Study Design and Participants
In this RCT, patients were recruited from six UK centers. Patients aged 2–75 years needing a renal transplant were included. We excluded those with an increased risk of bleeding and abnormal urinary tract anatomy or function. Patients with planned early use of mTOR (mammalian target of rapamycin) inhibitors were excluded. Abnormal donor urinary tract anatomy and donor stones were also excluded. Full details of all inclusion and exclusion criteria are available in our published protocol (ISRCTN09184595). Patients gave written informed consent in accordance with good clinical practice. The trial was approved by the London Research Ethics Committee (REC reference 10/H0718/5).
An online remote randomization system was provided by the Mental Health and Neurosciences Clinical Trials Unit, King's College London. Recruits were allocated at a 1:1 ratio to either early or late stent-removal groups. Randomization was block-stratified for age (child 2–16 years; adult 17–75 years) with randomly varying block sizes. Trained clinicians and research nurses enrolled participants at each site and were made aware of the allocated group at the time of randomization.
An extravesical Lich-Gregoir vesicoureteric anastomosis was performed in all patients over a double-J stent (6F/16 cm polyurethane stent in adults and 4.8–6F/16 cm in children; Cook Medical, Bloomington, IN). A urethral catheter was inserted intraoperatively. In patients randomized to early stent removal, the ureteric stent was attached to the urethral catheter using the string of the stent. An acceptable alternative was to pass a nonabsorbable suture through both the lower end of the stent and the distal drainage eye of the urethral catheter. This was subsequently removed as a single unit on day 5 postoperatively. In the late-removal group, the stent was not attached to the urethral catheter intraoperatively. The catheter was removed on day 5 postoperatively, and the stent was removed with a cystoscope at 6 weeks.
Baseline data were collected at time of randomization at each hospital site. Local research staff obtained follow-up data by use of participant questionnaires (EuroQol five dimensions questionnaire [EQ-5D] and Functional Assessment of Incontinence Therapy–Urinary symptom questionnaire [FAIT-U]) completed at weeks 1 and 6 and case report forms completed during clinic visits at 1, 3, and 6 mo. No additional clinical tests were mandatory during follow-up as part of the trial protocol.
Urine samples were routinely collected and tested according to current clinical practice. This included at admission for transplant, prior to urethral catheter removal, at each transplant outpatient clinic, and prior to cystoscopic stent removal. All patients routinely received prophylactic amikacin (7 mg/kg, single dose) at induction and cotrimoxazole (480 mg/day in adults and a weight adjusted dose for pediatric patients) for 6 mo.
Safety outcomes were recorded during clinic visits or hospital admissions and from patient and clinician reports as they happened. Adverse events were collected and recorded by the site local principal investigator. Serious adverse events (SAEs) were reported to the trial office and the chief investigator. Safety events were monitored by the sponsor, the research ethics committee, and the data monitoring committee.
The primary outcome was a composite of complications related to transplant ureteric stent (TUS) defined as pain requiring early removal, visible hematuria requiring catheterization with or without irrigation, migration confirmed on ultrasound or x-ray, fragmentation, and UTIs within 3 mo of transplantation. UTIs were defined as either symptomatic, with midstream urine (MSU) bacterial cultures of >102 and pyuria, or asymptomatic, with MSU bacterial cultures of >105 and pyuria.
Secondary outcomes assessed and reported were MUCs of vesicoureteric anastomotic leak and stenosis up to 6 mo after transplantation; transient ureteric obstruction, reported as an intermediate urological complication; health status with the EQ-5D and FAIT-U questionnaires, completed at weeks 1 and 6; and patient acceptability using diary cards completed at the time of stent removal. Health care resource use and procedural costs (health economic components) were also assessed and will be reported elsewhere. Other outcomes assessed to better interpret overall study impact included stent symptoms (discomfort, lower urinary tract symptoms, or mild visible hematuria), patient and allograft survival (graft loss was defined as estimated GFR [eGFR] <15 mL/min per 1·73m2 return to dialysis, or retransplantation), and renal allograft function (eGFR mL/min per 1.73 m2) at 6 mo.
Sample size. Based on audit data collected prior to the study that suggested a 15–20% TUS complication rate with cystoscopic stent removal 6 weeks after transplant (late removal), the trial was powered on the hypothesis that early stent removal would result in a 10–15% reduction in the stent complication rate. Using a χ2 test with 80% power and a two-sided type I error of 0.05, the minimum required sample size was 176 patients, assuming a baseline 20% TUS-related complication rate. The corresponding number was 320 patients, assuming a 15% baseline TUS-related complication rate.
Because there was a degree of uncertainty in the baseline TUS-related complication rate, a preplanned conditional sample size reestimation based on predictive power was performed in May 2013, at which time 105 adult patients and 11 pediatric patients had primary measures available. This inferred that a sample size of 176 adults would provide sufficient statistical power (>80%) to show significant differences as planned between the groups for stent complication rates by the end of the study (see statistical analysis plan for details).
Summary of statistics used. All participants were analyzed according to treatment received (per protocol [PP]) as well as randomization allocation (intention to treat [ITT]). Outcome data on patients who did not complete 6-mo follow-up were included in the analyses.
For the analysis of the primary end point, Mantel–Haenszel tests were used to stratify across age groups (adult or pediatric) to test for differences between the two study arms. Logistic regressions were also performed for the primary outcome so that an adjusted analysis could be obtained using the following potential covariates in addition to treatment arm: age (pediatric or adult), type of donor (living or deceased), previous recurrent UTIs, dialysis status, diabetes status, and previous transplantation.
Subgroup analyses were performed for the primary outcome using Mantel-Haenszel tests (adjusting for age group and treatment arm) for the following subgroups: donor type (living vs. deceased), surgeon expertise (consultant or other/trainee), and history of previous recurrent UTIs. Subgroup analyses were also performed in which χ2 or Fisher exact tests were used to determine differences between the two study arms for the primary outcome for adults and children separately. Sensitivity analyses were performed in which we removed patients who did not receive their allocated treatment. A sensitivity analysis was also performed to look at the incidence of the individual stent-related complications that composed the composite primary outcome.
For binary secondary outcomes, Mantel-Haenszel tests were used to stratify across age groups. For continuous secondary end points, generalized linear models (both adjusted and unadjusted) were used to test for differences between the two study arms. Nonparametric methods (Wilcoxon tests) were used for nonnormal data to test for treatment effect. SAEs, such as death and graft loss, were reported as secondary outcomes. Other adverse events are listed in full in the Supplementary Material.
Analyses were done with R version 3.2.2. The statistical analysis plan was approved by the chief investigator and the chair of the data monitoring committee.
American Journal of Transplantation. 2017;17(8):2129-2138. © 2017 Blackwell Publishing