Fluid Resuscitation in the Critically Ill: Crystalloid or Colloid?

Abstract & Commentary

Synopsis: In this meta-analysis of 37 prospective studies comparing crystalloid and colloid fluid infusion as resuscitation in critically ill patients with sepsis, trauma, or surgery, no differences in mortality were detected, either overall or in any subgroup of patients.

Source: Schierhout G, et al. BMJ 1998;316:961-964.

This meta-analysis by schierhout and associates at the University College London Medical School examined the effect on mortality of fluid resuscitation using crystalloid or colloid solutions. Only prospective randomized controlled trials involving crystalloid vs. colloid solution administration in adult or pediatric patients with sepsis, trauma, surgery, or burns were considered for inclusion. Studies were excluded if the purpose of fluid resuscitation was preparation for surgery or if there were confounding variables. Data bases searched for qualifying studies included MEDLINE, EMBASE/Excerpta Medica, and the Cochrane Controlled Trials Register. In addition, Schierhout et al individually searched the bibliographies of trials and reviews and the proceedings of international congresses in an attempt to identify appropriate trials.

Schierhout et al identified 48 trials in their search, of which 37 met the inclusion criteria. In 1315 patients selected from 19 of the trials, no differences in mortality were found in relation to the type of resuscitation fluid used. Overall all-cause mortality rate in these 1315 patients was 18.7% for those resuscitated with crystalloid solutions as compared with 19.4% for those who received colloid solutions (relative risk = 1.19; 95% confidence interval = 0.98 - 1.45; P = 0.3). Subgroup analysis of only those trials in which allocation of the type of solution infused was concealed from the clinicians managing the patients yielded essentially the same results. Analysis of subgroups according to the type of injury or illness failed to bring out significant mortality differences in any group.


Controversy continues both at the bedside and in the classroom as to whether colloid solutions such as albumin or dextrans are more effective than crystalloids (either isotonic or hypertonic) in restoring and maintaining circulating blood volume in critically ill patients. Although colloid solutions are substantially more expensive, their use continues to have strong advocates, and clinical practice varies both from institution to institution and within a given center. Schierhout et al’s rigorous examination of data from a large number of studies failed to show a survival benefit for the use of colloids, either in aggregate or in patients in any of the examined subgroups.

If infusing colloid solutions, in the circumstances examined in this analysis, is better for patients than using crystalloids, the difference must surely be a small one. Schierhout et al were unable to show a survival benefit for colloid infusion in subgroups of patients with trauma, surgery, burns, or other injuries. This does not mean that colloid infusion might not be preferable to the use of crystalloids in certain categories of patients within these subgroups or in certain clinical settings that were not examined in this study. However, just which patients or settings those might be remains unknown at present, at least on the basis of results from prospective, randomized controlled clinical trials. Using crystalloid rather than colloid for fluid resuscitation in the ICU is clinically acceptable and costs less.