Systematic review and economic modelling of the effectiveness and cost-effectiveness of non-surgical treatments for women with stress urinary incontinence

Continence mechanisms in health

Efficient mechanisms have evolved to ensure reliable urine storage and complete bladder emptying at socially convenient times. Mechanisms that prevent urine leakage involve the bladder, the urethra and the pelvic floor muscles, together with their controlling nerve pathways. The bladder is a highly compliant organ, allowing the storage of increasing quantities of urine without rise in pressure, a property underpinned by passive stretch and active relaxation of its smooth muscle (detrusor). Active central nervous control mechanisms in the pons and cerebrum inhibit detrusor contraction despite increasing sensation of bladder fullness until micturition is appropriate. Urethral mechanisms promoting continence are less well understood, but are thought to involve tonic contraction of smooth muscle in the urethral wall, together with a tight seal formed by the urethral lining (mucosa) and the highly vascular submucosal layer. Contraction of the pelvic floor striated muscles acts as an additional guarding mechanism, compressing the urethra and preventing leakage during actions that raise intra-abdominal pressure.

These features maintain continence during the storage phase of the micturition cycle by ensuring that bladder pressure is always lower than urethral closure pressure. Incontinence is therefore likely to result from deficiency of urethral closure mechanisms and/or involuntary detrusor activity, aggravated by factors that chronically increase intra-abdominal pressure. In general, maximal urethral closure pressure is higher in men and therefore incontinence is far more prevalent amongst women, who are the target population for this review. 1

Definition of urinary incontinence

The symptom of urinary incontinence is defined as the involuntary leakage of urine,2,3 and this can be subcategorised qualitatively according to the patient’s description (Box 1).

This symptom categorisation2,3 is based on a detailed history and provides a useful basis for discussion of the problem with the patient, identification of patient-centred treatment goals and initiation of treatment pathways.

When the patient first reports the problem to a clinician it is usual for the clinician to define possible causative factors and commonly associated problems by further questioning, physical examination and performance of simple tests. The severity of incontinence and the degree of bother it causes the individual can be estimated by appropriate direct questioning, including pad usage, or can be quantified more objectively using validated symptom scores4 or bladder diaries.

Further categorisation of incontinence according to the underlying functional or anatomical cause requires simultaneous measurement of bladder and rectal pressure, together with observation of urine loss during bladder filling. This invasive clinical test, filling cystometry, requires catheterisation of the bladder and is therefore generally performed only when more accurate categorisation is required, for example prior to surgical treatment in women with MUI. This test will differentiate urodynamic stress incontinence (USI) due to bladder outlet weakness from detrusor overactivity (DO) incontinence due to involuntary contraction of the bladder muscle (Box 2).

The test may be accompanied by radiographic visualisation of the bladder outlet to qualitatively subcategorise USI according to the degree of descent of the bladder neck on coughing (hypermobility) or loss of the sealing mechanism of the urethra (intrinsic sphincter deficiency). The diagnostic and prognostic usefulness of this additional imaging, together with related indices such as abdominal leak point pressure, is uncertain and currently not recommended for routine use by the International Continence Society. 2

Most people complaining of the symptom of SUI will have USI, which is demonstrable on cystometry, and this can be aggravated by conditions that chronically raise abdominal pressure, such as obesity and chronic obstructive pulmonary disease. In approximately 10–20% of cases, however, the symptom of SUI can result from DO provoked by coughing, for example, or from loss of bladder compliance due to fibrosis or neurological disease (low bladder compliance). It should therefore be noted that the symptom of stress incontinence does not always relate to weakness of the bladder outlet or urethral closure mechanism.

Practical definition of incontinence for outcome purposes requires a variable that can be measured before and after treatment. This presents a particular problem for evidence synthesis, as there is a lack of consensus on the most appropriate method, with a variety of variables being used to define improvement or cure (Box 3). It is also recognised that these variables may not capture outcomes of prime importance to individual women suffering SUI.

Prevalence

Prevalence estimates vary depending on population sampled, definition of incontinence, severity threshold and survey methodology. 5 A recent longitudinal study from one county in the UK surveyed a random sample of over 15,000 community-dwelling individuals aged ≥ 40 years and found prevalences of 34% in women and 14% in men. 6 These data were in line with prevalence rates among adults living in the UK, summarised from previous studies, showing a mean (range) of 40% (2–69%) for women and 10% (2–25%) for men. 7 These ranges were consistent with findings from other developed countries, which documented rates of 10–72% for women and 3–20% for men. 7

The EPINCONT study surveyed over 34,000 community-dwelling Norwegian women and found that the prevalence of urinary incontinence increased during young adult life, reached a broad peak between the ages of 45 and 55 years, and then showed a further steady increase in the elderly (Figure 1). 8–12 This study also found that about one-quarter of women with incontinence rated it as severe, a proportion in line with previous reports. 13

FIGURE 1.

Prevalence of female urinary incontinence by age group and symptomatic type of incontinence. Compiled from data in Hannestad and colleagues (2000). 8

This study also provided differential prevalence rates for symptoms of stress, urge and mixed incontinence. 10 Overall, SUI was the most common type, experienced by 50% of incontinent women, whereas 11% reported urgency incontinence alone and 36% reported mixed symptoms (Figure 1). The pattern does vary, however, according to age group, with SUI alone being most frequently reported in women who are younger than 50–55 years, after which urgency incontinence is reported most often, either alone or in combination, possibly reflecting menopausal status. 12 This pattern was also found in a large cross-sectional European study using validated questionnaires, for which SUI was most often reported by women under 60 years old. 11 Overall prevalence rates for urinary incontinence amongst older women who are living in supported accommodation are usually much higher, reaching 40–50%. 14

This high community prevalence does not tend to lead to equivalent rates of consultation with a clinician. Studies estimate that only 15% of the women identified as suffering from SUI in cross-sectional surveys have consulted a health professional about the problem. 15,16 The reasons for this are unclear but may relate to social class, mild symptoms, lack of bother, embarrassment, disinclination towards treatment options and perceived lack of effective treatment. 17

Natural history

In comparison with the many cross-sectional prevalence surveys, fewer longitudinal studies have examined the incidence and remission of urinary incontinence symptoms. One study followed a large cohort of community-dwelling middle-aged women (mean age 46 years) for 2 years and documented an average annual incidence of new incontinence of 9%. 18 The incidence increased with age, with the majority reporting mild, non-disabling leakage (Figure 2). Subanalysis of type of incontinence showed an annual incidence of frequent or severe stress incontinence of about 2% (Figure 3). This study also documented a 7% annual remission rate among those women reporting urinary incontinence at baseline. Similar results were found from a cohort of older, postmenopausal women (mean age 64 years), followed for 2 years, and a further large cohort study of women aged > 65 years, with rates of annual incidence for SUI of 9% and 9.5%, respectively, and annual remission rates of 7% and 8%, respectively. 19,20

FIGURE 2.

Two-year incidence of urinary incontinence by severity. Based on a cohort of 33,952 American women. Any incontinence defined as leaking at least once per month; occasional incontinence defined as leaking one to three times per month; frequent incontinence defined as leaking at least once per week; severe incontinence defined as frequent leaking of quantities at least enough to wet the underwear. Adapted from Townsend and colleagues (2007). 18

FIGURE 3.

Two-year incidence of frequent (at least once weekly) urinary incontinence by incontinence type. Cases of incident frequent incontinence; missing data on incontinence type symptoms are excluded from these calculations. Adapted from Townsend and colleagues (2007). 18

A number of studies have reported questionnaire follow-up of numerically smaller cohorts of younger women (mean age 26–30 years) before and after their first vaginal delivery and reported the annual incidence of new urinary incontinence to be 5%, 1% and 4% over periods of 4, 10 and 12 years, respectively. 21–23

Factors associated with SUI

The main risk factors for female SUI are pregnancy, vaginal delivery, increasing parity, increasing age, obesity and postmenopausal status. 24–26 In older women, particularly those requiring social care, age-related changes to the lower urinary tract (such as reduced bladder capacity) and comorbidity in other organ systems (such as cardiac or cognitive impairment treated with drugs such as diuretics) can precipitate or worsen incontinence. Consideration of these multiple factors, together with the increasing preponderance of mixed urgency and stress incontinence symptoms, makes effective management of the problem in the elderly more difficult. 14

Effect on well-being

Embarrassment associated with urinary incontinence may cause withdrawal from social situations and reduce quality of life. 32 Women with a severe or frequent problem find the leakage distressing and socially disabling. They may avoid going away from home, using public transport and sexual activity. 33 SUI does not generally lead to deterioration in physical health but can be associated with depression and other psychological morbidity. 33 The problem may also lead to withdrawal from regular physical activities, potentially harming general health. 34

Extent of problem in the UK

Assuming an overall prevalence for SUI of 15% among women aged over 20 years, it can be estimated that there are 3.3 million sufferers in the UK. 35

Cost to society

The high prevalence of urinary incontinence results in a high overall cost of treatment and containment. Precise cost is difficult to define but a recent study suggested an estimated figure for combined health care, personal and societal expenditure of £248 per person per year in the UK, which would equate to a total annual cost of £818M for SUI. 36 A further estimate, assuming that SUI accounts for 50% of cases, suggested a health care cost to the UK National Health Service (NHS) of £117M per year. 37

Existing guidelines

Epidemiological studies consistently demonstrate that proportionally few women who experience urinary leakage approach clinicians for advice and treatment. 15,16 It is likely that most women first seek advice from family, friends and the media and, for individually varying reasons, decide to manage the problem themselves. Those who present their problem to health-care professionals tend to have more severe symptoms, which cause interference to their social activities and they are therefore generally seeking active treatment. Most countries, such as the UK, have attempted to standardise the assessment and initial management of women with incontinence by publication of consensus documents and guidelines. 40–44 Despite this, uniformity of care remains lacking and will depend on individual clinician opinion and local service provision.

Current UK NHS care pathway

In the UK, the first port of call is likely to be the general practitioner (GP – primary care physician). An initial assessment will document the severity of the problem and the degree to which it bothers the women, and make sure that there are no more immediate health-threatening problems. Lifestyle advice, such as smoking cessation and weight loss, to modify risk factors may be offered. It is then possible that conservative therapy, in terms of bladder training (BT) or pelvic floor education and therapy, will be suggested, with referral to a practice nurse, physiotherapist or continence nurse specialist. Alternatively, or if these approaches subsequently fail, the woman will be offered referral to secondary care, to a urologist, urogynaecologist or gynaecologist, depending on local service arrangements. Such referrals will mostly result in further investigation, further conservative treatment including the use of drugs and eventually the offer of surgery to those with predominant SUI.

Pharmacotherapy

Serotonin–noradrenaline reuptake inhibitor

Experimental studies in animals suggest that noradrenaline and serotonin (5-hydroxytryptamine) act on efferent neurons in the sacral spinal cord (Onuf’s nucleus) to encourage contraction of the periurethral striated muscle of the urethral sphincter and relaxation of the bladder wall muscle (detrusor muscle), thereby promoting urine storage and continence. Duloxetine hydrochloride, a balanced serotonin–noradrenaline reuptake inhibitor (SNRI), was tested as an antidepressant but found to have an effect on reducing stress incontinence in women. It is now licensed as a continence-promoting drug in women with SUI. 55 Although it has been shown to improve symptoms of SUI, the usefulness of duloxetine is limited by side effects, particularly nausea, which result in up to 20% of women being unable to tolerate the drug. 56 There is some evidence to suggest that this drug may be more useful as an adjunct to other conservative therapies, such as PFMT. 57 Despite poor tolerability and modest efficacy, the prescribing of this drug appears to be increasing in the UK (Figure 4). Further details on the current use of SNRI’s were obtained from a survey of members of the Association of Continence Advisors (ACA). The ACA provided a copy of their e-mail distribution list and members were surveyed during January and February 2009. Out of approximately 650 ACA members on the distribution list, 86 responded (a 13.2% response rate). Of these, 57 provided details of their SUI caseload, of which 15/57 (26.3%) advisors reported that they had patients using duloxetine for the treatment of SUI. The number being treated was 92 out of a caseload of 1234 patients (7.5%); the majority of these, however, came from the caseloads of two respondents.

FIGURE 4.

Serotonin–noradrenaline reuptake inhibitor prescriptions for stress urinary incontinence. UK NHS prescribing data obtained from NHS information centre. 58

The drug is generally prescribed in divided doses, totalling 40–80 mg per day, and the response is assessed over a 12-week period. The relatively high risk of side effects requires initial close monitoring by the specialist or GP.

Setting

The poor tolerability profile of this drug has resulted in it mainly being prescribed by specialists through hospital clinics.

Personnel

The drug is predominantly initiated by hospital specialists for women with SUI, who are unsuitable for surgery or do not wish to undergo surgery.

Costs

The cost of a 56 × 40-mg tablet pack of duloxetine is £30.80. 59 This equates to an average annual cost per patient, for drugs alone, of £402. If two visits to the GP are included the total annual cost rises to £474.

Estrogen

Estrogens are thought to affect female continence through several modes of action, including maintenance of pelvic floor musculature and enhancement of urethra mucosal sealing. It was therefore considered that therapeutic estrogen supplementation in postmenopausal women would be beneficial delivered either locally or systemically. Overall, there is a high level of evidence that oral estrogen replacement therapy appears to increase incontinence symptoms amongst postmenopausal women,60–62 and its use for this indication is therefore not recommended. Recent appraisal of evidence for topical vaginal estrogen therapy suggests that it may benefit women with incontinence but not to any consistent degree,63 and it is currently only recommended for women with overactive bladder symptoms that are associated with mucosal atrophy. 43 The opinion from the expert group for the current review was in agreement with this guidance and so oral or topical estrogen were not included as a treatment option for SUI.

Adrenergic agonists

Despite a theoretical and experimental expectation that alpha-adrenergic drugs should improve urethral and bladder neck smooth muscle activity, no consistent clinical benefit was found in women with stress incontinence. 63,64 These drugs are neither licensed for this treatment indication nor recommended by current guidance,43 and were considered to be very rarely used in practice by our expert group. On this basis, the use of adrenergic agonists as a treatment option for SUI was not considered further.

Absorbent pads

The most commonly used method of incontinence management is absorbent pads. The total expenditure on such products is large and, although not altering the underlying condition, it can be a satisfactory option for women with minimal or predictable leakage, or for those who are unsuited to treatments or have failed to benefit from treatment. The products range from thin panty-liners to large nappy (diaper)-type pads. Despite their widespread use, little research has been conducted concerning patient satisfaction or effectiveness, although recent evidence summarised in a Cochrane review does suggest that specifically designed pant insert pads are superior to those designed for menstrual loss. 84,85 Containment-using pads was also the subject of a previous UK Health Technology Assessment (HTA) monograph. 86

Urinary diversion

Some women with severe intractable incontinence that has not responded to corrective methods of management may choose to undergo urinary diversion. The simplest method is transurethral or suprapubic insertion of an indwelling catheter, with continuous drainage into a collection bag. Summarised evidence regarding types of catheter and policies for their use has previously been published. 87,88 Generally, indwelling catheters are unsuited for women with long life expectancy, who are often better served with a formal surgical urinary diversion procedure. This can be performed using an ileal conduit or continent diversion. The techniques, benefits and adverse events have been described in a Cochrane review. 89

Patient treatment pathway 1

Patient treatment pathway 1, in Figure 5, details one plausible strategy of care for a woman presenting with SUI. The first line of treatment that women presenting with SUI are offered is lifestyle modification. Lifestyle modification may involve a combination of one or more elements. These are dietary factors, reduction of caffeine intake, fluid intake, smoking, weight, physical exercise, alcohol consumption and limiting heavy activity (it may be possible that there are several ways lifestyle modifications can be performed so that they can be considered as different treatments, therefore allowing someone to get more than one lifestyle modification treatment). There are three states that could arise after treatment: ‘continent’, ‘incontinent’ or ‘dead’ (from natural causes). The continent state includes women who report that they are completely cured and those that feel that their condition has improved to the extent that they require no further treatment at the present time. Women could remain continent and hence require no further treatment, but some of them may experience a return or worsening of symptoms of SUI at some point in the future and then require further treatment. The women who report that they are not cured are said to be in the incontinent state. These women could either require or not require further treatment. The women who require further treatment get the next treatment option in the pathway they are following. In the first care pathway, all of the women who need further treatment are offered the second line of treatment: physical therapy. There are different forms of physical therapies that women can receive. These include PFMT alone or in combination with an adjunct, such as BF, ES or VCs. Based on the number of sessions women receive, PFMT can be classed as ‘basic’, when a patient has a maximum of two sessions per month, or ‘PFMT with extra sessions’, when the patient receives more than two sessions per month. This pathway offers only basic PFMT without any adjunct. Women who are not cured and do not receive any further treatment may use containment products, such as pads, to manage symptoms.

If women are not cured after PFMT they progress to the third line of treatment, which is surgery. Those who are not cured and need further treatment are offered a second surgical procedure. If the second surgical procedure does not work then women use containment products until they die. It is appreciated that women may also use containment products even when they are receiving other possible interventions. Figure 6 provides a detailed description of pathway 1. Similar figures could have been developed for all of the other pathways.

FIGURE 6.

Description of alternative treatment strategies (note: women may choose not to seek further treatment at any point).

Patient treatment pathway 2

The treatment pathway option 2 is similar to that depicted in option 1. The only difference is that PFMT with extra sessions is offered as an additional non-surgical treatment between basic PFMT and surgery.

Patient treatment pathway 3

Pathway 3 extends pathway 2 by adding a drug therapy (an SNRI) between PFMT with extra sessions and surgery.

Patient treatment pathway 4

The treatment pathway 4 is substantially the same as pathway 1. The difference is that women receive PFMT with extra sessions straight away, instead of PFMT basic.

Patient treatment pathway 5

Pathway 5 extends pathway 4 by adding SNRI therapy between PFMT extra sessions and surgery.

Patient treatment pathway 6

In this pathway women would, as in all other pathways, initially receive the relevant advice or treatment aimed at modifying lifestyle. If they experienced a recurrence, had insufficient improvement or no improvement, they would proceed straight to surgery. Such a strategy might be appropriate for some subgroups of women. Alternatively, it may be popular with some women because it gives the prospect of relatively rapid relief from symptoms.

Patient treatment pathway 7

This treatment pathway is similar to pathway 1 but with VCs as a treatment option between basic PFMT and surgery.

Patient treatment pathway 8

This treatment pathway is similar to pathway 1 but with ES as a treatment option between basic PFMT and surgery.

These patient care pathways will be used in the study to inform the economic model reported in Chapter 9. These pathways will also assist in identifying outcomes of interest for the systematic review of effectiveness and identify the data required to populate the model. The key outcome of these pathways is that any need for further treatment is heavily influenced by whether the prior treatment has resulted in sufficient resolution of symptoms.

The questionnaire

The PGI is an individualised patient-reported health instrument that allows the respondent to select, weight and rate the importance of a particular health outcome. 96 The PGI was designed with the aim of producing a valid measure of outcome that reflected areas of importance to patients’ lives. 94 The PGI involves the respondent deciding what factors are important to her. Examples of the types of factors that may be important are included to provide guidance. The aim of the PGI is therefore to capture the diverse range of concerns or priorities of respondents. Using the PGI, respondents can vary the weight they attach to these concerns or priorities, which provides researchers with an insight into each respondent’s viewpoint. An overall score for the PGI for each respondent can then be calculated by multiplying the rating for each health area by the proportion of points allocated to that particular area.

The PGI is completed in three stages. In the first stage, respondents are asked to identify up to five areas of their life that are affected by their SUI. Respondents are given a list of outcomes to act as prompts to help them think about which areas of their life might be affected by their condition. Respondents can then choose from these options or provide their own examples. In addition to the five boxes, there is a sixth box that enables respondents to rate all other areas of their life affected by their SUI. Examples of the factors to include as prompts on the PGI were drawn from three sources. The first of these was the King’s Health Questionnaire,95 which was used to generate a list of outcomes under the broad headings of: ‘role limitations’, ‘physical limitations’, ‘social limitations’ and ‘personal relationships and emotions’. These outcomes were supplemented from Cochrane reviews of non-surgical treatments. 56,75,83,97 Finally, a general literature search was also conducted, although this did not provide further additions to the 17 different outcomes identified from the King’s Health Questionnaire and the Cochrane reviews. These outcomes were then narrowed down to broad categories of those considered most relevant by members of the project team. These were ‘work’, ‘household tasks’, ‘social activities’, ‘feeling depressed/anxious’, ‘personal hygiene’ and ‘affecting sleep’. In addition to the methods used to generate the prompt list, further qualitative work could have been conducted. This could have included a focus group of women with SUI to further refine the areas to include in the prompt list. This is an area that could be considered in future use of the PGI.

In stage two of the PGI, respondents were asked to score each area listed in stage one of the PGI on a scale ranging from 0 to 6. The score given in stage two was intended to reflect how the individual was affected by their SUI in the past month. A score of 0 would signify that the effect on their life was as bad as it could possibly be and a score of 6 would correspond to an effect that was as good as it could possibly be.

Finally, in stage three, respondents were asked to ‘spend’ 10 points to indicate the relative importance of each of the areas mentioned in stage one. Respondents were requested to spend more points on areas that were the most important to them.

As noted above, in addition to the PGI, the questionnaire also contained the King’s Health Questionnaire and the EQ-5D. The King’s Health Questionnaire is a condition specific questionnaire that aims to assess the impact of urinary incontinence on an individual’s quality of life. It contains questions set in nine domains relating to: general health perception, incontinence impact, role limitations, physical limitations, social limitations, personal relationships, emotions, sleep and energy, and severity. With the exception of the final part of the questionnaire (severity measures) scores can be calculated for each domain (0–100). The higher the score the worse off an individual feels they are and the lower they perceive their quality of life to be.

The EQ-5D is a standardised instrument used to measure quality of life. The EQ-5D is applicable to a wide range of health conditions and treatments and it provides a simple descriptive profile and a single index value for health status. 98 The EQ-5D has five dimensions (mobility, self-care, usual activities, pain/discomfort and anxiety/depression) that can be converted into a utility score.

Sample

The Bladder and Bowel Foundation (formerly ‘InContact’ and the ‘Continence Foundation’) is a national charity that provides information and support to people with bladder and bowel problems, representing the interests of people with continence problems with the aim of ensuring they have access to the latest information and services available. 99 In 2006 a survey conducted by InContact was completed by 755 people affected by bladder and bowel problems. 100 Of these, 188 women with SUI gave consent for future contact about relevant research and formed the sample for the current study. In July 2007 these women were sent questionnaires for the current study by InContact. Given that this is a self-selected sample of women suffering from SUI and not a random sample of the population, it is not known how representative this sample is of the wider population.

Ethical issues

The 2006 survey in which the participants were originally identified was a service evaluation in which The Bladder and Bowel Foundation surveyed people who had previously been in touch with the charity. As such, no ethical approval was necessary. The 2006 survey materials contained an explicit assurance that confidentiality would be maintained and that identifiable data would not be passed on to third parties. Respondents were asked if they were willing to be contacted in the future for research purposes. For this study questionnaires were sent in July 2007 to 188 women with SUI who gave their consent for further contact relating to research. The questionnaires were returned directly to the charity and, after screening only anonymous data, were subsequently forwarded to the authors in accordance with the Medical Research Council’s guidance on the use of personal information in medical research and the Data Protection Act 1998. 101,102

Types of studies

The types of studies considered were RCTs and quasi-RCTs (alternate allocation). Trial data reported in conference abstracts, as well as full-text papers, were included. For abstracts, solely those identified from the Cochrane Incontinence Group Specialised Register of trials were used.

Types of participants

The participants were women with SUI or incontinence that was predominantly SUI (however diagnosed). Classification of diagnoses was accepted as defined by the trialists.

Owing to the small number of studies per intervention available, a pragmatic decision was made to include studies where the majority (≥ 50%) of the sample consisted of women with SUI or predominant SUI. Studies were therefore included if:

• all women had SUI alone (type-1 population)

• at least 50% of women had SUI alone; the remainder could have UUI or MUI (type-2 population)

• under 50% of women had stress incontinence alone but the majority (50% or more) had MUI with stress symptoms as a predominant pattern; the remainder could have SUI, UUI or MUI (type-3 population).

Studies were excluded if the proportion of women with predominantly SUI was not reported, if the type of incontinence (stress, urge, mixed) was unknown or undiagnosed, or if predominant symptoms (stress or urgency) of women with MUI were not specified.

Women with urinary incontinence whose symptoms might be due to significant factors outside the urinary tract were excluded, for example neurological disorders, cognitive impairment, and lack of independent mobility. Studies investigating nocturnal enuresis in women were also excluded.

Incontinent women during pregnancy or in the early postpartum period were considered for inclusion. Data from these childbearing women were analysed separately on the assumption that the effect of PFMT might differ in this group due to the physiological changes of pregnancy and in the postpartum period. Studies investigating prevention of incontinence among childbearing women were excluded.

Studies that recruited mixed populations of men and women with different types of urinary incontinence were eligible for inclusion, providing that demographic and outcome data were reported separately for women with predominantly SUI.

Types of interventions

We defined non-surgical treatment as that which could be undertaken in a heath-care professional’s office or clinic and patients’ homes. Any of the following interventions, alone or in combination, were included. Treatment definitions used were as according to Cochrane Reviews where possible.

Types of outcome measures

The standardisation committee of the International Continence Society recommended that research investigating the effect of therapeutic interventions for urinary incontinence in adult women should incorporate outcome measurements in the following five domains: (1) the patient’s observations (symptoms); (2) quantification of symptoms (e.g. urine loss); (3) the clinician’s observations (anatomical or functional); (4) quality of life; and (5) socioeconomic measures. 108 Following this recommendation, the following measures of outcomes were sought in this review:

Primary outcomes

• Number of women cured.

• Number of women cured or improved (this outcome is henceforth referred to in the text as improvement).

• Adverse events.

• Condition-specific (and generic measures of health-related) quality of life.

Secondary outcomes

• Quantification of symptoms:

  1. – number of incontinent episodes over 24 hours

  2. – number of pad changes over 24 hours

  3. – mean volume or weight of urine loss on pad test

  4. – number of micturitions over 24 hours.

• Participant satisfaction or desire for further treatment.

• Long-term data:

  1. – number of women having incontinence surgery

  2. – return of symptoms/recurrence.

• Socioeconomic measures.

• Other intermediate, explanatory or treatment-specific outcomes:

  1. – measure of pelvic floor muscle function

  2. – treatment adherence

  3. – volume and type of fluid intake or change in body mass index.

The cure and improvement rate may be ascertained via women’s observation (self-report), quantification of symptoms (typically based on incontinence diaries or pad tests) or clinician’s observation. There was a considerable variability in the way these outcomes were defined by the trialists, which limited the possibilities for quantitative synthesis. We therefore chose to combine data on the cure and improvement rate from different sources. The women’s observation was given priority but for studies in which it was not reported, the rate based on diaries was used as a proxy. Where diary data were also not reported, the rate based on pad tests or any other definitions chosen by the trialists was used.

The choice of quality of life measures again varied across studies. In addition, scores were reported either as final values or changes from baseline. Final values were preferred, but where these were not reported, a change score was used. Quantitative synthesis was performed separately for final values and change scores.

Selection of studies

The titles and abstracts identified by the searches were assessed by one reviewer, having already been assessed by the Cochrane Incontinence Group Trials Search Co-ordinator. Full-text copies of all potentially relevant reports were obtained and independently assessed by two reviewers, using a form developed to determine whether the reports met the inclusion criteria (Appendix 3). Any disagreements were resolved by consensus or arbitration by a third person.

Data extraction

A data extraction form was developed to record details of study design, methods, participants, interventions and outcomes (Appendix 4). One reviewer extracted data and another reviewer checked the extracted data. Any disagreements that could not be resolved by discussion were referred to an arbiter.

For studies in which the majority of participants had stress incontinence alone (types 1 and 2 above), data were extracted for both primary and secondary outcomes. For studies where stress-predominant MUI was the majority (type 3) and studies of childbearing women, data were extracted for primary outcomes only. General background and methodological information was collected from all studies.

Assessment of risk of bias in included studies

Two reviewers independently assessed all of the studies that met selection criteria for potential risk of bias. The assessment used the adapted version of a checklist developed by the Cochrane Incontinence Group (Appendix 5). 109

Data synthesis

Data analysis was performed in two stages: (1) pairwise (head-to-head) comparison (Chapter 7), and (2) a mixed-treatments comparison (Chapter 8).

Number of studies identified

The results of the initial searches are summarised in Table 8. These results were then assessed for potential relevance to the project and to remove duplicates already present in the Specialised Register that could not be removed by reference manager duplicate checking. As described above, the numbers retrieved for CINAHL, EMBASE, BIOSIS and Science Citation Index include only the additional reports identified after excluding those identified in the Cochrane Incontinence Register. A total of 7103 titles and abstracts were identified, of which 378 were selected for full assessment.

Number and type of studies included

Of the papers selected for full text assessment the level of agreement between reviewers about whether a paper met the inclusion criteria was very high, and 176 papers from the search met the inclusion criteria for the review. These described 88 studies, which covered 37 distinct treatments (Table 9) and 68 treatment comparisons (Table 10). The list of included studies and associated references are listed in Appendix 6.

Excluded studies, with reasons for specific exclusions

A total of 199 papers were obtained but did not meet the inclusion criteria (Figure 7). Of these, 113 were excluded on the basis of study design, populations, interventions or outcomes. A further 86 papers (describing 56 studies) that did include relevant populations or interventions as part of the study were retained for further assessment and subsequently excluded through the consensus within the research team that it was impossible to attribute an effect to a particular intervention for women with SUI. Reasons for exclusions for the 86 papers excluded on further assessment are described in more detail in Appendix 7.

FIGURE 7.

Flow chart for study selection.

Ongoing studies

A list of ongoing trials is provided in Appendix 8.

Lifestyle change

No studies were found comparing lifestyle change with no treatment. The only relevant study119 was a crossover trial evaluating the effect of increasing or decreasing fluid intake on urinary symptoms in women (Table 13). Women were randomised in the order in which they increased or decreased fluids. All women were instructed to restrict caffeine. No information was available on any of the specified outcomes.

PFMT with and without BF

PFMT, with or without BF, vs no treatment

The characteristics of included studies comparing PFMT (with and without BF) and no treatment are summarised in Table 14. Fourteen studies were eligible for this comparison, in which a total of 958 participants were randomised. Studies varied widely in terms of the duration of treatment prescribed by the trialists, as well as supervisory intensity (measured in terms of the number of clinic visits or any face-to-face contacts with a health-care professional). No studies reported data on further follow-up after the end of the prescribed treatment phase. The comparison groups received no treatment,107,118,120–122,124,125 advice on continence device or pads,115,127 sham ES,126 placebo PFMT,128 imitation PFMT with placebo drug,57 a booklet including the entire behavioural training programme for self-administration123 or an educational leaflet, the equal number of clinic visits as the PFMT arm and encouragement to keep an exercise diary. 129

TABLE 14 - Background characteristics of studies comparing pelvic floor muscle training (plus biofeedback) with no treatment or comparing variation within comparators

NR, not reported.

None, no clinic visit for training, treatment or supervision; basic, up to two clinic visits per month; intensive, more than two clinic visits per month.

N randomised unclear due to poor reporting.

N in analysis; N randomised in each group unclear.

Study ID	Population type	Duration (month)	Comparator	N randomised	Age	Supervisory intensitya	Notes
Aksac 2003120	1	2	PFMT	20	52.5	Intensive
			PFMT + BF	20	51.6	Intensive
			NT	10	54.7	None
Bidmead 2002121	1	3.5	PFMT	40b	46.2	Basic
			PFMT + sham ES	42b	51.5	Basic
			NT	20b	47.5	None
Bø 1999115	1	6	PFMT	29	49.6	Intensive
			NT	32	51.7	None	Advice on Continence Guard™
Burns 1993122	2	2	PFMT	43c	63.0	Intensive
			PFMT + BF	40c	63.0	Intensive
			NT	40c	63.0	None
Ghoniem 200557	1	3	PFMT	50	54.0	Basic	Placebo drug
			NT	47	51.0	Basic	Placebo drug and imitation PFMT
Goode 2003123	3	2	PFMT	66	57.7	Basic	‘Behavioural training’
			NT	67	55.9	None	Self-administered behavioural training
Henalla 1989124	1	3	PFMT	26	NR	Intensive
			NT	25	NR	None
Henalla 1990125	1	1.5	PFMT	8	54	NR
			NT	7		NR
Hofbauer 1990126	1	1.5	PFMT	11	51.0	Intensive
			NT	10	59.8	Intensive	Sham ES
Kim 2007118	1	3	PFMT	35	76.6	Intensive
			NT	35	76.6	None
Lagro-Janssen 1991127	1	3	PFMT	33	46.1	Basic
			NT	33	44.6	None	Advice on continence pads
Miller 1998107	1	0.25	PFMT	13	68.4	Intensive	The Knack
			NT	14		None
Ramsay 1990128	1	3	PFMT	22	NR	NR
			NT	22	NR	NR	Placebo PFMT
Williams 2006129	2	3	PFMT	79	55.9	Basic
			NT	79	56.7	Basic	Leaflet, clinic visits and exercise diary

Pelvic floor muscle training plus sham ES was classified as being equivalent to PFMT. Where both PFMT and PFMT plus sham ES were present in a trial,121 the dichotomous data were combined (added up).

Cure and improvement rate

Figures 8 and 9 show the number of women who were either cured or improved, respectively, in the PFMT group (with or without BF) versus the no-treatment group at the end of the prescribed treatment phase. Pooled data for cure rates showed higher cure rates for PFMT with or without BF, but with significant heterogeneity (Figure 8: PFMT vs no treatment – 23% vs 7%, OR 5.41, 95% CI 1.64 to 17.82; PFMT with BF vs no treatment – 42% vs 2%, OR 21.54, 95% CI 3.65 to 126.98). The source of the heterogeneity appeared to be the inclusion of studies with women with SUI with or without UUI symptoms (population types 2 or 3). In these studies the direction and magnitude of effect varied across studies. Of note is that the two studies reporting the lowest effect size123,129 not only included women with different types of urinary incontinence, but also had the largest sample size, less intensive supervision (fewer clinic visits) and relatively substantial provision of care for the no-treatment group (e.g. self-administered behavioural training, clinic visits), which may have contributed to the relatively small effect size.

FIGURE 8.

Cure rates for pelvic floor muscle training, with or without biofeedback, versus no training.

FIGURE 9.

Improvement rates for pelvic floor muscle training, with or without biofeedback, versus no training.

Removing these studies reduced the statistical heterogeneity but also widened the CI (as fewer data were available for meta-analysis) (Appendix 18, Comparison 01: PFMT vs no treatment – 39% vs 3%, OR 15.15, 95% CI 5.50 to 41.75; PFMT with BF vs no treatment – 80% vs 0%, OR 77.00, 95% CI 3.75 to 1581.71).

Results for improvement rates similarly favoured PFMT with or without BF compared with no treatment, though statistical heterogeneity was again evident across studies (Figure 9: no BF, OR 11.75, 95% CI 3.49 to 39.55; with BF, OR 24.20, 95% CI 2.02 to 290.58). One additional cause of heterogeneity for this outcome was the greater variability in the way ‘improvement’ was defined between studies, as the heterogeneity remained even after removing studies with mixed diagnoses (Appendix 18, Comparison 02: OR 27.07, 95 CI 4.72 to 155.35). Crucially, the smallest effects were found in those studies with placebo or imitation PFMT57,128 in the no-treatment arm and with women with mixed types of urinary incontinence. 123,129

Adverse events

Adverse events were uncommon in the PFMT group (Table 15). Nevertheless, two studies127,129 reported that up to 12% of women experienced adverse events during PFMT.

TABLE 15 - Adverse events: pelvic floor muscle training versus no treatment

	PFMT		NT		Notes	Population type
	n/N	%	n/N	%
Lagro-Janssen 1991127	4/33	12	0/33	0	Pain, uncomfortable feeling during exercise	1
Williams 2006129	2/79	3	0/79	0	Urinary tract infection	2

Quality of life

Condition-specific quality of life was reported using various measures, including the Social Activity Index, the Urinary Incontinence Quality of Life (I-QoL) scale, the Leicester Impact Scale, the Incontinence Impact Questionnaire (IIQ) and the Bristol Female Lower Urinary Tract Symptoms (B-FLUTS) (Appendix 19, Comparison 01). In all but two trials,123,129 results were better for the PFMT group (with or without BF). One study115 reported this outcome using two instruments (Social Activity Index and B-FLUTS), with results consistently favouring PFMT. As the outcome measures varied between studies and not all studies reported data amenable to meta-analysis, quantitative synthesis was not performed.

Two studies115,123 reported general health-related quality of life (HRQoL) using SF-36 and the Norwegian version of the Quality of Life Scale (QoLS-N). One of these studies115 reported higher quality of life for the PFMT group, whereas the other123 reported no statistically significant difference between the groups (Appendix 17, Comparison 01).

PFMT vs PFMT plus BF

Fifteen studies120,122,146–158 evaluated the effect of adding BF to PFMT (Table 16). In two studies by Glavind and colleagues150 and Wilson and colleagues,157 the women randomised to use BF had more clinic visits than the PFMT group. There was also a potential difference in supervisory intensity in a further study by Pages and colleagues,154 in which the PFMT group had group therapy for 60 minutes, five times a week, whereas the BF group had individual therapy for 15 minutes, five times a week.

In addition to devices specifically designed to provide visual and/or audio BF, the use of intravaginal resistance device,149 an endotrainer151 and an exerciser155 were also classified as BF for the purpose of this review.

TABLE 16 - Baseline characteristics of studies comparing pelvic floor muscle training (PFMT) with PFMT plus biofeedback

IVRD, intravaginal resistance device.

None, no clinic visit for training, treatment or supervision; basic, up to two clinic visits per month; intensive, more than two clinic visits per month.

Treatment for 4 weeks, and measurement at 3 months.

Treatment for 6 weeks, and measurement at 3 months after end of treatment.

Study ID	Population type	Duration (month)	Comparator	N randomised	Age	Supervisory intensitya	Notes
Aksac 2003120	1	2	PFMT	20	52.5	Intensive
			PFMT + BF	20	51.6	Intensive
Aukee 2002146	1	3	PFMT	15	50.8	Basic
			PFMT + BF	15	51.8	Basic	Home BF
Berghmans 1996147	1	1	PFMT	20	50.4	Intensive
			PFMT + BF	20	46.4	Intensive
Burns 1993122	2	2	PFMT	43	63.0	Intensive
			PFMT + BF	40	63.0	Intensive	Clinic BF
Castleden 1984148	1	1	PFMT	19 (crossover)	55.0	NR
			PFMT + BF	19 (crossover)		NR	Home BF
Ferguson 1990149	1	1.5	PFMT	10	35.8	Basic
			PFMT + BF	10	37.1	Basic	IVRD
Glavind 1996150	1	3b	PFMT	20	45.0	Basic
			PFMT + BF	20		Basic	Additional visits for clinic BF
Klingler 1995151	1	3	PFMT	21	53.0	Intensive
			PFMT + BF	20	51.8	Intensive	Endotrainer
Laycock 2001152	1	3	PFMT	20	NR	Basic
			PFMT + BF	40	NR	Basic	Home BF
Mørkved 2002153	2	6	PFMT	50	45.4	Intensive
			PFMT + BF	53	47.8	Intensive	Home + clinic BF
Pages 2001154	1	3	PFMT	27	51.1	Intensive	Group therapy
			PFMT + BF	24		Intensive	Individual therapy; clinic BF
Shepherd 1983155	1	4.5c	PFMT	11	48.4	Basic
			PFMT + BF	11	48.2	Basic	Exerciser
Taylor 1986156	1	2.25	PFMT	13 (number in each group unclear)	NR	Intensive
			PFMT + BF		NR	Intensive	Clinic BF
			PFMT + BF		NR	Intensive	Clinic + home BF
			PFMT + BF		NR	Intensive	Home BF without vaginal sensor to be used as a resistive device
Wilson 1987157	1	1.5	PFMT	15	46.8	Basic
			PFMT + BF	15		Intensive	Clinic BF
Wong 1997a158	1	2	PFMT	7	48.2	Intensive
			PFMT + BF	10		Intensive

Cure and improvement rate

Pooled data showed that at the end of the prescribed treatment phase the addition of BF to PFMT resulted in significantly higher cure and improvements than with PFMT alone (Figures 10 and 11: cure rates 34% vs 49%, OR 0.48, 95% CI 0.30 to 0.77; improvement rates 76% vs 86%, OR 0.41, 95% CI 0.18 to 0.97). Among the studies showing the largest effect size were studies by Glavind and colleagues150 and Wilson and colleagues,157 which had additional supervisory visits in the BF group relative to the PFMT group. The study by Pages and colleagues,154 which was also a trial with a potential difference in supervisory intensity, also suggested a large treatment effect in the cure data.

FIGURE 10.

Cure rates: pelvic floor muscle training (PFMT) versus PFMT plus biofeedback.

FIGURE 11.

Improvement rates: pelvic floor muscle training (PFMT) versus PFMT plus biofeedback.

Two studies150,157 conducted a further follow-up after the end of the supervised treatment in which all women were advised to continue PFMT at home without BF or any close supervision by a health-care professional. The first study157 with 6-month follow-up (4.5 months after the end of 6-week treatment) reported that women who did not use BF in the treatment phase were significantly less likely to improve than those who trained with BF (Appendix 17, Comparison 02, 4/15 vs 9/14, OR 0.20, 95% CI 0.04 to 0.98). In the second study150 with 2.5 years of follow-up, the results for cure and improvement also favoured the BF group, although the differences were not statistically significant (Appendix 17, Comparison 02, cure 0/14 vs 5/19, OR 0.09, 95% CI 0.01 to 1.80; improvement 4/14 vs 8/19, OR 0.55, 95% CI 0.13 to 2.40).

Adverse events

Adverse events were uncommon but in two studies146,153 that did report incidents, more participants (15–27%) experienced adverse events if they were using BF. Some women found the device unpleasant or painful (Table 17).

TABLE 17a - Adverse events: pelvic floor muscle training (PFMT) versus PFMT plus biofeedback

	PFMT		PFMT + BF		Notes	Population type
	n/N	%	n/N	%
Aukee 2002146	3/15	20	4/15	27	Pain while training (of which 3/7 premenopausal)	1
Mørkved 2002153	3/46	7	7/48	15	PFMT + BF: 7/48 found use of apparatus ‘unpleasant’; PFMT: 3/46 found PFMT itself ‘unpleasant’; ‘However, they all followed the training protocol in spite of this’	2

TABLE 17b - Quality of life: pelvic floor muscle training (PFMT) versus PFMT plus biofeedback

NS, not statistically significant.

Lower scores reflect better quality of life.

	PFMT		PFMT + BF		Reported p-value	Notes	Population type
	N	Value	N	Value
Social Activity Index
Aksac 2003120	20	7.5 (1.2)	20	8.1 (0.8)		Score (median, SD)	1
Mørkved 2002153	34	9.5 (0.74)	36	9.6 (0.61)		Score (mean, SD) at 6 months	1
Modified PRAFAB
aBerghmans 1996147	20	13.1 (8.6)	20	11.1 (5.9)	NS	Score (mean, SD)	1
King’s Health Questionnaire
Laycock 2001152	16	8.13 (9.06)	22	6.14 (6.20)	NS	Change in score (mean increase, SD)	1
Incontinence Impact Questionnaire
Wong 1997a158	7	24.5 (10.8)	10	8.5 (19.9)	< 0.05	Change in score (mean reduction, SD)	1

Quality of life

Five studies120,147,152,153,158 reported condition-specific quality of life using four different measures (Table 17). In four of these studies results were similar for both groups, but in one study,158 women reported statistically significantly better quality of life in the PFMT group than with those using BF. Quantitative synthesis did not demonstrate a statistically significant difference between the groups (Appendix 19, Comparison 01, SMD for total score, –0.29, 95% CI –0.62 to 0.03; SMD for change in score, 0.49, 95% CI –0.14 to 1.12).

PFMT vs PFMT with additional sessions

The characteristics of included studies comparing PFMT with and without additional supervisory clinical sessions are summarised in Table 18.

TABLE 18 - Baseline characteristics of studies comparing pelvic floor muscle training (PFMT) and PFMT with additional supervisory clinical sessions

None, no clinic visit for training, treatment or supervision; basic, up to two clinic visits per month; intensive, more than two clinic visits per month.

Study ID	Population type	Duration (month)	Comparator	N randomised	Age	Supervisory intensitya
Bø 1990159	1	6	PFMT	31	45.9	Basic
			PFMT with additional sessions	26	44.9	Intensive
Konstantinidou 2007116	1	3	PFMT	15	47.8	Basic
			PFMT with additional sessions	15		Intensive
Wong 1997b160	1	1	PFMT	26	48.8	Basic
			PFMT with additional sessions	21		Intensive
Zanetti 2007161	1	3	PFMT	21	54.0	Basic
			PFMT with additional sessions	23	56.0	Intensive

Cure and improvement rate

Cure and improvement rates were consistently higher for women who received additional supervisory sessions (Figures 12 and 13: cure rate, 15% vs 43%, OR 0.11, 95% CI 0.03 to 0.43; improvement rate, 54% vs 97%, OR 0.05, 95% CI 0.01 to 0.28).

FIGURE 12.

Cure rates: pelvic floor muscle training (PFMT) versus PFMT with additional sessions.

FIGURE 13.

Improvement rates: pelvic floor muscle training (PFMT) versus PFMT with additional sessions.

One small study159 with 15 years of follow-up also reported that the number of women who remained continent (cured) after they were left to continue PFMT on their own was higher in the group who had additional sessions during the initial supervised treatment phase, although this difference was not statistically significant (Appendix 17, Comparison 03, 16% vs 30%, OR 0.44, 95% CI 0.11 to 1.87).

Adverse events

No studies reported any adverse events.

Quality of life

Condition-specific quality of life measures varied, but all studies consistently reported that PFMT with additional sessions was associated with better quality of life (Table 19). Two studies116,159 provided data amenable to quantitative synthesis; results showed a statistically significant difference between the groups favouring PFMT with additional sessions (Appendix 19, Comparison 02, SMD –1.07, 95% CI –1.98 to –0.15).

TABLE 19 - Pelvic floor muscle training (PFMT) versus PFMT with additional sessions

Lower scores reflect better quality of life.

	PFMT		PFMT + additional sessions		Reported p-value	Notes	Population type
	N	Value	N	Value
Social Activity Index
Bø 1990159	29	8.2 (2.06)	23	9.3 (0.73)	< 0.01	Sum score (mean, SD) at 6 months	1
Quality-of-life index
aKonstantinidou 2007116	10	3.6 (1.5)	12	1.7 (0.8)	0.000	Score (mean, SD)	1
Incontinence Quality of Life
Zanetti 2007161	21	79	23	89	0.0456	Score (median)	1

Other physical and behavioural interventions

The characteristics of included studies comparing physical or behavioural interventions other than PFMT (with and without BF) are summarised in Table 21.

Electrical stimulation vs no treatment

Eight studies115,124,126,130–134 compared ES with no active treatment. Five studies provided treatment by means of a device for home use, and compared it with either sham treatment130,131,133,134 or no treatment. 115 The other studies provided treatment at clinic, compared with sham treatment126,132 or no treatment. 124

Cure and improvement rate

Pooled data showed no statistically significant difference between the groups in the cure rate (6% vs 6%, OR 1.10, 95% CI 0.41 to 2.94), but ES showed a significantly higher rate for improvement (37% vs 13%, OR 3.93, 95% CI 1.43 to 10.80) compared with no active treatment, although there was some evidence of heterogeneity (Figures 14 and 15). The source of heterogeneity is unclear but it appears to be caused by the study by Bø and colleagues. 115 Removal of this study reduced statistical heterogeneity and the difference between the groups was still statistically significant (OR 2.46, 95% CI 1.14 to 5.30, figure not shown).

FIGURE 14.

Cure rates: electrical stimulation versus no treatment.

FIGURE 15.

Improvement rates: electrical stimulation versus no treatment.

Adverse events

Only two studies115,134 reported any incidence of adverse events (Table 22). All recorded cases were attributed to the treatment device, whether it was used for active or sham ES. In the study by Bø and colleagues,115 seven of the 32 participants in the ES group stopped treatment due to adverse events.

TABLE 22a - Adverse events: electrical stimulation versus no treatment

	ES		NT		Type of AEs	Population type
	n/N	%	n/N	%
N experiencing AEs
Bø 1999115	10/32	31	0/32	0	Smarting (tenderness, bleeding, discomfort), motivation problem, difficulty in using the stimulator	1
Sand 1995134	14/35	40	7/17	41	All cases added up, although unclear if the same patient experienced more than one adverse event; vaginal irritation: ES 5/35, NT 2/17; occasional pain: ES 3/35, NT 1/17; vaginal infection: ES 4/35, NT 2/17; urinary tract infection: ES 1/35, NT 2/17	1

TABLE 22b - Quality of life: electrical stimulation versus no treatment

Lower scores reflect better quality of life.

	ES		NT		Reported p-value	Notes	Population type
	N	Value	N	Value
Social Activity Index
Bø 1999115	25	0.6 (1.02)	30	–0.2 (1.68)		Change in score (mean, SD)	1
Incontinence Impact Questionnaire
aJeyaseelan 2000131	12	–4.1 (16.4)	12	–9.1 (17.1)	NS	Change in score (mean, SD)	1
Urogenital Distress Inventory
aJeyaseelan 2000131	12	–11.8 (15.9)	12	–3.3 (8.3)	0.01	Change in score (mean, SD)	1

Quality of life

Two studies115,131 reported condition-specific quality of life, with one of these studies131 using two instruments (questionnaires). Results were inconsistent across and within studies (Table 22 and Appendix 19, Comparison 03).

There were no statistically significant differences in general HRQoL scores (SF-36) between the groups in two studies (Appendix 17, Comparison 10). 131,134

VCs vs no treatment

Cure and improvement rate

Two studies115,129 comparing VCs with no treatment reported conflicting data on improvement (Figure 16). This may stem from a range of factors such as difference in study populations (SUI with or without UUI symptoms) and sample size, duration of treatment and supervisory intensity. The intervention for the ‘no-treatment’ group also differed, with one study by Bø and colleagues115 offering instructions on a disposable vaginal device (Continence Guard™) only, whereas the other study by Williams and colleagues129 provided the same number of clinic visits as the treatment group, with leaflets giving advice on the pelvic floor muscles. Cure rates were not reported.

FIGURE 16.

Improvement rates: vaginal cones versus no treatment.

Adverse events

In one study115 62% (18/29) of the women who used weighted VCs reported abdominal pain and vaginitis, as well as difficulty in using cones and motivational problems. In the other study,129 3% (2/80) in the cones group reported urinary infection. No adverse events were reported for women receiving no treatment (Table 23).

TABLE 23a - Adverse events: vaginal cones versus no treatment

	VC		NT		Notes	Population type
	n/N	%	n/N	%
Bø 1999115	18/29	62	0/32	0	Abdominal pain, vaginitis, bleeding, motivation problems, trouble in using the cones	1
Williams 2006129	2/80	3	0/79	0	Urinary tract infection	2

TABLE 23b - Quality of life: vaginal cones versus no treatment

Lower scores reflect better quality of life.

	VC		NT		Reported p-value	Notes	Population type
	N	Value	N	Value
Social Activity Index
Bø 1999115	27	0.1 (1.06)	30	–0.2 (1.68)		Change in score (mean, SD)	1
The Leicester Impact Scale
aWilliams 2006129	79	2 (0.0 to 5.0)	75	1.5 (0.0 to 5.0)	0.658	Score (0–42, median, interquartile range)	2

Quality of life

There was no evidence of a difference between the groups in either study (Table 23).

SNRI drug therapy

The characteristics of included studies comparing SNRI drug therapy (duloxetine) with no treatment are summarised in Table 25.

SNRI vs no treatment

All but one57 of the 12 trials compared SNRI (duloxetine) with placebo. The other study by Ghoniem and colleagues57 was a four-arm trial comparing duloxetine with placebo, both combined with either active or imitation PFMT. For cure and improvement as well as quality of life, data from two arms were compared (SNRI with imitation PFMT vs placebo with imitation PFMT), while for adverse events, data from all four arms were combined (SNRI with active or imitation PFMT vs placebo with active or imitation PFMT). All but three studies57,136,137 included women with SUI with UUI symptoms (population types 2 and 3).

The majority of studies used a daily dose of 80 mg, although participants in the SNRI group in the Cardozo and colleagues study137 ingested 80 mg daily for 4 weeks, escalating to 120 mg daily for another 4 weeks. Another four-arm trial by Castro-Diaz and colleagues138 varied a starting dose in the first 2 weeks (20 mg twice a day, 40 mg once a day or 40 mg twice a day), with all participants in the SNRI group taking 40 mg twice daily in the subsequent weeks. In this study, cure and improvement rates and quality of life were measured at the end of 8 weeks, whereas adverse events were reported at the end of 4 weeks when all participants had at least 2 weeks of ingesting 40 mg twice a day. In the study by Kinchen and colleagues,140 the participants were allowed to reduce or suspend study drug or use other modalities of treatment (e.g. PFMT) simultaneously with SNRI. One study by Manning and colleagues142 did not specify dosage and this was assumed to be 80 mg per day.

Cure and improvement rate

Cure and improvement rates were consistently higher across studies for SNRI than with placebo, regardless of the dosage (80, 40 or 20 mg per day) (Figures 17 and 18). Pooled data for a daily dosage of 80 mg showed that cure was reported in 11% (67/609) of the participants using SNRI compared with 8% (53/683) for placebo, and improvement was reported in 65% (1254/1939) for SNRI compared with 46% (803/1733) for placebo. The difference between the groups in improvement reached statistical significance (Figure 18: OR 2.02, 95% CI 1.67 to 2.44) but for cure this was not the case. The results also appeared to show a noticeable placebo effect. The study by Cardozo and colleagues137 showed a higher OR for improvement than other studies. The reason for this is unclear, but may partly stem from the fact that the study recruited women with SUI only and also provided a higher drug dosage.

FIGURE 17.

Cure rates: serotonin–noradrenaline reuptake inhibitors versus no treatment.

FIGURE 18.

Improvement rates: serotonin–noradrenaline reuptake inhibitors versus no treatment.

While most studies provided treatment for a period of 2–3 months, one study140 continued treatment for a total of 9 months and provided data on improvement; the results favoured the treatment group, although the difference was not statistically significant (Appendix 17, Comparison 14, 49% vs 41%, OR 1.37, 95% CI 0.93 to 2.01).

A small number of studies using a daily dosage of 40, 30 or 20 mg reported similar results. Pooled data for improvement from two studies144,145 using a daily dosage of 40 and 20 mg showed statistical heterogeneity, which may be partly explained by different sample populations, with one study145 recruiting women with stress incontinence alone and the other144 including women with different types of incontinence.

Adverse events

Table 26 shows the number of participants who experienced any treatment-related adverse events and also lists all adverse events that were defined by trialists as being significant (in terms of quantity rather than severity): the latter is not a list of all adverse events. adverse events were experienced by 52–93% of participants in the SNRI group. 137,138 Notably, 32–64% of participants in the placebo group also reported adverse events. 117,141 Nausea was the most commonly reported adverse event with SNRI. Castro-Diaz and colleagues138 reported that at 4 weeks (when all SNRI-treated participants had had at least 2 weeks’ use of 40 mg twice a day) a low starting dose of 20 mg, twice daily, significantly reduced the incidence of nausea and dizziness compared with other drug regimens.

TABLE 26 - Adverse events: serotonin–noradrenaline reuptake inhibitors versus no treatment

	SNRI		NT		Notes	Population type
	n/N	%	n/N	%
N experiencing adverse events
SNRI 80 mg vs NT
Cardozo 2004137	43/46	93	37/52	71	Adverse events that occurred in more than 10% of participants: nausea, constipation, headache, dry mouth, fatigue, dizziness, insomnia, somnolence and vomiting; serious adverse events: cardiovascular (not significantly different in both arms); increasing the dose from 80 mg to 120 mg daily did not increase efficacy or side effects	1
Ghoniem 200557	85/104	82	58/97	60	Adverse events that were significantly more common with SNRI than with placebo (with PFMT or imitation PFMT): nausea, dizziness, dry mouth, constipation, insomnia, somnolence, aesthesia	1
Norton 2002144	102/140	73	84/138	61	Adverse events that occurred in ≥ 5% of subjects in any treatment arm: nausea, headache, diarrhoea, constipation, dry mouth, dizziness, insomnia, sinusitis, fatigue, nasopharyngitis	2
Castro-Diaz 2007138 (40 mg b.i.d. starting dose)	87/136	64	53/120	44	Adverse events that occurred in ≥ 2 patients in first 4 weeks: nausea, dry mouth, constipation, somnolence, dizziness, insomnia, fatigue, headache, diarrhoea	3
Castro-Diaz 2007138 (40 mg q.d. starting dose)	76/127	60	53/120	44	As above	3
Castro-Diaz 2007138 (20 mg b.i.d. starting dose)	69/133	52	53/120	44	As above	3
Dmochowski 2003139	255/344	74	170/339	50	Adverse events significantly more common with SNRI and occurring in ≥ 5% of subjects on SNRI: nausea, fatigue, insomnia, dry mouth, constipation, somnolence, dizziness, headache, diarrhoea	3
Kinchen 2005140	198/224	88	159/227	70	Adverse events for which there are statistically significant differences between groups: nausea, fatigue, insomnia, dizziness, headache, somnolence, dry mouth, constipation, diarrhoea, vomiting, increased sweating, decreased appetite, anxiety, tremor, decreased libido, lethargy, nightmare, fungal infection	3
Mah 2006141	50/61	82	19/60	32	Adverse events that occurred in ≥ 5% of the women randomised to the SNRI group or which occurred significantly more often with SNRI than with placebo: nausea, dizziness, anorexia, fatigue, lethargy, abdominal discomfort, somnolence, constipation, headache, dry mouth	3
Millard 2004143	173/227	76	137/231	59	Adverse events: significantly more common with, and occurring in, ≥ 5% of subjects with SNRI: nausea, headache, insomnia, constipation, dry mouth, dizziness, fatigue, somnolence, anorexia, vomiting, increased sweating, anxiety	3
van Kerrebroeck 2004117	200/247	81	158/247	64	Adverse events occurring in at least 5% of patients on SNRI or occurring significantly more often with SNRI than placebo: nausea, dry mouth, constipation, fatigue, insomnia, dizziness, headache, increased sweating, vomiting, somnolence, tremor	3
SNRI 40 mg vs NT
Norton 2002144	93/137	68	84/138	61	Adverse events that occurred in ≥ 5% of subjects in any treatment arm: nausea, headache, diarrhoea, constipation, dry mouth, dizziness, insomnia, sinusitis, fatigue, nasopharyngitis	2
SNRI 20 mg vs NT
Norton 2002144	86/138	62	84/138	61	Adverse events that occurred in ≥ 5% of subjects in any treatment arm: nausea, headache, diarrhoea, constipation, dry mouth, dizziness, insomnia, sinusitis, fatigue, nasopharyngitis	2
Discontinuation due to adverse events
SNRI 80 mg vs NT
Cardozo 2004137	18/55	33	3/54	6		1
Castro-Diaz 2007138 (40 mg b.i.d. starting dose)	22/136	16	7/120	6		3
Castro-Diaz 2007138 (40 mg q.d. starting dose)	15/127	12	7/120	6		3
Castro-Diaz 2007138 (20 mg b.i.d. starting dose)	10/133	8	7/120	6		3
Dmochowski 2003139	83/344	24	14/339	4		3
Ghoniem 200557	28/104	27	8/97	8		1
Kinchen 2005140	20/224	9	5/227	2		3
Mah 2006141	21/61	34	5/60	8		3
Manning 2005142	53/306	17	9/311	3		3
Millard 2004143	39/227	17	4/231	2		3
Norton 2002144	21/140	15	7/138	5		2
van Kerrebroeck2004117	53/247	21	12/247	5		3
SNRI 40 mg vs NT
Norton 2002144	17/137	12	7/138	5		2
SNRI 20 mg vs NT
Norton 2002144	13/138	9	7/138	5		2

It should be noted that the discontinuation rate among participants due to adverse events was high, at 8–34% in the SNRI group, compared with 2–8% in the placebo group (Table 26).

Serious adverse events were reported by three studies. One of these studies57 reported one case of rectal bleeding in the SNRI group but this was not attributed to the study drug. The second study144 reported that five subjects had adverse events that required hospitalisation (one event before randomisation, one event in the group taking SNRI 20 mg per day, two events in the group taking SNRI 40 mg per day, and one event in the group taking SNRI 80 mg per day): only one of these events (a rash) was judged by the study authors to be related to the study drug. The third study140 recorded serious adverse events (details not reported) in eight of the 224 participants (16 events) in the SNRI group and seven of the 227 participants (eight events) in the placebo group. Of these, two of the events in the placebo group but none in the SNRI group was considered by the study author to be related to the study drug (active or placebo).

Quality of life

A total of 11 studies provided information on condition-specific quality of life using various outcome measures and drug doses (Table 27). With respect to studies using a daily dosage of 80 mg, all 10 studies favoured SNRI-treated participants. This difference was reported to be statistically significant in all but two studies117,140 that performed a statistical test. One study138 using two outcome measures reported consistent results favouring duloxetine. Only three studies reported data amenable to meta-analysis, with one reporting a total score at the final evaluation,143 and the other two reporting a change in score from baseline to the final evaluation. 137,139 Pooled data from the two studies137,139 reporting a change in score found a statistically significant difference between the groups (Appendix 19, Comparison 04, OR 0.35. 95% CI 0.16 to 0.55).

TABLE 27 - Quality of life: serotonin–noradrenaline reuptake inhibitors versus no treatment

ICIQ-UI SF, International Consultation on Incontinence Questionnaire – Urinary Incontinence Short Form; I-QoL, Urinary Incontinence Quality of Life Scale.

No significant difference among the three groups with different starting dose.

Lower scores reflect better quality of life.

	SNRI		NT		Reported p-value	Notes	Population type
	N	Value	N	Value
I-QoL
SNRI80 vs NT
Cardozo 2004137	52	10.6 (19.1)	52	2.4 (9.4)	0.003	Change in score (mean, SD)	1
Ghoniem 200557	50	8.3	45	4.8		Mean percentage score increase	1
Norton 2002144	130	9.3	132	5.8	0.03	Change in score (mean)	2
aCastro-Diaz 2007138	344	12.9	112	5.7	< 0.001	Change in score (mean).	3
Dmochowski 2003139	334	11.1 (14.8)	332	6.8 (13.8)	< 0.001	Change in score (mean, SD)	3
Kinchen 2005140	208	13	218	10.4	0.07	Change in score (mean)	3
Mah 2006141	56	63.41	57	60.23		Total score (mean)	3
Millard 2004143	220	69.2 (23.8)	229	64.7 (24.9)		Total score (mean, SD)	3
van Kerrebroeck 2004117	240	72.2	245	68.5	0.127	Total score (mean)	3
SNRI40 vs NT
Zinner 1998145	33	8.2 (10.8)	34	2.6 (8.8)	< 0.05	Change in score (mean, SD)	1
Norton 2002144	129	7.8	132	5.8	0.16	Change in score (mean)	2
SNRI30 vs NT
Zinner 1998145	26	10 (6.4)	34	2.6 (8.8)	< 0.05	Change in score (mean, SD)	1
SNRI20 vs NT
Zinner 1998145	34	12 (16)	34	2.6 (8.8)	< 0.05	Change in score (mean, SD)	1
Norton 2002144	132	5.3	132	5.8	0.6	Change in score (mean)	2
King’s Health Questionnaire
SNRI80 vs NT
bManning 2005142	306	–9.2	311	–2.6	< 0.0001	Change in score (mean)	3
ICIQ-UI SF
SNRI80 vs NT
bCastro-Diaz 2007138	344	–2.8	112	–1.7	0.004	Change in score (mean)	3

One study140 provided information for a longer treatment duration of 9 months; no significant difference between the groups was reported (Appendix 17, Comparison 14).

Two studies144,145 used a daily dose of 40, 30 and 20 mg, with one study145 favouring duloxetine and the other144 reporting no statistically significant difference between SNRI and placebo (Table 27).

PFMT with adjunct treatment

The characteristics of included studies comparing PFMT with adjunct treatment versus no treatment are summarised in Table 28.

PFMT plus ES vs no treatment

Cure and improvement

Three studies121,123,126 compared PFMT with adjunct ES versus no treatment (Figures 19 and 20). Pooled data from two of these studies with data123,126 showed a higher improvement rate (but not cure rate) in the intervention group, although the CI was wide (Figure 20: OR 8.69, 95% CI 1.87 to 40.32).

FIGURE 19.

Cure rates: pelvic floor muscle training and electrical stimulation versus no treatment.

FIGURE 20.

Improvement rates: pelvic floor muscle training and electrical stimulation versus no treatment.

Adverse events

Only one study123 reported adverse events: 6% (4/67) of the participants using ES experienced vaginal irritation (Table 29).

TABLE 29a - Adverse events: pelvic floor muscle training plus electrical stimulation versus no treatment

	PFMT + ES		NT		Notes	Population type
	n/N	%	n/N	%
Goode 2003123	4/67	6	0/67	0	Vaginal irritation	3

TABLE 29b - Quality of life: pelvic floor muscle training plus electrical stimulation versus no treatment

	PFMT + ES		NT		Notes	Population type
	N	Value	N	Value
Incontinence Impact Questionnaire
Goode 2003123	67	No difference	67	No difference	Total score	3

Quality of life

The same study123 found no significant difference between the groups in condition-specific quality of life (Table 29) and general quality of life (Appendix 17, Comparison 17).

PFMT vs ES

Seven studies provided at least one of the specified primary outcomes. 115,124,126,174–177 Of these, five studies, namely Bernardes and colleagues (2000),174 Bø and colleagues (1999),115 Henalla and colleagues (1989),124 Hofbauer and colleagues (1990)126 and Laycock and colleagues (1988),176 provided intensive supervision (more than two sessions per month) for both groups. In addition, the study by Bø and colleagues (1999)115 also provided intensive supervision for the PFMT group but basic supervision (two or fewer sessions per month) for the comparator group.

Cure and improvement

Pooled data for cure and improvement indicated that PFMT was more effective than ES in terms of both cure (Figure 21: OR 2.65, 95% CI 0.82 to 8.60) and improvement (Figure 22: OR 2.18, 95% CI 0.76 to 6.28). The direction and magnitude of effects varied across studies, particularly for the improvement rate which displayed statistical heterogeneity at the 10% level (p = 0.070). The source of heterogeneity was unclear. However, it is noteworthy that all but one study (Laycock 1988)176 that provided intensive supervision for the PFMT group showed ORs (point estimate) of greater than one (favouring PFMT on average) for both cure and improvement data.

FIGURE 21.

Cure rates: pelvic floor muscle training versus electrical stimulation.

FIGURE 22.

Improvement rates: pelvic floor muscle training versus electrical stimulation.

PFMT, with or without BF, vs VC

Seven studies provided information on at least one of the specified primary outcomes. 115,129,152,178–181 Three of these studies, namely Bø and colleagues (1999),115 Cammu and colleagues (1998)181 and Peattie and colleagues (1988),180 provided intensive supervision (more than two sessions per month) for the PFMT group (with or without BF) but basic supervision (two or fewer sessions per month) for the VCs group. The other studies provided basic supervision for both groups.

Cure and improvement

There appeared to be no clear differences between PFMT, with or without BF, and VCs (Figures 23 and 24). One study by Bø and colleagues (1999)115 stands out, showing a relatively higher OR favouring PFMT. It is not clear why this is the case. One explanation may be a longer training duration and more supervisory sessions for the PFMT group relative to other studies.

FIGURE 23.

Cure rates: pelvic floor muscle training, with or without biofeedback, versus vaginal cones.

FIGURE 24.

Improvement rates: pelvic floor muscle training, with or without biofeedback, versus vaginal cones.

PFMT, with or without BF, vs BT

PFMT vs SNRI (80 mg)

One study57 compared PFMT (with placebo drug) with SNRI (with imitation PFMT). The results showed greater improvement and better condition-specific quality of life in the PFMT group, although the difference was not statistically significant (Table 35).

PFMT vs surgery

Data on cure and improvement from two studies184,185 comparing PFMT with surgery are shown in Figures 25 and 26. One of these studies184 compared PFMT with either Burch colposuspension, vaginal repair or combined procedures, whereas the other study185 compared PFMT with Burch colposuspension. The results for both cure and improvement favoured surgery (Figures 25 and 26: OR for cure 0.08, 95% CI 0.03 to 0.23; OR for improvement 0.19, 95% CI 0.04 to 0.77). However, no information was reported on adverse events or quality of life.

FIGURE 25.

Cure rates: pelvic floor muscle training versus surgery.

FIGURE 26.

Improvement rates: pelvic floor muscle training versus surgery.

ES vs VC: cure and improvement

Figures 27 and 28 show pooled data from three studies115,187,188 reporting the number of women cured or improved after treatment with ES compared with VCs. No statistically significant difference between the groups was found (Figures 27 and 28: OR for cure 1.00, 95% CI 0.26 to 3.91, OR for improvement 1.30, 95% CI 0.59 to 2.84).

FIGURE 27.

Cure rates: electrical stimulation versus vaginal cones.

FIGURE 28.

Improvement rates: electrical stimulation versus vaginal cones.

One of these studies187 provided data at 6 months after the initial treatment phase had finished and found no statistically significant difference between the groups in terms of cure (OR 0.93, 95% CI 0.31 to 2.78) or improvement (OR 0.54, 95% CI 0.17 to 1.68) (Appendix 17, Comparison 19).

ES vs VC: adverse events

One study115 indicated that more participants experienced adverse events in the VC group (62%) than in the ES group (31%), although discontinuation due to adverse events was more common in the ES group (7/32) than in the VC group (1/29) (Table 36). Another study,187 which used cones to assess pelvic floor muscle strength in both groups, reported that nine women withdrew because of a failure to tolerate the cones during pretreatment assessment. In seven of these women, the vagina was too narrow and the cones ‘wedged’. The same study187 reported a further two adverse events (one psychiatric disorder and one death) in the cones group, which were considered by the study author to be unrelated to the study intervention.

ES vs VC: quality of life

The information on condition-specific quality of life was available from one study only. 115 The results were similar for both groups (Table 36).

PFMT (with or without BF) vs PFMT (with or without BF) and adjunct treatment

PFMT (with or without BF) vs PFMT (with or without BF) and ES

Cure and improvement

Pooled data on cure and improvement showed that all trial estimates overlapped, and there was significant statistical heterogeneity for the cure rate at the 10% level (Figures 29 and 30: OR for cure without BF 1.02, 95% CI 0.29 to 3.55; OR for improvement without BF 0.84, 95% CI 0.34 to 2.07; OR for improvement with BF 0.86, 95% CI 0.36 to 2.08). The source of heterogeneity appears to be the study by Blowman and colleagues,189 which suggested that adding ES to PFMT was significantly better than PFMT alone. The reason for this is unclear.

FIGURE 29.

Cure rates: pelvic floor muscle training (PFMT) [± biofeedback (BF)] versus PFMT (± BF) plus electrical stimulation.

FIGURE 30.

Improvement rates: pelvic floor muscle training (PFMT) [± biofeedback (BF)] versus PFMT (± BF) plus electrical stimulation.

Two studies157,172 reported the improvement rate at 6 months after the end of the initial treatment phase (Appendix 17, Comparison 20). No differences between the treatment groups were found.

Adverse events

Only one study123 reported any adverse events (Table 38). The combined treatment group recorded four (6%) occurrences of vaginal irritation due to application of ES devices, compared with none in the PFMT group.

TABLE 38a - Number experiencing adverse effects: pelvic floor muscle training (PFMT) versus PFMT plus electrical stimulation

	PFMT		PFMT + ES		Notes	Population type
	n/N	%	n/N	%
Goode 2003123	0/66	0	4/67	6	Vaginal irritation	3

TABLE 38b - Quality of life: pelvic floor muscle training (PFMT) versus PFMT plus electrical stimulation

	PFMT		PFMT + ES		Notes	Population type
	N	Value	N	Value
Incontinence Impact Questionnaire
Goode 2003123	66	No difference	67	No difference	Total score	3

Quality of life

Condition-specific (Incontinence Impact Questionnaire) (Table 38) and general HRQoL (SF-36) (Appendix 17, Comparison 20) was reported by one study only. 123 The results did not differ by treatment group.

PFMT plus adjunct treatment vs the adjunct treatment