Twenty-one RCTs (n=2,590) were included in meta-analyses. The mean Jadad score of the trials was 2.76. Thirteen trials were judged as high quality. Median follow-up varied from 0 to 67 months.
Compared with the steroid-free group, the steroid group was associated with significant increases of post-transplant diabetes (RR 1.86, 95% CI 1.43 to 2.41; 11 RCTs), cholesterol levels (WMD 19.71, 95% CI 13.7 to 25.7; five RCTs) and severe acute rejection (RR 1.71, 95% CI 1.14 to 2.54; 12 RCTs) and a non-significant increase of cytomegalovirus infection (RR 1.47, 95% CI 0.99 to 2.17; nine RCTs).
The steroid group was associated with a significant increase of overall acute rejections when steroids were replaced with another immunosuppression agent in the steroid-free arm (RR 1.31, 95% CI 1.09 to 1.58; 10 RCTs) and a significant decrease of overall acute rejections when steroids were not replaced (RR 0.75, 95% CI 0.58 to 0.98; 13 RCTs).
There were no significant differences in one-year patient survival and one-year graft survival between the steroid and steroid-free groups.
For studies on transplanted HCV patients exclusively, the steroid group was associated with significant increases of HCV recurrence compared with the steroid-free group (RR 1.15, 95% CI 1.01 to 1.13; 14 RCTs), acute graft hepatitis (OR 3.15, 95% CI 1.18 to 8.40; three RCTs) and treatment failure (OR 1.87, 95% CI 1.33 to 2.63; five RCTs).
Significant publication bias was observed only for the outcome of acute rejection. Significant heterogeneity was observed only for the outcomes of cytomegalovirus infection (I2= 50%) and overall acute rejection when steroids were not replaced in the steroid-free arm (I2=61%). Sensitivity analyses showed that two studies contributed to the significant heterogeneity.
Meta-regression showed no significant differences except for the comparison of acute rejection between studies that used replacement of steroid and those without replacement in the steroid-free group (p=0.036).