Crystalloids versus colloids: exploring differences in fluid requirements by systematic review and meta-regression.

BACKGROUND: Positive fluid balance has been associated with worse outcomes, and knowledge of differences in the amounts of different types of fluid needed to achieve the same end points may have important clinical implications. Large molecules persist longer in the blood vessels than smaller molecules, such that less IV colloid may be needed to achieve similar hemodynamic end points compared with crystalloid. Recent clinical data have, however, challenged this physiological concept, with investigators reporting lower-than-expected crystalloid/colloid ratios in various populations.

METHODS: We performed a systematic search in MEDLINE, EMBASE, and CENTRAL up to December 18, 2013, to retrieve all studies comparing (any) crystalloid with (any) colloid in all types of patients. The crystalloid/colloid ratio was calculated for each study. Descriptive analysis was performed for all studies, and a meta-analysis was performed in those studies reporting full data (in terms of means and standard deviations) of infused fluid volumes. Studies were grouped according to study and population characteristics. A meta-regression analysis was then performed to evaluate some of the possible reasons for differences in crystalloid/colloid ratios across studies.

RESULTS: From 976 studies, 48 were retained for the final analysis; 24 of the studies had sufficient data for meta-analysis. The crystalloid/colloid ratio across all the studies included in the meta-analysis was 1.5 (95% confidence interval, 1.36-1.65) with marked heterogeneity among studies (I = 94%). From the meta-regression analysis, decade of publication across all publications (P = 0.001) and concentration (tonicity) in the subgroup of albumin studies (P = 0.001) were associated with the administered crystalloid/colloid ratio. The reduction in heterogeneity among studies for all publications in the meta-regression was minimal, with the maximal decrease obtained when decade of publication was considered (R = 12%).

CONCLUSIONS: Greater fluid volumes are required to meet the same targets with crystalloids than with colloids, with an estimated ratio of 1.5 (1.36-1.65), but there is marked heterogeneity among studies. The crystalloid/colloid ratio seems to have decreased over the years, and differences in ratios are correlated with the concentration of albumin solutions; however, the main reasons behind the high heterogeneity among studies remain unclear.