Ten RCTs were included in the review (n=1,589 total participants, range 11 to 550).
Neutrophil engraftment: There was no significant difference in the rate of neutrophil engraftment between matched and 1-Ag mismatched cord blood treatment groups (seven studies, n=493 participants). 2-Ag mismatched treatments showed a significant reduction in engraftment in comparison to matched (OR 2.28, 95% CI 1.22 to 4.23; seven studies, n=413 participants). There was no significant heterogeneity in either analysis.
Platelet engraftment: The rate of platelet engraftment was significantly lower in 1-Ag mismatched (OR 1.72, 95% CI 1.02 to 2.89; six studies, n=419 participants) or 2-Ag mismatched (OR 2.31, 95% CI 1.36 to 3.94; six studies, n=352 participants) treatments compared to the matched group. There was no significant heterogeneity in the analyses.
Graft versus host disease: For grade 0 to 1 there was no significant difference between the three treatment groups. The occurrence of grade >II disease was significantly increased in the 2-Ag mismatched group (OR 2.00, 95% CI 1.04 to 3.85; five studies, n=323 participants), but not by the 1-Ag mismatched group (five studies, n=307 participants) compared to matched groups. There was no significant heterogeneity in the analyses.
Early transplant related mortality: Mortality rates were significantly higher with 2-Ag mismatched treatments (OR 3.08, 95% CI 1.15 to 8.23; four studies, n=395 participants), but not 1-Ag mismatched treatments (four studies, n=424 participants) compared to matched groups. Heterogeneity was significant in both analyses (2-Ag mismatch, I2=54%; 1-Ag mismatch, I2= 56%).
Disease free survival: There was no significant difference in survival rates between matched and 1-Ag mismatched treatments (nine studies, n=845 participants). There was a significantly lower rate of survival in the 2-Ag mismatched treatments (OR 1.70, 95% CI 1.17 to 2.48; nine studies, n=565 participants), in comparison to the matched treatment. There was no significant heterogeneity in the analyses.
The studies were symmetrically distributed on the funnel plot, which indicated that publication bias was low.