The overlying histogram plot shows the distribution of annual surgeon volume. Predicted probability (solid line) was calculated using a logistic regression model with the same covariates as the primary analysis and was based on the observed sample means and proportions for each of the covariates. Dashed lines represent the 95% CI.
eTable 1. STROBE Checklist for Cohort Studies
eTable 2. CCI Codes Used to Identify Any Synthetic Mesh-Based Procedure for SUI
eTable 3. CCI Codes Used to Define Our Composite Outcome of Mesh Removal or Revision
eTable 4. Coding Definitions for Study Covariates
eTable 5. One-Year Rate of Mesh Removal or Revision by Operative Year
eTable 6. Summary of Sensitivity Analyses for Primary and Secondary Exposures
eFigure 1. Flow Diagram of Cohort Selection
eFigure 2. Number of Mesh-Based Procedures for SUI Performed Yearly, 2002-2012
eFigure 3. Number of Surgeons Considered High Volume by Specialty, 2002-2012
Welk B, Al-Hothi H, Winick-Ng J. Removal or Revision of Vaginal Mesh Used for the Treatment of Stress Urinary Incontinence. JAMA Surg. 2015;150(12):1167-1175. doi:10.1001/jamasurg.2015.2590
Synthetic mesh slings are the most common surgical treatment for female stress urinary incontinence (SUI). However, the US Food and Drug Administration has released warnings that question the safety of vaginal mesh.
To measure the incidence of mesh removal or revision after SUI procedures and to determine whether significant surgeon and patient risk factors exist.
Design, Setting, and Participants
Population-based retrospective cohort study that included all adult women undergoing an incident procedure for SUI with synthetic mesh in Ontario, Canada, from April 1, 2002, through December 31, 2012 (N = 59 887). The end of potential follow-up was March 31, 2013. Data were analyzed from November 1, 2014, through February 28, 2015.
Yearly volume of mesh-based procedures for SUI performed by the treating surgeons and their surgical specialty.
Main Outcomes and Measures
The primary outcome was a composite of surgical procedures related to removal or revision of mesh slings (owing to erosion, fistula, pain, or retention). We hypothesized a priori that surgeon volume would be inversely correlated with complications.
Among the identified 59 887 women who underwent a mesh-based procedure for SUI, the median age was 52 (interquartile range [IQR], 45-63) years. High-volume surgeons (≥75th percentile of yearly mesh-based procedures) were less likely to perform a simultaneous hysterectomy (performed in 11.5% vs 16.5% of patients; standardized difference, 0.14), were more likely to work in an academic center (28.9% vs 16.3% of patients; standardized difference, 0.30), and saw the patient less frequently in the year before the procedure (median, 2 [IQR, 1-3] vs 3 [IQR, 2-4] visits; standardized difference, 0.26). Complications were treated in 1307 women (2.2%), and the 10-year cumulative incidence rate was 3.29 (95% CI, 3.05-3.53). In our multivariable survival model, patients of high-volume surgeons had a significantly lower risk (95% CI) for experiencing our composite outcome (hazard ratio [HR], 0.73 [0.65-0.83]; absolute risk reduction, 0.63% [0.36%-0.92%]; P < .01). Gynecologists were not significantly associated with more complications compared with urologists (HR, 0.94 [95% CI, 0.83-1.08]; P = .38). Among our secondary exposures of interest, multiple mesh-based SUI procedures increased the risk for complications (HR, 4.73 [95% CI, 3.62-6.17]; P < .01). However, traditional high-risk patient features did not increase the risk (HR, 0.58 [95% CI, 0.08-4.13]; P = .59).
Conclusions and Relevance
Ten years after SUI mesh surgery, 1 of every 30 women may require a second procedure for mesh removal or revision. Patients of lower-volume surgeons have a 37% increased likelihood of having a complication. These findings support the recommendations of the US Food and Drug Administration related to the use of vaginal mesh for treatment of SUI.
Female stress urinary incontinence (SUI) is a common condition1 that is often treated with surgery when conservative management options are unsuccessful. An estimated 1 in 7 women will undergo surgery for SUI during their lifetime.2
Synthetic mesh has been used since the 1960s for the treatment of SUI. However, the introduction of commercial midurethral sling kits in the mid-1990s led to the rapid adoption of this procedure and an overall increase of use in procedures to treat SUI.3 Midurethral sling implants constitute the most prevalent surgical treatment for SUI, and they have supplanted previously common operations, such as retropubic suspensions and bladder-neck sling implants.3,4 Although the efficacy is similar among all of these procedures,5 the midurethral sling is preferable owing to a significantly shorter operative time and hospital stay and a quicker patient recovery.6 Long-term follow-up of some of the first women to receive a midurethral sling (>17 years ago) suggests that patients have excellent results and minimal complications.7
However, in recent years, significant concerns have been raised about the safety of vaginal mesh (used for procedures to treat SUI and pelvic organ prolapse).8 Several case series have suggested that mesh-based SUI procedures can lead to chronic pain,9 urethral fistula,10 significant voiding dysfunction,11 and mesh erosions into the urethra or the vagina.6 The treatment of these complications may require surgical revision or removal of the mesh and can leave the patient with substantial residual symptoms and emotional distress.8,12 In the United States, more than 50 000 women have joined class action lawsuits for transvaginal mesh complications resulting from SUI and prolapse procedures.13
The US Food and Drug Administration (FDA)14 and Health Canada15 have released warnings stating that complications caused by vaginal mesh are not uncommon, that surgeons should obtain specialized training in their use, and that the risk factors for complications have not been well identified. In light of these statements, we conducted a population-based cohort study with the following objectives: (1) to measure the incidence of mesh removal or revision after a mesh sling procedure for SUI, (2) to assess the potential effect of surgeon volume and specialty, and (3) to determine whether specific risk factors exist for mesh removal or revision.
We conducted this study through the Institute for Clinical Evaluative Sciences according to a prespecified protocol. This study was approved by the research ethics board at Sunnybrook Hospital, Toronto, Ontario, Canada. Individual patient consent was waived by the research ethics board, in keeping with provincial privacy regulations. All data were deidentified.
We performed a population-based, retrospective cohort study of all adult women who underwent a mesh sling implant for SUI from April 1, 2002, through December 31, 2012, in the province of Ontario (which has a population of approximately 13 million people, all of whom have unlimited access to a single universal health care system). The STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines were followed (eTable 1 in the Supplement).16
Our study population, outcomes, exposures, and covariates were obtained using 3 linked databases. The Discharge Abstract and Same Day Surgery databases of the Canadian Institute for Health Information (CIHI-DAD/SDS) capture all diagnostic and procedural information for patients in Ontario who have a hospital admission or a surgical procedure (inpatient and outpatient). The Ontario Health Insurance Plan database captures all health claims for inpatient and outpatient care rendered by physicians. Finally, the Registered Persons Database contains demographic information on all individuals in Ontario. The databases are considered complete for all study variables, aside from physician specialty, which was unknown for 150 of 1068 of relevant physicians (14.0%). All databases have patient-level data available from October 1, 1992, through March 31, 2013. The accuracy of these databases has been previously measured,17- 19 and data elements are valid and reliable.
Canadian Classification of Health Intervention (CCI) codes from the CIHI-DAD/SDS database were used to identify our cohort of adult women who underwent a mesh-based surgical procedure for SUI during our study period (eTable 2 in the Supplement). The date of the procedure was considered the index date, and patients were followed up until death, emigration from the province, the first occurrence of an outcome, or the end of the study (March 31, 2013). We excluded women who may have had a mesh-based procedure for SUI before April 1, 2002 (n = 56), and women who had a mesh-based procedure for pelvic organ prolapse before or during our study period (n = 1101) (eFigure 1 in the Supplement).
We used CCI codes to identify our composite outcome of the first reoperation for SUI mesh-related complications (eTable 3 in the Supplement). We included specific codes related to the surgical removal or revision of vaginal or urethral mesh or removal of a foreign body, endoscopic treatment of a urethral foreign body or mesh encrustation, urethrolysis, or repair of a urethrovaginal fistula. Rare acute surgical complications of SUI mesh procedures, such as adjacent organ or major-vessel injury, hematoma, and necrotizing infections, have been characterized previously11 and were not included in this study.
We prespecified surgeon volume and specialty as the primary exposures of interest, with a hypothesized inverse relationship between volume and complications and a hypothesized difference in the complication rate between urologists and gynecologists. Surgeon volume was defined as the number of mesh-based procedures for SUI performed per year and was determined using unique physician identifiers in the CIHI-DAD/SDS database. High-volume surgeons were defined as being at the 75th percentile or greater for mesh implants for SUI in a given year, and surgeons could move between high- and low-volume categories with each successive year. Surgeon specialty was classified as urology, obstetrics/gynecology, or undetermined using the CIHI-DAD/SDS database. Secondary exposures included patients at high risk for mesh-related complications (a binary variable indicating a previous urinary fistula, urethral diverticulum, urethral injury, or pelvic radiotherapy20,21) and those with multiple mesh slings. Other covariates (eTable 4 in the Supplement) were age, obesity (body mass index [calculated as weight in kilograms divided by height in meters squared] >40), diabetes mellitus,22 combined hysterectomy or non–mesh-based repair of pelvic organ prolapse, provincial region, academic or community hospital, rural residence,23 general medical comorbidity (health care usage served as a proxy for overall health and was determined from the Aggregated Diagnostic Groups resource utilization bands from the validated Johns Hopkins University Adjusted Case Groups case-mix system24,25), socioeconomic status (using the Ontario Marginalization Index, a measure of regional marginalization, which served as a proxy for individual socioeconomic status26), and health care usage in the year before the mesh-based procedure for SUI.
Data analysis was performed from November 1, 2014, through February 28, 2015. All baseline characteristics are reported as frequencies (number [percentage]) or medians (interquartile range [IQR]). We compared patient baseline characteristics between high- and low-volume surgeons using standardized differences.27 This method identifies clinically meaningful differences (standardized difference, >10%) as opposed to statistically significant differences (which are likely with large sample sizes).27
Our primary analysis was a multivariable survival analysis performed using the PROC PHREG procedure in SAS software (version 9.3; SAS Institute Inc). Results are reported as hazard ratios (HRs) with 95% CIs and P values (P < .05 was considered significant). We included surgeon volume and specialty, high-risk patients, and the number of mesh slings (modeled as a time-varying covariate to account for different periods during which the patient would be at risk owing to an increasing number of consecutive procedures). We adjusted for age, obesity, diabetes mellitus, other pelvic surgery, Aggregated Diagnostic Groups resource utilization bands (as an ordinal variable), provincial region, hospital type, rurality, and socioeconomic status. The Cochrane-Armitage test for linear trends28 was used to assess for significant changes in the 1-year event rate over time.
A number of post hoc sensitivity analyses were performed to confirm the effects of our primary and secondary exposures. First, we restricted the patient population to those with CCI codes suggesting a midurethral sling implant (as opposed to other potential mesh-based surgical procedures for SUI, such as bladder-neck mesh sling implants). Second, we censored patients on future non–mesh-based SUI surgery. Third, we varied our definition of high-volume surgeon to those who performed procedures at the 50th percentile or greater among all surgeons performing the mesh-based procedures in a given year. Fourth, we explored the volume outcome relationship graphically. Fifth, we considered multiple mesh slings as a count variable. Finally, we accounted for patient clustering within individual surgeons using a hierarchical model.
We identified 59 887 women who underwent mesh-based procedures for SUI during the study period. Median follow-up was 4.4 (IQR, 2.4-6.9) years. One thousand seven hundred forty women (2.9%) were censored for death; 915 women (1.5%), for emigration.
The surgical procedures were performed by 1068 unique surgeons (293 urologists [27.4%], 625 gynecologists [58.5%], and 150 unspecified [14.0%]). The overall number of procedures increased from 2002 through 2009 and then declined (eFigure 2 in the Supplement). The numbers of high- and low-volume surgeons are shown in eFigure 3 in the Supplement. Patients were classified as having received a sling from a high-volume surgeon (≥75th percentile for the yearly volume of surgeons performing mesh-based procedures for SUI) or a low-volume surgeon (<75th percentile), and the baseline characteristics of our cohort are shown in Table 1. The median cutoff for the number of yearly mesh-based procedures for SUI that defined high-volume surgeons during each year of the 10-year study was greater than 16 (IQR, 13-18). High-volume surgeons were less likely to perform a simultaneous hysterectomy (11.5% vs 16.5%; standardized difference, 0.14), were more likely to perform surgery in an academic center (28.9% vs 16.3%; standardized difference, 0.30), and saw the patient less frequently in the year before the surgery (median, 2 [IQR, 1-3] vs 3 [IQR, 2-4] visits; standardized difference, 0.26). We found significant regional variation, with patients in some regions being more or less likely to be treated by a low-volume surgeon.
Overall, 1307 women (2.2%) underwent mesh removal or revision a median of 0.94 (IQR, 0.35-2.49) years after receiving a mesh implant for SUI (Table 2). The sling complication was treated by the same surgeon responsible for the original procedure in 812 of the 1307 cases (62.1%). The cumulative incidence rate of the complications of interest increased from 1.17 (95% CI, 1.09-1.27) at 1 year to 3.29 (95% CI, 3.05-3.53) at 10 years (Table 3). Unadjusted analysis demonstrated that patients of low-volume surgeons had a 37% (95% CI, 17%-49% [P < .01]) higher relative risk and a 0.63% (95% CI, 0.36%-0.92%) increased absolute risk for mesh removal or revision compared with patients treated by high-volume surgeons (HR, 0.73 [0.65-0.83]); the effect of surgical specialty was not significant. Similarly, in our multivariable model, patients of low-volume surgeons were significantly more likely to experience the composite outcome (HR, 1.37 [95% CI, 1.21-1.55]; P < .01), and the difference between surgical disciplines was not significant (Table 4).
For our secondary exposures, 1252 women (2.1%) had multiple mesh sling implants (of whom 1191 [95.1%] had 2 and 61 [4.9%] had ≥3), and 73 women (0.1%) had potential risk factors for mesh-related complications. Among women with multiple mesh-based procedures for SUI, the absolute risk for mesh removal or revision was 4.87% (95% CI, 3.86%-6.06%). In multivariable modeling, these women had a 4.73-fold increased hazard of this complication (95% CI, 3.62-6.17 [P < .01]) and an absolute risk increase of 2.8% (95% CI, 1.7%-4.1%). The HR among women who had risk factors for mesh removal or revision was not significant (0.58 [95% CI, 0.08-4.13; P = .59]). The rate of intervention for SUI-related mesh revision or removal during the first postoperative year increased significantly during the study period from 9.74 per 1000 person-years in 2002 to 14.84 per 1000 person-years in 2012 (P < .001; eTable 5 in the Supplement).
The results of the sensitivity analyses are described in eTable 6 in the Supplement. The statistical significance and direction of effect for our primary and secondary exposures across all of these analyses were consistent. The probability of experiencing an SUI-related mesh complication based on the surgeon’s annual procedure volume is shown in the Figure.
Mesh-based slings are appropriate for most female patients with SUI.29,30 For women with complications, surgical intervention is considered the standard of care for mesh erosions into the urinary tract or urinary fistula and for most women with vaginal mesh exposure, significant voiding dysfunction, or severe pelvic pain after surgery.6,9- 11 Over time, early surgical interventions have become more common (eTable 5 in the Supplement), perhaps as a result of increased patient awareness31 and improved physician confidence and experience with operative management of mesh complications. We have demonstrated that among our cohort of women undergoing vaginal mesh-based procedures solely for SUI, 2.2% underwent operative intervention for these complications (cumulative incidence rate, 3.29 at 10 years). Our results are consistent with an analysis of data from US health maintenance organizations32 that reported a 9-year rate of urethrolysis or mesh removal of 3.7% and a meta-analysis of clinical trial data33 that determined an overall proportion of patients requiring any secondary surgery of 3.2%. Although the FDA in the past has treated all vaginal mesh implants as equivalent, the intervention rates for mesh-based complications in procedures for SUI appear to be lower than those associated with procedures for pelvic organ prolapse.34
A surgeon volume of mesh-based procedures for SUI of less than the 75th percentile was associated with a 37% increased risk for surgically managed mesh-related complications. Although surgeon volume has been shown to be important for complex operations, such as oncologic and cardiac surgery,35 its role in less demanding procedures is not well defined. The vaginal mesh regulatory notifications and expert opinion suggest that surgeons should obtain specialized training and experience in vaginal mesh surgery8,14,15; the demonstrated volume-outcome relationship (Figure) in this study supports this assertion. Urologists and gynecologists have very different surgical training and day-to-day practices, and previous studies36 have suggested that differences exist in SUI outcomes between these groups of physicians. We hypothesized that complication rates may differ between these 2 groups; however, our hypothesis was not proved. This finding suggests that procedure-specific knowledge and experience is important for surgery to treat SUI rather than general operative training.
The prospective assessment of risk factors for SUI-related complications of mesh implants requiring surgical intervention is challenging given their rarity and prolonged time to presentation. Registries, such as the FDA Manufacturer and User Facility Device Experience database, and case series have significant reporting bias.11 However, observational administrative data studies are well suited to address this question. We demonstrated that undergoing 2 or more mesh-based procedures for SUI is associated with an almost 5-fold increased risk for complications. This novel finding should temper the enthusiasm of case series that suggest that the use of multiple synthetic slings is safe and efficacious.37,38 This finding is particularly important because undergoing additional mesh-based sling implant procedures (presumably for recurrent SUI) is the primary practice pattern in our cohort (and others39): after an initial mesh sling implant, and before surgery for a mesh-related complication, 1307 women underwent an additional mesh sling implant procedure, whereas only 147 women underwent a secondary non–mesh-based procedure for SUI. We did not demonstrate a significant increased risk among women who are considered to be at higher risk for complications from a mesh-based procedure for SUI.20 This finding is likely a result of the very small number of these patients within our cohort, which in itself is reassuring in terms of patient selection.
The other covariates in our study also deserve discussion. Being younger was associated with a higher likelihood of mesh removal or revision, possibly as a result of more aggressive surgical management of complications32 and an increased likelihood of dyspareunia. Increased comorbidity was associated with increased mesh removal or revision, which may be caused by impaired healing and preexisting voiding dysfunction from concurrent medical conditions. The additional dissection, trauma to the pelvic nerve plexus, and postoperative change to the vaginal anatomy may account for the increased risk for complications observed with a simultaneous hysterectomy and mesh-based procedure for SUI.32,34 The increased risk for mesh removal or revision associated with academic hospitals may be owing to an increased clinical vigilance for complications or to unadjusted case-mix differences compared with nonacademic hospitals. The involvement of surgical trainees may also play a role because an increased risk for morbidity with trainee involvement has been demonstrated across a large spectrum of surgical procedures40 and specifically among urology residents performing urogynecologic procedures.41 The reduced risk observed with simultaneous prolapse surgery may be a marker of more favorable vagina exposure at the time of sling placement. The significant regional variation across Ontario (despite multivariable adjustment) may be owing to regional preferences for specific commercial mesh slings, different training and mentoring for practicing surgeons learning the procedure, and varying access to tertiary referral practices for complex issues, such as pain or voiding dysfunction after mesh sling procedures.
Our study has several strengths. This population-based study addresses multiple key questions from the FDA and Health Canada and substantiates their recommendation that surgeons should obtain training and experience in the use of vaginal mesh implants for SUI. Ontario’s health care databases and universal health care system provide a large and inclusive population of women treated by a variety of surgeons and with a variety of mesh types and mesh-based procedures. This setting maximizes the precision and generalizability of our results compared with the results of randomized clinical trials that rely on highly selected patients treated at specialist centers. Across several sensitivity analyses, the magnitude and significance of our primary and secondary exposures were consistent.
The limitations of this study should also be acknowledged. Despite measuring several important covariates, observational studies may have residual confounding; for example, the degree of incontinence, smoking status, and type of mesh could not be measured. Although the CCI codes as a whole defined our cohort and primary outcome well, we could not reliably identify specific sling types (such as retropubic or transobturator slings or mini-slings) or specific complications. Some outcomes in our study may not have had a direct causal link with the mesh-based procedure for SUI. Finally, we relied on surgical removal or revision of mesh as our outcome. This measure captures the most serious complications; however, we have likely underestimated the true absolute rate of mesh-related complications because we were not able to include asymptomatic patients, those with nonsurgical complications, or those treated outside the operating room.
Among adult women undergoing mesh-based procedures for SUI, the risk for secondary surgery for specific complications is low (cumulative incidence rate, 3.29 at 10 years). Patients of lower-volume surgeons are 37% more likely to require surgery for mesh complications. These findings support the regulatory statements that suggest that patients should be counseled regarding serious complications that can occur with mesh-based procedures for SUI and that surgeons should achieve expertise in their chosen procedure. Multiple mesh-based procedures for SUI are a novel risk factor associated with an almost 5-fold higher rate of mesh removal or revision, and the safety of this practice should be studied further.
Accepted for Publication: May 27, 2015.
Corresponding Author: Blayne Welk, MD, MSc, Department of Surgery, Western University, Room B4-667, St Joseph’s Health Care, 268 Grosvenor St, London, ON N6A 4V2, Canada (firstname.lastname@example.org).
Published Online: September 9, 2015. doi:10.1001/jamasurg.2015.2590.
Author Contributions: Dr Welk and Ms Winick-Ng had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Welk.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Welk.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Welk, Winick-Ng.
Study supervision: Welk.
Conflict of Interest Disclosures: Dr Welk reported receiving investigator-initiated grant funding from Astellas Canada. No other disclosures were reported.
Funding/Support: The Institute for Clinical Evaluative Sciences (ICES) is supported by an annual grant from the Ontario Ministry of Health and Long-term Care; the ICES site at Western University is supported by an operating grant from the Academic Medical Organization of Southwestern Ontario.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The opinions, results, and conclusions are those of the authors, and no endorsement by the ICES, Ontario Ministry of Health and Long-term Care, or Academic Medical Organization of Southwestern Ontario is intended or should be inferred.