Statistical analysis

The baseline characteristics of all patients were compared using the Pearson chi-square test for categorical variables and one-way analysis of variance (ANOVA) with Scheffe posthoc test for continuous variables. The effect of BMI on the time to survival was evaluated first by treating BMI as a quantitative variable, and second by discretizing BMI. For the analysis that regards BMI as a quantitative variable, the relationship between BMI and all-cause mortality was evaluated through a Cox proportional hazard model with restricted cubic spline functions to capture potential nonlinear effects. The adjusted covariates included age, gender, diabetes mellitus as baseline comorbidity, hemoglobin, serum creatinine, albumin, and dialysis vintage. The survival analysis that regards BMI as an ordinal variable was conducted by discretizing the BMI with cut-offs at 19.9, 21.6, 23.0, and 25.1 kg/m², corresponding 20^th, 40^th, 60^th, and 80^th percentiles, respectively. Survival curves were estimated by the Kaplan-Meier method and compared by the log-rank test according to BMI category (<25.1 kg/m² and ≥25.1 kg/m²). Treating kidney transplantation as competing risk, the analysis based on the subdistribution hazard approach was used to evaluate the effect of BMI category (<25.1 kg/m² and ≥25.1 kg/m²) on survival. The Competing risk analysis based on the subdistribution hazard approach was used to calculate the subdistributional hazard ratio (SHR) with 95% confidence interval (CI) for mortality with kidney transplantation as competing event, with BMI range (21.6–23.0 kg/m², third quintile) serving as the reference. Next, we calculated the 1-year changes in BMI and had stratified all patients to 5 groups by BMI change (≥+3 kg/m², +1 to +3 kg/m², -1 to +1 kg/m², -3 to -1 kg/m², and <-3 kg/m²). The association between BMI change and mortality with kidney transplantation as competing event was evaluated using the Competing risk analysis with subdistribution hazard approach. Finally, we fit the Competing risk analysis with subdistribution hazard approach to estimate the SHR of mortality with kidney transplantation as competing event for serum creatinine levels at baseline that divided into 5 categories. A p value <0.05 was considered statistically significant. SAS for Windows, version 9.3 (SAS Institute Inc., Cary, NC) and R (R Foundation for Statistical Computing, Vienna, Austria; www.r-project.org) were used for statistical analysis.