality in critically ill patients. Volume assessment in the intensive care unit (ICU) is thus of great importance, but there are currently few methods to obtain an accurate and timely assessment of hydration status. Our aim was to evaluate the hydration status of ICU patients via bioelectric impedance vector analysis (BIVA) and to investigate the relationship between hydration and mortality.
METHODS: We evaluated 280 BIVA measurements of 64 patients performed daily in the 5 days following their ICU admission. The observation period ranged from a minimum of 72 h up to a maximum of 120 h. We observed the evolution of the hydration status during the ICU stay in this population, and analyzed the relationship between mean and maximum hydration reached and mortality--both in the ICU and at 60 days--using logistic regression.
RESULTS: A state of overhydration was observed in the majority of patients (70%) on admission, which persisted during the ICU stay. Patients who required continuous renal replacement therapy (CRRT) were more likely to be overhydrated starting from the 2nd day of observation. Logistic regression showed a strong and significant correlation between mean/maximum hydration reached and mortality, both independently and correcting for severity of prognosis.
CONCLUSIONS: Fluid overload measured by BIVA is a frequent condition in critically ill patients--whether or not they undergo CRRT--and a significant predictor of mortality. Hence, hydration status should be considered as an additional prognosticator in the clinical management of the critically ill patient. KEY MESSAGES: (i) On the day of ICU admittance, patients showed a marked tendency to overhydration (>70% of total). This tendency was more pronounced in patients on CRRT. (ii) Hyperhydration persisted during the ICU stay. Patients who underwent CRRT showed significantly higher hyperhydration from the 2nd day of hospitalization. (iii) Nonsurvivors showed worse hyperhydration patterns in comparison to survivors in logistic univariate analysis (p < 0.05). This relationship between hydration and mortality is confirmed even when controlling for the effect of a worse prognosis approximated by any of three ICU scoring systems (APACHE II, SAPS II and SOFA). Mean and maximum hydration levels present a stronger correlation with mortality than with mean and maximum cumulative fluid balance reached during the observation period.