Fluid therapy and outcome: a prospective observational study in 65 German intensive care units between 2010 and 2011

Christian Ertmer, Bernhard Zwißler, Hugo Van Aken, Michael Christ, Fabian Spöhr, Axel Schneider, Robert Deisz, Matthias Jacob

Outcome data on fluid therapy in critically ill patients from randomised controlled trials may be different from data obtained by observational studies under "real-life" conditions. We conducted this prospective, observational study to investigate current practice of fluid therapy (crystalloids and colloids) and associated outcomes in 65 German intensive care units (ICUs). In total, 4545 adult patients who underwent intravenous fluid therapy were included. The main outcome measures were 90-day mortality, ICU mortality and acute kidney injury (AKI). Data were analysed using logistic and Cox regression models, as appropriate. In the predominantly post-operative overall cohort, unadjusted 90-day mortality was 20.1%. Patients who also received colloids (54.6%) had a higher median Simplified Acute Physiology Score II [25 (interquartile range 11; 41) vs. 17 (7; 31)] and incidence of severe sepsis (10.2 vs. 7.4%) on admission compared to patients who received exclusively crystalloids (45.4%). 6% hydroxyethyl starch (HES 130/0.4) was the most common colloid (57.0%). Crude rates of 90-day mortality were higher for patients who received colloids (OR 1.845 [1.560; 2.181]). After adjustment for baseline variables, the HR was 1.666 [1.405; 1.976] and further decreased to indicate no associated risk (HR 1.003 [0.980; 1.027]) when it was adjusted for vasopressor use, severity of disease and transfusions. Similarly, the crude risk of AKI was higher in the colloid group (crude OR 3.056 [2.528; 3.694]), after adjustment for baseline variables OR 1.941 [1.573; 2.397], and after full adjustment OR 0.696 [0.629; 0.770]), the risk of AKI turned out to be reduced. The same was true for the subgroup of patients treated with 6% HES 130/0.4 (crude OR 1.931 [1.541; 2.419], adjusted for baseline variables OR 2.260 [1.730; 2.953] and fully adjusted OR 0.800 [0.704; 0.910]) as compared to crystalloids only. The present analysis of mostly post-operative patients in routine clinical care did not reveal an independent negative effect of colloids (mostly 6% HES 130/0.4) on renal function or survival after multivariable adjustment. Signals towards a reduced risk in subgroup analyses deserve further study. Trial registration ClinicalTrials.gov Identifier: NCT01122277, registered May 11th, 2010.