Hypothesis / aims of study
Less than 30 years after their appearance, mid-urethral slings (MUS) became the most indicated therapy for the treatment of urinary incontinence. However, serious long-term side effects have escaped post-marketing studies. Healthcare databases offer the opportunity to investigate this issue in real-life setting. In France, the Système National des Données de Santé (SNDS) cover all practices, almost the entire target population, with a long follow-up.
The main objective was to compare the incidence of surgical procedures related to sling removal or sling section following the implantation of a mid-urethral sling of two different approaches of mid-urethral slings in the treatment of stress urinary incontinence in adult women: tension-free vaginal tape (TVT) and trans-obturator tape (TOT).
To answer this question, we propose the emulation of a target trial. This methodology was first described in 2016 by Hernan and Robins [1]. We considered a target clinical trial that would have been the best design to answer our question. Then, we emulated its design in real world data. Indeed, randomized trials might not be feasible are particularly not feasible when the goal is to evaluate long-term effects.
Study design, materials and methods
In this study, we emulated in real-world data a hypothetical target trial.
The target trial is a randomized controlled trial aiming to compare retropubic (TVT) and transobturator (TOT) approach in terms of safety. Participants are randomly assigned to TVT or TOT. Participants and their treating physicians are aware of the assigned treatment strategy. The eligibility criteria include adult women who were scheduled for the implantation of a mid-urethral sling for stress urinary incontinence with no history of previous sling implantation. The treatment strategies to be compared are the TVT or TOT approaches.
We used a large observational database (SNDS) to emulate this target trial. Eligibility criteria and treatment strategies were the same as for the target trial, except that implantation of the slings was already performed at the inclusion. Participants were classified at baseline accorded to the sling implanted. Randomization was emulated by Inverse Probability Weighting on a propensity score. The outcomes were also the same as for the target trial. Patients were followed from baseline (sling implantation) to the occurrence of the outcome of interest, death, administrative censoring or to the end of the observation period (at least one year of follow-up).
The inclusion criteria are adult women who underwent first mid-urethral slings between 2011 and 2018, identified with code procedures.
The primary endpoint was a composite outcome: sling removal or sling section, i.e. the occurrence of surgical procedures related to the implanted sling. These surgical procedures were identified on hospital discharge summaries, with procedure codes and reflect severe complications directly linked to the sling.
The secondary endpoints were the rate of sling removals and the rate of slings sections. We also considered other severe (i.e. leading to hospitalization), directly linked to the implantation of mid-urethral slings: obstruction, erosion, infection of the device. There were identified through ICD-10 codes. Reoperation for urinary incontinence identified by diagnoses codes were also considered. Other complications (chronic pain addressed through the dispensation of painkillers, overactive bladder addressed through the dispensation of antimuscarinic drugs, urinary tract infection addressed through the realization of CBEU and dispensation of antibiotics, urinary retention addressed through the dispensation of urethral catheter and urine collectors) were also identified.
The SNDS is the French Healthcare database. It includes the national health insurance schemes and the national hospital discharge database, linked by a unique patient identifier. The SNDS collects data about drugs deliveries, examinations, and procedures received during ambulatory cares and hospitalizations. Information such as date of birth or place of residence are also available. Since 2005, these data are prospectively and exhaustively recorded covering 99% of the entire French population (~67 million inhabitants). This database does not have access to clinical details or medical records.
Cumulative incidence rates of outcomes were reported using the reverse Kaplan Meier
Results
From January 2011 to December 2018, 215,187 women were included: 170,803 underwent TOT (79.4%) and 44,384 TVT (20.6%).
The mean age of patients was 58 years (DS=13) and respectively for TVT and TOT 59 years (DS=13) and 58 years (DS=13). The distribution was homogeneous over the inclusion period and on French territory. 48,488 (22.5%) patients had a chronic disease and 76,525 (70.8%) patients had a Charlson score between 1 and 2. In the TVT group, 5,580 patients (20.1%) had a Charlson score higher than 5, and in the TOT group, 20,310 (18.9%) had a Charlson score higher than 5. In all 3,173 (1.5%) died during follow-up.
Figure 1 represents the cumulative incidence of the primary outcome. At one year, 3,503 patients (2.05%) had undergone a reoperation in the TOT group and 1,324 in the TVT group (2.98%). At five years 5,119 patients (3.24%) had undergone a reoperation in the TOT group and 1,744 in the TVT group (4.15%).
At eight years 5,326 patients (3.72%) had undergone a reoperation in the TOT group and 1,836 in the TVT group (4.83%).
Interpretation of results
We have reported some preliminary results of our cohort.
In the TVT group, patients were lightly older and with more comorbidities than in the TOT group.
There were more reoperations in the TVT group than in the TOT group. The gap between the two groups occurred during the first year after implantation and was next stable during the follow-up.
To emulate the randomization of a target trial, we will use inverse probability of treatment weighting on a propensity score. The estimation of TOT vs TVT implantation effect on the different outcomes of interest will be assessed by use of Cox models with a cluster effect for the healthcare facility.