Six trials were included in the review. Sample sizes ranged from 20 to 200 patients. Three trials reported randomisation procedures but none of the studies reported methods to conceal allocation. One trial reported blinding of outcome assessors and two trials conducted analyses using intention-to-treat and per-protocol principles. The mean Jadad score was 2.33 (range 2 to 3 points).
Circuit Survival: There were no significant differences in circuit survival time observed between the citrate-treated patients and patients treated with heparin anticoagulation therapy (five trials). Considerable heterogeneity was observed (Ι²=100%). Subgroup analyses conducted on the basis of pre- or post-dilution found circuit survival times to be similar between the two treatment groups.
Major bleeding: Statistically significant reductions in major bleeding were observed with citrate treatment compared to heparin anticoagulation (RR 0.34, 95% CI 0.17 to 0.65; six trials; Ι²=0%; NNT=6.87). Subgroup analyses showed that lower risks of bleeding were observed in the citrate group compared to the low-molecular weight heparin group in one trial (RR 0.40, 95% CI 0.16 to 0.99). Incidence of hypocalcaemia was significantly less in the heparin anticoagulation group (RR 3.51, 95% CI 1.17 to 10.60; five trials; Ι²=0%) but no adverse events related to hypocalcaemia were reported in the studies that evaluated this outcome.
Metabolic alkalosis and thrombocytopaenia: No differences were observed between the treatment groups for incidence of metabolic alkalosis (four trials; Ι²=38%) although marginally less alkalosis occurred in one trial with citrate treatment than with low molecular weight heparin (RR 0.48, 95% CI 0.23 to 1.00). No differences in heparin-induced thrombocytopenia were observed between the treatment groups (two trials; Ι²=0%).