Colloids versus crystalloids as priming solutions for cardiopulmonary bypass: a meta-analysis of prospective, randomised clinical trials.

Using Cochrane methodology a review was performed of prospective randomized clinical trials comparing colloidal pump priming solutions for cardiopulmonary bypass. Dextrans were not considered. Database searches from 1966 through December 2002 delivered 265 articles. Seventeen studies finally met the eligibility criteria involving 997 patients. Summary odds ratio estimates from the 5 studies reporting mortality were 1.46 (n = 326; 95%-Confidence-Interval: .55 to 3.85; p = .49) for crystalloids against colloids and .74 (n = 150; 95%-Confidence-Interval: .17 to 3.36; p = .49) for albumin versus synthetic colloids. Most commonly used outcome measures further included postoperative blood loss, platelet-count, fluid-balance and, colloid osmotic pressures from which Standardized Mean Differences (SMD) and their 95%-Confidence-Interval (95%CI) were extracted. Colloids produced significantly higher oncotic pressures and less positive fluid-balances. Although across 9 studies postoperative bleeding between colloids and crystalloids did not differ (n = 663; SMD: -.03, 95%CI: -.18 to .12; p = .69), platelet counts significantly favoured crystalloids (n = 465; SMD: -.42; 95%CI: -.68 to -.16; p = .00). However, compared to albumin platelet counts were significantly disfavoured only by starches (n = 321; SMD: -.55; 95%CI: -.77 to -.32; p = .00). To conclude, using mere crystalloids produced more pronounced positive fluid balances and their avoidance as a single pump-prime component can be suggested. Since albumin is not necessarily associated with better outcomes and is more expensive, it is hard to continue its use. However, there is still insufficient evidence available to allow definitive conclusions.