Dr JE Jelovsek, Department of Obstetrics and Gynecology/A81, Cleveland Clinic, 9500 Euclid Avenue, Cleveland OH 44195, USA. Email firstname.lastname@example.org
Objective To compare the long-term efficacy of laparoscopic Burch colposuspension with tension-free vaginal tape (TVT) for the treatment of urodynamic stress urinary incontinence (SUI).
Design Long-term follow up from a prospective randomised trial.
Setting Academic tertiary referral centre.
Sample Seventy-two women with urodynamic SUI from two institutions.
Methods Subjects were randomised to either laparoscopic Burch or TVT from August 1999 to August 2002. Follow-up evaluations occurred 6 months, 1 year, 2 years, and 4–8 years after surgery.
Main outcome measures Subjects completed the Incontinence Severity Index, Urogenital Distress Inventory 6 (UDI-6), Incontinence Impact Questionnaire (IIQ-7), and Patient Global Impression of Improvement (PGI-I) scales.
Results Median follow-up duration was 65 months (range 12–88 months) with 92% completing at least one follow-up visit. Seventy-four percent of subjects had long-term (4–8 years) follow up. Fifty-eight percent of subjects receiving laparoscopic Burch compared with 48% of TVT subjects reported any urinary incontinence 4–8 years after surgery (Relative Risk (RR):1.19; 95% CI: 0.71–2.0) with no significant difference between groups. Bothersome SUI symptoms were seen in 11 and 8%, respectively, 4–8 years after surgery (P= 0.26). There was significant improvement in the postoperative UDI-6 and IIQ-7 scores in both groups at 1–2 years that were maintained throughout follow up with no significant differences between the groups.
Conclusions TVT has similar long-term efficacy to laparoscopic Burch for the treatment of SUI. A substantial proportion of subjects have some degree of urinary incontinence 4–8 years after surgery; however, the majority of incontinence is not bothersome.
Laparoscopic Burch colposuspension was introduced in the early 1990s for the surgical treatment of stress urinary incontinence (SUI) with advantages of avoiding larger incisions, shorter hospital stay, and quicker return to normal activities when compared with open Burch colposuspension. In 1996, the tension-free vaginal tape (TVT) midurethral sling was introduced and purportedly offered similar advantages including being easier to perform, less costly, and as effective as the colposuspension.1
We previously reported a randomised trial comparing the laparoscopic Burch with TVT that found a higher rate of urodynamic SUI at 1 year in subjects receiving laparoscopic Burch for the treatment of SUI.2 Postoperative subjective symptoms of all types of urinary incontinence were reported more often in the laparoscopic Burch group than in the TVT group. A recent Cochrane review compared laparoscopic Burch with tension-free midurethral sling for the treatment of SUI.3 Subjects were less likely to leak on objective testing if they received a tension-free midurethral sling when compared with subjects receiving Burch (RR: 0.92; 95% CI: 0.85–0.99), and there were no significant differences in subjective cure rates between the two procedures. However, a major limitation addressed by the Cochrane reviewers was that postoperative follow up was limited to 18 months. The Cochrane reviewers concluded that the long-term performance of laparoscopic Burch remains uncertain and while there is a trend in favour of the sling procedures in the short-term, longer term data are needed. Therefore, the objective of our study was to compare the long-term efficacy of laparoscopic Burch with TVT for the treatment of urodynamic SUI.
This is a supplemental study to our original report published in 2004.2 Both the original study and this supplemental study were approved by the institutional review boards at the Cleveland Clinic (Cleveland, OH, USA) and Good Samaritan Hospital (Cincinnati, OH, USA). All subjects were recontacted and consents obtained again for this long-term follow-up study. As described in our original report, women planning surgery for primary urodynamic SUI were invited to participate in this trial. Eligible women were enrolled between August 1999 and August 2002 at one of the two tertiary pelvic floor centres. All women provided informed consent to participate. Subjects were included if they had urodynamic SUI with abdominal leak point pressures greater than or equal to 60 cm H2O (or a positive cough stress test if the woman did not leak with the catheters in place); urethral hypermobility, defined as maximal straining cotton-tipped swab angle of greater than or equal to 30°; ability to undergo general anaesthesia and laparoscopy; no previous anti-incontinence surgery; no detrusor overactivity on urodynamics defined by a rise in the true detrusor pressure not due to bladder compliance of more than 15 cm H2O; no anterior vaginal wall prolapse to or beyond the hymen; and willingness to complete postoperative follow up and testing. A detailed description of study methods including inclusion criteria, method of randomisation, baseline evaluation of surgical interventions, and 6 month, 1-year and 2-year outcomes has been previously reported.2
For this long-term assessment, all subjects were contacted by research nurses through telephone to obtain reconsents for this long-term follow-up study and to assess subjective urinary incontinence, quality of life, and overall improvement since surgery. Subjects who were unable to be reached by telephone were mailed up to two mailings, including one certified mailing.
The primary outcome in our original report was urodynamic SUI 1 year after surgery. We chose to use subjective cure of any urinary incontinence at long-term follow up as the primary outcome of this report. We felt that the use of a subjective outcome measure as the primary outcome for this long-term analysis rather than the objective measure of urodynamics would be more relevant to clinical practice, would be more in line with the recent recommendations of National Institutes of Health (NIH) for outcomes of urinary incontinence trials,4 and would increase our ability to obtain long-term follow up. Subjects were characterised as having subjective cure by answering ‘never’ to the question, ‘How often do you experience urine leakage?’ on the Incontinence Severity Index (ISI).5 The ISI is a valid, reliable, and responsive severity index comprised of two questions: ‘How often do you experience urine leakage?’ and ‘How much urine do you loose?’ Each response receives a score and scores are multiplied together to provide an index that is categorised as dry or having slight, moderate, and severe urinary incontinence.6 The ISI was not used in the original study because it had not yet been validated; we chose to use it for this assessment given its validated ability was not only to assess the presence of urinary incontinence but also to assess the severity of incontinence. Secondary outcome measures included quality of life and overall improvement since surgery. Quality of life was determined using the condition-specific questionnaires, Urogenital Distress Inventory 6 (UDI-6) and Incontinence Impact Questionnaire 7 (IIQ-7).7 Subjects also completed a Patient Global Impression of Improvement (PGI-I) scale.8 Subjects were considered to have subjective stress and urge urinary incontinence based on their response to the UDI-6 questions, ‘Do you experience urine leakage related to physical activity, coughing, or sneezing’ and ‘Do you experience urine leakage related to a feeling of urgency’, respectively. Bothersome symptoms were defined as subjects being bothered ‘moderately’ or ‘quite a bit’. Surgical outcomes also included reoperation rates including injections of bulking agent, bladder surgery, and use of medications. All methods, definitions, and units conform to the standards recommended by the International Continence Society unless otherwise noted.9
The proportion of subjects reporting any or bothersome urinary incontinence was compared using Pearson chi-square statistic. Only subjects who had data available at long-term follow up were included for this primary analysis. Kaplan–Meier survival curves were generated for the development of subjective incontinence symptoms (any and bothersome SUI), and comparisons were made using the log-rank test. Subjects were considered censored if they were dry or had nonbothersome SUI at their most recent follow up in each survival analysis. All follow-up assessments were considered in the survival analysis. Other continuous normally distributed data were compared using Student’s t test, the Wilcoxon rank sum test for continuous non-normally distributed data, and Pearson chi-square statistic or Fisher’s exact test for categorical data including any or bothersome symptoms, reoperation rates, and use of medications. The paired t test was used for within-group comparisons while repeated measures analysis of variance (ANOVA) were used for between-group comparisons including comparing quality-of-life scores between groups. P < 0.05 was considered statistically significant. Subjects were analysed in the group to which they were randomised consistent with the intent-to-treat principle.
A sample size calculation was performed in our original report demonstrating that 65 women in each group were necessary to have 80% power to detect a 20% difference in continence rates with an α of 0.05 and a loss to follow-up rate of 10%.2 The original trial was stopped early because of slow recruitment and lack of funding. Statistical analysis was performed using JMP 6.0 (SAS Institute, Cary, NC, USA).
Seventy-two subjects were randomised; 36 into each group (Figure 1). One subject in the laparoscopic Burch group withdrew from the study and one subject was lost to follow up, leaving 34 subjects in the laparoscopic Burch group. In the TVT group, three subjects were lost to follow up. There was one unrelated death secondary to cardiovascular disease prior to follow up that was not associated with the surgery leaving 32 subjects in the TVT group. Twenty-eight subjects in the laparoscopic Burch group and 25 subjects in the TVT were contacted for this supplementary study. Median follow-up duration was 65 months (range 12–88 months) with 92% completing at least one follow up visit and 74% of subjects with long-term (4–8 years) follow up. Baseline demographic data were similar in both groups and have been previously reported with no significant differences between groups.2 Mean (±SD) age was 54 ± 10 years with 92% of subjects being Caucasian and 5% having had previous urogynaecological surgery.
Fifty-seven percent (16/28) of subjects receiving laparoscopic Burch compared with 48% (12/25) of TVT subjects (RR: 1.19; 95% CI: 0.71–2.0) reported any urinary incontinence 4–8 years after surgery with no significant difference between the groups (Table 1). However, only 11% (3/28) of subjects in the laparoscopic Burch group and 8% (2/25) of subjects in the TVT group reported bothersome SUI 4–8 years after surgery (RR: 1.3; 95% CI: 0.24–7.4). Based on the ISI, 22% (6/28) of subjects in the laparoscopic Burch group and 32% (8/25) in the TVT group had moderate to severe incontinence of urine 4–8 years after surgery (RR: 0.67; 95% CI: 0.27–1.66) with no significant difference between the groups.
Table 1. Severity of urinary incontinence, pad use, and therapy 4–8 years after subjects receive TVT or laparoscopic Burch colposuspension for SUI
TVT (n = 25)
Laparoscopic Burch (n= 28)
Data are presented as n (%).
Reoperations included: transurethral collagen injections (1) and urethrolysis (1).
The time to the development of any urinary incontinence and bothersome SUI after surgery is demonstrated in Figures 2 and 3, respectively. Subjects who received a laparoscopic Burch developed any recurrent urinary incontinence at a median time of 52 months as against more than 87 months in the TVT group with no significant difference between groups (P= 0.6). Subjects who received a laparoscopic Burch developed recurrent bothersome SUI at a median time of more than 86 months versus more than 87 months in the TVT group with no significant difference between groups (P= 0.23).
Seventy-two percent of subjects in the laparoscopic Burch group and 68% in the TVT group reported they were, ‘much better’ or ‘very much better’ compared with before their surgery, while 7 and 4% reported they were much or very much worse on the PGI-I. Eighty-eight percent of subjects in the laparoscopic Burch group reported that if they had to do it all over again, they would choose the same treatment versus 79% in the TVT group. Only 4% of subjects in both groups underwent reoperation for SUI at some point during the follow-up period. Overall, 25% of subjects were using pads postoperatively for urinary leakage and approximately 10% of subjects were taking anticholinergic medication with no significant differences between groups. Quality of life significantly improved on the UDI-6 and IIQ-7 in both groups by 1–2 years after surgery with no differences between groups. This improvement was maintained throughout the length of follow up (Figures 4 and 5).
Our data suggest that only 52% of subjects receiving TVT and 43% of subjects receiving laparoscopic Burch are completely dry 4–8 years after surgery. These proportions appear low at first glance; however, they are consistent with larger randomised trials for surgical treatment of SUI. In 2004, Ward and Hilton10 reported that the TVT was as effective as open colposuspension for the treatment of urodynamic SUI with 63% of subjects in the TVT group and 51% of the colposuspension group having a negative 1-hour pad test 2 years after surgery (OR: 1.67; 95% CI: 1.09–2.58). Five years after surgery, only 39% in the TVT group and 46% in the open colposuspension group reported no leakage under any circumstances.11 Carey et al.12 reported that 66% of subjects undergoing either a laparoscopic or open Burch did not demonstrate postoperative stress incontinence or detrusor overactivity. Cure rates from this and other randomised trials appear to be slightly lower than the long-term results of a prospective cohort from Nilsson et al.,13 in which 81% of subjects were subjectively cured 7.6 years after receiving TVT. In our study, 48 and 57% of subjects in the TVT and laparoscopic Burch groups, respectively, reported ‘any’ leakage of urine 4–8 years after surgery. Importantly, however, 50% of those with leakage were classified as having slight urinary incontinence (‘leakage of drops one or several times per month’ or ‘more than drops less than once per month’) and only 18% had bothersome SUI. Moreover, approximately 80% of subjects at long-term follow up reported that if they had to do it all over again, they would choose the same treatment, and reoperations were exceedingly low. This study suggests that while a substantial proportion of subjects have some urinary leakage 4–8 years after a TVT or laparoscopic Burch, it tends to be mild and nonbothersome. Most subjects have a substantial improvement in their incontinence symptoms and health-related quality of life, which is maintained for 4–8 years.
In our initial report, TVT had greater objective and subjective cure rates than laparoscopic Burch 1 year after surgery.2 Some differences between these long-term follow-up results and our 1-year results may be explained by using different definitions of cure. The primary outcome of our 1-year results was objective cure, defined as no evidence of stress leakage during postoperative urodynamic studies. Additionally, the subjective outcomes in our initial report were assessed using the UDI only and therefore were not stratified by severity or degree of bother as they were in this report using the ISI.2 Another explanation is that differences do exist between laparoscopic Burch and TVT in the short-term but with longer follow up, these differences diminish.
The benefits of the laparoscopic approach are that it is a modification of an existing, well-established procedure that offers similar advantages to TVT by avoiding a large incision, quicker return to activities, and shorter hospital stay when compared with the open approach. Since the open colposuspension is a familiar procedure to most gynaecologists, many surgeons attempt to gain the benefits of minimally invasive approach by performing colposuspension through the laparoscopic approach. The main disadvantages are that the laparoscopic Burch is not easily learnt by surgeons due to the steep learning curve required for laparoscopic suturing. It also includes the requirement for general anaesthesia, abdominal entry, pneumoperitoneum, and three or four abdominal incisions. These barriers, in combination with our previously published experience demonstrating a higher rate of urodynamic SUI at 1 year in the subjects undergoing laparoscopic Burch compared with TVT, dramatically reduced the number of laparoscopic colposuspensions performed at our institution.2
However, the laparoscopic colposuspension continues to be a useful minimally invasive alternative to midurethral slings in certain scenarios, such as when the surgeon needs to obtain intraperitoneal visualisation so as to evaluate an adnexal mass or chronic pelvic pain; when a paravaginal defect repair and/or vaginal vault suspension is being performed to surgically correct mild to moderate, symptomatic, anterior and/or apical pelvic organ prolapse; to avoid bowel perforation during blind TVT trocar passage in a woman with an unrepaired inguinal hernia; during performance of laparoscopic sacral colpopexy to reduce the risk of postoperative SUI;14 or when the woman does not desire permanent mesh placement.
The strengths of this study lie in its randomised study design, our use of up-to-date valid and reliable outcome measures, and the long-term follow-up times. The use of subjective outcome measures rather than objective measures allowed us to obtain a 74% follow-up rate 4–8 years after surgery and is more in compliance with recent NIH recommendations. A difference between the groups was demonstrated in our original report despite recruiting only half the number of subjects in each group before stopping the trial early due to poor recruitment and funding despite the original power calculation requiring 65 women in each group. It is possible that we were underpowered to detect a difference in subjective symptoms between groups. A posthoc power calculation demonstrated an 80% power to detect a 31% difference in the rates of any urinary incontinence or 27% difference in the rates of bothersome SUI between both groups. However, our statistical tests were underpowered to detect statistically significant real differences in cure rates between the two groups, if they were less than 31 and 27%, respectively. The implications are that we may conclude there is no difference when in fact a real difference between groups exists.
TVT has similar long-term subjective efficacy compared with laparoscopic Burch for the treatment of urodynamic SUI. Quality of life significantly improves after both procedures and reoperations are uncommon. A substantial proportion of subjects have some degree of urinary incontinence 4–8 years after surgery; however, the majority of incontinence is not bothersome.
Grant from the Minimally Invasive Surgery Center at the Cleveland Clinic.
Disclosure of interests
M.M.K. is the consultant for Ethicon Women’s Health.
M.F.R.P. is a paid instructor for Ethicon Women’s Health.
Contribution to authorship
J.E.J., participated in the design, implementation, and writing of the manuscript and that I have seen and approved the final version. M.D.B., participated in the design, implementation, and writing of the manuscript and that I have seen and approved the final version. M.M.K., participated in the design, implementation, and writing of the manuscript and that I have seen and approved the final version. M.D.W., participated in the design, implementation, and writing of the manuscript and that I have seen and approved the final version. M.F.R.P., participated in the design, implementation, and writing of the manuscript and that I have seen and approved the final version.
Commentary on ‘Randomised trial of laparoscopic Burch colposuspension versus tension-free vaginal tape: long-term follow up’
The trial by Jelovsek et al. is underpowered. So why did we publish it? The researchers’ sample size calculation demonstrated that they needed 130 women in the trial to have an 80% power to detect a 20% difference in continence rates with an alpha of 0.05 and a loss to follow-up rate of 10%. However, the original trial was stopped after the randomisation of 72 women because of slow recruitment and lack of funding. The researchers, thus, managed to recruit only just over half (72/130, 55%) of the required numbers. Two of the three peer-reviewers and some of the editors felt that such an underpowered trial may not be suitable for publication. They are not alone in this view: prominent methodologists have commented that an underpowered study may be ‘scientifically useless, and hence unethical in its use of subjects and resources’ (Altman, BMJ 1980;281:1336). We, in BJOG, do not necessarily agree, and we are not alone either (Edwards, Lancet 1997;350:804).
While we encourage adequately powered trials and confirm that such trials will generally have a better chance of publication with BJOG, an underpowered trial will not be robotically rejected. There are at least three reasons for this.
First, when several small (underpowered) trials are combined in a systematic review, they may acquire sufficient power to answer the research question with precision (i.e. narrow confidence interval) and thus minimise the risk of ‘false-negative’ trial results from the individual small studies. It is for this reason that we encourage systematic reviews in BJOG. However, we appreciate that publication bias can skew the pooled summaries in a systematic review; in a bid to minimise this, methodologists recommend trial registration, and BJOG offers researchers the opportunity to register and publish trial protocols.
The second reason is that even as we recognise a clear-cut answer (from a large trial) to be better than an uncertain answer (from a small trial), we welcome an uncertain answer (from a small trial) compared with no answer at all (from the absence of any trials). The choice is often not between a large versus a small trial, but a small versus no trial. In this situation, to deny small trials the opportunity to add to the pool of medical knowledge is unreasonable.
The third reason relates to feasibility. Often, small trials are conducted by enthusiastic clinicians or academic clinicians who work with limited resources, and ‘soft monies’ and pursue with vigour single centre small trials. Many of us in BJOG belong to this stock, and we feel we should encourage such clinicians. Our response to mega-trialists who may want to pour cold water on such endeavours is this: remind yourselves that the path to megatrials is often paved by a series of small trials from enthusiastic clinicians (Edwards, Lancet 1997;350:804). Without small (underpowered) trials, it is unlikely that there would be as many robust megatrials.
This mini commentary is not intended to open the flood gates on underpowered trials; we will always favour adequately powered studies to underpowered equivalents; however, if we are convinced that a study remained underpowered despite concerted efforts at recruitment, such a trial is likely to get a fair hearing in BJOG. The trial by Jelovsek et al. is judged to fit this principle.