Single-incision mini-slings versus standard synthetic mid-urethral slings for surgical treatment of stress urinary incontinence in women: The SIMS RCT

Burch colposuspension

Up to the mid-1990s, Burch colposuspension (BC) was the most commonly performed continence procedure worldwide. BC is performed via a transverse lower-abdominal incision and had a reasonable success rate of ≈ 80% at 5 years’ follow-up. 44 BC is associated with an up to 30% risk of development of posterior wall prolapse, a 25% risk of postoperative voiding difficulties and an 18% risk of de novo OAB symptoms (urgency and UUI). 45

The first laparoscopic colposuspension (LC) was described in 1991. Several studies have reported patient-reported and objective success rates of 70–98%. 46–48 Kitchener showed that LC has a similar effectiveness to the open colposuspension, with shorter operating time. 49 LC has the advantage of less postoperative pain, a shorter hospital stay and shorter recovery time. 50 Interestingly, laparoscopic skills were not widely available in urology or gynaecology at the time LC was introduced, hence LC was offered only in certain centres. At the same time (1996), synthetic MUSs, namely retropubic tension-free vaginal tapes (RP-TVTs), were introduced.

Traditional slings

Traditional slings were first described by Aldridge51 in 1942; they require a combined abdominal and vaginal approach. Several studies comparing traditional MUSs with BC showed that the patient-reported cure rate was lower with the traditional slings at 1 year of follow-up [risk ratio (RR) 0.75, 95% confidence interval [CI], 0.62 to 0.90). BC was associated with fewer perioperative complications, shorter duration of use of indwelling catheter and less long-term voiding dysfunction (VD). 52–54

Synthetic mid-urethral slings, mesh, tapes

In 1996, Ulmsten et al. 55 presented the first MUS: the RP-TVT (Figure 2), which was revolutionary to continence surgery worldwide. MUS was developed with the aim of transforming continence surgery into a day case procedure. MUS primarily depended on the integral theory for continence which was first described by Petros and Ulmsten and later upheld by DeLancey’s hammock hypothesis. 56,57 In both, the pubourethral ligaments and the vaginal hammock structure constitute the main continence mechanism, disrupted in women with SUI, and require re-enforcement during surgical treatment.

FIGURE 2.

The RP-TVT needle and the position of its hammock.

The RP-TVT is considered the first generation of standard-length MUSs. It utilises a type-1 polypropylene mesh strip to create a suburethral hammock at the mid-urethral level. One main advantage of this procedure is that it is placed in a tension-free fashion (i.e. supporting the urethra). The RP-TVT procedure was easier to learn than LC, and soon evidence accumulated to show that it had a similar success rate and comparable pattern of postoperative complications to BC, and showing that subsequent prolapse development was much higher with BC. 58 RP-TVT was introduced as a day procedure under local anaesthetic (LA). However, the British Society of Urogynaecology (BSUG) surgical database in 2010 showed that 97% of MUS procedures in the UK were performed under general anaesthetic (GA). 59 The main concern with RP-TVT is bladder injury, with a reported rate of ≈ 6.3%, and is mainly attributed to the blind retropubic trajectory of the insertion needles/trocars. 58

The second generation of standard-length MUSs, the transobturator tension-free vaginal tape (TO-TVT), (Figure 3) was developed as an outside–in TO-TVT by Delorme60 in 2001, followed by the introduction of the inside–out tension-free vaginal tape – obturator (TVT-O) TO-TVT by de Leval and Waltregny in 2003. 61 The main aim was to keep the concept and benefits of the RP-TVT (i.e. tension-free strip of polypropylene mesh supporting the mid-urethra), but to avoid the blind retropubic trajectory to reduce the risk of bladder injury and the more serious, but rare, bowel and vascular injury. In TO-TVT procedures, the insertion trocars pass in a more horizontal fashion (than they do with RP-TVT procedures) through the bilateral obturator complexes, with the skin incisions in the upper medial thighs. These theoretical benefits of TO-TVT materialised, with very low bladder injury rates and lower voiding difficulty rates than with RP-TVT. 62 However, more patients experienced groin/thigh pain with TO-TVTs, especially with the inside–out technique (i.e. TVT-O). 63 This was attributed to the passage of the mesh sling through the adductor muscles and the obturator complexes in the upper thigh (see Figure 3).

FIGURE 3.

The TO-TVT.

At the time of the single-incision mini-sling (SIMS) trial design, a number of systematic reviews and the Cochrane review reported no evidence of significant differences, at 12 months’ follow-up, in the patient-reported and objective cure rates between TO-TVT (RR 1.01, 95% CI 0.96 to 1.05) and RP-TVT (RR 0.96, 95% CI 0.93 to 0.99). 63–69 Similarly, there were no significant differences in hospital stay (median 0.01 days, 95% CI –0.09 to 0.11 days) or recovery time (median 0.00 weeks, 95% CI –0.14 to 0.13 weeks). The TO-TVT procedure was significantly shorter in operative time (17 minutes, compared with 27 minutes for RP-TVT). Groin pain was more common (12%) with TO-TVT than with RP-TVT, but postoperative VD was significantly less with TO-TVT than with RP-TVT (4% vs. 7%, respectively, RR 0.63, 95% CI 0.44 to 0.89). No bladder injury occurred with TO-TVT; with RP-TVT, the rate of bladder injury was 7%. 70

Mid-urethral slings (both RP-TVT and TO-TVT) rapidly became the most common continence procedures worldwide. 30 Between April 2008 and March 2017, 100,516 MUS procedures were performed in England, compared with 1195 for all other procedures. 32 The Scottish independent review on transvaginal mesh implants71 analysed routinely collected data in Scotland and showed immediate postoperative adverse events (AEs) of 3.7% for RP-TVT, 2.5% for TO-TVT and 7.8% for colposuspension, with similar rates for repeat surgery or for later complications, when comparing standard mid-urethral sling (SMUS) procedures with open colposuspension. 71 The report recommended RP-TVT as the preferred mesh-based procedure for surgical treatment of SUI among women.

Single-incision mini-slings

Single-incision mini-slings were introduced in 2006 with the aim of keeping the advantages of SMUSs, but avoiding both the retropubic trajectory and the perforation of the adductor muscles to reduce the risks of bladder injury and upper thigh pain. 72 It also involves less surgical dissection and shorter mesh length (8–14 cm, compared with 17 ± 2.87 cm for TO-TVT and 20.4 ± 0.8 cm for RP-TVT). A number of small studies showed that the SIMS procedure was more likely to be performed under LA, and had lower incidence of immediate postoperative pain, shorter hospital stay and quicker recovery. 73–76

In 2008, the National Institute for Health and Care Excellence (NICE) produced interventional procedures guidance (IPG) on the SIMS procedure. It found no RCTs evaluating its clinical effectiveness and cost-effectiveness, compared with those of other continence procedures. The NICE IPG recommended that SIMS procedures be confined to research and/or performed under special governance conditions (NICE IPG 262). 77 A Cochrane systematic review in 2011 reported lower patient-reported and objective success rates for SIMSs than for SMUSs, with 6–12 months’ follow-up, and higher incidences of repeat continence surgery and de novo UUI. However, SIMSs were associated with less operative time, less immediate postoperative groin pain and shorter recovery. 78

The introduction of SIMSs was associated with great enthusiasm as a truly ambulatory procedure. A number of SIMS devices (Figure 4) were introduced into clinical practice rather quickly, without any robust assessment of their effectiveness or safety. At the time of design of the SIMS trial, a number of SIMS procedures were used in clinical practice, such as Minitape^® (Mpathy Medical Devices Ltd, Glasgow, UK), MiniArc^® (American Medical Systems, Inc., Minnetonka, MN, USA), Ophira Mini Sling System (Promedon, Córdoba, Argentina), Zippere™ (ProSurg, Inc., San Jose, CA, USA), Contasure-Needleless^® (NeoMedic Ltd, Watford, UK), Solyx™(Boston Scientific, Marlborough, MA, USA), Ajust™ (C.R. Bard, Inc., New Providence, NJ, USA), Altis^® (Coloplast A/S, Humlebæk, Denmark) and Tissue Fixation System^® (TFS) (Adelaide, SA, Australia) (see Figure 4). TVT Secur™ (Ethicon, Inc., Bridgewater, NJ, and Cincinnati, OH, USA) was withdrawn from clinical practice in 2013 by its manufacturer for ‘commercial reasons’.

FIGURE 4.

Types of SIMSs: TVT Secur (Johnson & Johnson, New Brunswick, NJ, USA), Minitape, Mini sling, Zippere and MiniArc, Solyx, Ajust, Epilog, Ophira, Needleless and Altis.

Single-incision mini-slings fundamentally differ from SMUSs because they have a shorter trajectory of insertion, and therefore need a robust anchoring mechanism to the obturator complex with a strong post-insertion pull-out force. 79,80 All clinically used SIMSs share the same mesh material (type-1 polypropylene) and the same insertion technique, through a single vaginal incision; however, they differ in the type/robustness of the anchorage mechanism used. 81 A number of more recently developed SIMSs, such as Ajust and Altis, have an added advantage in that they allow post-anchorage adjustment of the sling tension and have been shown in independent animal studies, assessing their immediate and delayed extraction forces, to be associated with the strongest and most robust anchoring mechanism to the obturator complex. 79,80

At the time of SIMS trial design, several observational studies have shown promising results for SIMSs. The objective and patient-reported success rates were 82–91% and 80–85%, respectively, at 12 months’ follow-up. SIMSs were associated with rapid recovery, low levels of postoperative pain and a short hospital stay. 73,74,82,83 There were, however, reports of potentially higher rates of postoperative voiding difficulty, vaginal exposure, de novo urgency and reoperation rate. 83–85

To our knowledge, our group was the first in the UK to evaluate the adjustable anchored SIMS (Ajust) in a series of interlinked projects. A multicentre prospective cohort study of the adjustable anchored SIMS Ajust among 100 women has shown its acceptability (75%) and feasibility (97%) under LA. 74 A multicentre prospective RCT comparing the SIMS Ajust with the SMUS TO-TVT, with a minimum of 12 months of follow-up, showed no significant differences in the patient-reported success rates [odds ratio (OR) 0.895, 95% CI 0.344 to 2.330; p = 1.000], the objective success rates (OR 0.929, 95% CI 0.382 to 2.258; p = 1.00) or the reoperation rates (OR 0.591, 95% CI 0.136 to 2.576; p = 0.721) between the two groups. 86 Comparable numbers of women in both groups reported significant improvement in QoL (p = 0.190) and sexual function (p = 0.699). Similar results were reached by a Dutch group in a similar small RCT. 87 In addition, a number of observational studies assessing adjustable anchored SIMSs, across multiple countries (UK, France, Italy, USA and Israel) and with varying cohort sizes and lengths of follow-up (6–12 months), have shown similar patient-reported and objective success rates of 85–91%. 73,88,89

Evidence on the longer-term outcomes of adjustable anchored SIMSs emerged. In July 2012, one RCT reported its 5-year follow-up comparing an adjustable anchored SIMS (TFS) with a SMUS. 90 The objective and patient-reported success rates were 83% and 89%, respectively, in the SIMS (TFS) group, compared with 75% and 78%, respectively, in the SMUS group (p = 0.16). Naumann et al. 91 reported their prospective observational study of 51 women who underwent a SIMS procedure (Ajust) with 20–29 months’ follow-up; the patient-reported success rate was 86%.

We conducted the first health economic analysis of the adjustable anchored SIMS Ajust, compared with the SMUS TO-TVT. 92 Results have shown an incremental total cost saving to the health service of £142 per procedure with the adjustable anchored SIMS, not counting the further potential economic gain of earlier return to work among these women. There were no significant differences in quality-adjusted life-years (QALYs) generated, compared with the SMUS.

An updated systematic review and meta-analysis comparing the effectiveness and complications of SIMSs with those of SMUSs for the surgical management of female SUI included a total of 26 RCTs (n = 3308 women). 81 The results showed that, after excluding RCTs evaluating TVT Secur, which was clinically irrelevant having been excluded from clinical practice, there was no evidence of significant differences between SIMSs and SMUSs in patient-reported success rates (RR 0.94, 95% CI 0.88 to 1.00) and objective success rates (RR 0.98, 95% CI 0.94 to 1.01) at a mean follow-up of 18.6 months. These results were sustained on comparing the SIMS with TO-TVT and RP-TVT separately.

In the same review, meta-analyses showed that SIMSs lead to significantly earlier return to normal activities and to work [weighted mean differences (WMDs) of –5.08 (95% CI –9.59 to –0.56) and –7.20 (95% CI –12.43 to –1.98), respectively], and to lower immediate postoperative pain scores (WMD –2.94, 95% CI –4.16 to –1.73).

Single-incision mini-slings had numerically higher rates of repeat continence surgery (RR 2.00, 95% CI 0.93 to 4.31), but this difference was not statistically significant. We urged caution in interpretation of results because of the heterogeneity of the small trials, including lack of blinding of the assessors, which can be a source of bias; the level of incomplete data, leading to attrition bias; and the relatively short term of follow-up.

The Cochrane review in 2014 included data from 3290 patients from 31 studies and showed that SIMSs were less effective than other tapes, but once data from the withdrawn TVT Secur were excluded, the difference was no longer statistically significant. 93 The authors concluded that there was not enough evidence to show difference between the SIMS and the SMUS, and recommended an adequately powered RCT with long-term follow-up to evaluate the clinical effectiveness and cost-effectiveness of SIMSs, compared with those of SMUSs.

The SIMS trial compares the patient-reported success rate and cost-effectiveness of SIMSs with those of SMUSs.

Adjustable anchored single-incision mini-slings

The choice of adjustable anchored SIMS (Ajust or Altis) was dependant on the device used as standard in the collaborating centre and/or surgeon preference and experience.

A standard combination of fast- and delayed-action LA (dose was dependant on participant’s body weight) was infiltrated vaginally into either side of the mid-urethra, the vaginal angles (sulci) and behind the inferior pubic ramus into the obturator complex (e.g. using a curved black spinal needle and/or pudendal block needle). When possible, women were accompanied throughout the procedure by a health-care professional for support. The women’s bladders were emptied with a catheter. An adjustable anchored SIMS (meeting the prespecified criteria described previously) was used. The standard insertion steps for the adjustable anchored SIMS (Ajust and Altis) were as follows: women were positioned in lithotomy position with hips flexed at 90–100 °; LA infiltration was conducted as above; a suburethral vertical vaginal incision (≈ 1.5 cm) was made; and bilateral paraurethral tunnels were created reaching to the posterior margin of the inferior pubic ramus, but without piercing the obturator membrane. Further infiltration of LA into the obturator complex was carried out; the SIMS, with the ‘fixed anchor’ end mounted on the applicator, was introduced through the pre-dissected paraurethral tunnel until reaching behind the inferior pubic ramus. The applicator then pivoted slowly behind the ramus, allowing the fixed anchor to maintain its position in the obturator complex (membrane and obturator internus muscle) at points equivalent to 10 and 2 o’clock in relation to the urethral orifice. The insertion steps were repeated on the other side, allowing the ‘adjustable anchor’ to be fixed in the contralateral side. With the SIMS now robustly anchored, the tension was then adjusted as required to achieve continence while avoiding voiding difficulty. The cough stress test (CST) was conducted when possible. For Ajust, the adjustable anchor was then locked (this was not required with Altis). A cystoscopy was then performed to exclude lower urinary tract (LUT) injury and the vaginal incision was closed.

Standard tension-free mid-urethral slings

The choice of RP-TVT or TO-TVT was dependant on the standard procedure and device used in the collaborating centre and/or surgeon preference and experience.

Clinical centres

The trial was conducted in 21 secondary and tertiary care acute hospital settings across the UK. NHS Grampian was the clinical co-ordinating centre, housing the chief investigator.

Each collaborating centre had at least one participating surgeon who was competent in performing SIMS procedures under LA prior to enrolling in the RCT. This experience was demonstrated by the surgeons having performed an average of 12 adjustable anchored SIMS procedures (with six or more procedures performed under LA) in the preceding year. Clinical experts in the trial team watched the surgeons performing two SIMS procedures under LA in their local hospitals and deemed them eligible for the trial. All collaborating centres also had at least one participating surgeon who was experienced in at least one type of SMUS (RP-TVT or TO-TVT) and had performed an adequate workload (an average of 20 procedures) in the preceding 2 years. In 20 out of 21 centres, the same surgeon was experienced in both procedures (SIMS and SMUS procedures). In five centres, at least one additional surgeon participated in the trial and was experienced in either procedure.

Population

The population comprised women aged ≥ 18 years with SUI, who had been referred to one of the collaborating centres from across the UK, and for whom MUS surgery had been indicated. Women had completed their families and failed or declined conservative treatment: PFMT. All women had either urodynamic stress incontinence or urodynamic mixed incontinence with predominant SUI bothering symptoms. Women with pure symptoms and signs of SUI, and no symptoms of OAB or voiding difficulties, were included without urodynamic investigations, as per NICE clinical guidelines 171,19 at the time. Patients were discussed in the local multidisciplinary team meetings as per standard local practice.

Preoperative urodynamic investigations included free uroflowmetry, post-voiding residual urine volume assessment and subtracted filling cystometry. Other tests, such as urethral pressure profile and leak point pressures, were not mandatory.

We excluded women if they had one or more of the following:

• anterior or apical prolapse ≥ stage 2 on the Pelvic Organ Prolapse Quantification system

• previous incontinence surgery (for SUI or OAB)

• MUI with predominant OAB symptoms (defined as OAB failed to be controlled on conservative treatment, such as bladder retraining, PFMT and/or antimuscarinic treatment)

• neurological conditions (e.g. multiple sclerosis, spinal cord injuries)

• concomitant surgery at time of SUI surgery

• previous pelvic irradiation

• pregnancy or planning for a family

• inability to understand the information leaflet and consent form in English.

There were 14 minor breaches to the exclusion criteria of concomitant surgery at the time of SUI surgery. These are detailed in Appendix 5, Table 35.

Study Management Group

The Study Management Group was responsible for the day-to-day management of the trial. This group was chaired by the chief investigator and consisted of the trial manager, senior trial manager, data co-ordinator, health economist and statistician.

Project Management Group

The PMG was responsible for overseeing the management of the trial. This group consisted of the Study Management Group plus grant applicants, a patient and public involvement (PPI) representative and a senior programmer. Membership of the PMG is listed in the Acknowledgements.

Trial Steering Committee

The TSC was responsible for monitoring and supervising the progress of the SIMS trial. The committee met seven times between April 2014 and September 2020, at intervals agreed by the TSC. The TSC consisted of independent experts, a PPI representative, the chief investigator and key members of the PMG. Membership of the TSC is given in the Acknowledgements.

Data Monitoring Committee

The DMC was independent of the trial and responsible for monitoring safety and data integrity. The committee met nine times between April 2014 and September 2020, at intervals agreed by the committee. The trial statistician provided the data and analyses requested by the DMC prior to each meeting. The committee consisted of three independent experts. Membership of the DMC is given in the Acknowledgements.

Patient and public involvement

One of the trial PPI representatives was a grant holder and an active member of the PMG. As part of this role, she contributed extensively to the development and review of trial materials, including the protocol, PIL, questionnaires and participant newsletters, as well as to the trial processes, the 6-monthly funder reports and the final report. There was also an active PPI member of the TSC.

Subgroup analyses of the primary outcome

Subgroup analyses were performed to compare the following:

• urodynamic diagnoses – SUI versus MUI

• age – < median (i.e. 48 years) versus ≥ median

• age – < 65 years versus ≥ 65 years (post hoc)

• PFMT versus no PFMT (post hoc)

• device availability – still available versus withdrawn from the market (post hoc).

The forest plot for the subgroups is shown in Figure 9, which shows the effect sizes in all of the subgroups. Table 8 provides a summary of the subgroups and includes the size of the interaction terms. The interaction is not statistically significant for any of the subgroups.

FIGURE 9.

Differences between treatments, by subgroup.

Most participants had urodynamic stress incontinence [SIMS group, 79% (n = 235); SMUS group, 78% (n = 231)], rather than urodynamic mixed incontinence [SIMS group, 12% (n = 36); SMUS group, 11% (n = 33)]. In both subgroups, the success rates were higher among those receiving a SIMS, but the difference is smaller in the urodynamic mixed incontinence subgroup.

The median age of participants was 48 years. Of those aged > 48 years, there was a higher success rate among those who received SIMSs, whereas, of those aged < 48 years, the success rate among those who received a SMUS was slightly higher. We presented another subgroup analysis according to age, which is clinically relevant: among those aged < 65 years, it appears that SIMSs have a higher success rate than SMUSs, whereas, among those aged ≥ 65 years, it appears that there is very little difference between the devices.

Only a small number of participants had not received PFMT. Of the women who did receive PFMT, there is a higher success rate in the SIMS group. Table 8 shows a large amount of uncertainty around the size of this interaction, so the difference in effects is due to the small sample size.

The post hoc subgroup analysis compared devices still available for clinical practice with those that are no longer available (see Table 8). In both subgroups, non-inferiority of the SIMS was confirmed.

Quality of life and sexual function

The EQ-5D-3L scores increased from baseline to peak at 3 months; at 36 months, the EQ-5D-3L scores in both groups were lower than at baseline. Across all the ICIQ-LUTSqol outcomes, the pattern was similar: small differences favouring the SIMS, but with considerable uncertainty and no clear signal that one treatment was better than the other.

The PISQ-IR sexual function scores showed a small improvement from baseline to 15 months in both groups, although this improvement then diminished at 24 and 36 months. The effect size favours the SMUS group, although the difference is small with a CI that excludes a significant difference. The patient-reported secondary outcomes are shown in Table 10.

Other urinary questionnaires’ scores

For all ICIQ-FLUTS domains, the between-group differences were small, and CIs were incompatible with a worthwhile difference favouring either treatment. For both the filling and incontinence domains, there were sizeable improvements at 36 months, compared with baseline, in both groups. However, the voiding domain did not show this trend, and there was a slight deterioration from baseline in the SMUS group.

Urgency perception as assessed by the UPS at 15, 24 and 36 months is shown in Table 11. At all time points, participants in the SIMS group reported less urgency. For urgency perception at 15 months, the OR was 1.3 (95% CI 0.8 to 2.0; p = 0.26); at 36 months, the effect size was (OR) 1.1 (95% CI 0.7 to 1.6; p = 0.81). These effect sizes favour the SIMS group, suggesting less urgency, but the CI excludes a significant effect.

Of the women with preoperative urgency at baseline, the SIMS group participants appear more likely to report postoperative cure/improvement in urgency at almost all follow-up time points. However, the effect sizes rule out a significant difference between groups (see Table 11).

New-onset urgency was observed for only 89 participants. The SIMS participants appear more likely to develop new-onset urgency; however, the effect sizes rule out a significant difference between groups (see Table 11).

Operative outcomes

Table 12 shows operative outcomes. The procedure time for those receiving a SIMS was slightly shorter than for those receiving a SMUS. The postoperative stay was shorter for the SIMS group and the analysis of pain scores over the 14 days post operation also shows significantly lower pain scores in the SIMS group. Although the difference is not statistically significant, the blood loss for those receiving a SIMS does appear to be lower.

Table 12 also shows that women in the SIMS group were slightly more likely to have satisfactory voiding and to have returned to normal activities within 28 days, but there is a level of uncertainty around the estimates of the effect sizes.

Adverse events

Adverse events during surgery and self-reported at follow-up are shown in Table 17. The numbers of RP-TVT and TO-TVT procedures are similar, but the table shows a higher frequency of bladder injuries for those receiving a RP-TVT. Instances of blood loss of > 200 ml and vaginal buttonholes are comparable across all four groups. Needing to use more than one kit is more common for SIMS devices.

A higher rate of CISC at earlier time points is seen in the RP-TVT group. By 36 months, CISC frequency is low in all groups (≤ 2%).

Groin and thigh pain have the highest frequency at both 15 and 24 months among the Altis participants; however, by 36 months, the rates are comparable in all groups [RP-TVT, 16.7% (n = 21); Altis, 14.6% (n = 29); TO-TVT, 14.3% (n = 21); and Ajust, 11.3% (n = 7)]. Analgesia use is observed to be highest in the Altis and RP-TVT groups, whereas the TO-TVT was least associated with use of analgesia [Altis, 8.5% (n = 17); RP-TVT, 6.3% (n = 8); Ajust, 4.8% (n = 3); and TO-TVT, 3.4% (n = 5)].

The rates of dyspareunia and coital incontinence were low in the TO-TVT and RP-TVT groups at 15 and 36 months’ follow-up. In the Altis group, there was almost a 50% reduction in dyspareunia rates at 36 months, compared with 15 months: 11.1% (n = 11) versus 20.2% (n = 20), respectively.

Further hospital consultations and procedures are reported in Table 18.

No RP-TVT participants required additional treatment for either pain or mesh exposure; the other three groups had comparable frequencies for this treatment.

Selection of attributes and levels for the discrete choice experiment

Discrete choice experiment attributes and levels were selected following a review of the literature to identify processes and outcomes of care that were of greatest value to service users and that were also collected in the SIMS trial. The attribute and level selection process involved engagement with clinical and patient representatives on the trial team to ensure that the attributes and levels selected for the DCE were realistic and captured the most important patient outcomes. This iterative approach was used to draft and revise the DCE questionnaire to ensure that the choices presented included attributes and levels that were realistic, tradeable, meaningful to patients and clinically relevant. The DCE included a cost attribute framed as a one-off out-of-pocket expense and the cost attribute to enable calculation of WTP. Appendix 7, Table 42, describes the attributes and levels included in the DCE.

Experimental design

Based on a DCE with a total of six attributes, (1 × 2 levels, 4 × 4 levels, 1 × 5 levels), there are a total of (2¹ × 4⁴ × 5¹ =) 2560 different potential combinations of attributes and levels in an unrestricted full factorial design. This would lead to > 3 million unique choice sets. Therefore, we created a main-effects D-optimal experimental design using Ngene experimental design software (ChoiceMetrics, Sydney, NSW, Australia), reducing the number of potential choice tasks to 40. 108 The optimal experimental design of 40 choice tasks was further split into four blocks of 10 choice tasks to minimise respondent burden. Two further choice tasks were added to each block. Choice one was a warm-up exercise to familiarise respondents with the structure of the questions and a further repeated choice task was added as a consistency check. Each respondent therefore completed one block of 12 choice tasks.

The design was informed using priors for complications and level of symptom improvement obtained from a pilot version of the survey sent to 136 trial respondents. Ninety participants responded to the pilot: a response rate of 66%. Two further restrictions were placed on the combinations of attributes and levels that were presented to respondents within any one choice task alternative after inspection of the pilot data. The restrictions prevented respondents from being presented with ‘symptoms: very much improved’ in combination with ‘complications: intermittent self-catheterisation’ or ‘avoid activities: frequently’. These restrictions were imposed following further clinical and patient expert feedback to improve the plausibility of the choice tasks. The full experimental design code for the final survey can be found in Appendix 7.

Each choice task asked respondents to choose between two different surgical procedures, labelled ‘treatment A’ and ‘treatment B’, and an opt-out alternative, labelled ‘no treatment’, which varied in terms of the attributes and levels described in Appendix 7, Table 42. The opt-out ‘no treatment’ alternative was fixed across all choice tasks and included no surgery, but with no improvement in symptoms and no surgical-related complications, and it was assumed that women would occasionally avoid activities because of a fear of leaking. As no treatment was offered, the cost attribute for the opt-out alternative was £0. An example choice task is provided in Figure 10.

FIGURE 10.

Example choice task.

Questionnaire design

There were three sections in the questionnaire. Section 1 provided guidance for completing the survey, including a detailed description of all the attributes and levels. Respondents were first asked to tell us if they had experienced any of the complications included in the DCE and how long it took them to recover from the surgical procedure. They were then asked to imagine a baseline reference scenario to consider when completing the choice tasks, described as follows:

You leak a moderate amount of urine several times a day. You leak when you cough, sneeze or are physically active. Your urinary problem causes you to occasionally avoid activities due to fear of leaking. You always use pads to keep dry from your stress urinary incontinence.

Respondents were then asked to rate how acceptable improvements from this scenario would be to them (improved, much improved, very much improved). They were also asked to consider how often they would avoid activities because of a fear of urinary leakage in this scenario. The questions in section 1 were designed to encourage respondents to think about the attributes they would make choices about in the DCE stage of the survey.

Section 2 described the process of making choices and provided respondents with an example choice task. Respondents then proceeded to complete 12 choice tasks, indicating which of the three alternatives they would choose (treatment A, treatment B or no treatment) in each choice task. Immediately following the choice tasks, respondents were asked to agree or disagree (on a five-point scale ranging from ‘strongly disagree’ to ‘strongly agree’) with six statements describing their experience of the DCE survey in terms of understanding, level of information provided, relevance, plausibility, complexity and clarity.

The survey concluded in section 3 with demographic questions about education and income. These questions were included because evidence shows that ability to pay can have an impact on WTP estimates in DCEs. The income question was therefore used to adjust preferences and WTP for ability to pay. Respondents were also provided with an opportunity to give any further additional comments about the questionnaire.

The survey was sent as a postal questionnaire to SIMS trial participants who completed the follow-up questionnaire at 3 years. One reminder was sent to participants to encourage survey completion.

Discrete choice experiment data analysis

Responses to the DCE data were collated, merged with the experimental design and analysed using best-practice methods, according to random utility theory. 109 Under the random utility theory framework, each survey respondent (n) chooses their preferred treatment package (j) in each of the 10 different choice tasks (t). Data were analysed using conditional and mixed logistical regression models, allowing for multiple choices per respondent, to estimate the relative importance of the included attributes and levels. 110 The observable component of the utility function (V_njt) is a linear additive function of the attributes and levels, such that:

The alternative specific constant (ASC) is represented by α, included as a normally distributed random parameter. The ASC describes the latent utility associated with choosing any surgical treatment package, compared with none (i.e. opting in). Type of anaesthetic (reference category: local), complication (reference category: none), symptoms (reference category: no improvement) and avoidance of activities (reference category: never) are included as categorical variables in the model, whereby the set of β parameters represents the marginal utility of each dummy-coded attribute level, compared with the reference category. Time to return to pre-surgery usual activities and cost are included in the model as continuous variables, where β₆ and β₁₃ describe the impact on utility of a 1-day reduction in time to return to usual activities and a £1 increase in cost, respectively. Mixed logit models were estimated with the ‘mixlogit’ command in Stata^® version 14 (StataCorp LP, College Station, TX, USA), using maximum simulated likelihood, fitted with 50 Halton draws. 111

In total, we estimated six different DCE models. M1, M3 and M5 are conditional logit, errors component logit (random ASC) and random parameters logit (all random except cost, with a normal distribution), respectively. M2, M4 and M6 are the same as M1, M3 and M5, respectively, but with the inclusion of interaction terms to account for the impact of type of anaesthesia received on preferences for anaesthesia and the impact of income on preferences for the cost attribute.

The estimated utility parameters from M1 to M6 describe the impact on utility, but, to compare changes in all attributes in a single unit, it is necessary to calculate marginal rates of substitution between each attribute level (β_k) and the cost attribute (β₁₃) to estimate marginal WTP, calculated as:

The delta method, using the ‘NLCOM’ command in Stata, was used to compute CIs surrounding calculated marginal WTP.

Analysis of data quality and perceptions of the discrete choice experiment survey

As described in the questionnaire design, respondents were asked to agree or disagree on a five-point scale with six statements regarding the process of making choices within the survey. These data are tabulated and plotted graphically for illustration of the results.

Sample characteristics

Appendix 7, Table 43, describes the sample characteristics of the women who returned the DCE questionnaire, compared, for reference, with the corresponding baseline data for all trial participants. Those who returned completed DCE questionnaires closely matched the demographics of the participants who took part in the trial. This provides reassurance that the DCE sample was, in general, a good representation of the preferences of those in the larger trial sample who received an index surgery.

Preferences for treatments and willingness to pay

The results of the respective DCE analysis models are provided in Table 19 and corresponding WTP values are provided in Table 20. M5 generates the best model fit and so these results will be considered the base-case WTP values for carrying forward to a cost–benefit analysis in Chapter 8. Across all six models analysed, women indicated a preference to have a surgical treatment as opposed to none, with the consequence of remaining in their current health state. GA was preferred to LA, with those who had GA in the trial indicating a stronger preference for procedures conducted under GA than those who had LA. Women prefer shorter times to return to normal activities and are willing to pay between £70 and £100 per day of reduction in recovery time following surgery. As would be expected, women attach the greatest value to achieving improvement in their UI symptoms and the avoidance of complications. In general, the complications with the greatest negative impact on utility were the need for intermittent self-catheterisation, and the experience of mesh exposure, although avoidance of dyspareunia and new-onset urgency incontinence were also valued. Women placed an approximately equal value to achieving an outcome of ‘very much improved’ and avoiding the need for intermittent self-catheterisation or experiencing mesh exposure. Women did not have a statistically significant preference against treatments for which they would only rarely have to avoid their usual activities, compared with never having to avoid activities. However, they were willing to pay between £1700 and £2200 for treatments that improved their frequency of avoiding daily activities because of a fear of leaking from occasionally to never. They were willing to pay more (between £4500 and £5700) to improve from frequently avoiding usual activities to never avoiding usual activities.

Assessment of data quality and views on the survey

Figure 11 describes the participants’ responses to six debriefing questions about the survey content, asking about perceptions of survey complexity, learning through the survey, plausibility, relevance to policy, level of information provided and clarity of the choice tasks. Almost all respondents understood the concept of choosing between different treatment options (i.e. DCE choice tasks), which suggests that respondents were willing to make hypothetical trade-offs between benefits and risks. Of the respondents, 26% reported that they found the choice tasks to be confusing, but most felt that the process became easier the more choice tasks they answered. The majority of the respondents felt that the treatment options made sense and that their responses would have an impact on the future availability of treatments, suggesting a high degree of perceived survey consequentiality. Regarding the level of information provided in the choice tasks, about half of the respondents felt that they needed more information to inform their decision-making. Although this proportion would appear low, it must be considered that the volume of information provided is a trade-off between complete information and survey complexity.

FIGURE 11.

Participants’ perceptions of the DCE survey.

Economic evaluation overview

The aim of this chapter is to present the cost-effectiveness of adjustable anchored SIMSs, compared with that of tension-free SMUSs, in the surgical management of female SUI. The primary economic objective was to conduct an ITT analysis alongside a RCT to determine the cost-effectiveness of SIMSs, compared with SMUSs, from a UK NHS perspective. Cost-effectiveness was measured using incremental cost-effectiveness ratios (ICERs) based on QALYs derived from responses to the generic EQ-5D-3L and the UI-specific ICIQ-LUTSqol measures over various time points during the 36-month follow-up period. The secondary economic objectives were to estimate costs from a societal perspective, to determine patient preferences for the processes and outcomes of alternative surgical procedures using a DCE (see Chapter 7) and to use the DCE results to conduct a cost–benefit analysis. Costs and QALYs accrued in the 24 and 36 months were discounted by 3.5% per annum, in line with NICE recommendations. 113 The base-case analysis was conducted as an imputed analysis owing to the impact of missing data. 114 The choice of method for handling missing data (multiple imputation) was grounded in the assumed missing data mechanism, missing at random, which was supported by the data. The methods and results were reported following the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) recommendations and the analyses followed the health economics analysis plan. 115

Resource use and cost collection

Operation resource use

The base-case analysis used a component costing approach. Several surgical devices from different manufacturers were used in the trial as long as they met the prespecified criteria in the trial protocol. 1 Data on the type of device used for each woman was collected on the CRF. The unit cost of the device was based on the list price that the sites purchased them for. The unit costs were derived from personal communication with four sites. The intervention device costs were applied at the individual participant level. The grade of the operating surgeon and whether or not they were supervised, and the grade of the anaesthetist, were also collected on the CRF. An assumption was made about grade and the number of supplementary staff present during a typical surgical procedure (nurses and theatre assistants) based on the clinical opinion of experts working on the trial team. The unit costs of staff salaries were derived from a published source. 116 The resource use data for the different anaesthesia were based on a published study117 and the unit costs of different anaesthesia were based on the costs of the drugs administered. 118 Although most interventions were day cases, some women spent some days in hospital. The unit cost of overnight stay was based on the cost of elective inpatient excess bed-days for vaginal tape/MUS operations.

Follow-up costs

The follow-up primary care visit costs were derived from a published source116,118,119 and the secondary diagnostic and procedure costs were based on Healthcare Resource Group tariffs. The unit costs were measured in Great British pounds (£). The costs were based on 2018/19 prices. Unit costs/prices were obtained from the following published sources: the British National Formulary,118 reference costs119 and Unit Costs of Health and Social Care 2019,116 as indicated in Table 21.

Participant-incurred unit costs

Participant resource use comprised three main elements: travel costs for making return visit(s) to NHS health care, time costs of travelling and attending NHS health care, and self-purchased health care. The time resource and unit cost of making a return journey to each type of health-care provider were obtained from the Participant Time and Travel Cost Questionnaire.

The participants reported how long they spent travelling to and attending their last visit to each type of health-care provider. Participants were asked what activity they would have been undertaking (e.g. paid work, leisure, housework) had they not attended the health-care provider. These data were presented in their natural units (e.g. hours) and costed using standard economic conventions (e.g. the Department of Transport estimates for the value of leisure time). Participants were asked if they were accompanied by a friend or a relative and, if so, to provide details of their time and travel. These unit time costs were multiplied by the number of health-care contacts derived from the health-care resource use questions to generate the opportunity cost of time and travel from the patient perspective. Details of unit costs applied to the various activities are included in Table 22.

Data collected through the patient questionnaire were used to estimate the costs of self-purchased health-care items, including pads bought by participants, prescription costs and over-the-counter medications. All self-purchased health care related to items purchased for the management or treatment of UI.

Estimates of resource use were combined with unit costs to derive total costs for each item of resource use and each patient. The costs for each item of resource use for each patient were summed to produce a total cost for each patient and an average total cost per patient in each intervention arm.

Derivation of quality-of-life measures

Health-related QoL was measured using a standard generic tool and a disease-specific tool: the EQ-5D-3L and the ICIQ-LUTSqol, respectively. For both these tools, higher utility scores indicate better health as perceived by the participants.

Trial participants completed the EQ-5D-3L at baseline; at 4 weeks and at 3 months after their intervention; and at 15, 24 and 36 months after randomisation. The responses to the EQ-5D-3L questionnaire were valued using UK general population tariffs, based on the time trade-off technique to generate a utility score for every participant at each follow-up time point in the trial. 126 QALYs were calculated as the area under the curve defined by the utility values at baseline and at each follow-up. 127 Most women had to wait for some time for their surgery. Therefore, the base-case analysis QALY calculation assumed that utility remained constant between randomisation and surgery (i.e. women could not generate QALY gains before the point of surgery). The waiting time between randomisation and surgery was calculated using the dates recorded on the CRF. For the remaining time points, a linear interpolation between utility measurement time points was assumed.

The 4-week and 3-month questionnaires were triggered by the operation date. Women who did not get surgery did not receive the 4-week or 3-month questionnaires; therefore, QALY calculation area under the curve was estimated using data from baseline and 15, 24 and 36 months. Overall self-rated health status was also collected using the EuroQol Visual Analogue Scale (EQ-VAS) score (0 equals worst health and 100 equals best health). QoL data were also collected using items from the condition-specific tool ICIQ-LUTSqol for sensitivity analysis. These data were collected at baseline and at 3, 15, 24 and 36 months.

The ICIQ-LUTSqol is a psychometrically robust patient-completed questionnaire evaluating QoL in urinary incontinent patients for use in research and clinical practice across the world. The ICIQ-LUTSqol is the King’s Health Questionnaire (KHQ) adapted for use within the ICIQ structure and provides a measure to assess the impact of UI on QoL with particular reference to social effects. ICIQ-LUTSqol responses were converted into a utility index using a published algorithm. 128 There was no need to impute utility data as no deaths were recorded during the trial follow-up period.

Data analysis

All components of costs were described with the appropriate descriptive statistics when relevant: mean and SD for continuous and count outcomes, and numbers and percentages for dichotomous and categorical outcomes (e.g. numbers reporting problems on the EQ-5D-3L). All analyses were conducted using Stata version 14.1. Investigations were carried out for skewed cost data (i.e. a small proportion of participants incurring very high costs), using GLMs to test alternative model specifications for appropriate fit to the data. The GLM models allow for heteroscedasticity by selecting and specifying an appropriate distributional family for the data. This family offers alternative specifications to reflect the relationship between the mean and variance of the estimates under consideration (Glick 2015). 129,130 Two methods were used to identify the most appropriate distributional family: (1) a modified Park test and (2) the Akaike information criterion. For costs, the post-prediction test statistics indicated that the most appropriate family was inverse Gaussian. The goodness-of-fit statistics based on multiple tests (Pregibon link, modified Hosmer–Lemeshow and Pearson’s correlation) indicated that the log was the appropriate link. The QALY data were analysed using a Gauss family and an identity link. Analysis models were run to estimate the incremental effect of treatment group on costs and QALYs. Models were adjusted using a fixed effect for the minimisation covariate: previous supervised PFMT within the preceding 2 years (PFMT: yes/no), age and baseline QoL data (EQ-5D-3L or ICIQ-LUTSqol as appropriate); furthermore, the analyses used robust standard errors, clustered by centre.

Incremental cost per quality-adjusted life-year

Results of the cost–utility analysis are reported as incremental cost per QALY for each treatment group. ICERs were computed to determine the cost-effectiveness of SIMSs, compared with that of SMUSs. The point estimate of the ICER is calculated as:

where C_i and C_j are the mean NHS costs for the SIMS and the SMUS arms, respectively. Similarly, E_i and E_j are the mean QALYs in the SIMS and SMUS arms, respectively. The ICER is assessed against the NICE-recommended cost-effectiveness WTP threshold of £20,000 per QALY gained; additional information on the probability of cost-effectiveness for £0 and £30,000 WTP thresholds is also provided. 113

The joint density of incremental costs and incremental QALYs was derived using non-parametric bootstrapping. From the results of the bootstrapping, cost-effectiveness acceptability curves (CEACs) were created. CEACs are used to display the inherent uncertainty surrounding cost-effectiveness at various WTP threshold values for society’s WTP for a QALY. CEACs present results when the analysis follows a net benefit approach. This approach utilises a straightforward rearrangement of the cost-effectiveness decision rule used when calculating ICERs to create the net monetary benefit (NMB) for each bootstrapped iteration at increasing values of WTP per QALY:

where λ represents a decision-maker’s WTP for incontinence avoided or a QALY gained. If the above expression holds true for a given iteration and WTP threshold value (λ), then the intervention is considered cost-effective for that iteration. As society’s WTP is unknown, the NMB was calculated for several possible λ values, including the usual £20,000–30,000 range often adopted by policy-makers in the NHS. 113

Missing data

Missing data are a frequent problem in economic evaluations undertaken within a RCT setting, driven by the requirement to use multiple resource use items to calculate costs, and the repeated measures nature of questionnaires required to derive total costs and the QALY area under the curve. There are several possible methods that can be employed to account for missing data, such as mean or multiple imputation. For the analysis of costs, the mechanism of data missingness was investigated using logistical regression analysis, whereby the dependent binary variable (missing or not) was regressed on age, minimisation variables, baseline EQ-5D-3L utility score and baseline measures of type of UI. Imputation analysis was conducted as > 5% of both cost and QALY data needed for primary analysis were missing. The handling of missing data was based on the premise that the data were missing at random, and multiple imputation was employed. All imputation was completed using Stata’s multiple imputation procedure. Components of cost (primary and secondary care costs were imputed for each year) and utility data were imputed for each measurement time point, based on GLMs that were adjusted for minimisation variables, age and baseline utility. To preserve the allocation of participants’, the imputation was done on each randomised group. Imputations were generated using predictive mean matching drawn from the five k-nearest neighbours (k-NN = 5); predictive mean matching preserves distribution of the data and is more robust to violations of the normality assumption. The multiple imputation was run 20 times, generating 20 complete data sets, to ensure that stable results were combined using Rubin’s rules to generate estimates of costs. 131

Sensitivity analyses

The estimation of the QALYs used in the base-case analysis considered the time participants waited for surgery and incorporated a bottom-up approach in costing the intervention. Sensitivity analyses were performed to gauge the impact of varying these assumptions and/or parameter values on the base-case analysis.

• The base-case analysis used multiple imputation of missing data. A complete-case analysis was performed as a sensitivity analysis to highlight the importance of this base-case assumption.

• Sensitivity analyses on the QoL measure (i.e. the EQ-5D-3L) were conducted without adjusting for the time the women waited for the initial surgery after randomisation.

• A sensitivity analysis was conducted using the utility values from the condition-specific QoL measure, ICIQ-LUTSqol.

• Analyses were conducted to explore the impact of changing the discount rate used for 24- and 36-month costs and QALYs in accordance with NICE best-practice recommendations; the discount rate was varied from 0% to 6% per annum.

• An analysis was conducted to explore the relaxation of the assumption that women who did not have surgery and did not have any CRF completed for follow-up visits had zero costs (see Appendix 8, Table 47).

Subgroup analysis

Depending on the availability of data, the following subgroup analyses were conducted:

• urodynamic mixed incontinence versus urodynamic stress incontinence

• age – above and below the observed median age of the recruited women

• a post hoc analysis for age – < 65 years compared with those aged ≥ 65 years

• a post hoc comparison between those who had received supervised PFMT in the previous 2 years and those who had not.

Cost–benefit analysis

The cost–benefit analysis compared SIMSs with SMUSs using a definition of net benefit whereby both costs and outcomes are measured in monetary values. This allows a direct calculation of net benefits (benefit, measured as WTP – cost) for each participant in the trial. To obtain monetary valuation of benefits, marginal WTP tariffs, calculated from the DCE base-case model (random parameters logit model; see M5 in Table 19), were mapped to clinically important and patient-relevant characteristics of the surgical procedure (type of anaesthesia) and trial outcomes (number of recovery days after surgery, complications, PGI-I scale and avoidance of activities). The mapping process and associated assumptions made to match DCE attributes with trial outcomes is described in Table 23.

Benefits, measured as total WTP, were calculated for each woman in the trial by summing the marginal WTP tariffs from the DCE for each attribute level mapped to the corresponding patient outcomes from the trial. An average total WTP for women in each treatment arm (SIMS and SMUS) was then calculated for use in the cost–benefit analysis.

Total WTP was compared with total costs measured from both the NHS perspective and the DCE to generate estimates of net benefit and calculate benefit-to-cost ratios. All surgical treatments with a benefit-to-cost ratio of > 1 should be implemented on the grounds that they offer value for money from a societal welfare maximisation perspective. Similarly, all surgical treatments with a benefit-to-cost ratio of < 1 should not be implemented, or should be phased out if already in routine use, as the benefits would be insufficient to offset the costs of treatment and would lead to inefficiency.

We also report incremental net benefit for SIMSs versus SMUSs. Positive incremental net benefit values indicate an improvement in efficiency by adopting SIMSs over SMUSs, whereas negative incremental net benefit values indicate that SIMSs would reduce overall efficiency. As with the cost–utility analyses, results are reported based on multiple imputation of missing data, with imputations for WTP using multiple imputation with chained equations and using predictive mean matching, with 20 imputed data sets. Total WTP data were analysed using standard ordinary least squares regression models, with robust standard errors, clustered by centre. Bootstrapping was used to generate recycled predictions of cost and benefit (total WTP) pairs, and results are reported using scatterplots of the cost–benefit plane and cost–benefit acceptability curves to illustrate uncertainty around net benefit estimates.

Intervention and follow-up costs (NHS perspective costs)

The following sections report descriptive statistics and cost difference results based on complete-case data. These costs include intervention costs up to the point of discharge and primary care and secondary care follow-up costs. The total NHS costs were calculated by multiplying the resource use (see Appendix 8, Table 49) by the appropriate unit costs outlined in Table 21.

Intervention costs

Details of costs are reported in Table 24. The cost of the intervention was, on average, lower for the SIMS group. On average, there were significant cost savings in the SIMS group in several of the resources that contributed to the operation costs: the type of anaesthesia administered, the availability of an anaesthetist in the index intervention and recovery time. The women in the SMUS group spent more time in theatre and recovery and more had GA for their surgery. The average cost of the intervention device was similar in both groups. Overall, the total cost of the initial intervention was significantly lower in the SIMS group, at –£180 (95% CI –£287 to –£73), than in the SMUS group.

Health service costs over trial follow-up period

The follow-up costs at 3, 15, 24 and 36 months were not significantly different between the groups. Details of secondary care resource use and costs are given in Appendix 8, Table 50. Overall, over the 36 months of follow-up, the mean cost saving for the SIMS group was –£254 per person (95% CI –£587 to £79). The results of the complete-case cost analysis should be interpreted cautiously, and in the light of the substantial amount of missing information to generate cost profiles.

Participant perspective costs

This analysis includes participant-incurred costs for attending their SIMS trial surgery and any follow-up visits that related to UI. Table 25 reports the costs of attending the index surgery, return to usual activities, appointments at their GP surgery and hospital outpatient and inpatient departments, over-the-counter purchases, payments to private health-care providers and purchases of pads. Women reported their return to usual activities in the questionnaire at 4 weeks after surgery; therefore, the maximum number of days (28) was used for those who reported that they had not returned to usual activities to estimate the opportunity cost of days away from usual activities. These costs were then summed across all the available cost data for participants and companion time and travel cost data to estimate the total cost over the 3-year follow-up period.

On average, women in the SMUS group took longer to return to usual activities than those in the SIMS group. Of the women who returned the questionnaire at 4 weeks, 61 out of 246 (27%) in the SIMS group and 66 out of 226 (29%) in the SMUS group had not returned to usual activities. For those who returned the questionnaire, the average time to return to usual activities was 12 (SD 7) days in the SIMS group and 15 (SD 8) days in the SMUS group. The cost difference in opportunity costs for return to usual activities was lower for the SIMS group (–£145, 95% CI –£275 to –£14) than for the SMUS group. The opportunity costs of time and travel, purchase of over-the-counter medicines and activities related to their urinary symptoms for the intervention, general practice visits, outpatient visits and inpatient stay were similar across the groups. The mean cost for the women who had complete cost data in the SIMS group [129/298 (43%)] was £1462 (SD £739); for the SMUS group [106/298 (36%)], it was £1560 (SD £718). The total cost difference was –£89 (95% CI –£280 to £102). The high SDs in both groups indicate that some women had high out-of-pocket costs.

Combining the women’s and the total NHS costs increased the cost difference to –£328 (95% CI –£804 to £147); however, it was not statistically significant. The wide CIs suggest that there was great variation in the costs incurred by the women. Some women had no costs and a few women had very high costs.

Table 26 provides descriptive data of complete-case utility scores and QALYs generated by combining utilities with duration of follow-up. The differences are based on linear regression models (GLMs), with adjustment for minimisation covariate, age and baseline utility score. The utility scores derived using the EQ-5D-3L indicate that the QoL at all time points was higher for the SIMS group, including at baseline. However, when calculating incremental QALYs adjusting for baseline imbalances in EQ-5D-3L utilities, SIMSs were associated with fewer QALYs gained over the 3 years (mean difference –0.089, 95% CI –0.156 to –0.023). There was little difference in ICIQ-LUTSqol utilities at any of the measurement time points, and there was no evidence of a difference in the derived QALYs between the groups. The EQ-VAS scores were higher in the SIMS group, but the differences in the scores were not statistically significant. These results need to be interpreted in the context that they are based on complete-case data across all utility measurement time points, of which between 43% and 47% are missing. It is important therefore to consider multiple imputation for the base-case cost-effectiveness analysis.

Numbers reporting problems on the EuroQol-5 Dimensions, three-level version, questionnaire

The proportion of women reporting problems was based on the number of women who reported that they had some or severe problems in the domains in the EQ-5D-3L questionnaire. The details of the number and percentage of women reporting problems are in Figures 12–17. The domain in which the lowest proportion of women reported problems, over all the time points that data were collected, was self-care (see Figures 12–17 and Appendix 8, Figure 24). At baseline, the proportion of women reporting problems was evenly distributed among the domains, apart from anxiety or depression, for which the proportion of women in SIMS group was lower (19%) than that in the SMUS group (33%) (see Figure 12). More women reported having problems with usual activities (SIMS group, 31%; SMUS group, 40%) at 4 weeks post surgery (see Figure 13) than at all the other time points, and more women in SIMS group (33%) than in the SMUS group (16%) reported having problems with pain or discomfort at 4 weeks post surgery. The proportion of women reporting problems with pain or discomfort and anxiety or depression at 15, 24 and 36 months ranged between 24% and 31% (see Figures 15–17). At 24 and 36 months, the number of women reporting problems in all domains apart from usual activities was higher than at baseline.

FIGURE 12.

Number and percentage of women reporting problems at baseline.

FIGURE 13.

Number and percentage of women reporting problems at 4 weeks.

FIGURE 14.

Number and percentage of women reporting problems at 3 months.

FIGURE 15.

Number and percentage of women reporting problems at 15 months.

FIGURE 16.

Number and percentage of women reporting problems at 24 months.

FIGURE 17.

Number and percentage of women reporting problems at 36 months.

Incremental cost-effectiveness analysis

Results of the within-trial analysis are reported as ICERs, calculated as the difference in costs divided by the differences in QALYs (SIMS vs. SMUS) over the 36-month follow-up period. To account for the substantial proportion of missing cost and QALY pair data (72%), the base-case and all sensitivity analyses, apart from the complete-case analysis, were conducted using multiple imputation of missing data. The base-case results in Table 27 show no differences in costs (–£6, 95% CI –£228 to £208) or QALYs (0.005, 95% CI –0.068 to 0.073) between the groups. The uncertainty in differences in costs and QALYs is illustrated by the width of the CIs and in Figure 18, which shows cost and QALY differences scattered in all the quadrants of the cost-effectiveness plane. The cost and QALY differences are in Figure 18 are well distributed in all quadrants of the cost-effectiveness plane. The CEAC (Figure 19) illustrates that there is a 56% probability that SIMSs will be considered cost-effective at the £20,000 WTP threshold value for a QALY.

FIGURE 18.

Incremental costs and QALYs for the SIMS group, compared with the SMUS group, using imputed costs and EQ-5D-3L QoL scores.

FIGURE 19.

The CEACs for the SIMS group vs. the SMUS group using imputed costs and EQ-5D-3L QoL scores.

Deterministic sensitivity analysis

The results of the sensitivity analysis are presented in Table 27, and the scatterplots and CEACs are shown in Appendix 8, Figures 25–38. Some of the estimates of incremental cost and QALYs were sensitive to the approach taken.

The results of the complete-case sensitivity analysis suggest that, on average, SIMSs cost less than SMUSs (–£209, 95% CI –£493 to £76), but had fewer QALYs (–0.096, 95% CI –0.227 to 0.027). The estimates of cost and QALY differences fall mainly in the south-west quadrant of the cost-effectiveness plane (see Figures 25 and 26). SIMSs had an ICER of £2187 cost saving per QALY loss and a 10% chance that they would be considered cost-effective if society would require £20,000 cost savings to justify a QALY loss.

When the assumption was relaxed that the secondary care costs were zero for those who did not have surgery, the cost difference was higher for the SIMS group (–£89, 95% CI –£192 to £9) than for the SMUS group, and the QALY difference was 0.005 (base case) to 0.035 (95% CI –0.018 to 0.082). The probability that SIMSs would be considered cost-effective at the £20,000 WTP threshold was higher than for the base-case analysis (61% vs. 56%) (see Figures 35 and 36).

The results that used ICIQ-LUTSqol data indicate that SIMSs cost less (–£6, 95% CI –£228 to £208) and were less effective (–0.001, 95% CI –0.029 to 0.023) than SMUSs. The ICER was £4120 cost saving per QALY loss and the probability that SIMSs are cost-effective at the £20,000 WTP threshold was 48% (see Figures 27 and 28).

The assumptions made in the calculation of the QALY, and discount rates applied to the costs, did not seem to have an impact on the overall results, with SIMSs always costing less than SMUSs and having more QALYs than SMUSs. The probability that SIMSs would be cost-effective at the £20,000 WTP threshold ranged between 56% and 62% for these analyses (see Figures 29–34). The societal perspective analysis that combined participant costs and NHS data did not have a substantial effect on overall findings for the cost–utility analysis (see Figures 37 and 38). SIMSs cost –£11 (95% CI –£267 to £241) less than SMUSs and had 0.005 (95% CI –0.068 to 0.073) more QALYs than SMUSs.

Subgroup analyses

Subgroup analyses were conducted; the results are presented in Table 27 for the analyses for which data were available.

Most of the women (78%) had a diagnosis of urodynamic stress incontinence. Results for this subgroup analysis indicate that SIMSs cost –£117 (95% CI –£226 to –£19) less and had 0.024 (95% CI –0.030 to 0.077) more QALYs than SMUSs. The probability that SIMSs would be cost-effective at the £20,000 WTP threshold was 85%.

Most of the women (87%) were aged < 65 years. Results for this subgroup analysis indicate that SIMSs cost –£69 (95% CI –£179 to £39] less and had 0.006 (95% CI –0.055 to 0.064) more QALYs than SMUSs. The probability that SIMSs would be considered cost-effective at the £20,000 WTP threshold was 62%.

The number of women who had PFMT was equal in both group [254/298 (85%)]. Results for this subgroup analysis indicate that SIMSs cost –£42 (95% CI –£141 to £53) less and had 0.0003 (95% CI –0.066 to 0.061) more QALYs than SMUSs. The probability that SIMSs would be considered cost-effective at the £20,000 WTP threshold was 54%.

For women who were aged ≤ 48 years, SIMSs cost –£49 (95% CI –£169 to £70) less and had 0.0003 (95% CI –0.066 to 0.061) more QALYs than SMUSs. The probability that SIMSs would be considered cost-effective at the £20,000 WTP threshold was 97%. For women aged > 48 years, SIMSs cost –£77 (95% CI –£229 to £83) less and had –0.062 (95% CI –0.147 to 0.018) fewer QALYs than SMUSs. The probability that SIMSs would be considered cost-effective at the £20,000 WTP threshold was 10%.

Cost–benefit analysis

The incremental cost–benefit analysis results are reported in Table 28 and in Figures 20 and 21. The cost difference is the same as for the base-case analysis and the incremental net benefit for SIMSs, compared with SMUSs, is –£941. The probability that SIMS would be considered cost-effective at the £20,000 WTP threshold is 9%.

FIGURE 20.

Scatterplot of incremental costs and incremental benefits (WTP) for SIMSs, compared with SMUSs.

FIGURE 21.

Cost–benefit acceptability curves: SIMS vs. SMUS using imputed costs and net benefit.

The Patient Global Impression of Improvement scale

The Patient Global Impression scale is a global index that is widely used to rate the response of a condition to a therapy (transition scale). The PGI-I scale is a simple, direct and easy-to-use scale that is intuitively understandable to clinicians and patients. 146 The PGI-I scale has excellent construct validity, compared with various assessment variables: incontinence episode frequency, the Incontinence Quality of Life Questionnaire and the fixed-volume (400 ml) stress pad test. 147 In a benchmark study, Yalcin and Bump102 reported a secondary analysis of data from two double-blind, placebo-controlled clinical trials (n = 1133) that evaluated the treatment of women with predominant SUI. The authors showed that significant correlations (p < 0.0001) were observed between the PGI-I scale response categories and three independent measures of improvement in SUI (0.49, 0.33 and –0.43 for incontinence episode frequency, stress test and Incontinence Quality of Life Questionnaire results, respectively). This important study established the construct validity of the PGI-I scale for the evaluation of the baseline severity of, and treatment response among women with, SUI.

Since that study,102 the PGI-I scale has been widely used in clinical trials assessing surgical and conservative interventions for UI among women.

• The botulinum toxin A versus placebo for refractory detrusor overactivity in women (RELAX) trial148 evaluated the PGI-I scale as an outcome assessment tool among women with refractory OAB symptoms requiring Botox treatment. The authors commented that their results showed that ‘the PGI-I scales are robust and valid instruments to assess disease severity, bother and improvement after treatment in women with detrusor overactivity’. 148

• Brubaker et al. 149 used the PGI-I scale to measure the primary outcome in their RCT evaluating Botox treatment among women with refractory OAB. The authors showed that the PGI-I scale was able to detect differences in responses between groups.

• The PGI-I scale was used to measure the primary outcome in our previous small RCT (n = 137)86 comparing the SIMS Ajust with the SMUS TO-TVT for the surgical treatment of SUI in women. 86 The results showed an excellent response rate at 1 and 4 years of follow-up.

• The PGI-I scale was used to measure the primary outcome in the E-TOT study comparing inside–out and outside–in TO-TVTs, which reported outcomes at 1, 3 and 9 years. 65,139,140

• The 10-year outcomes were recently (in 2016 and 2017) reported for two clinical trials of surgical treatment of SUI among women. 150,151 They both used the PGI-I scale to measure their primary patient-reported outcomes.

• The PGI-I scale is currently used as the primary outcome comparing surgical interventions for the treatment of UUI in the Health Technology Assessment (HTA)-funded Female Urgency, Trial of Urodynamics as Routine Evaluation (FUTURE) study. 152

Women with urodynamic mixed urinary incontinence

The International Continence Society definition of MUI emphasises the presence of SUI and components of OAB, that is urgency with or without frequency, nocturia and urgency incontinence. 8 MUI is considered more difficult to treat because of the need to mutually manage SUI and OAB symptoms, with the latter often being unpredictable, with evidence of flaring up and remission of symptoms over time. 165 There is a paucity of primary research in this area and the management of women with urodynamic mixed incontinence remains a subject of much debate. Duckett and Tamilselvi166 have shown that 63% of women with urodynamic mixed incontinence experience complete resolution of urgency symptoms after a RP-TVT procedure, with objective cure rates of 47% and 92% for DO and urodynamic stress incontinence, respectively. Similarly, Lee et al. 167 have shown a 79.4% cure rate among women with MUI, compared with 89.8% among women with SUI, with up to 6 years of follow-up; the difference was not statistically significant. It has been previously shown that, among women with SUI-predominant urodynamic mixed incontinence, TO-TVTs are associated with good patient-reported success rates of 75%, 73.8% and 65% at 1, 3 and 9 years, respectively. 168–170

In the SIMS trial, most participants had urodynamic stress incontinence (SIMS group, 79%; SMUS group, 78%), rather than urodynamic mixed incontinence (SIMS group, 12%; SMUS group, 11%). In both subgroups, the success rates were higher for those receiving SIMSs than for those receiving SMUSs; however, the difference was smaller in the urodynamic mixed incontinence subgroup.

Age

Age is another clinically relevant subgroup, so we conducted a subgroup analysis of the primary outcome according to age. The median age was 48 years; patient success rates were better for participants aged > 48 years in the SIMS group (see Table 8). We also conducted another post hoc subgroup analysis based on age: < 65 years versus ≥ 65 years. We chose 65 years as it is the retirement age in the UK; hence, day-to-day activities can significantly change at this age. There seems to be no significant differences between SIMSs and SMUSs in this subgroup.

In 2020, Ahn et al. 171 reported their study assessing the impact of age on outcomes of SMUS procedures among 262 women who underwent the procedure in 2010–2015. They divided women into three age groups (≤ 50 years, 51–59 years and ≥ 60 years) and found no significant differences in patient-reported success rates between RP-TVT and TO-TVT. In 2018, Engen et al. 172 presented the outcomes of 21,832 women with SUI or MUI who underwent MUS procedures between 1998 and 2016. Data were obtained from the Norwegian Female Incontinence Registry. The ‘very satisfied’ rates at 6–12 months post procedure (other options were ‘moderately satisfied’, ‘neither satisfied nor dissatisfied’, ‘moderately dissatisfied’ and ‘very dissatisfied’), per age decade, were as follows: 18–29 years – 79.4%, 30–39 years – 86.3%, 40–49 years – 87.1%, 50–59 years – 86.2%, 60–69 years – 81.1%, 70–79 years – 71.7% and 80–99 years – 61.8%. This is consistent with earlier reports from Malek et al. 173 in their retrospective study of nearly 700 women undergoing MUS procedures. The mean age of patients aged ≥ 70 years (n = 160) was 75.4 ± 4.5 years; the mean age for those aged < 70 years (n = 536) was 56.2 ± 9.4 years. A multivariable analysis revealed no difference in SUI failure rates among older cohorts, compared with younger cohorts (adjusted OR 1.7, 95% CI 0.9 to 3.1). Women aged < 70 years reported a greater impression of improvement than women aged ≥ 70 years (67.7% vs. 56.6%, respectively; p = 0.01).

Different definitions of patient-reported success

The results of a post hoc analysis for patient-reported ‘cure’ using the strict definition of a response of ‘no leakage’ on the ICIQ-UI-SF were as follows: SIMS group, 38.6%; SMUS group, 33.2%. Ward et al. 174 used a similar 2-year strict definition, and found 25% and 20% cure rates for those undergoing RP-TVT and colposuspension, respectively. Interestingly, in one observational study for the mid-term safety and efficacy of the SIMS Altis, using the same strict definition, the patient-reported cure rate was 88.2% among 110 women. 175,176

Some studies use different definitions of patient-reported success. Hence, we presented the full analysis of itemised PGI-I scale responses (very much improved, much improved, improved, same, worse, much worse and very much worse) between both groups at all time points of assessment (i.e. 3, 15, 24 and 36 months). It gives confidence in the results to see that, at 36 months, comparable percentages of women in the SIMS and SMUS groups reported their continence status to be the ‘same’ (9.3% vs. 7.7%, respectively), compared with before their procedure, whereas 7.3% and 10.2% in the SIMS and SMUS groups, respectively, reported their continence status to be ‘worse’, ‘much worse’ or ‘very much worse’. By 36 months, 4.3% and 2.3% in the SIMS and SMUS groups, respectively, had received further invasive intervention for UI (including both SUI and UUI).

Excluding devices that were withdrawn from clinical practice

We included a post hoc subgroup analysis for the primary outcome excluding the devices that were withdrawn from clinical practice. The findings did not change.

Unlike other studies in the literature, the SIMS RCT is, to our knowledge, the first to compare two MUS ‘technologies’, and not specific types of slings/devices, that is it compared standard-length tension-free MUSs (both retropubic and transobturator approaches) with adjustable anchored mini slings (SIMSs). Within each trial arm (SMUS and SIMS), the choice of device used in any specific patient/hospital was down to the surgeon experience/preference and availability at the specific hospital. Hence, the trial results are generalisable to any type of free SMUS and any type of adjustable anchored SIMS that were available at the time of the trial, or similar devices that may be available in the future, or those that might be reintroduced.

Intraoperative injuries

In the SIMS trial, there were no cases of major visceral injuries. Data from the BSUG surgical database showed ureteric and bowel injuries of 0.02% (i.e. one in 5000) and 0.006% (i.e. one in 17,000) for RP-TVT, that is rare and extremely rare, respectively. 185 In our study, only one case of retropubic haematoma following RP-TVT insertion occurred; it was diagnosed in the immediate postoperative period and confirmed by radiological imaging. Diagnostic laparoscopy was performed on day 1 post operation, and the haematoma was managed conservatively. It is debatable if the laparoscopy was absolutely necessary as the haematoma is retroperitoneal and imaging such as magnetic resonance imaging and computerised tomography could have sufficed to confirm stability in the haematoma size and exclude active bleeding. However, one advantage of the laparoscopy was to exclude any other organ injuries.

In the SIMS trial, intraoperative LUT injuries occurred exclusively in the SMUS group (3.8% of participants; RP-TVT 7.1% vs. TO-TVT 0.7%). Similarly, the Trial Of Mid-Urethral Slings (TOMUS) reported LUT injury rates in its RP-TVT and TO-TVT groups of 5.3% and 0%, respectively. 187 A large retrospective study has shown a 1% risk of LUT injuries with TO-TVT procedures. 188 Data from the BSUG database showed LUT injury rates of 3.8% and 0.6% for RP-TVT and TO-TVT procedures, respectively. 185 In the literature, 20 RCTs (n = 2613 participants) compared SIMSs with SMUSs with regard to LUT injuries. 86,90,138,141,154,155,157,158,160,162,164,189–197 Pooled data favoured neither SIMSs nor SMUSs (RR 0.69, 95% CI 0.35 to 1.36).

Mesh exposure

Our results showed higher tape/mesh exposure rates among the SIMS participants than the SMUS participants, with nine out of 276 (3.3%) in the SIMS group and five out of 261 (1.9%) in the SMUS group (mean difference 1.3, 95% CI –1.7 to 4.4; p = 0.373) reporting tape exposure over the 36 months of follow-up. Management of tape exposure included partial tape removal/limited excision of exposed mesh (SIMS group, 1.4%; SMUS group, 1.1%) and recovering of the exposed mesh using vaginal walls (SIMS group, 1.1%; SMUS group, 0%). In a large retrospective study of 316 women, the vaginal mesh exposure rate for TO-TVT was 5%. 198 Data from the BSUG database showed overall mesh complication rates of 2.9% for RP-TVT and 2.3% for TO-TVT. 185 In the current literature, 12 RCTs reported on vaginal mesh exposure (n = 1528 participants). 86,90,138,155,156,160,162,191–193,196,197 Pooled data did not favour either SIMSs (excluding TVT Secur) or SMUSs (RR 0.98, 95% CI 0.44 to 2.20).

Groin/thigh pain

We asked participants to report groin/thigh procedure-related pain, analgesia use and any other pain treatments received, but did not measure the impact of pain on day-to-day activities. The absence of preoperative pain measurement is a limitation, as we are unable to measure baseline pelvic pain from other pathology. The rates of groin or thigh pain and subsequent use of analgesics were higher in the SIMS group at 15 months: SIMS group, 14.9%; SMUS group, 11.9% (mean difference 3.0%, 95% CI –2.8% to 8.8%). However, by 3 years, there was a slightly higher rate of pain in the SMUS group: SIMS group, 14.1%; SMUS group, 14.9% (mean difference –0.8%, 95% CI –4.1% to 2.5%). Despite no statistically significant difference in pain rates at 15 months, all four participants receiving tape removal for pain up to 15 months were in the SIMS group. By 3 years, two women in the SMUS group underwent partial removal for pain, whereas there were no further removals for pain in the SIMS group. This might be attributed to the patients’/surgeons’ lower threshold for earlier removal of SIMSs, as the newer non-standard procedure, or the reluctance of the surgeons to remove the SMUS as the standard procedure. Furthermore, the shorter mesh in the SIMS gives the perception that most of the mesh could be removed vaginally without groin or retropubic dissection. However, the authors are aware of one report in the trial of difficulties in removing the SIMS anchors with vaginal dissection only.

In the TOMUS (n = 597),187 the neurological (pain) symptoms rate was higher in the TO-TVT group than in the RP-TVT group (9.4% vs. 4%, respectively; p = 0.01). Sabadell et al. 158 compared 60 women randomly assigned to either Ajust or TO-TVT. At 1 year of follow-up, three women in the Ajust group reported persistent thigh pain 1 year after surgery; none in the TO-TVT group reported pain. The authors concluded that, at 1 year, the Ajust SIMS showed non-inferior effectiveness, compared with TO-TVT, and added ‘Although not statistically significant, unexpectedly, more women reported persistent thigh pain in the Ajust group’. 158 In the literature, three RCTs (n = 585 participants) have compared immediate postoperative pain between those receiving SIMSs (excluding TVT Secur) and those receiving SMUSs at postoperative day 7. 138,155,161 Pooled data of postoperative pain favoured SIMSs over SMUSs (mean difference –0.58, 95% CI –0.73 to –0.39).

Dyspareunia

More women in the SIMS group than in the SMUS group reported dyspareunia at baseline (17.2% vs. 14.5%, respectively). The rate of dyspareunia continued to be higher in the SIMS groups at all follow-up time points. By 3 years, more women in the SIMS group than in the SMUS group reported dyspareunia (11.7% vs. 4.8%, respectively; RD 7.0%, 95% CI 1.9% to 12.1%). The reason for this higher rate in the SIMS group is unclear, but it is possible that the anchoring mechanism of the SIMS (compared with that of the tension free SMUS) potentially contributes to higher rates of dyspareunia.

Shah and Badlani199 explained that dyspareunia may occur because of mesh-related infection, exposure or abnormal healing leading to scarring. Alexandridis et al. 141 showed no significant difference in rates of dyspareunia between SIMS and SMUS groups at 3 years. Masata et al. 161 compared the outcomes of 100 women randomised to either Ajust or TO-TVT. 161 At 2 years of follow-up, two participants in the Ajust group mentioned de novo dyspareunia. The authors explained that clinical examination revealed palpable painful anchor in the obturator membrane in both women, and one patient required surgical removal of the anchor.

In the current literature, seven RCTs (n = 726 participants) have reported on dyspareunia at 1–3 years of follow-up. 154,157,158,161,178,192,200 Pooled data regarding the occurrence of dyspareunia did not favour SIMSs nor SMUSs (RR 1.26, 95% CI 0.57 to 2.82).

Repeat continence surgery

In the SIMS trial, over 3 years of follow-up, 4.3% and 2.3% of SIMS and SMUS participants, respectively, received further invasive intervention for UI (including both SUI and UUI). In the SMUS group, 1.2% (n = 3) received further SUI surgery, compared with 2.6% (n = 7) in the SIMS group. Five of the women receiving further SUI surgery underwent urethral bulking. Urethral bulking was a popular choice presumably because of the suspension of mesh procedures in the UK, whereas major surgery (colposuspension/autologous slings) could be less favoured by women and/or surgeons. It may be less favoured by surgeons because of lack of experience.

In the current literature, 13 RCTs (n = 1703 participants) comparing SIMSs (excluding TVT Secur) with SMUSs have reported on participants receiving repeat continence surgery at 1–3 years of follow-up. 86,90,142,155,157,161,162,178,192,196,200–202 Pooled data did not favour SIMSs nor SMUSs (RR 1.92, 95% CI 1.18 to 3.12). In a population-based retrospective study,183 the authors studied the records of 95,057 women in England, with a median follow-up of 5.5 years after a primary SMUS procedure (RP-TVT, n = 60,194; TO-TVT, n = 34,863). The results showed that the risk of reoperation for SUI was 1.3% (95% CI 1.3% to 1.4%) at 1 year, 3.5% (95% CI 3.4% to 3.6%) at 5 years and 4.5% (95% CI 4.3% to 4.7%) at 9 years after the index procedure.

Mesh removal

In a population-based retrospective study, the authors studied the records of 95,057 women in England, with a median follow-up of 5.5 years after a primary SMUS procedure (RP-TVT, n = 60,194; TO-TVT, n = 34,863). The results showed that the rate of SMUS removal was 1.4% (95% CI 1.3% to 1.4%) at 1 year, 2.7% (95% CI 2.6% to 2.8%) at 5 years and 3.3% (95% CI 3.2% to 3.4%) at 9 years. In the SIMS trial, the total rates for mesh removal (partial/complete) for any indication were comparable to those of the large study [SIMS group, 3% (n = 8); SMUS group, 2% (n = 5)]. However, we reported on mesh removals within the relevant clinical categories for better informing the clinical decision-making of both women and clinicians. Complete/partial mesh removal for pain was 1.5% in the SIMS group, compared with 0.8% in the SMUS group, as described in Chapter 5, Additional surgical treatment. In addition, partial removal/limited excision of mesh exposure was reported by 1.4% of participants in the SIMS group, compared with 1.1% of the SMUS group. None underwent complete tape/mesh removal for the sole indication of mesh exposure. The total rates for mesh removal (partial/complete) for any indication were comparable (SIMS group, 2.9%; SMUS group, 1.9%).

Quality of life

According to the World Health Organization, QoL is defined as an individual’s perception of their position in life, in the context of the culture and value systems in which they live, and in relation to their goals, expectations, standards and concerns. 203

Quality of life can be quite complicated to assess at long-term follow-up, as many other confounding factors may have developed in participants’ lives. The use of disease-specific questionnaires can help to overcome this issue to an extent. A previous RCT (n = 341)140 assessed the long-term outcomes (9 years) of SMUSs (TO-TVT) using the KHQ, which is the predecessor of the ICIQ-LUTSqol used in the SIMS trial;204 clinically significant improvement in QoL was seen in 76.8% of participants. Women reported improvement in all KHQ domains except the general health domain. Similarly, other studies by Ulrich et al. 151 and Serati et al. 150 reported a significant improvement in most QoL domains at 10 years of follow-up following a SMUS (TO-TVT) procedure. 150,151 The evidence, therefore, tends to confirm that the successful outcome of SMUSs can have a long-lasting positive effect on women’s QoL.

However, the impact of SIMSs on QoL is less clear. In a RCT (n = 137)136 comparing an adjustable anchored SIMS (Ajust) with a SMUS (TO-TVT), 82% of women showed clinically significant improvement in total KHQ score, compared with baseline, with no significant differences between the groups (Ajust group, 76.9%; TVT-O group, 87.7%, OR 2.143, 95% CI 0.805 to 5.704; p = 0.19). 136 All KHQ domains except general health showed significant improvement after both operations, compared with baseline, with no significant differences between groups. Pascom et al. 155 compared SIMSs with SMUSs among 130 women. Secondary outcomes included assessment of QoL using the Incontinence Quality of Life questionnaire and the Urogenital Distress Inventory-6 items (UDI-6). 205,206 Women in the SMUS group had better outcomes regarding the avoidance and limiting behaviour domain of the Incontinence Quality of Life questionnaire (p = 0.021) and UDI-6 scores (p = 0.026); otherwise, there was no evidence of significant differences between the groups in any of the other domains. The authors suggested that the better QoL improvement in the SMUS group could be explained by the significantly higher objective success rate in this group.

In the SIMS trial, scores of the QoL questionnaires generally improved compared with baseline, with no statistically significant differences between the SIMS and SMUS groups. We used the ICIQ-LUTSqol, a condition-specific validated multidimensional tool that is widely used for assessing QoL in clinical trials evaluating interventions in UI and pelvic organ prolapse, allowing comparison between our results and those of other RCTs in the literature. We also used the EQ-5D-3L, a general QoL assessment tool. EQ-5D-3L scores increased from baseline to peak at 3 months, but, at 3 years, the EQ-5D-3L score in both groups was lower than at baseline and there were no significant differences between groups (difference 0.013, 95% CI –0.030 to 0.056; p = 0.55). The agreement of the results between the general and disease-specific QoL tools provides reassurance in the results that there is no strong signal favouring SIMSs or SMUSs with regard to impact on women’s QoL at up to 3 years of follow-up.

Our results agree with the current evidence. In 2019, Alexandridis et al. ,141 in a multicentre RCT (n = 279) comparing the SIMS Ajust with the SMUS TO-TVT, assessed the impact on QoL using the same tool used in the SIMS trial, the ICIQ-LUTSqol. At 3 years’ follow-up, a similar improvement in scores, compared with baseline, was observed between the groups (2.8 ± 3.6 vs. 3.0 ± 3.9 in the Ajust and TO-TVT groups, respectively). Similar results were reported by Masata et al. 161 and Xin et al. 138 in RCTs comparing the SIMS Ajust with the SMUS TO-TVT. In the current literature, five RCTs (n = 713 participants) have reported on the impact of SIMSs on participants’ QoL at 2 and 3 years’ follow-up. 141,155,161,178,194 Pooled data did not favour either SIMSs or SMUSs (standardised mean difference 0.19, 95% CI –0.06 to 0.43). A significant limitation is the substantial heterogeneity: studies used a mixture of questionnaires (ICIQ-UI-SF, Incontinence Impact Questionnaire-7 items, UDI-6 and KHQ) to assess QoL.

Sexual function

Female sexual function is complex and multifactorial. It is estimated that 60% of women with UI experience female sexual dysfunction. 207 Several studies found that women with UI report lower intercourse frequency, coital incontinence, decreased libido, vulval and vaginal irritation from persistent urine leakage, dyspareunia, and avoidance or even abstinence from sexual activity. 208 It is therefore reasonable to assume that the resolution of UI and coital incontinence will positively influence a patient’s sexual life. However, the available data on the sexual activity of people who underwent SMUS procedures are inconsistent. 208 Some studies report improvement, whereas others show equivocal results or even deterioration. 209–214 In 2020, a study assessed the impact of SMUSs (RP-TVT) on sexual function using the PISQ-IR among 171 women at 1 year of follow-up. 208 They found that coital incontinence was reported by 56% of women before the surgery, and by 8.6% of women at the 1-year follow-up. The authors concluded that successful treatment of SUI with a SMUS significantly improves a patient’s sexual life. On the other hand, persistent incontinence appears to be the most probable cause of lack of improvement in sexual life.

One RCT140 assessed the long-term outcomes (9 years) of a SMUS (TO-TVT) using the Pelvic Organ Prolapse/Urinary Incontinence Sexual Questionnaire-12 items (PISQ-12),215 which is the predecessor of the PISQ-IR used in the SIMS trial. The sexual function score showed improvement in 61% of participants and deterioration in 34.5% of participants. The authors commented that several confounding factors may have occurred in that decade of follow-up that would inevitably affect women’s sexual function, such as advancing age, development of prolapse and menopausal vaginal dryness.

In their systematic review, Kim et al. 177 reported that four RCTs showed better QoL postoperatively with a SMUS procedure than with a SIMS procedure (OR 2.02, 95% CI 1.18 to 2.85). However, Kim et al. 177 also found no significant difference between groups in sexual function after SIMS or SMUS surgery in four RCTs (OR –0.39, 95% CI –1.87 to 1.08). The PISQ-12 and the Female Sexual Function Index were used to assess sexual function among pooled studies.

In the SIMS trial, the scores of the sexual function questionnaire (i.e. the PISQ-IR) generally improved from baseline, with no statistically significant differences between the SIMS and SMUS groups. The International Urogynecological Association revised the PISQ-IR as a condition-specific validated multidimensional tool that assesses arousal, orgasm, partner-related and condition-specific global quality, condition impact and desire. The PISQ-IR is widely used for assessing sexual function in clinical trials evaluating interventions in UI and pelvic organ prolapse; hence our use of it allows comparison of our results with those of other RCTs in the literature. 216

However, at time of the SIMS trial design, some members of the trial team felt that the PISQ-IR was quite long and intrusive. Hence, participants were randomly assigned to either complete the PISQ-IR or answer direct questions on sexual function (dyspareunia and coital incontinence). Interestingly, a relatively high percentage of participants described dyspareunia and coital incontinence at baseline. However, participants undergoing a SIMS procedure were significantly more likely to report dyspareunia and coital incontinence than SMUS participants at all follow-up time points up to 3 years [dyspareunia: 11.7% vs. 4.8% (RD 7.0%, 95% CI 1.9% to 12.1%; p = 0.010) for the SIMS and SMUS groups, respectively; coital incontinence: 11% vs. 4.8% (RD 6.0%, 95% CI –0.9% to 12.9%; p = 0.084) for the SIMS and SMUS groups, respectively]. The discrepancy between these findings and the insignificant differences in the PISQ-IR scores, between groups, casts doubt on the ability of the PISQ-IR to reliably capture these specific outcomes.

Other urinary symptoms

We used the validated ICIQ-FLUTS, which assess three domains: voiding, filling and incontinence. The results showed that the between-group differences were small: CIs were incompatible with a worthwhile difference favouring either SIMSs or SMUSs. It was reassuring to see that the filling and incontinence symptoms have shown sizeable improvements across both groups at 3 years, compared with baseline. However, the voiding domain did not show this trend: there was a slight deterioration from baseline in the SMUS group. These are not surprising findings. Voiding symptoms are not typically impaired in women with UI, and hence are not expected to vastly improve following continence surgery. In addition, continence surgery including a MUS can be potentially obstructive, leading to new voiding symptoms such as slow stream, prolonged voiding and incomplete bladder-emptying. In a relatively smaller RCT (n = 137)86 comparing an adjustable anchored SIMS (Ajust) with a SMUS (TO-TVT), at 1 year of follow-up, analysis of all domains of the ICIQ-FLUTS showed no evidence of significant differences between the groups in any of the storage, voiding or continence domains.

Our results showed that participants in the SIMS group were more likely to report improvement or cure in preoperative urgency than those in the SMUS group. We used the UPS, which is a validated simple tool for assessment of urgency. Eighteen RCTs (n = 1876 participants) in the literature compared the occurrence of de novo, or worsening, urgency symptoms between those receiving a SIMS and those receiving a SMUS, and favoured neither treatment (RR 1.03, 95% CI 0.76 to 1.38). 86,90,154,155,157–162,189,191,192,194,196,197,217,218 Pooled data of these 18 RCTs should be interpreted with caution, as the results are confounded by significant heterogeneity because of the different assessment tools used.

Observational data

We presented descriptive data on the primary outcome and secondary outcomes, and safety data, by the type of procedure received by the 537 participants who underwent surgery (Ajust, n = 62; Altis, n = 199; RP-TVT, n = 126; and TO-TVT, n = 147) (the participant who received an autologous fascial sling and the two participants who received MiniArc slings are not included). These are observational data with purely descriptive analyses and no adjustment of underlying risk factors; therefore, they should be interpreted with caution. As per trial protocol, all surgeons were asked to identify their standard procedure (SMUS: RP-TVT or TO-TVT; SIMS: Altis, Ajust or other) in advance of starting recruitment.

The patient-reported success rates observed in all groups fell over the 3-year follow-up, which is not unexpected. At 15 and 36 months, the observed success rates were 84% and 71%, respectively, for Ajust; 80% and 66%, respectively, for TO-TVT; 79% and 74%, respectively, for Altis; and 73% and 69%, respectively, for RP-TVT. The decline in success rates overtime is not unexpected. The E-TOT RCT (n = 341)65 showed a 73% patient-reported success rate for the TO-TVT at the 3-year follow-up. This was a sizeable deterioration from the 81% success rate reported for this same cohort at 1 year.

In our observational data, the impact of the procedures on urinary urgency seemed to start quite early. At 3 months, the percentages of those who reported cure of/improvement in preoperative urgency were relatively high in all groups: Ajust, 54%; Altis, 62%; RP-TVT, 62%; and TO-TVT, 58%. At 36 months, the urgency cure/improvement rates were stable for the Ajust and RP-TVT groups, and relatively reduced for the Altis and TO-TVT groups (Ajust, 59%; Altis, 55%; RP-TVT, 61%; and TO-TVT, 48%).

In the SMUS group, the observational data showed that a higher percentage of women in the RP-TVT group than in the TO-TVT group reported groin/thigh pain and use of analgesia at 3 years (groin/thigh pain: 16.7% vs. 14.3% for the RP-TVT and TO-TVT groups, respectively; use of analgesia: 6.3% vs. 3.4% for the RP-TVT and TO-TVT groups, respectively). A slightly higher percentage of women in the TO-TVT group than in RP-TVT group reported using CISC (2% vs. 0.8%, respectively), but similar percentages in both groups required surgery for voiding difficulties. This is different from the current literature, in which groin/thigh pain is believed to be predominately associated with TO-TVT. Latthe et al. 219 showed that there was a significantly higher incidence of groin/hip pain among women receiving a TO-TVT (12%) than among those receiving a RP-TVT (1%). In the TOMUS (n = 597 women),187 there was no significant difference in patient-reported success rates between the RP-TVT (62.2%) and the TO-TVT (55.8%) groups at 12 months; the incidence of voiding difficulties necessitating surgery was higher in the RP-TVT group than in the TO-TVT group (2.7% vs. 0%, respectively; p = 0.004), whereas the rate of neurological (pain) symptoms was higher in the TO-TVT group than in the RP-TVT group (9.4% vs. 4% respectively; p = 0.01). 187 The Cochrane review on SMUSs showed few RCTs reporting medium-term (1–5 years, n = 683) and longer-term (> 5years, n = 714) follow-up. 220 The patient-reported success rate at medium-term follow-up was similar between RP-TVT and TO-TVT groups (RR 0.97, 95% CI 0.87 to 1.09, and RR 0.95, 95% CI 0.80 to 1.12, respectively). In the long term, patient-reported success rates were as follows: TO-TVT, 43–92%; RP-TVT, 51–88%. Postoperative VD was less frequent following a TO-TVT procedure (RR 0.53, 95% CI 0.43 to 0.65; moderate-quality evidence). Overall rates of groin pain were higher in the TO-TVT group than in the RP-TVT group (6.4% vs. 1.3%, respectively; RR 4.12, 95% CI 2.71 to 6.27), whereas the rate of suprapubic pain was lower in the TO-TVT group than in the RP-TVT group (0.8% vs. 2.9%, respectively; RR 0.29, 95% CI 0.11 to 0.78); both were of short duration. Differences between our results and the aforementioned literature are most likely due to the observational nature of our data on RP-TVTs versus TO-TVTs and the inherent selection bias that can arise. Another potential reason is that, in the SIMS trial, surgeons used their SMUS procedure of preference (RP-TVT vs. TO-TVT), which may reflect higher level of expertise in performing that procedure.

Our observational data also showed that Altis was associated with relatively higher rates of groin/thigh pain (18%) and dyspareunia (20%) at 15 months. However, by 36 months, the groin/thigh pain rates were comparable in all groups (RP-TVT, 16.7%; Altis, 14.6%; TO-TVT, 14.3%; and Ajust, 11.3%). There were differences in analgesia use across groups at 36 months, but TO-TVT was least associated with use of analgesia (Altis, 8.5%; RP-TVT, 6.3%; Ajust, 4.8%; and TO-TVT, 3.4%). In the Altis group, there was almost a 50% reduction in dyspareunia rates at 36 months, compared with 15 months (11% vs. 20%, respectively).

In 2020, a number of small observational-size studies reported on the short-term effectiveness of various types of SIMS. 175,176,221–225 Most of the studies had small cohorts and some were not independent from the industry. One prospective cohort study, of 116 women receiving Altis, reported groin/hip/thigh pain (8%), dyspareunia (1%) and tape exposure (3.5%) at 12 months’ follow-up, but, interestingly, no further new AEs at 24 months’ follow-up. 176,225 In contrast, in a 2014 observational study175 of the safety and efficacy of Altis, 22% reported AEs at 24 months: none had mesh exposures and 6.3% had self-limiting pain. 175

Discrete choice experiment

The purpose of the DCE is to estimate trial participant preferences for different characteristics of surgical treatment for SUI (type of anaesthesia, surgery recovery time, impact on SUI symptoms, impact on daily activities, complications). Inclusion of a cost attribute within the DCE allows for an assessment of the value to women of any service configuration that varies in terms of the attributes and levels included in the DCE. This allows for an assessment of how women value the different benefit/risk trade-offs of treatment and may be useful for shared decision-making. For example, using the WTP tariffs calculated from M5 (base-case model) in Table 20, consider two different possible surgical options being evaluated (treatment A and treatment B).

Treatment A (WTP = £5721) is provided under GA (WTP = £1632), leads to no complications (WTP = £0), takes 7 days to recover from surgery (WTP = –£76 × 7 = –£532), leads to an improvement in incontinence symptoms (WTP = £8173) and means that the patient occasionally has to avoid daily activities because of a fear of leaking (WTP = –£1981). The total WTP for treatment A is £13,013.

Consider an alternative, treatment B, in which surgery is also provided (WTP = £5721), under LA (WTP = £0), with a shorter recovery time of 2 days (WTP = –£76 × 2 = –£152). Symptoms of UI are much improved (WTP = £9855), meaning that the patient rarely avoids usual activities because of a fear of leaking (WTP = –£967), but, in this case, the woman experiences mesh extrusion/erosion that requires additional treatment to resolve (WTP = –£10,351). The total WTP for treatment B is £4106.

This hypothetical example illustrates the potential value trade-offs that the DCE can inform. In this case, treatment A is preferred to treatment B because there are no complications, despite treatment B having better outcomes overall. This example illustrates that it is not always the most effective procedure in terms of preventing UI that would be preferred by women, and therefore the benefit/risk trade-offs of different procedures require careful collaborative consideration between women and their health-care providers.

To the best of our knowledge, ours is the only DCE study worldwide that allows estimation of WTP for surgical treatment of SUI. We are aware of one other study (Brazzelli et al. 35) that uses a DCE to explore UK women’s preferences for the processes and outcomes of treatment for SUI, but it does not include a cost attribute to enable WTP estimation. Brazzelli et al. 35 report the findings of an online DCE with UK patients (45% of sample) and non-patients (55% of sample), exploring women’s preferences for different surgical treatments for SUI. Although some of the attributes included in the DCE were comparable to ours (AEs and time to return to normal activities), others were not (chronic pain, risk of recurrence). For those attributes that can be compared between the studies, similar conclusions were drawn, in that patients prefer to have surgery as opposed to none. However, women in our study attached greater value to avoiding AEs than appears to be the case in Brazzelli et al. 35 Caution is required when making direct comparisons. Our study did not include ‘chronic pain’ as a specific attribute. Chronic pain is likely to be caused by AEs, for example mesh extrusion. This may explain the high WTP values associated with the avoidance of adverse events in our DCE. However, our study may not capture women’s preferences for the avoidance of all possible sources of chronic pain. Future research exploring whether or not women’s preferences for the avoidance of chronic pain are determined by the source cause of that pain would help inform future DCEs in this area. As Brazzelli et al. 35 did not include a cost attribute in their DCE, it is not possible to directly compare the valuations (WTP) elicited in our study with those elicited in theirs.

Strengths and limitations of the discrete choice experiment study

Our DCE study was based on rigorous iterative methodology, and piloting with a sample of the trial population, which led to some refinement and improvement of the final experimental design. Our study has a unique strength, compared with other DCEs or valuation exercises in the literature. The inclusion of a cost attribute in the DCE allows for the calculation of marginal rates of substitution, enabling a comparison of value attached to different characteristics of care and outcomes using a single metric (money). This enables an assessment of the benefit/risk trade-offs that women are willing to make, and is useful for shared decision-making, involving women directly in the resource allocation process. The DCE can be used to describe the total value of any treatment package that varies according to the characteristics included in the valuation exercise. Furthermore, including cost as an attribute allows for the results of the DCE to be integrated with the trial findings to enable a cost–benefit analysis to be conducted, which allows women’s preferences to be considered directly in the economic evaluation process and ensures that women’s views can be closely integrated into developing patient-centred and cost-effective policy recommendations.

A potential limitation of the DCE study is that preferences for the cost attribute are lower than has been observed in many other DCEs. This suggests that the values of the cost attribute may not have been sufficiently high enough to induce trade-offs against cost. The result is that WTP estimates for several attribute levels are beyond the maximum level value included in the DCE. This may be a cause for concern that outcomes are overvalued, and it is important to consider the validity of the estimates against other studies in the literature. As described in the literature review section, there are no other DCEs with which to compare our results. However, we are aware of a contingent valuation study that estimates WTP for resolution of incontinence problems. 226 Our DCE shows that women are willing to pay a one-off payment of £11,706 for the maximum improvement score on the PGI-I scale (very much improved). By comparison, a US contingent valuation study found that women were willing to pay US$70 (2005 values) per month for complete resolution of incontinence symptoms. Assuming an average age of 60 years and an average life expectancy of 80 years, this would suggest that women were willing to pay 70 × 12 × 20 = US$16,800. Using an online tool to inflate and convert to UK values, this would suggest that women in that study would be willing to pay £14,612 over their remaining lifetime to obtain full resolution of incontinence symptoms, a value somewhat higher that what we have estimated in our DCE. 227 Although there are many differences in study design, framing of cost and heterogeneity in payment for health care, the results nonetheless provide some reassurance that our WTP estimates are not unreasonable.

Cost-effectiveness

The base-case analysis results indicate that, over the 36-month follow-up period, on average, SIMSs cost less (–£6, 95% CI –£228 to £208) than SMUSs and SIMSs had, on average, 0.005 (95% CI –0.068 to 0.073) more QALYs than SMUSs. The QALY difference equates to just under an additional 2 days in perfect QoL accrued over the 36 months of follow-up. Neither cost nor QALY differences are statistically significant. There is a high level of uncertainty attached to these results because of the small differences in cost and, more importantly, small differences in QALYs between the groups. The cost and QALY differences are distributed in all the quadrants of the cost-effectiveness plane and there is no clear trend in these results. When the decision-maker is not willing to pay anything for an additional QALY, there is 51% chance that SIMSs are cost-effective. This increases to a 56% chance at a £20,000 WTP threshold. These results should be interpreted taking into consideration all the factors. There is no evidence that SIMSs or SMUSs are superior on the grounds of cost-effectiveness.

The intervention costs were lower for the SIMS group for most of the resources that were used in the index surgery, and the differences were statistically significant in the cost of the surgeons who were supervised, the time spent in surgery, the type of anaesthesia used (more women in the SMUS group had GA) and the time spent in recovery. The difference in costs over the 36-month follow-up period was mainly driven by the index surgery cost, as SIMSs, on average, cost more than SMUSs at the 24- and 36-month follow-ups. However, these differences in costs were not statistically significant.

The results were robust to several deterministic sensitivity analyses around the discount rate, method used to account for time from randomisation to surgery for QALY calculation and broadening the costing perspective to a societal one. In all cases, on average, SIMSs cost less, had more QALYs and thus were dominant over SMUSs. The probability that SIMSs would be considered cost-effective at the £20,000 WTP threshold value for a QALY gained ranged between 56% and 62% across the analyses conducted.

The results of the complete-case data, the estimation of QoL utilities using the condition-specific instrument ICIQ-LUTSqol and the relaxation of the assumption that women who did not have surgery had zero secondary resource use costs (i.e. these values were considered missing) indicate that they were sensitive to the assumptions made in the base-case analysis. The complete-case analysis results reported that SIMSs cost less but were less effective than SMUSs. The ICER was £2187 for the QALY loss and the probability that SIMSs would be cost-effective at the £20,000 WTP for a QALY threshold was 10%. However, these results were based on < 30% of the women in each group: 87 out of 298 in the SIMS group and 77 out of 298 in the SMUS group. SIMSs had fewer QALYs than SMUSs in the analysis that used ICIQ-LUTSqol estimates for QoL. However, the probability that SIMSs would be cost-effective remained close to that reported in the baseline analysis. The QALY values were higher than those reported in the generic instrument (EQ-5D-3L); this could be explained by the fact that the range of values that can be derived from these instruments is different. The preference-based index produced from this instrument has a substantially smaller range of values than the established generic preference-based measures of health. 128 The lowest value is 0.77, whereas that of EQ-5D-3L is negative.

The results of most of the subgroup analyses suggest that SIMSs cost less and have more QALYs than SMUSs, and the probability that SIMSs would be considered cost-effective at the £20,000 WTP threshold ranged from 10% to 97%. The probability that SIMSs would be considered cost-effective at the £20,000 WTP threshold was 54% among women who had received PFMT and 62% among women aged ≤ 65 years. For women with a urodynamic stress incontinence diagnosis, the probability that SIMSs would be considered cost-effective at the £20,000 WTP threshold was 85%; for women aged ≤ 48 years, this probability was 97%. However, for women aged > 48 years, the probability that SIMSs would be considered cost-effective at the £20,000 WTP threshold was 10%.

The cost–benefit analysis incorporates a valuation of outcomes that goes beyond health outcomes only, such as those captured in QoL measures such as the EQ-5D-3L, instead providing a more holistic measure of value that encapsulates women’s valuation of having surgery, weighing the procedural outcomes against recovery time, improvements in symptoms, risk of complications and impact on daily activities. When comparing these benefits with the NHS perspective costs over the 3 years of follow-up, the results show that both procedures generate positive net benefits, with benefit-to-cost ratios of 4.67 for SIMSs and 5.21 for SMUSs. Assuming that all interventions that generate a benefit-to-cost ratio of > 1 improve the efficiency of the health-care system, either surgery could be considered to offer value for money. However, in terms of identifying the optimal surgical procedure, we can consider that the incremental net benefits for SIMSs, compared with SMUSs (–£941) indicate that replacing SMUSs with SIMSs would not be the most efficient use of scarce NHS resources. The results are further illustrated using the scatterplots of the cost–benefit plane and the cost–benefit acceptability curves, indicating that SIMSs would be unlikely to be considered the optimal treatment strategy over the 3-year time horizon of the analysis.

Strengths and limitations of the health economic study

A key strength is that, to our knowledge, this is the first economic analysis performed with women who were followed up over 36 months, which allowed us to capture the costs and benefits related to the SIMS and SMUS interventions over a longer period. The economic analysis was also undertaken alongside the multicentre design that included women from all over the UK, thus increasing the generalisability and the validity of the results in the UK. A comprehensive microcosting that included detailed costing of the intervention device was undertaken, which adds to the generalisability of the results. Incorporation of a wider economic perspective on costs as a secondary analysis added value in terms of understanding the impact of non-health-care costs on women, their families and the economy, as a result of urodynamic stress incontinence symptoms and problems. The economic analysis was conducted using two perspectives of benefits (QALYs and WTP values). The DCE provided useful information in the valuation of UI treatment outcomes.

One of the challenges of the study was the number of missing cost and QALY data. Exploratory analysis conducted to predict missingness indicated that data were missing at random and the base-case analysis was conducted using multiple imputation data based on best practice. Another limitation of the study was that several of the women reported that they had not returned to usual activities at 4 weeks. An assumption was made to cap the days to 28 days (4 weeks after surgery), when they completed the questionnaire. This may be a conservative estimate that may have underestimated the cost of return to normal activities.

A previous cost-effectiveness analysis92 reported 12-month results that were similar to the results of this study: low and non-significant QALY differences and cost-savings for SIMSs. The study reported that SIMSs had an ICER of £48,419 cost saving per QALY loss, with an 80% probability that SIMSs would be cost-effective at the £20,000 WTP threshold. The findings of this study suggest that the SIMS follow-up costs at both 24 and 36 months were higher than those of the SMUS, and, at 36 months, the probability that SIMSs would be cost-effective reduced to 56%.

Another study35 that used a decision-analytic model to evaluate the cost-effectiveness of nine different surgical interventions for the treatment of women with SUI or stress-predominant MUI concluded that the RP-TVT was less costly and more effective than all other surgical interventions over a lifetime time horizon; therefore, it was a dominant strategy. The probabilistic results showed that RP-TVT and traditional sling have the highest probabilities of being cost-effective across all WTP thresholds over a lifetime time horizon. RP-TVT remains dominant over a 10-year time horizon in the cure model. The only major deviation from these findings was when the time horizon was reduced to 1 year. The most cost-effective surgical intervention was single-incision sling, which was similar to the results that were reported in Boyers et al. 92 There is need for long-term follow-up of these women to evaluate the long-term effectiveness of SIMSs.

The results of this study suggest that the probability that SIMSs would be cost-effective at the £20,000 WTP threshold is 56%. However, there is still some uncertainty over the longer time complication and failure rates for the devices used in the treatment of UI; therefore, long-term follow-up would be needed to establish the cost and QoL implications for these events.