This will be a non-inferiority randomized controlled trial (RCT) comparing the new proposed short-term (7 day) urethral catheterization to longer-term (14 day) urethral catheterization in terms of predicting fistula repair breakdown, measured 3 months following fistula repair surgery. This study will be conducted among 512 women with simple fistula presenting at eight study sites for fistula repair surgery (Table 1). Figure 1 depicts the flow of study participants.
Intervention
The study intervention will entail removal of the urethral catheter at 7 days after surgery in the intervention arm and 14 days after surgery in the control arm. While 7 and 14 days are the target days for urethral catheter removal in the two groups, should it be necessary to remove the catheter either one day earlier or one day later (e.g. holiday, surgeon unavailable) in either group (i.e. 6-8 days or 13-15 days) these women will still be considered as compliant with the group assignment. Each patient will receive the same type of catheter, and with the exception of the timing of catheter removal, all other procedures related to the removal of the catheter will remain the same across both study arms. No drugs or new devices will be examined as part of this study.
Outcome measures
The primary study endpoint is fistula repair breakdown any time after day 7 after urethral catheter removal up to three months following surgery. Repair breakdown will be assessed using a dye test; a urinary catheter will be inserted into the bladder, saline colored with dye will be introduced into the bladder via the catheter, and the suture line will be checked for leaks.
Secondary outcomes will include: repair breakdown one week following indwelling urethral catheter removal, the need for intermittent catheterization to manage urinary retention, and the occurrence of septic or febrile episodes, prolonged hospitalization, catheter blockage, and self-reported residual incontinence. Prolonged hospitalization will be defined as a stay at the facility beyond one week following initial catheter removal for medical reasons.
Study population
All women who present at study sites for fistula repair surgery will be potentially eligible. Women will be randomized if:
-
they have a "simple" fistula, based on the surgeon's perception during surgery
-
they have a closed fistula at completion of surgery
-
they have a closed fistula 7 days after surgery
-
they understand study procedures and requirements
-
they agree to return to the facility for one follow-up visit three months after the date of surgery
-
they provide informed consent to participate in the study
-
they have no contraindications precluding their participation
Women will be excluded if they have a fistula that is:
-
determined to be "not simple"
-
radiation-induced, associated with cancer or due to lymphogranuloma venereum
-
not closed immediately after surgery or 7 days after surgery
Generation of allocation sequence
Randomization will be done using permuted blocks within site to ensure similarity of groups with regard to potential confounding factors that might influence repair outcomes within a particular study site. The random allocation sequence will be generated centrally at the World Health Organization (WHO) Headquarters using computer generated random numbers. Randomization will be to two groups and stratified by site. Blocking with randomly varying groups will be used to restrict randomization within the strata. Each site will recruit 64 participants.
Random allocation technique and allocation concealment
Allocation of the random generated sequence will be by consecutively numbered envelopes. Allocation concealment will be achieved by using sealed opaque envelopes. Allocation will take place 7 days after fistula repair surgery.
Rationale for the non-inferiority hypothesis and for sample size estimation
The research question of interest is whether short-term catheterization is not worse by more than a minimal relevant difference than longer-term catheterization in terms of achieving fistula closure. This question lends itself to a non-inferiority design.
The choice of a non-inferiority margin, i.e. the smallest clinical difference that is acceptable between the two treatments, is based on a combination of clinical judgment and statistical reasoning. Because there are no data from previous trials to help define the clinical difference between treatments, we have relied on our own and outside experts' clinical judgment to determine that a margin of inferiority of 10% is an irrelevant small difference. In other words, if the two-sided 95% confidence interval (95% CI) for the difference in fistula repair breakdown rates ("7-day" minus "14-day") lies fully to the left of the 10% non-inferiority margin, we will have proved non-inferiority of the "7-day" procedure at the level of significance α = 0.025; superiority (as a bonus) will be demonstrated at the level of significance α = 0.05 if the two-sided 95% CI lies fully to the left of 0.
Analyses were conducted using preliminary data from the Fistula Care/USAID prospective cohort study examining fistula repair outcomes in order to determine the probability of successful closure in women with simple fistula catheterized for longer periods of time (i.e. the equivalent of the "standard" treatment group in the study outlined here). Among the women with simple repairs in the prospective study for whom follow-up data were available (n = 145), 87% had a closed fistula at 3 months follow up. Thus, we believe that it is reasonable to expect the failure rate (e.g. proportion of fistula that are not closed) to be between 10 to 15%.
Assuming 13% failure rate in the control group, non-inferiority will be demonstrated within the margin of 10% at a one-sided significance level of 0.025 and a power of 80% (calculated when failure rates in both arms are the same), with a sample size of 177 per arm (354 women in total). Adjusting by 20% for loss to follow-up and 10% for protocol violations and withdrawals, this would result in a sample size of 507 women. Each site will randomize 64 women, for a total sample size of 512.
Type of data and collection procedure
For all women who have consented to participate, data will be collected, at hospital level, at three time points using case report forms (CRFs) developed for the study: i) prior to surgery, detailed information on socio-demographic characteristics of the women, duration of the fistula, previous attempts at repairing the fistula and pre-operative care procedures will be recorded; ii) at the time of surgery, a detailed clinical examination will be conducted to assess the characteristics of the fistula, and to allow the surgeon to determine if he/she would consider the fistula "simple" or not. For all participants, detailed information will be recorded on the anatomical and clinical characteristics of the fistula(s), including the presence of scarring, location and type of the fistula and length and width of the fistula; iii) post-surgery, at 1, 3, 7 and 90 days after surgery. Completed CRFs will be sent periodically to the data management centre at the WHO in Geneva for data entry.
Analysis plan
The primary outcome of the trial is repair breakdown at 3 months or earlier in an intent-to-treat (ITT) analysis. The primary analysis will be conducted for the ITT population. The ITT population would include patients for whom the protocol is violated (i.e. patients who are randomized to short-term catheterization but are in fact catheterized for longer durations, or patients who are randomized to the long-term catheterization group but are in fact catheterized for only 7 days) and withdrawals.
Unlike a superiority trial, where the treatment effect is the primary parameter of interest, the interpretation of a non-inferiority trial's results depends on the location of the CI for the effect of the experimental treatment relative to the margin of non-inferiority (Δ) and a null effect. A range of possible scenarios are depicted in Figure 2, where error bars indicate 2-sided 95% CIs, and the tinted area indicates the zone of inferiority. Thus, primary analyses will be interpreted as follows: if the whole 95% CI lies to the right of the non-inferiority margin of 10% (scenario H), the experimental intervention will be declared "inferior." If the whole 95% CI lies below the non-inferiority margin, the intervention will be considered to be non-inferior to the standard treatment (scenarios B-D). If the 95% CI includes the non-inferiority margin Δ (scenarios E, F and G) study results will be deemed inconclusive. Finally, in scenario A, since the 95% CI lies completely to the left of zero, the new treatment may be considered superior to the standard treatment [11].
Once non-inferiority is demonstrated, it is acceptable to then test the hypothesis that the new treatment is superior to the active control, with a significance level defined a priori and with an ITT analysis [12].
Interim data analysis
A Data Safety and Monitoring Board (DSMB) with no direct involvement in the trial will be appointed. The role of the DSMB will be to deal with any ethical issues that may arise while the trial is in progress, and to scrutinize an interim analysis. An interim analysis will be conducted to be reviewed by the DSMB after one-third of the study participants have returned for the three-month follow-up visit. At the time of the interim analyses, event and recruitment rates will be provided to the DSMB by unmasked treatment group.
Stopping the trial
The DSMB will be asked to give advice regarding stopping the trial if they have proof beyond doubt of an important advantage or disadvantage for one of the treatment groups, and they consider that the results are likely to affect clinical practice. For the outcomes of the trial, the following stopping guidelines are proposed for the DSMB:
(1) At the time of the interim analysis, the DSMB may recommend stopping the study or temporarily halting recruitment if there are significantly more repair breakdowns in one catheterization group compared to the other; a difference between the two treatment arms will be considered significant if the p value for the difference is less than 0.001.
(2) In the event that the interim analysis shows notably more adverse events in either study arm, the DSMB may recommend stopping the study or temporarily halting recruitment.
(3) The DSMB may also recommend stopping the study, temporarily halting recruitment or adjusting study sites if it seems that recruitment is not proceeding at rates that will allow the study to reach its target sample size in a reasonable timeframe.
It should be noted that if there appears to be an unexpectedly high number of repair breakdowns among short-term catheterization cases compared to long-term catheterization cases either as reported by study site staff or the study monitor, or based on the interim analysis, any one of the study Institutional Review Boards (IRBs) may also temporarily or permanently halt the study at any time.
If the study is stopped temporarily or permanently for any reason, follow-up of women already enrolled will continue as originally scheduled and all women already enrolled will receive continued care, appropriate to their clinical condition and circumstances, in line with each sites standard practice.
Duration of project
It is anticipated that the recruitment into the trial and all follow-up of participants can be completed in approximately 2 years. Recruitment will begin in December 2011 after trial procedures have been tested, study staff trained and materials have been distributed to the study sites.