Fluid resuscitation for hypovolaemia is a mainstay of the medical management of critically ill patients, whether as a result of trauma, burns, major surgery or sepsis. Although recent studies (Bickell 1994) have suggested that the timing of volume replacement deserves careful consideration, when it comes to selecting the resuscitation fluid, clinicians are faced with a range of options. At one level the choice is between a colloid or crystalloid solution. Colloids are widely used, having been recommended in a number of resuscitation guidelines and intensive care management algorithms (Armstrong 1994; Vermeulen 1995).
The US Hospital Consortium Guidelines recommend that colloids are used in haemorrhagic shock prior to the availability of blood products, and in non-haemorrhagic shock following an initial crystalloid infusion. A 1995 survey of US academic health centres, however, found that the use of colloids far exceeded even the Hospital Consortium recommendations (Yim 1995). Surveys of burn care in the US (Fakhry 1995) and in Australia (Victorian DUAC 1991) found that the use of colloids for resuscitation varied without a set pattern.
The choice of fluid has considerable cost implications. Volume replacement with colloids is considerably more expensive than with crystalloids. Clinical studies have shown that colloids and crystalloids have different effects on a range of important physiological parameters. Because of these differences, all-cause mortality is arguably the most clinically relevant outcome measure in randomised trials comparing the two fluid types.
Why it is important to do this review
Although there have been previous meta-analyses of mortality in randomised trials comparing colloids and crystalloids (Bisonni 1991; Velanovich 1989), neither of these satisfy the criteria that have been proposed for scientific overviews (Oxman 1994), and they predate most of the trials that have been conducted using synthetic colloids, and hypertonic crystalloid solutions. The purpose of this systematic review is to identify and synthesise all available unconfounded evidence of the effect on mortality in critically ill patients of colloids compared to crystalloids for volume replacement.
To assess the effects on mortality of using colloids compared to crystalloids, during fluid resuscitation in critically ill patients.
Criteria for considering studies for this review
Types of studies
Controlled trials in which participants were randomised to treatment groups (colloid or control) on the basis of random allocation. As the comparison between fluid type was in terms of effects on mortality, we excluded randomised cross-over trials.
Types of participants
Critically ill patients (excluding neonates) who required volume replacement. We included patients who were critically ill as a result of trauma, burns, were undergoing surgery, or had other critical conditions such as complications of sepsis.
We excluded pre-operative elective surgical patients.
Types of interventions
We considered the following colloids: Dextran 70, hydroxyethyl starches, modified gelatins, albumin or plasma protein fraction.
There is overlap between albumin given for volume replacement and albumin given as a nutritional supplement, and many patients with a critical illness have low serum albumin. Where the trial was of total parenteral nutrition with or without albumin, we excluded it. We included trials where the albumin was given as part of volume replacement guided by colloid osmotic pressure or albumin levels.
The control group received crystalloid (isotonic or hypertonic) for fluid replacement. We included trials in which both groups received blood.
We excluded trials of fluids used for other purposes. For example, we excluded trials of pre-loading in preparation for elective surgery, and trials in patients undergoing fluid loading before cardiopulmonary bypass.
Types of outcome measures
The principal outcome measure was mortality from all causes, assessed at the end of the follow-up period scheduled for each trial.
Search methods for identification of studies
The searches were not restricted by date, language or publication status.
We searched the following electronic databases:
- Cochrane Injuries Group Specialised Register (searched 30 Sept 2008)
- CENTRAL (The Cochrane Library 2008, Issue 3)
- MEDLINE (1966 to September 2008)
- PubMed (searched 30 September, last three months)
- EMBASE (1980 to September 2008)
- ISI Web of Knowledge (1970 to September 2008)
- National Research Register (2006, Issue 4)
- Controlled Trials metaRegister (www.controlled-trials.com) (searched 30 Sept 2008)
The search strategy can be found in Appendix 1.
Searching other resources
We checked the reference lists of all identified trials and review articles, and contacted the trialists to identify any studies that may have been missed.
Data collection and analysis
Selection of studies
We independently examined titles, abstracts, and keywords of citations from electronic databases for eligibility. We obtained the full text of all relevant records and independently assessed whether each met the pre-defined inclusion criteria. We resolved disagreement by discussion.
Assessment of risk of bias in included studies
We scored allocation concealment as described by Higgins 2008, assigning 'No' to poorest quality and 'Yes' to best quality (the presence of solutions in identical containers was only taken to mean adequate concealment if the fluid containers were used sequentially).
- Yes = trials deemed to have taken adequate measures to conceal allocation (that is, central randomisation; serially numbered, opaque, sealed envelopes; or other description that contained elements convincing of concealment).
- Unclear = trials in which the authors either did not report an allocation concealment approach at all or reported an approach that did not fall into one of the other categories.
- No = trials in which concealment was inadequate (such as alternation or reference to case record numbers or to dates of birth).
We collected but did not score information on blinding and loss to follow up.
As a result of comments on the previous version of this review, we have stratified trials by type of fluid rather than type of original injury.
We calculated relative risks (RRs) and 95% confidence intervals (95% CI) for each study using a fixed-effect model. We then inspected each comparison visually for evidence of heterogeneity and performed a Chi
We then excluded trials with allocation concealment judged as inadequate and repeated the calculations.
Description of studies
We identified 65 trials meeting the inclusion criteria for study design, participants and interventions. We were able to obtain mortality data for 56 of these. We have reported details of the included trials in the 'Characteristics of included studies' table.
Reasons for exclusion of trials were: the use of a cross-over design, testing a resuscitation algorithm, giving the control group oral fluids, the intervention being directed to the maintenance of serum albumin levels, for haemodilution, for fluid loading and for the reduction of intracranial pressure (see 'Characteristics of excluded studies' table).
Of the 56 trials with data on deaths, the quality of allocation concealment was adequate in seven trials and unclear in most of the others.
There were 60 comparisons of colloids and crystalloids (add-on colloid), nine comparisons of colloid in hypertonic crystalloid with isotonic crystalloid, and three comparisons of colloid with hypertonic crystalloid.
Risk of bias in included studies
In general, the design of studies was not well reported. This is reflected in the number of unclear scores given for allocation concealment. We also collected information on blinding and loss to follow up. Blinding was not well reported and loss to follow up was generally small. The characteristics for each trial are listed in the 'Characteristics of included studies' table.
Effects of interventions
Colloids compared to crystalloids
Albumin or plasma protein fraction
Twenty-three trials reported data on mortality, including a total of 7754 patients. The pooled relative risk (RR) was 1.01 (95% confidence interval (95% CI) 0.92 to 1.10). When we excluded the trial with poor quality allocation concealment (Lucas 1978), pooled RR was 1.00 (95% CI 0.91 to 1.09).
Seventeen trials compared hydroxyethyl starch with crystalloids, including a total of 1172 randomised patients. The pooled RR was 1.18 (95% CI 0.96 to 1.44).
Eleven trials compared modified gelatin with crystalloid, including a total of 506 randomised patients. The pooled RR was 0.91 (95% CI 0.49 to 1.72).
Nine trials compared dextran with a crystalloid, including a total of 834 randomised patients. The pooled RR was 1.24 (95% CI 0.94 to 1.65).
Colloids in hypertonic crystalloid compared to isotonic crystalloid
One trial compared albumin and hypertonic saline with isotonic crystalloid. The RR of death was 0.50 (95% CI 0.06 to 4.33).
Eight trials compared dextran in hypertonic crystalloid with isotonic crystalloid, including 1283 randomised patients. The pooled RR was 0.88 (95% CI 0.74 to 1.05).
Colloids in isotonic crystalloid compared to hypertonic crystalloid
Three trials compared colloids in isotonic crystalloid with hypertonic crystalloid. In two of these, where the colloid was either gelatin or starch, there were no deaths in either group. In the remaining trial, with 38 patients, there was a RR of death of 7.00 (0.39 to 126.93) for use of colloid, based on three deaths in the treatment group and none in the control group.
This systematic review synthesises the evidence from RCTs comparing colloid and crystalloid fluid resuscitation across a wide variety of clinical conditions. The review has been updated and extensively revised to take into account the comments made since it was first published. In particular, several commentators pointed out that it is inappropriate to combine effect estimates from studies of different colloids. For example, it was argued that large molecular weight colloids such as hydroxyethyl starch may be better retained in the vascular compartment than albumin and gelatins, and would therefore be more likely to show a favourable effect on mortality (Gosling 1998). In response to these concerns, the review has been stratified by type of colloid. However, the pooled relative risks fail to show a mortality benefit for resuscitation with any type of colloid.
There was a trend towards a favourable effect on mortality for colloids in hypertonic crystalloid, compared to isotonic crystalloids. Nevertheless, the results are compatible with the play of chance.
Common to all meta-analyses, this systematic review may have included studies whose interventions and patient characteristics are sufficiently incomparable that the calculation of a summary effect measure may be questioned. The resuscitation regimen differed between trials. Some trials randomised participants to an initial quantity of colloid or crystalloid, and then proceeded with some form of standard resuscitation for all participants. Other trials resuscitated with the allocated fluid to pre-determined end-points, either resuscitation end-points, or in the case of trauma, until corrective surgery. In addition, the type of colloid or crystalloid, the concentration, and the protocol to determine the quantity of fluid varied. Despite these differences, all participants were in need of volume replacement, and we believe that this variation in the intervention would have an impact on the size of the effect, rather than on its direction.
As regards the effects of albumin versus crystalloid, most of the information (as indicated by the weighting in the meta-analysis) was provided by the SAFE trial (SAFE 2004). The SAFE trial used central randomisation with a minimisation algorithm to ensure balance on known potential confounders. Blinding was assured through the use of specially designed masking cartons and specially designed and manufactured administration sets. The trial authors report that the effectiveness of the blinding was confirmed in a formal study before the trial was initiated. In brief, this was a well-conducted, high-quality trial. There were 726 deaths (20.9%) in the albumin-treated group and 729 deaths (21.1%) in the saline-treated group (RR of death 0.99; 95% CI 0.91 to 1.09). Although even this large trial was unable to confirm or refute the possibility of a modest benefit or harm from albumin, it has provided some reassurance that any hazard from albumin, if indeed there is any, is unlikely to be as extreme as was suggested by the results from the previously published (now here updated) meta-analysis of much smaller trials. The pooled RR for death with albumin in this updated meta-analysis is now 1.02 (95% CI 0.93 to 1.11). It is important to note that the effect estimate from the SAFE trial is entirely consistent with the results of previous trials of albumin in hypovolaemia and there is no significant heterogeneity (I
The results of this updated meta-analysis have important policy implications. There is still no evidence that colloids are superior to crystalloids as a treatment for intravascular volume resuscitation in critically ill patients. Importantly, the SAFE trial also provided no evidence of any other clinical advantages from using albumin. It also debunked the belief, from pathophysiological inference, that very large volumes of crystalloid must be administered to reach the same resuscitation end-points as can be achieved using much smaller volumes of colloid. In the SAFE trial, the ratio of albumin administered to saline administered was approximately 1:1.4. Colloids, in particular albumin, are considerably more expensive than crystalloids, and albumin is a blood product and so carries at least a theoretical infectious disease risk. The economic opportunity cost of ongoing colloid use, particularly albumin use, is likely to be considerable and for this reason its ongoing use in this context is unjustified.
Implications for practice
There is no evidence from RCTs that resuscitation with colloids, instead of crystalloids, reduces the risk of death in patients with trauma, burns or following surgery. As colloids are not associated with an improvement in survival, and further, colloids are considerably more expensive than crystalloids, it is hard to see how their continued use outside the context of RCTs in subsets of patients of particular concern, can be justified.
Implications for research
Future trials may need to concentrate on specific subgroups of patients to identify people who may benefit from colloids rather than crystalloids.
We acknowledge the contribution of Phil Alderson, Frances Bunn, Paul Chinnock and Gillian Schierhout, who were authors of earlier versions of this review.
We would like to acknowledge the Intensive Care National Audit and Research Network in London, for assistance with identification of trials for this review.
We thank Dr. Frank M. Brunkhorst for providing the Supplementary Appendix to the paper Brunkhorst 2008.
Data and analyses
- Top of page
- Authors' conclusions
- Data and analyses
- What's new
- Contributions of authors
- Declarations of interest
- Sources of support
- Index terms
Appendix 1. Search strategy
Cochrane Injuries Group’s Specialised Register (searched 30 Sept 2008), PubMed (searched 30 September; last three months), Controlled Trials metaRegister (www.controlled-trials.com) (searched 30 Sept 2008)
colloid* or hydrocolloid* or crystalloid*
MEDLINE 1950 to Sept 2008, EMBASE 1980 to Sept 2008
1.exp Fluid Therapy/
2.exp Rehydration Solutions/
4.exp Plasma Substitutes/
8.exp Isotonic Solutions/
10.((fluid$ or volume or plasma or rehydrat$ or blood or oral) adj3 (replace$ or therap$ or substitut$ or restorat$ or resuscitat$ or rehydrat$)).ab,ti.
11.((fluid$ or volume or plasma or rehydrat$ or blood or oral) adj3 (challenge or perfusion or volume or intravenous or shock)).ti,ab.
12.(isotonic saline solution$ or Blood substitute$ or blood expander$ or plasma volume expander$ or volume expander$).mp. [mp=title, original title, abstract, name of substance word, subject heading word]
13.(colloid$ or crystalloid$ or albumin$ or albumen$ or plasma$ or starch$ or dextran$ or gelofus$ or hemaccel$ or haemaccel$ or hydrocolloid$ or serum$ or hetastarch or isotonic or ringer$ or gelatin$ or gentran$ or pentastarch$ or pentaspan$ or hartman or sodium or potassium or salin$ or hypertonic or hypotonic or hemodilution or haemodilution or ringer lactatae).ti.
16.randomized controlled trial.pt.
17.controlled clinical trial.pt.
19.clinical trials as topic.sh.
24.22 and 23
25.14 and 24
26.colloid* or hydrocolloid* or crystalloid*
28.26 or 27
29.25 and 28
ISI Web of Science: Science Citation Index Expanded (SCI-EXPANDED) 1970 to Sept 2008, ISI Web of Science: Conference Proceedings Citation Index- Science (CPCI-S) 1990 to Sept 2008
Topic=(colloid* or hydrocolloid*) AND Topic=(crystalloid*) AND Topic=(randomised OR randomized OR randomly OR random order OR random sequence OR random allocation OR randomly allocated OR at random OR randomized controlled trial* OR controlled clinical trial* OR randomized controlled trial*) NOT Topic=(animal model* OR Animal* OR Animal Experiment* OR Animal disease model* OR Laboratory Animal*)
CENTRAL (The Cochrane Library Issue 3, 2008), National Research Register (to 2006, Issue 4)
#1MeSH descriptor Albumins explode all trees
#2MeSH descriptor Plasma Substitutes explode all trees
#3MeSH descriptor Plasma explode all trees
#4MeSH descriptor Plasma Volume explode all trees
#5MeSH descriptor Fluid Therapy explode all trees
#6MeSH descriptor Colloids explode all trees
#7(#1 OR #2 OR #3 OR #4 OR #5 OR #6)
#8(crystalloid* or ringer* or hartman* or sodium* or potassium* or salin*):ti or (crystalloid* or ringer* or hartman* or sodium* or potassium* or salin*):ab
#9Isotonic saline solution* OR Blood substitute* OR blood expander* OR plasma volume expander* OR volume expander*
#10(colloid* OR crystalloid* OR albumin* OR albumen* OR plasma OR starch* OR dextran* OR gelofus* OR hemaccel* OR haemaccel* OR OR serum OR hetastarch OR isotonic OR ringer* OR gelatin* OR gentran* OR pentastarch* OR pentaspan* OR hartman OR sodium OR potassium OR saline):ti
#11(fluid* OR volume OR plasma OR rehydrat* OR blood OR oral) AND (replace* OR therapy OR substitut* OR restorat* OR resuscitat* OR rehydrat*):ab
#12MeSH descriptor Rehydration Solutions explode all trees
#13MeSH descriptor Serum explode all trees
#14MeSH descriptor Isotonic Solutions explode all trees
#15MeSH descriptor Hetastarch explode all trees
#16(#1 OR #2 OR #2 OR #4 OR #5 OR #6 OR #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15)
#17MeSH descriptor Colloids explode all trees
#18colloid* OR crystalloid* OR hydrocolloid*
#19#17 OR #18
#20#16 AND #19
Last assessed as up-to-date: 29 September 2008.
Protocol first published: Issue 4, 1997
Review first published: Issue 4, 1997
Contributions of authors
PP and IR examined trials for inclusion or exclusion, reaching agreement by discussion. PP and IR extracted data from the new studies. PP amended the text of the review.
IR and MP examined trials for inclusion or exclusion, reaching agreement by discussion. IR and MP extracted data from the new study. MP amended the text of the review. PP edited the final version.
Declarations of interest
Sources of support
- Institute of Child Health, University of London, UK.
- UK Cochrane Centre, NHS R&D Programme, UK.
- NHS R&D Programme: Mother and Child Health, UK.
- Cochrane Review Incentive Scheme, Department of Health, UK.
Medical Subject Headings (MeSH)
*Rehydration Solutions; Colloids [*therapeutic use]; Critical Illness [*therapy]; Fluid Therapy [*methods]; Plasma Substitutes [*therapeutic use]; Randomized Controlled Trials as Topic; Resuscitation [methods]
MeSH check words
* Indicates the major publication for the study