Table 3. Association between time-varying hyperuricemia and graft outcomes.
| Models | HR (95% CI) | P value |
|---|---|---|
| Time varying Cox without imputationa | ||
| OGF | 2.07(1.70–2.52) | <0.001 |
| DCGF | 2.08 (1.61–2.69) | <0.001 |
| Composite event | 2.73 (2.32–3.21) | <0.001 |
| Time varying Cox with imputationb | ||
| OGF | 1.96 (1.59–2.43) | <0.001 |
| DCGF | 2.02 (1.55–2.63) | <0.001 |
| Composite event | 3.15 (2.6–3.81) | <0.001 |
| Marginal structural modelc | ||
| OGF | 2.27(1.33–3.78) | 0.002 |
| DCGF | 2.38 (1.09–4.90) | 0.03 |
| Composite event | 3.05 (1.64–5.49) | 0.004 |
UA level (high or normal) was considered a time-varying dichotomous variable
aAdjusted for transplant era, age, sex, body mass index, donor type, donor age, donor sex, pre-transplant diabetes mellitus, duration of pre-transplant dialysis, retransplantation, number of human leukocyte antigen mismatches, tacrolimus use, delayed graft function, biopsy-proven acute rejection within 1 year, baseline systolic/diastolic blood pressure, and time-varying eGFR
bMultiple imputation using chained equations for the missing UA level and eGFR values
cPerformed using the imputed dataset and the model created with consideration of all confounders, including time-varying eGFR
OGF, overall graft failure; DCGF, death-censored graft failure; HR, hazard ratio; CI, confidence interval; eGFR, estimated glomerular filtration rate; UA, uric acid