We conducted a historical cohort study for patients with a history of kidney transplantation in Shahid-Hasheminejad Hospital from 2001 to 2014. We measured the outcome variables at baseline, 1 month, 6 months, and 12 months after transplantation.
We used patient charts throughout the study period and conducted phone calls to collect the missing data. We excluded individuals from the study in cases in which contact via telephone was not possible. A single surgery unit was in charge of all transplantation throughout the study period.
Exposure:
The exposure of interest in this study was the source of transplant, which was either live or cadaver donors. There were multiple harvest units in Tehran to collect cadaver kidney. To mitigate the likelihood of cold ischemia, we transplanted all the kidneys immediately after we received them, as they were prepared within 3 hours of harvesting.
For live donor recipients, we assessed underlying disease, immunological evaluations, and kidney artery assessment with CT-angiography (computed tomography angiography). In a few cases, we performed angiography for artery assessment.
We excluded donors with multiple arteries and individuals with a body mass index (BMI) greater than 30. For left nephrectomy, we used the transperitoneal method (open transperitoneal donor nephrectomy) with a midline incision, and for right nephrectomy, we used a right flank incision.
The immunosuppression protocol in both study groups was comprised of mycophenolate mofetil, cyclosporine.
Outcome:
The glomerular filtration rate (GFR) was the primary outcome. GFR is a powerful indicator for estimating long-term patient survival. To calculate the GFR, we used creatinine clearance based on the Cockcroft-Gault formula[10]. We excluded patients without GFR information in months 6 and 12 of the study.
Confounders:
We adjusted the analysis for potential confounding variables, which we extracted from the charts or obtained via telephone interviews.
We used age, BMI at the time of transplantation, and gender as potential demographic confounder variables. A history of diabetes mellitus was also included as a factor that might strongly affect survival after transplantation. Lastly, we included variables that addressed the severity of kidney disease, including a history of nephrectomy, re-operation, previous transplantation, and a history of delayed graft function (DGF). The baseline (preoperative) GFR was also used as a covariate to adjust the model.
Statistical analysis:
SAS Enterprise Guide (Version 6.1, Cary, NC, USA) was used for all analyses. Generalised linear model (GLM) with generalised estimating equations were used to account for the clustered nature of the data (four observation units for the same patient). All aforementioned exposures of interest and confounders were defined as independent variables in the model. GFR was defined as a dependent variable, and a normal distribution was assumed.
We used the multiple imputation (MI) method to handle missing variables. The units of time in this study were ordinal variables, which represented the time of measurement at month 1, 6, and 12 after the date of transplantation.
For the secondary objective, we calculated the rate of rejection using a GLM model with a binomial distribution and logit function. This model generated the odds ratio (OR) of rejection in the cadaver donor group compared to that in the living donor group (cadaver assigned as the reference group). To evaluate the effect of time on the rejection rate among the groups, we performed an interaction of time and donor type, and the coefficient for this interaction indicated whether the likelihood of rejection increased or decreased over the study period.