Fluid therapy is one of the most frequent interventions given to hospitalised patients, and one-third of all patients in intensive care units (ICUs) worldwide receive fluid for resuscitation each day [1]. As these patients have considerable mortality rates, any differences in outcome between different types of fluids will have a marked effect on the overall mortality of critically ill patients. Appropriate fluid therapy is therefore likely to improve global health.

For decades this has been one side of a difficult dilemma faced by clinicians, guideline committee members and regulatory authorities. The other grim side was the lack of high-quality data to support the choice of fluid therapy for resuscitation. In the recent ESICM task force guideline on colloid therapy, seven out of the nine recommendations were based on low quality of evidence and none were based on high quality of evidence [2]. Now this is changing through the publication in 2012 of several randomised trials on fluid therapy [3–5]. Following the well-established track of evidence-based medicine, these trial data are now being added to the cumulated evidence in updated meta-analyses on fluid therapy.

In the current issue of Intensive Care Medicine, David Gattas and colleagues present the results of their systematic review and meta-analysis on behalf of the CHEST trial management committee [6]. They have aggregated the data of 35 trials assessing the effect of 6 % hydroxyethyl starch (HES) with molecular weight of 130 and substitution ratio of approximately 0.4 for resuscitation in more than 10,000 acutely ill adults. This is highly relevant, because HES was the most frequently used colloid solution in the latest assessment of fluid resuscitation in ICUs worldwide [1] and 6 % HES130 has for years been the preferred HES solution [7]. The results of the meta-analysis are convincing, as the heterogeneity of the included trial data was low, high numbers of events were analysed for the main outcomes and the trials contributing with most events had crystalloid as comparator and low risk of bias [4, 5]. Therefore, the results should, together with those from the recent large trials on 6 % HES130 [4, 5], inform clinicians, guideline committees and regulatory authorities.

So what did Gattas and colleagues’ meta-analysis on 6 % HES130 [1] show? In acutely ill surgical and intensive care patients, fluid resuscitation with 6 % HES130 increased the use of renal replacement therapy [relative risk (RR) in the 6 % HES130 group 1.25, 95 % confidence interval (CI) 1.08–1.44] and mortality (RR 1.08, 95 % CI 1.00–1.17) compared with other fluid solutions. As can be seen, the latter was borderline significant with the lower range of the 95 % CI touching the no-difference point. There are limitations to the systematic review, as the authors did not adhere to all the recommendations from the Cochrane Collaboration. Thus the protocol was not pre-published, only trials that reported specific outcomes were included, authors were not contacted to try and obtain unpublished data and not all patient-relevant outcomes were assessed (e.g. pruritus). In addition, the mortality data may have been skewed by the inclusion of trials with short follow-up time, which may not capture delayed harm induced by HES. The trials on HES using follow-up beyond 28 days have shown either statistically significant increased mortality with HES [4] or point estimates trending towards increased mortality [3, 5, 8–11]. The reporting in the included trials of data on transfusion and bleeding differed so much that meta-analyses of these data were deemed inappropriate.

Appropriately, the authors’ conclusions were conservative. Clinicians, on the other hand, may not have to be as conservative, because they should give the fluid that is likely to benefit their patients. Presently there are no data from high-quality trials showing that 6 % HES130 improves any patient-important outcome, and there are clear signals of harm. In addition, the cost of 6 % HES130 is several times higher than that of crystalloid solutions. Subsequent analyses and trial data may inform us if there are subgroups of patients who may have net benefit from resuscitation with 6 % HES130. Until then, we should rely on the results of the meta-analysis by Gattas et al. [6] and those of the recent large trials [4, 5] and avoid 6 % HES130 in critically ill patients.