Sixty studies involving 19,387 patients were included:
colorectal cancer, 28 studies with 8,374 patients;
breast cancer, 8 studies with 3,814 patients;
head and neck cancer, 7 studies with 766 patients;
lung cancer, 6 studies with 1,362 patients;
prostate cancer, 6 studies with 2,829 patients;
stomach cancer, 5 studies with 2,242 patients.
Twenty-eight (46.7%) of the 60 reports concluded that perioperative transfusion was associated with an adverse clinical outcome. The 'average' adverse transfusion effect was significant (P<0.05) at all sites except for breast cancer. The 'average' transfusion effect ranged from a 6% increase in the risk of a negative outcome in cases of breast cancer to a 262% increase in risk for patients undergoing resection of head and neck cancer. Calculated RRs and 95% confidence intervals were:
colorectal; 1.49 (95%CI: 1.23, 1.79);
breast; 1.06 (95%CI: 0.90, 1.24);
head and neck; 3.62 (95%CI: 2.15, 6.08);
lung; 1.30 (95%CI: 1.02, 1.66);
prostate; 1.51 (95%CI: 1.13, 2.01);
gastric; 2.44 (95%CI: 1.60, 3.71).
When studies of colorectal cancer were separated into two design groups (retrospective and prospective), the RR was 1.60 (95% CI:1.27, 2.02) and 1.18 (95% CI: 0.93, 1.51) respectively.