Review: Fluid resuscitation with colloids does not reduce mortality more than crystalloids in critically ill patients

Question In critically ill patients, is fluid resuscitation with colloids (alone or with crystalloids) better than resuscitation with crystalloids for mortality? Methods Data sources MEDLINE, Web of Science, and MetaRegister (to December 2006); EMBASE/Excerpta Medica (to wk 50, 2006); CENTRAL (Cochrane Library) and National Research Register (Issue 4, 2006); PubMed (Oct to Dec 2006); Cochrane Injuries Group's specialized register; references of relevant trials and reviews; and trial authors. Study selection and assessment Randomized controlled trials (RCTs) that compared colloids (dextran 70, hydroxyethyl starches, modified gelatins, albumin, or plasma protein fraction [PPF]) with crystalloids (isotonic or hypertonic) for volume replacement in critically ill patients, including those with trauma or burns, having surgery, or with such conditions as complications of sepsis. Crossover trials and trials in neonates or preoperative, elective surgical patients were excluded. 58 RCTs and 6 quasi-RCTs met the selection criteria. 8 RCTs reported adequate allocation concealment, 38 had no loss to follow-up, and blinding was not well reported. Outcome Mortality. Main results Meta-analyses showed that fluid resuscitation with colloids did not differ from resuscitation with crystalloids for mortality; results were consistent for different colloidsalbumin or PPF, hydroxyethyl starch, modified gelatin, and dextran (Table). Resuscitation with dextran in hypertonic crystalloid did not differ from resuscitation with isotonic crystalloid (Table). Conclusion Fluid resuscitation with colloids does not reduce mortality more than resuscitation with crystalloids in critically ill patients. Fluid resuscitation with colloids vs crystalloids for mortality in critically ill patients* Comparisons Number of trials (n) Weighted event rates RRR (95% CI) NNT Modified gelatin vs crystalloid 11 (506) 4.3% vs 5.3% 9% (72 to 51) NS Albumin in hypertonic crystalloid vs isotonic crystalloid 1 (14) 15% vs 29% 50% (333 to 94) NS Dextran in hypertonic crystalloid vs isotonic crystalloid 8 (1283) 26% vs 29% 12% (5 to 26) NS RRI (CI) NNH Albumin or PPF vs crystalloid 22 (7750) 20% vs 20% 1% (8 to 10) NS Hydroxyethyl starch vs crystalloid 16 (637) 6.9% vs 6.9% 5% (37 to 75) NS Dextran vs crystalloid 9 (834) 18% vs 14% 24% (6 to 65) NS Albumin or PPF in isotonic crystalloid vs hypertonic crystalloid 1 (38) 16% vs 0% 600% (61 to 12592) NS *NS = not significant; PPF = plasma protein fraction; other abbreviations defined in Glossary. RRR, RRI, NNT, NNH, and CI calculated from data in article using a fixed-effects model. Information provided by author. Commentary Controversy persists over the role of colloids in fluid resuscitation, one of the most common interventions in critical care. The review by Perel and colleagues synthesized a large body of literature that is fraught with limitations related primarily to dated standards of RCT methodology and reporting. For instance, Perel and colleagues judged trial quality based solely on allocation concealment, which is perhaps a suboptimal surrogate for overall quality; only 8 trials had adequate concealment, suggesting substantial potential for bias in most studies. Aside from the limitations of included trials, this review is controversial in the decision to pool data. Despite statistical homogeneity, trials differed fundamentally in their objectives (assessing physiologic responses to brief interventions vs clinical outcomes with alternative management strategies), patient populations (e.g., sepsis vs elective hip repair), resuscitation protocols, and co-interventions. Many trials administered a single bolus of colloid vs crystalloid; these studies seem misplaced in a meta-analysis investigating mortality. In addition, 15 trials in the albumin analysis were conducted before 1990 and may have outdated clinical protocols for resuscitation and clinical care. The Saline versus Albumin Fluid Evaluation (SAFE) trial (1) had the greatest influence on the analysis of albumin vs crystalloids. With excellent methodology, investigators found no differential effect by treatment for mortality in a heterogeneous population of nearly 7000 critically ill patients. In a predefined subgroup analysis, trauma patients, specifically those with brain injury, had higher mortality with albumin. Some may agree with Perel and colleagues that there is no role for colloid therapy in routine resuscitation; others may continue to use colloids for specific indications. The limitations and heterogeneity of early trials and the subgroup hypotheses supported by the SAFE trial suggest that the debate is not entirely over.