Mixed-Effects Model for 1-year Treated-Acute Rejection Rate
In model A, which assessed the effect of each center as a random
parameter, the standard deviation for the center effect for drug-treated
rejection within 1-year post OHT was 0.66 [95% confidence interval
(CI) 0.56, 0.78, p<0.001] (Table 2). This variability is
further evidenced by the finding that 42 of the CIs from the 100 centers
did not overlap the average center effect line (Supplementary Figure 1).
After adjusting for year of transplantation and recipient and donor risk
factors as fixed parameters (model B), the standard deviation for the
center effect increased to 0.68 (95% CI 0.57, 0.79, p<0.001).
The variables transplant year, recipient age, recipient gender, and HLA
mismatch were significantly associated with 1-year treated rejection. In
Model C, after introducing induction therapy at the patient-level as a
fixed parameter, no significant association was identified between
induction therapy use and 1-year rejection rate (OR 0.93, 95% CI
0.83-1.03; p=0.20). The VPC for model B and C was 12.5%, further
suggesting there is substantial center variation in rejection rates
attributable to other unmeasured centers characteristics without any
impact by the inclusion of induction therapy parameter.