Serotonin and noradrenaline reuptake inhibitors (SNRI) for stress urinary incontinence in adults

Abstract

Background

To date, standard recommendations for the management of stress urinary incontinence (SUI) would be either pelvic floor muscle training (PFMT) or surgery. A new form of drug treatment with a serotonin‐noradrenaline reuptake inhibitor (SNRI), duloxetine, may now have a place in treatment of this condition.

Objectives

To determine whether a SNRI is better than placebo (or no treatment, other pharmacological and non‐pharmacological therapies, or surgery) in the treatment of women with SUI, or mixed urinary incontinence that includes stress incontinence (MUI), or both and which doses should be used.

Search methods

We searched the Cochrane Incontinence Group Specialised Register (searched 5th March 2007), CENTRAL (The Cochrane Library 2006, Issue 4), MEDLINE (January 1966 to January 2007), MEDLINE In‐Process & Other Non‐Indexed Citations (7 February 2007) and the reference lists of relevant articles.

Selection criteria

All randomised or quasi‐randomised controlled trials of treatment for SUI or MUI, in which at least one management arm involved a SNRI.

Data collection and analysis

Two authors evaluated the trials for appropriateness for inclusion and methodological quality. Three authors performed the data extraction using predetermined criteria. Analyses were performed using the Cochrane Review Manager software, RevMan.

Main results

Ten randomised trials were included, involving 3944 adults with predominantly SUI, randomised to receive duloxetine or placebo and/ or PFMT. All arms in individual trials were comparable for various baseline characteristics. Treatment duration was between three weeks and 12 weeks.

Duloxetine was significantly better than placebo in terms of improving patients' quality of life (weighted mean difference 5.26, 95% confidence interval 3.84 to 6.68. P less than 0.00001) and perception of improvement. Individual studies demonstrated a significant reduction in the Incontinence Episode Frequency (IEF) by approximately 50% during treatment with duloxetine. With regard to objective cure, however, meta‐analysis of stress pad test and 24 hour pad weight change failed to demonstrate a benefit for duloxetine over placebo though data were relatively few. Subjective cure favoured duloxetine, albeit with a small effect size (3%). One trial suggested that duloxetine was better than pelvic floor muscle training alone in reducing IEF (P less than 0.05) based on median percentage decrease in IEF per week. Although significant side effects were commonly associated with duloxetine, they were reported as acceptable.

Authors' conclusions

The available evidence suggests that duloxetine treatment can significantly improve the quality of life of patients with stress urinary incontinence, but it is unclear whether or not benefits are sustainable. Adverse effects are common but not serious. About one in three participants allocated duloxetine reported adverse effects (most commonly nausea) related to treatment, and about one in eight allocated duloxetine stopped treatment as a consequence.

Plain language summary

Duloxetine can improve the quality of life of patients with stress urinary incontinence, long‐term effects of the treatment are unclear.

Stress urinary incontinence is involuntary urine leakage on coughing or exertion. The trials reviewed compared duloxetine against dummy placebo tablets and also pelvic floor muscle training in women with predominantly stress urinary incontinence. Duloxetine reduced the frequency of episodes of incontinence and improved quality of life scores. However, it had little impact on the numbers cured and commonly had side effects, especially nausea. More studies comparing a serotonin and noradrenaline reuptake inhibitor with placebo and surgery are required, especially if conducted independently of pharmaceutical companies.

Background

Urinary incontinence (UI) as defined by the International Continence Society (ICS) is the complaint of any involuntary leakage of urine. It is a common and socially debilitating problem. UI can be clinically classified as: (a) stress urinary incontinence (SUI) which is incontinence on effort, exertion, coughing or sneezing; (b) urge urinary incontinence (UUI) which is incontinence accompanied by or immediately preceded by urinary urgency and (c) mixed urinary incontinence (MUI) which is a combination of SUI and UUI. Most large population based surveys have shown the prevalence of UI in women to range between 20% and 50%. The variability may be explained by factors such as differences in the definition of urinary incontinence or methods of collection of study samples or survey procedures or effects of interventions. The large EPINCONT survey conducted on 27,936 women aged 20 years and above in Norway, revealed an overall prevalence of 25% of whom 50%, 36% and 11% had SUI, MUI and UUI respectively (Hannestad 2000). The reported prevalence of UI in institutionalised patients is 6% to 72% and the prevalence in men is half that of women (Hunskaar 2002). SUI is more prevalent in middle aged women whereas UUI is seen more often in older, post menopausal women.

Continence of urine is dependent upon the ability of the bladder to store urine at low pressures and a mechanism that maintains the pressure at the bladder outlet above that of bladder pressure. Low pressure storage is achieved by the viscoelastic properties of the bladder wall combining with active relaxation of the bladder muscle (detrusor) to allow a high degree of compliance as the bladder fills with urine. As the bladder fills, detrusor (bladder muscle) contraction is suppressed by inhibitory interneurons in the sacral spinal cord. In women, closure pressure at the bladder outlet is maintained by the proximal urethral mucosa and vascular submucosa aided by appropriate and coordinated contraction of smooth and striated (rhabdosphincter) muscle sphincters. These constitute the intrinsic sphincter mechanism. Further continence at the outlet during raised intra‐abdominal pressures is supported by the pelvic floor musculature that maintain the position of the urethra in relation to the bladder neck.

The theories behind the pathogenesis of SUI include urethral hypermobility (Einhorning 1961), intrinsic sphincter deficiency (ISD) (McGuire 1981), 'backboard' or 'hammock' concept, whereby the tissues posterior to the proximal urethra are unable to provide strong support to coapt the urethra when intra‐abdominal pressure rises (DeLancey 1994), and the integral theory, wherein the laxity of pubourethral ligaments are thought to cause SUI (Petros 1990).

Based on these theories, the treatments for SUI are: correction of urethral hypermobility, enhancing/strengthening the urethral support (Backboard), or strengthening the intrinsic sphincter mechanism. Current management options for SUI include pelvic floor muscle training (PFMT) with or without biofeedback to strengthen pelvic floor musculature, injection of urethral bulking agents to provide urethral coaptation, colposuspension and urethral sling surgery. The choice of treatment depends on the severity of the symptoms, presence of urethral hypermobility or intrinsic sphincter deficiency, patient preference, potential child bearing, associated co‐morbidity and surgeon's experience. Surgical interventions have a reported cure rate of up to 90%, but they have risks (Lapitan 2005) which include failure to correct incontinence, de novo detrusor overactivity, voiding difficulty and urethral erosion if a sling is used. Although not clearly understood, surgical options (colposuspension and sling surgery) are thought to correct either urethral hypermobility or strengthen the 'backboard' of the urethra. Urethral bulking agents are aimed at correcting ISD albeit with short‐term success of 60% to 70% only.

Drug treatment in the management of UI is aimed at either maintaining bladder relaxation or improving sphincter function (Campell's 2002). These effects can be achieved by the use of anti‐muscarinic agents to inhibit detrusor contraction in the treatment of UUI and by alpha adrenergic agents to increase urethral closure pressure but have a limited role in SUI. These groups of drugs have been the subject of previous Cochrane reviews (Alhasso 2005; Nabi 2006).

There are no universally accepted pharmacological treatments for SUI. Recently, however, experiments on cats have shown that dual action serotonin (5‐HT) and noradrenaline reuptake inhibitors (SNRI) suppress parasympathetic activity as well as increase sympathetic and somatic neural activity in the lower urinary tract (Katofiasc 2002; Thor 1995). The inhibition of serotonin and noradrenaline reuptake during bladder storage is thought to increase pudendal nerve output, resulting in increased tone of the rhabdosphincter and subsequently improved urethral closure. Following experimental evidence, phase II and phase III trials have been carried out to determine the efficacy and safety of SNRI in the treatment of SUI.

The focus of this review is the new SNRI, duloxetine.

Objectives

To determine the effects of serotonin and noradrenaline reuptake inhibitors (SNRI) in the management of stress (SUI) and mixed urinary incontinence (MUI) that includes stress incontinence in adults.

The following comparisons will be addressed:

(1) a SNRI versus placebo or 'open' no treatment; 
 (2) a SNRI versus conservative non‐pharmacological therapies; 
 (3) a SNRI versus surgery; 
 (4) a SNRI versus other classes of drugs; 
 (5) a particular SNRI versus another (if available); 
 (6) a particular SNRI given orally versus given via another route (rectal, intra‐vesical, topical); 
 (7) a higher dose of a SNRI versus a lower dose; 
 (8) a SNRI in combination with another drug versus the other drug alone; 
 (9) a SNRI in combination with conservative non‐pharmacological therapy versus the conservative non‐pharmacological therapy alone.

Methods

Criteria for considering studies for this review

Types of studies

All randomised or quasi‐randomised controlled trials of treatment for stress urinary incontinence or mixed urinary incontinence in which at least one management arm involved a SNRI.

Types of participants

Adult women and men with urinary incontinence diagnosed as having any of the following:

(1) stress incontinence (clinical diagnosis); 
 (2) urodynamic stress incontinence (urodynamic diagnosis); 
 (3) mixed incontinence (stress incontinence plus urge incontinence).

Types of interventions

At least one arm of a study will use a SNRI.

Comparison interventions include the use of placebo, no treatment, other pharmacological and non‐pharmacological therapies, surgery, and different types, doses and routes of administration of SNRI.

Types of outcome measures

The primary measure of outcome is the number of participants whose symptoms are not 'cured' whilst on treatment. Data for the following outcome measures have been sought:

A. Participants' observations

1. Number not cured (worse, unchanged or improved versus cured) during treatment (self‐reported, subjective) 
 2. Number not cured (worse, unchanged or improved versus cured) after treatment (self‐reported, subjective) 
 3. Number not improved (worse or unchanged versus improved or cured) during treatment (self‐reported, subjective) 
 4. Number not improved (worse or unchanged versus improved or cured) after treatment (self‐reported, subjective)

B. Quantification of symptoms

1. Number of pads over 24 hours (from self‐reported number of pads used) 
 2. Number of incontinent episodes over 24 hours or per week (from self‐completed urinary diary) 
 3. Mean volume or weight of urine loss on pad test (stress pad test ‐SPT, 24 hr pad weight ‐24HPW) 
 4. Number of micturitions over 24 hours (from self‐completed urinary diary)

C. Clinicians' observations

1. Number not cured (worse, unchanged or improved versus cured) during treatment (objective test) 
 2. Number not cured (worse, unchanged or improved versus cured) within first year of treatment (objective test) 
 3. Number not cured (worse, unchanged or improved versus cured) after first year of treatment (objective test) 
 4. Number not cured (worse, unchanged or improved versus cured) after five years of treatment (objective test) 
 5. Stress incontinence diagnosed by urodynamics

D. Health status/quality of life

1. General health status measures e.g. Short Form 36 (Ware 1993)

2. Condition‐specific health measures (specific instruments designed to assess impact of urinary problems)

3. Measures of psychological health

E. Socioeconomic measures

1. Health economic measures ‐ costs of treatments; differential costs of treatment effect differences; formal cost effectiveness measures

F. Adverse events

1. Number experiencing adverse effects 
 2. Number withdrawing from treatment or trial arm 
 3. Number changing dose of treatment

G. Other outcomes

Non pre‐specified outcomes judged important when performing the review

Search methods for identification of studies

This review drew on the search strategy developed for the Incontinence Group as a whole. Relevant trials were primarily identified from the Cochrane Incontinence Group Trials Register. The methods used to derive this, including the search strategy, are described under the Group's details in The Cochrane Library (For more details please see the ‘Specialized Register’ section of the Group’s module in The Cochrane Library). The register contains trials identified from the Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE and CINAHL in addition to trials identified by hand searching of journals and conference proceedings. 
 The date of the most recent search of the specialised register for this review was: 5th March 2007. 
 The trials in the Incontinence Group Specialised Register are also contained in CENTRAL.

The Incontinence Group Specialised Register was searched using the Group's own keyword system, the search terms used were: 
 topic.urine.incon* 
 AND 
 ({design.cct*} OR {design.rct*}) 
 AND 
 ({intvent.chem.drug.SNRI}) 
 Key: * = trunctation symbol. 
 (All searches were in the keywords field of Reference Manager 9.5 N, ISI ResearchSoft).

For this review extra specific searches were performed by one of the reviewers. These are detailed below. 
 The following electronic bibliographic databases were searched:

• CENTRAL ‐ The Cochrane Library 2006, Issue 4 was searched on 14 November 2006;

• MEDLINE (on OVID) ‐ the years January 1966 to January week 4 2007 were searched on 7 February 2007;

• MEDLINE In‐Process & Other Non‐Indexed Citations (on OVID) ‐ the issue covering 6 February 2007 was searched on 7 February 2007.

The search terms used to search these databases were: ((urinary incontinence OR stress incontinence) AND (duloxetine OR serotonin norepinephrine reuptake inhibitors) AND (clinical trial)). On OVID these terms were used to search in the default 'map' fields.

The reference lists of relevant articles were searched for other possible relevant trials.

We did not impose language or other restrictions on any of these searches.

Data collection and analysis

Trials under consideration for inclusion in the review were assessed independently for their appropriateness by two review authors without prior consideration of their results. Any disagreements that could not be resolved by discussion were considered by a third person.

The review authors independently made an assessment of methodological quality using the Incontinence Group's quality assessment tool ‐ disagreements were resolved by discussion with a third person.

Data were independently abstracted by at least two review authors and cross‐checked. Where data have been, or may have been, collected but were not reported, further clarification was sought from the trialist.

Included trial data were processed as described in the Cochrane Reviewers' Handbook (Alderson 2004). Where appropriate, data were combined quantitatively using the Cochrane statistical package Review Manager. Synthesis was performed using a fixed‐effect model. We reported relative risks (RR) for dichotomous data and weighted mean differences (WMD) for continuous data, accompanied by 95% confidence intervals (CI). Evidence of heterogeneity came from visual inspection of the data or from the chi squared test for heterogeneity (at 10%) and the I squared test. If evidence of significant heterogeneity was identified, potential sources of heterogeneity were explored within populations, interventions, outcomes and settings. If there was no obvious plausible explanation for heterogeneity a random‐effects model was also to be used for combining data and sensitivity of the results to the choice of model discussed. If synthesis was inappropriate, a narrative overview was to be undertaken.

Statistical advice was sought regarding the use of data from cross‐over trials. This might include using data from the first arm only, entering the data as 'Other Data' only without applying statistical tests of significance, the use of individual patient data or using approximations of the standard deviations from other properly reported trials.

If possible, sensitivity analyses were to be done to assess the impact of study quality. Planned subgroup analyses were to consider differences with types of incontinence (stress, mixed).

Results

Description of studies

Twenty two reports of studies were identified, 17 were eligible for analysis and five studies were excluded because they did not describe individual trials (reasons for exclusion are detailed in the table of 'Characteristics of excluded studies' ). Ten primary studies were identified (see list of included studies). There were seven additional reports related to one or two of the primary references (Zinner 2002 was a part of Dmochowski 2003; Bump 2003, Bump 2002, Bump 2001 and Norton 2000 were part of Norton 2002; whilst Van Leeuwen 2004 was part of Ghoniem 2005). The primary trials reviewed were all supported by Eli Lily and Company.

Ethics committee approval was clearly stated in the full text journal articles included. Intention to treat analysis (ITT) and completers (per protocol) analysis were also clearly stated in these articles.

All full text studies included were multicentre RCTs, except (Mulcahy 1996). The selection criteria varied between trials with detailed selection criteria available in the full text articles (see Characteristics of included studies' ). In seven of the ten primary references, patients included had predominantly SUI based on a clinical diagnosis with or without urodynamics in some of the patients. The inclusion criteria for clinical SUI were predominant symptoms of SUI with incontinence episode frequency (IEF) per week ranging from equal to or greater than four (Norton 2002) to equal to or greater than 14 (Cardozo 2004), diurnal frequency equal to or below seven (others) to below nine (Millard 2003), nocturia below two (others) to below three (Dmochowski 2003), positive cough stress test (CST) and positive stress pad test (SPT greater than 2 g per hour). The exclusion criteria for clinical SUI were predominant urge urinary incontinence (UUI), first sensation on bladder filling of less than 100 ml, maximum bladder capacity less than 400 ml. Cardozo 2004 included only women with urodynamic stress incontinence (USI) whereas (Mulcahy 1996) included patients with MUI and UUI as well. Kinchen 2005 on the other hand, included different types of incontinence in women and stratification into SUI, UUI or MUI was made based on an unvalidated questionnaire. These authors excluded patients who were pregnant, breast‐feeding, with active UTI, previous duloxetine trial participants, and patients with arrhythmias, liver disease, seizures and unstable cardiac illnesses. 
 
 The trial design in the full text publications were clear. Cardozo 2004; Dmochowski 2003; Kinchen 2005; Millard 2003 and Van Kerrebroeck 2004 described clear baseline comparability of treatment and placebo arms for age, race, body mass index (BMI), use of hormone replacement therapy, prior continence surgery, performing PFMT (Millard 2003) and severity of incontinence at baseline. Norton 2002 described balance for all of the above except for prior continence surgery, PFMT and hormone replacement therapy. The baseline characteristics of participants was described in all full text publications.

The trials available addressed only four out of the nine hypotheses considered in this review (see 'Objectives'). There were no studies comparing duloxetine with surgery, with other classes of drugs, with other routes of administration, with another type of SNRI or in combination with another drug.

Duloxetine was the only SNRI used in these trials. All studies randomised patients to receive duloxetine 80 mg/day or placebo except for Mulcahy 1996 (duloxetine 20 mg/day versus placebo); Norton 2002 (duloxetine 20 mg, 40 mg and 80 mg per day versus placebo); Cardozo 2004 (duloxetine 80 mg/day for 4 weeks then escalating to 120 mg/day versus placebo) and Ghoniem 2005 (duloxetine 40 mg twice daily versus PFMT versus combined duloxetine 40 mg twice daily and PFMT versus placebo). In the trial by Kinchen 2005, the patients were allowed to use other modalities of treatment (anticholinergics, PFMT and oestrogens) simultaneously with trial drugs.

In general, from the full text articles (except Cardozo 2004 and Kinchen 2005), the patients underwent a two‐week screening period, followed by a two‐week placebo lead‐in and then were randomised (double‐blind) to 12 weeks of duloxetine or placebo or alternative treatment. They were seen at intervals of four weeks throughout the treatment period. Patients were required to complete a bladder diary (over a week) at no drug lead in (baseline), placebo lead‐in, four weeks, eight weeks and 12 weeks of treatment. The Incontinence Quality of Life (I‐QoL) and Patient Global Impression of Improvement (PGI‐I) were completed at each visit (PGI‐I was done only after randomisation) and Patient Global Impression of Severity (PGI‐S) was done at baseline. Adverse events and compliance were assessed during the visits. In Mulcahy 1996, the treatment was for three weeks only, whereas Kinchen 2005 addressed long‐term effects of duloxetine by treating patients for 36 weeks.

The primary outcome measure in all studies was incontinence episode frequency per week (IEF) except Kinchen 2005, where I‐QoL was used. Manning 2005 described the changes using the King's Health Questionaire (KHQ) score as well. IEF was reported as the median of percentage change in IEF. Where data were given in percentages only (e.g. percentage of patients feeling 'better' or 'much better' on the PGI‐I scale), the actual number was calculated and adjusted to the closest figure.

Some outcome measures were reported in a way that could not be formally used in the review (such as reporting median percentage change using median values, reporting P values only and not reporting means and standard deviations). However, they are described in a narrative form in the results section.

Outcome measures not in our protocol, comparing duloxetine versus placebo were: 
 (a) Willingness to consider surgery (Cardozo 2004) 
 (b) Beck Depression inventory ‐II (BDI‐II) (Dmochowski 2003) 
 (c) Increase in voiding interval (Dmochowski 2003; Millard 2003; Norton 2002; Van Kerrebroeck 2004)

The following definitions were used for improvement in incontinence or responders to treatment in the trials: 
 (a) Cardozo 2004 ‐ at least a 50% decrease in IEF with treatment; 
 (b) Dmochowski 2003 ‐ 50% to 100% decrease in IEF with treatment; 
 (c) Millard 2003 ‐ 50% to 100% decrease in IEF with treatment; 
 (d) Mulcahy 1996 ‐ equal to or greater than 30% decrease in volume of stress pad test; 
 (e) Van Kerrebroeck 2004 ‐ 50% to 100% decrease in IEF with treatment; 
 (f) Zinner 1998 ‐ greater than 70% improvement from baseline‐to‐endpoint.

The following subgroup analyses were performed in some of the included trials. 
 (a) Efficacy assessment for patients with MUI and SUI were reported in Bump 2003 (Norton 2002). 
 (b) Efficacy assessment based on severity of symptoms at baseline, i.e. patients with greater than 14 IEF/ week (severe) (Cardozo 2004; Dmochowski 2003; Kinchen 2005; Millard 2003; Norton 2002; Van Kerrebroeck 2004). 
 (c) Efficacy assessment based on severity of symptoms at baseline using PGI‐S (Cardozo 2004). 
 (d) Changes in I‐QoL based on the different domains of the I‐QoL questionnaire (Avoidance/ limited behaviour, Psychosocial and social embarrassment) (Cardozo 2004; Dmochowski 2003; Millard 2003). 
 (e) Efficacy assessment based on presence or absence of Intrinsic Sphincter Deficiency (ISD) (Cardozo 2004). 
 (f) Changes in IEF and I‐QoL were analysed according to visit (4 weeks, 8 weeks and 12 weeks of treatment) (Dmochowski 2003; Kinchen 2005; Van Kerrebroeck 2004). 
 (g) Changes to the Kings Health questionnaire (Manning 2005)

Risk of bias in included studies

The methodology of individual trials is summarised in the table of 'Characteristics of Included Studies'.

All trials were described as double blind, randomised placebo controlled trials and description of randomisation and concealment of allocation into groups were found to be adequate. CONSORT guidelines (CONSORT 2001) were adhered to in some full text publications (Cardozo 2004; Norton 2002; Van Kerrebroeck 2004). Description of sample number calculation, drop‐outs, withdrawals and adverse events was clear in all the full text publications.

Several trials failed to provide measures of dispersion, such as standard deviations, for outcome measures using continuous variables and provided only median and P values (Cardozo 2004; Dmochowski 2003; Ghoniem 2005; Manning 2005; Millard 2003; Norton 2002; Van Kerrebroeck 2004).

One trial (Manning 2005) failed to provide the dosage of Duloxetine used, precluding analyses.

Effects of interventions

Demographics and clinical characteristics:

The total number of participants in the ten included trials was 3944, with 2132 randomised to a placebo arm. Patients randomised to the treatment arms were: 1280 patients to duloxetine 80 mg/day, 170 patients to duloxetine 40 mg/day, 26 patients to duloxetine 30 mg/day, 227 patients to duloxetine 20 mg/day, 306 unspecified dosage, 50 patients to PFMT and 52 patients to a combination of duloxetine 80 mg/day and PFMT. Participants were recruited from Africa, North America, South America, Australia and Europe. The mean age of patients ranged from 49.4 years to 54 years in the included studies and the majority of patients were Caucasian (range 86.2% to 99%).

At baseline, MUI was reported in 30.9% (Bump 2003) and 47% (Mulcahy 1996) of the recruited patients. Patients with MUI had apparently more severe symptoms compared with patients having pure SUI (Bump 2003). From the studies which had sub‐categorised IEF/ week into severe (greater than or equal to 14 per week) and less than 14 per week, the prevalence of the severe UI was 50.6% (Van Kerrebroeck 2004), 54.% (Millard 2003) and 29.5% (Norton 2002). Compliance in the placebo arm was better than in the duloxetine arm, 78% (duloxetine) and 83% (placebo) (Norton 2002).

Comparison 1: The use of a SNRI versus placebo or no treatment (Comparison 01)

Eight eligible trials addressed this comparison (Cardozo 2004; Dmochowski 2003; Ghoniem 2005; Millard 2003; Mulcahy 1996; Norton 2002; Van Kerrebroeck 2004; Zinner 1998). All eight compared duloxetine (treatment arm) with placebo. Zinner 1998 and Norton 2002 assessed different doses of duloxetine (20 mg/day, 30 mg/day (only Zinner 1998), 40 mg/day and 80 mg/day) against placebo, whereas Cardozo 2004 used duloxetine 80 mg/day versus placebo for 4 weeks then escalating to 120 mg/day versus placebo. The other trials only used duloxetine 80 mg/day versus placebo. None of the trials compared duloxetine with an open no treatment group.

Subjective cure

The numbers 'not cured' in the groups allocated duloxetine 80 mg daily were lower than in the placebo group in the three trials with data (98.2% versus 92.8%, overall relative risk 0.97; 95% confidence interval 0.93 to 1.0; P equals 0.04, Comparison 01.01). The estimated size of effect was small: about three more cured amongst every 100 treated. Data from the 40 mg dose groups were consistent with this, although there appeared to be little or no difference between treated and control groups in the 20 mg stratum. The results for improvement also favour duloxetine (Comparison 01.03) although the effect size estimates are larger (about 15 more improved amongst every 100 treated) and the differences more highly significant. The reason for heterogeneity in this comparison is the larger effect estimate in Cardozo 2004.

The data from the lower dose comparisons are consistent with those for the 80 mg comparison.

Objective parameters

The few available data from Zinner 1998; Comparison 01.10, 01.11 show overall reduction in urine loss in favour of duloxetine, although generally this is not statistically significant. Data describing objective cure using pad tests or stress tests were available from only one trial (Norton 2002) and showed no clear difference between the groups (Comparisons 01.13 and 01.14). The results for cure based on bladder diaries were similar to those for subjective cure: most participants were not cured. The difference was marginally statistically significant and equivalent to about three more cured for every 100 treated (Comparison 01.15).

Quality of life domain

The groups allocated duloxetine had significantly greater overall improvement in Incontinence Quality of Life (I‐QoL, Comparison 01.16) weighted mean difference for 80 mg 4.5; 95% confidence interval 2.83 to 6.18 (Z=5.27, P less than 0.00001) and no significant heterogeneity. General health status (Patient Global Impression of Improvement, PGI‐I) (MetaView 01.18) also improved significantly more in the duloxetine groups (80 mg: relative risk for better health status 1.24, 95% confidence interval 1.14 to 1.36; Z=4.75, P less than 0.00001).

Analyses of continuous data on IEF (Zinner 1998) showed better results in duloxetine groups: typical difference in per cent change about 20% (see Other Data Table 01.09 and Comparison 01.20).

Adverse events and discontinuation

The majority of those allocated duloxetine (86%) in five trials with data reported adverse effects. However, more than half (58%) of those allocated placebo also reported adverse effects (Comparison 01.19.01), suggesting that about one in three in the duloxetine group reported adverse effects that were due to duloxetine. The overall relative risk was 1.31 (95% confidence interval 1.24 to 1.39, Comparison 01.19.01) (Cardozo 2004; Dmochowski 2003; Ghoniem 2005; Kinchen 2005; Norton 2002; Van Kerrebroeck 2004).

Across the seven trials, 18% in the duloxetine groups withdrew compared with 4.1% in the placebo groups (RR 4.5; 95% CI 3.44 to 5.89, Comparison 01.19.02) (Cardozo 2004; Dmochowski 2003; Ghoniem 2005; Kinchen 2005; Manning 2005; Norton 2002; Van Kerrebroeck 2004). Thus, about one in eight allocated duloxetine stopped treatment due to an effect of duloxetine.

Nausea is the commonest side effect across the trials (ranging from 22.7% to 25.1%) in the duloxetine arm and the main cause for discontinuation.

Other outcome measures

(1) Number of micturitions per 24 hrs (Other Data Table 01.12) 
 In the single trial with data (Norton 2002) duloxetine 20 mg, 40 mg and 80 mg/day were associated with greater reduction in the number of micturitions per 24 hrs when compared with placebo, but the differences were not statistically significant.

(2) Increase in voiding interval (Other Data Table 01.21) 
 In the same trial groups treated with duloxetine 20 mg/day, 40 mg/day and 80 mg/day had significantly longer intervals between voids (range = 15 to 20 minutes for duloxetine) compound with two to nine minute for placebo) (Dmochowski 2003; Millard 2003; Norton 2002; Van Kerrebroeck 2004).

(3) Willingness to consider surgery (Cardozo 2004). 
 Of the women who were unsure or somewhat interested in, or strongly interested in surgery for SUI prior to randomisation to the duloxetine arm, 20.4% indicated that they were somewhat or strongly not interested in surgery after treatment with duloxetine. This was in contrast with none of the patients in the placebo arm feeling they were not interested in surgery (P equals 0.001).

Comparison 2: The use of a SNRI versus conservative non‐pharmacological therapies (Comparison 02)

Only one trial addressed this comparison (Ghoniem 2005) and the only outcome measure available for analysis is the median percentage change in IEF ('Other data tables' section 02.01). The group treated with duloxetine 80 mg/day had significantly reduced IEF compared with PFMT (57% decrease versus 35% decrease P less than 0.05).

Comparison 3: The use of a SNRI versus surgery (Comparison 03)

No eligible trials were found.

Comparison 4: The use of a SNRI versus other classes of drugs (Comparison 04)

No eligible trials were found.

Comparison 5: The use of a particular SNRI versus another (if available) (Comparison 05)

No eligible trials were found.

Comparison 6: The use of a particular SNRI given orally versus given via another route (rectal, intra‐vesical, topical) (Comparison 06)

No eligible trials were found.

Comparison 7: The use of a higher dose of a SNRI versus a lower dose (Comparison 07)

Two of the primary trials addressed this comparison (Norton 2002 ‐ duloxetine 20 mg/day, duloxetine 40 mg/day and duloxetine 80 mg/day; Zinner 1998 ‐ duloxetine 20 mg/day, duloxetine 30 mg/day and duloxetine 40 mg/day).

Subjective parameters

In terms of subjective cure, a higher dose of duloxetine was not significantly better than a lower dose (Comparison 07.01). Similarly, for patients reporting 'improvement', a higher dose was not significantly better than a lower dose (Comparison 07.03). However, the confidence intervals include the size of difference observed in the placebo‐controlled trial.

Analysis of continuous data on IEF showed a greater improvement with higher doses (see Norton 2002, Other Data Tables 07.09 and Comparison 07.22, Zinner 1998).

Objective parameters

Data from Zinner 1998 and Norton 2002 failed to demonstrate a significant benefit of a higher dose of duloxetine when the following objective parameters are assessed: stress pad test (Comparison 0.7.10 and 07.13), 24 hour pad weights (Comparison 07.11), cough stress test (Comparison 07.14) and bladder diary (Comparison 07.15). Confidence intervals were wide, however.

When quality of life was assessed, a higher duloxetine dose was not significantly better than a lower dose. The doses investigated ranging from 20 to 40 mg/day (Zinner 1998) (Comparison 07.16). However, using PGI‐I (Comparison 07.18), the higher dose duloxetine groups had better scores, significantly so for 80 mg/day versus 20 mg/day (Z = 2.22, P equals 0.03).

There were no significant differences in the overall adverse effects (Comparison 07.19) and discontinuation rates (Comparison 07.20) between a higher and lower dose of duloxetine, although in all the comparisons a higher dose of duloxetine was associated with more withdrawals (Norton 2002).

Other outcome measures

(1) Number of micturitions per 24 hrs (Other Data Table 07.12, Comparison 07.12) 
 Results for duloxetine 80 mg/day were not significantly better than for duloxetine 20 mg/day or 40 mg/day and for duloxetine 40 mg/day was not significantly better than for 20 mg/day (Norton 2002).

(2) Increase in voiding interval (Other Data Table 07.21, Comparison 07.21) 
 It was not possible to compare duloxetine 80 mg/day, 40 mg/day and 20 mg/day doses directly. In the trial with data (Norton 2002), the increase in voiding interval was 16 to 18 minutes in all groups.

Comparison 8: The use of a SNRI in combination with another drug versus the other drug alone (Comparison 08)

No eligible trials were found.

Comparison 9: The use of a SNRI in combination with conservative non‐pharmacological therapy versus the conservative non‐pharmacological therapy alone (Comparison 09)

One of the trials addressed this comparison (Ghoniem 2005) but the only outcome measure available for analysis is the median percentage change in IEF. Combination of duloxetine 80 mg/day and PFMT was reported to have significantly better results than PFMT alone with regards to reduction in IEF (57% versus 35% P less than 0.05).

Discussion

The WHO sponsored International Consultation on Incontinence 2002 (ICI 2002) does not recommend pharmacological agents for the treatment of SUI. However, since surgery has recognisable complications, effective drug therapy with an acceptable adverse event profile would be a welcome addition to current management options. Alpha‐adrenergic agonists have been previously reviewed (Alhasso 2005) and it was concluded that they do not have a role in current management of stress incontinence. Duloxetine, on the other hand, is the first serotonin‐noradrenaline reuptake inhibitor (SNRI) to be extensively investigated in phase III trials. The authors in most of these trials report an improvement in incontinence episode frequency (IEF) of about 50%, which was sustained throughout the course of treatment, compared to placebo with a reported median percentage reduction of 26.3% to 40%.

All trials included in this review were described as double blind, randomised placebo controlled trials and description of randomisation and concealment of allocation into groups were found to be adequate. CONSORT guidelines (CONSORT 2001) were clearly adhered to in some full text publications (Cardozo 2004; Norton 2002; Van Kerrebroeck 2004). Description of sample number calculation, drop‐outs, withdrawals and adverse events was clear in all the full text publications. Selection bias is unlikely given the double blind randomised study design in all trials. All trials were conducted under the auspices of a single pharmaceutical company. 
 
 Meta‐analysis of the reviewed articles revealed that duloxetine 80 mg daily groups did better than placebo groups in terms of subjective cure and improvement in incontinence (based on individual trial definitions), although the estimated effect on cure was modest ‐ only about 3% extra patients cured. Other outcomes favouring duloxetine were the Incontinence quality of life (I‐QoL) and Patient Global Impression of Improvement (PGI‐I) measures. Median percentage change in IEF was used as the primary outcome measure by most authors but the data were not presented in a way that would allow a meta‐analysis to be performed (data not reported as mean and standard deviation, except Zinner 1998). Thus, although almost all the trials showed a significant reduction in the median percentage of IEF change, we were not able to derive a typical effect size across the trials. We do realise, however, that the trialist may have had valid reasons for reporting the IEF data in this manner such as skewed distribution of data, but these are not clear to us. There was no significant benefit attributable to duloxetine in terms of change in 24 hour pad weight, cough stress test or stress pad test, but the data describing these were relatively sparse.

Most participants in the duloxetine groups reported an adverse effect. Around 30% of these could be directly attributed to duloxetine because over half in the placebo group also reported adverse effects. Overall 17% stopped treatment in the duloxetine group, about 75% of whom were attributable to duloxetine. This is a higher rate than in trials of anticholinergic treatment for urgency. This adverse effect profile was judged to be acceptable by the authors of the trial reports; however, further longer‐term experience is required to rule out rare potentially serious complications.

There was little evidence with which to judge whether a higher dose of duloxetine versus a lower dose. Patients' global impression of improvement suggested benefit of higher doses, especially duloxetine 80 mg/day over 20 mg/day. This was not supported by the report from Zinner 1998, which suggested a higher dose of duloxetine had no added benefit although it is important to note that only duloxetine 20 mg/day, duloxetine 30 mg/day and duloxetine 40 mg/day were investigated. The trial groups treated with larger doses all had higher rates of adverse events and withdrawals, but the confidence intervals around these were all wide and none was statistically significant. An initial dose of 80 mg is now generally recommended but it was not clear from the evidence reviewed why this dose has been chosen.

There were no trials comparing duloxetine with surgery nor other drugs (e.g. anticholinergics). Limited evidence from a single trial suggested that duloxetine alone or in combination with PFMT might have an advantage over PFMT alone (Ghoniem 2005). The authors, however, were not explicit about compliance rates for those treated with PFMT even though it is well recognised that compliance plays a significant role in success of PFMT. 
 
 It is important to take into account the differences in the duration of study between the trials ‐ Mulcahy 1996 (three weeks), Zinner 1998 (six weeks), others (12 weeks) and Kinchen 2005 (36 weeks). Dmochowski 2003 and Millard 2003 found an enhanced placebo effect after the fourth week of the study which would not have been seen in the shorter duration studies, however we found no significant heterogeneity between trials with regard to the primary outcomes analysed. From available data, it is not clear what the duration of treatment should be or whether the efficacy of duloxetine is sustainable after long‐term treatment. Collecting long‐term data to determine sustained efficacy should now be a priority.

The trialist recruited patients with moderate to severe symptoms (based on PGI‐S). In the trial by Kinchen 2005, patients had less severe symptoms and although the response to duloxetine was similar to the phase III trials, the placebo response was higher. Thus there was no clear difference between duloxetine and placebo in this trial. Patients currently on pelvic floor exercises ranged from 5% to 19% of participants from the studies where these data were available. It was not made clear, in most trials (except Kinchen 2005), what proportion of these women had failed (or tried) PFMT or behavioural modification prior to recruitment. The trialist also failed to provide evidence that the recruited patients found these measures to be unsatisfactory, and thus required alternative treatment.

No specific drug interactions were mentioned in the trial reports. SNRI do have recognised interactions such as with anti‐depressants, non‐steroidal anti‐inflammatory drugs and barbiturates. Duloxetine's use for treating depression will be considered in a separate Cochrane review; central nervous system effects may, in part at least, be a reason for the improvement in measures of quality of life described in this review.

Authors' conclusions

Implications for practice.

This review suggests that duloxetine versus placebo improves incontinence episodes and quality of life in patients with predominantly SUI, although the differences were small. There were, however, only limited usable objective data in favour of duloxetine and it is uncertain if these benefits are sustained long‐term. Data from the trials included in this review would suggest that duloxetine commonly has side effects ‐ most often nausea ‐ and that these lead to between 1 in 6 and 1 in 8 stopping treatment. This raises questions about its acceptability in normal clinical practice. There is no RCT‐based evidence at present with which to judge whether or not duloxetine is a cost‐effective alternative to current approaches to care such as pelvic floor muscle training and surgery.

Implications for research.

Research is needed to clarify whether management policies incorporating duloxetine are clinically effective and cost‐effective compared with current approaches such as pelvic floor muscle training, injection therapy and more invasive surgery in patients with varying severity of stress urinary incontinence. In trials that include participants with mixed incontinence, attempts should be made to assess the effects of duloxetine, if any, on the urge incontinence component. Such RCTs will need to be long‐term to clarify when treatment is prolonged and after it has stopped (a) whether the short‐term efficacy of duloxetine is sustained, and (b) whether the safety profile remains acceptable.

What's new

Date	Event	Description
10 October 2008	Amended	Converted to new review format.

History

Protocol first published: Issue 2, 2004
 Review first published: Issue 3, 2005

Date	Event	Description
16 August 2007	New search has been performed	minor update, one study added
25 May 2005	New citation required and conclusions have changed	Substantive amendment

Data and analyses

Comparison 1. A SNRI is better than placebo or no treatment.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Numbers not cured during treatment	3		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.1 Duloxetine 80 mg daily	3	1396	Risk Ratio (M‐H, Fixed, 95% CI)	0.97 [0.93, 1.00]
1.2 Duloxetine 40 mg daily	1	255	Risk Ratio (M‐H, Fixed, 95% CI)	0.89 [0.79, 1.01]
1.3 Duloxetine 20 mg daily	1	260	Risk Ratio (M‐H, Fixed, 95% CI)	0.99 [0.89, 1.09]
2 Numbers not cured after treatment	0		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3 Numbers not improved during treatment	6		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3.1 Duloxetine 80 mg daily	4	1733	Risk Ratio (M‐H, Fixed, 95% CI)	0.74 [0.68, 0.81]
3.2 Duloxetine 40 mg daily	1	67	Risk Ratio (M‐H, Fixed, 95% CI)	0.64 [0.45, 0.90]
3.3 Duloxetine 20 mg daily	2	160	Risk Ratio (M‐H, Fixed, 95% CI)	0.55 [0.40, 0.75]
3.4 Duloxetine 30 mg daily	1	60	Risk Ratio (M‐H, Fixed, 95% CI)	0.45 [0.27, 0.75]
4 Number with nocturia during treatment	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
5 Number with nocturia after treatment	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
6 Number with urgency during treatment	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
7 Number with urgency after treatment	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
8 Number of pad per 24hrs	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
9 Number of incontinence episodes per week ‐ IEF (quantification of symptoms)			Other data	No numeric data
9.1 Duloxetine 20 mg daily			Other data	No numeric data
9.2 Duloxetine 40 mg daily			Other data	No numeric data
9.3 Duloxetine 80 mg daily			Other data	No numeric data
9.4 Duloxetine 120 mg daily			Other data	No numeric data
10 Mean reduction in volume/weight of urine loss on pad test on Stress Pad Test	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
10.1 Duloxetine 20 mg daily	1	68	Mean Difference (IV, Fixed, 95% CI)	8.8 [‐1.43, 19.03]
10.2 Duloxetine 30 mg daily	1	60	Mean Difference (IV, Fixed, 95% CI)	0.60 [‐7.55, 8.75]
10.3 Duloxetine 40 mg daily	1	67	Mean Difference (IV, Fixed, 95% CI)	7.50 [‐1.55, 16.55]
10.4 Duloxetine 80 mg daily	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
11 Mean reduction in volume/weight urine loss on pad test (quantification of symptoms ‐ 24HPW)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
11.1 Duloxetine 20 mg daily	1	68	Mean Difference (IV, Fixed, 95% CI)	31.4 [5.15, 57.65]
11.2 Duloxetine 30 mg daily	1	60	Mean Difference (IV, Fixed, 95% CI)	8.20 [‐15.66, 32.06]
11.3 Duloxetine 40 mg daily	1	67	Mean Difference (IV, Fixed, 95% CI)	10.20 [‐10.08, 30.48]
11.4 Duloxetine 80 mg daily	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
12 Number of micturiations per 24hrs (quantification of symptoms)			Other data	No numeric data
12.1 Duloxetine 20 mg daily			Other data	No numeric data
12.2 Duloxetine 40 mg daily			Other data	No numeric data
12.3 Duloxetine 80 mg daily			Other data	No numeric data
13 Numbers not cured (objective ‐ SPT)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
13.1 Duloxetine 80 mg daily	1	227	Risk Ratio (M‐H, Fixed, 95% CI)	0.97 [0.79, 1.18]
13.2 Duloxetine 40 mg daily	1	225	Risk Ratio (M‐H, Fixed, 95% CI)	0.90 [0.73, 1.11]
13.3 Duloxetine 20 mg daily	1	224	Risk Ratio (M‐H, Fixed, 95% CI)	1.08 [0.89, 1.30]
14 Numbers not cured (objective ‐ CST)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
14.1 Duloxetine 80 mg daily	1	232	Risk Ratio (M‐H, Fixed, 95% CI)	0.96 [0.87, 1.07]
14.2 Duloxetine 40 mg daily	1	230	Risk Ratio (M‐H, Fixed, 95% CI)	0.88 [0.78, 0.99]
14.3 Duloxetine 20 mg daily	1	230	Risk Ratio (M‐H, Fixed, 95% CI)	1.03 [0.94, 1.13]
15 Numbers not cured (objective ‐ Diary)	3		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
15.1 Duloxetine 80 mg daily	3	1396	Risk Ratio (M‐H, Fixed, 95% CI)	0.97 [0.93, 1.00]
15.2 Duloxetine 40 mg daily	1	255	Risk Ratio (M‐H, Fixed, 95% CI)	0.89 [0.79, 1.01]
15.3 Duloxetine 20 mg daily	1	260	Risk Ratio (M‐H, Fixed, 95% CI)	0.99 [0.89, 1.09]
16 General health status (I‐QoL)	6		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
16.1 Duloxetine 80 mg daily	5	1835	Mean Difference (IV, Fixed, 95% CI)	4.50 [2.83, 6.18]
16.2 Duloxetine 40 mg daily	1	67	Mean Difference (IV, Fixed, 95% CI)	5.6 [0.87, 10.33]
16.3 Duloxetine 30 mg daily	1	60	Mean Difference (IV, Fixed, 95% CI)	7.4 [3.55, 11.25]
16.4 Duloxetine 20 mg daily	1	68	Mean Difference (IV, Fixed, 95% CI)	9.4 [3.26, 15.54]
17 General health status (PGI‐I)	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
18 General Health Status ('very much better', 'much better' or 'a little better' from PGI‐I)	6		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
18.1 Duloxetine 80 mg daily	6	1879	Risk Ratio (M‐H, Fixed, 95% CI)	1.25 [1.14, 1.36]
18.2 Duloxetine 40 mg daily	1	275	Risk Ratio (M‐H, Fixed, 95% CI)	1.39 [0.98, 1.97]
18.3 Duloxetine 20 mg daily	1	276	Risk Ratio (M‐H, Fixed, 95% CI)	1.16 [0.80, 1.68]
19 Adverse events	7		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
19.1 Numbers experiencing adverse events	6	2480	Risk Ratio (M‐H, Fixed, 95% CI)	1.31 [1.24, 1.39]
19.2 Numbers withdrawing from treatment/trial arm	7	3108	Risk Ratio (M‐H, Fixed, 95% CI)	4.50 [3.44, 5.89]
19.3 Number changing dose of treatment	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
20 Reduction in incontinence episodes per week ‐ IEF (quantification of symptoms)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
20.1 Duloxetine 20 mg daily	1	68	Mean Difference (IV, Fixed, 95% CI)	8.0 [3.00, 13.00]
20.2 Duloxetine 30 mg daily	1	60	Mean Difference (IV, Fixed, 95% CI)	2.90 [‐0.57, 6.37]
20.3 Duloxetine 40 mg daily	1	67	Mean Difference (IV, Fixed, 95% CI)	1.70 [‐2.92, 6.32]
21 Increase in voiding interval (minutes)			Other data	No numeric data
21.1 Duloxetine 20 mg daily			Other data	No numeric data
21.2 Duloxetine 40 mg daily			Other data	No numeric data
21.3 Duloxetine 80 mg daily			Other data	No numeric data
22 Numbers not cured after 1 year (objective)	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
23 Number not cured after 5 years (objective)	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
24 Frequency of sensation of urgency (quantification of symptoms)	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
25 Socioeconomic measures	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

1.1. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 1 Numbers not cured during treatment.

1.3. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 3 Numbers not improved during treatment.

1.9. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 9 Number of incontinence episodes per week ‐ IEF (quantification of symptoms).

Number of incontinence episodes per week ‐ IEF (quantification of symptoms)
Study	% change ‐ Dulox.	% change ‐ placebo	p value
Duloxetine 20 mg daily
Norton 2002	‐44	‐40	0.6
Duloxetine 40 mg daily
Norton 2002	‐59	‐40	0.02
Duloxetine 80 mg daily
Cardozo 2004	‐54.7	‐26.3	< 0.002
Dmochowski 2003	‐50	‐27.5	< 0.001
Ghoniem 2005	‐56.5	28.9	0.05
Millard 2003	‐53.6	‐40	0.05
Norton 2002	‐58	‐40	0.04
Van Kerrebroeck 2004	‐50	‐29.3	0.002
Zinner 1998
Duloxetine 120 mg daily
Cardozo 2004	‐64	‐26.3	< 0.001

1.10. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 10 Mean reduction in volume/weight of urine loss on pad test on Stress Pad Test.

1.11. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 11 Mean reduction in volume/weight urine loss on pad test (quantification of symptoms ‐ 24HPW).

1.12. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 12 Number of micturiations per 24hrs (quantification of symptoms).

Number of micturiations per 24hrs (quantification of symptoms)
Study	no. of voids(Dulox)	no. of voids(Plac)	p value
Duloxetine 20 mg daily
Norton 2002	‐0.8	‐.05	0.4
Duloxetine 40 mg daily
Norton 2002	‐1.0	‐0.5	0.09
Duloxetine 80 mg daily
Norton 2002	‐0.8	‐0.5	0.3

1.13. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 13 Numbers not cured (objective ‐ SPT).

1.14. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 14 Numbers not cured (objective ‐ CST).

1.15. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 15 Numbers not cured (objective ‐ Diary).

1.16. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 16 General health status (I‐QoL).

1.18. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 18 General Health Status ('very much better', 'much better' or 'a little better' from PGI‐I).

1.19. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 19 Adverse events.

1.20. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 20 Reduction in incontinence episodes per week ‐ IEF (quantification of symptoms).

1.21. Analysis.

Comparison 1 A SNRI is better than placebo or no treatment, Outcome 21 Increase in voiding interval (minutes).

Increase in voiding interval (minutes)
Study	Treatment	Placebo	p value
Duloxetine 20 mg daily
Norton 2002	16	5	0.05
Duloxetine 40 mg daily
Norton 2002	18	5	0.02
Duloxetine 80 mg daily
Dmochowski 2003	20	1.7	< 0.001
Millard 2003	20.4	8.5	< 0.001
Norton 2002	18	5	0.006
Van Kerrebroeck 2004	15	3.8	0.0001

Comparison 2. SNRI is better than conservative non‐pharmacologic therapies.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Number of incontinence episodes per week ‐ IEF (quantification of symptoms)			Other data	No numeric data
1.1 Duloxetine 20 mg daily			Other data	No numeric data
1.2 Duloxetine 40 mg daily			Other data	No numeric data
1.3 Duloxetine 80 mg daily			Other data	No numeric data
2 General Health status (PGI‐I)	1	92	Odds Ratio (M‐H, Fixed, 95% CI)	2.44 [1.05, 5.65]
3 General health status (I‐QoL)	1	99	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]

2.1. Analysis.

Comparison 2 SNRI is better than conservative non‐pharmacologic therapies, Outcome 1 Number of incontinence episodes per week ‐ IEF (quantification of symptoms).

Number of incontinence episodes per week ‐ IEF (quantification of symptoms)
Study	med.% decrease Dulox	med.% decrease PFMT	p value
Duloxetine 80 mg daily
Ghoniem 2005	57	35	<0.05

2.2. Analysis.

Comparison 2 SNRI is better than conservative non‐pharmacologic therapies, Outcome 2 General Health status (PGI‐I).

2.3. Analysis.

Comparison 2 SNRI is better than conservative non‐pharmacologic therapies, Outcome 3 General health status (I‐QoL).

Comparison 7. A higher dose of a SNRI is better than a lower dose.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Numbers not cured during treatment (quantification of symptoms)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
1.1 Duloxetine 80 mg daily versus Duloxetine 40 mg daily	1	246	Risk Ratio (M‐H, Fixed, 95% CI)	1.08 [0.94, 1.23]
1.2 Duloxetine 80 mg daily versus Duloxetine 20 mg daily	1	251	Risk Ratio (M‐H, Fixed, 95% CI)	0.97 [0.87, 1.09]
1.3 Duloxetine 40 mg daily versus Duloxetine 20 mg daily	1	251	Risk Ratio (M‐H, Fixed, 95% CI)	0.90 [0.80, 1.03]
2 Numbers not cured after treatment (quantification of symptoms)	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 Numbers not improved during treatment (quantification of symptoms)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
3.1 Duloxetine 30 mg daily versus 20 mg daily	1	60	Risk Ratio (M‐H, Fixed, 95% CI)	0.73 [0.41, 1.30]
3.2 Duloxetine 40 mg daily versus 20 mg daily	1	67	Risk Ratio (M‐H, Fixed, 95% CI)	1.03 [0.66, 1.61]
3.3 Duloxetine 40 mg daily versus 30 mg daily	1	59	Risk Ratio (M‐H, Fixed, 95% CI)	1.42 [0.80, 2.53]
3.4 Duloxetine 80 mg daily versus 20 mg daily	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.5 Duloxetine 80 mg daily versus 30 mg daily	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.6 Duloxetine 80 mg daily versus 40 mg daily	0	0	Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Number with nocturia during treatment (quantification of symptoms)	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
5 Number with nocturia after treatment (quantification of symptoms)	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
6 Number with urgency during treatment (quantification of symptoms)	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
7 Number with urgency after treatment (quantification of symptoms)	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
8 Number of pad changes per 24hrs (quantification of symptoms)	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
9 Number of incontinence episodes per 24hrs ‐ IEF (quantification of sysmptms)			Other data	No numeric data
9.1 Duloxetine 80 mg daily versus 40 mg daily			Other data	No numeric data
9.2 Duloxetine 80 mg daily versus 20 mg daily			Other data	No numeric data
9.3 Duloxetine 40 mg daily versus 20 mg daily			Other data	No numeric data
10 Mean reduction in volume/weight of urine loss on pad test (quantification of symptoms ‐ SPT)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
10.1 Duloxetine 30 mg daily versus 20 mg daily	1	60	Mean Difference (IV, Fixed, 95% CI)	‐8.2 [‐19.01, 2.61]
10.2 Duloxetine 40 mg daily versus 20 mg daily	1	67	Mean Difference (IV, Fixed, 95% CI)	‐1.30 [‐12.81, 10.21]
10.3 Duloxetine 40 mg daily versys 30 mg daily	1	59	Mean Difference (IV, Fixed, 95% CI)	6.90 [‐2.80, 16.60]
11 Mean reduction in volume/weight of urine loss on pad test (quantification of symptoms ‐ 24HPW)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
11.1 Duloxetine 30 mg daily versus 20 mg daily	1	60	Mean Difference (IV, Fixed, 95% CI)	‐23.20 [‐52.10, 5.70]
11.2 Duloxetine 40 mg daily versus 20 mg daily	1	67	Mean Difference (IV, Fixed, 95% CI)	‐21.20 [‐47.22, 4.82]
11.3 Duloxetine 40 mg daily versus 30 mg daily	1	59	Mean Difference (IV, Fixed, 95% CI)	2.0 [‐21.60, 25.60]
12 Number of micturiations per 24hrs (quantification of symptoms)			Other data	No numeric data
12.1 Duloxetine 80 mg daily versus 40 mg daily			Other data	No numeric data
12.2 Duloxetine 80 mg daily versus 20 mg daily			Other data	No numeric data
12.3 Duloxetine 40 mg daily versus 20 mg daily			Other data	No numeric data
13 Numbers not cured (objective ‐ SPT)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
13.1 Duloxetine 80 mg daily versus 40 mg daily	1	224	Risk Ratio (M‐H, Fixed, 95% CI)	1.08 [0.86, 1.34]
13.2 Duloxetine 80 mg daily versus 20 mg daily	1	223	Risk Ratio (M‐H, Fixed, 95% CI)	0.90 [0.74, 1.09]
13.3 Duloxetine 40 mg daily 20 mg daily	1	221	Risk Ratio (M‐H, Fixed, 95% CI)	0.83 [0.68, 1.02]
14 Numbers not cured (objective ‐ CST)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
14.4 Duloxetine 80 mg daily versus 40 mg daily	1	226	Risk Ratio (M‐H, Fixed, 95% CI)	1.10 [0.96, 1.25]
14.5 Duloxetine 80 mg daily versus 20 mg daily	1	226	Risk Ratio (M‐H, Fixed, 95% CI)	0.93 [0.84, 1.03]
14.6 Duloxetine 40 mg daily 20 mg daily	1	224	Risk Ratio (M‐H, Fixed, 95% CI)	0.85 [0.76, 0.96]
15 Numbers not cured (objective ‐ Diary)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
15.4 Duloxetine 80 mg daily versus 40 mg daily	1	246	Risk Ratio (M‐H, Fixed, 95% CI)	1.08 [0.94, 1.23]
15.5 Duloxetine 80 mg daily versus 20 mg daily	1	251	Risk Ratio (M‐H, Fixed, 95% CI)	0.97 [0.87, 1.09]
15.6 Duloxetine 40 mg daily 20 mg daily	1	251	Risk Ratio (M‐H, Fixed, 95% CI)	0.90 [0.80, 1.03]
16 General health status (I‐QoL)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
16.1 Duloxetine 30 mg daily versus 20 mg daily	1	60	Mean Difference (IV, Fixed, 95% CI)	‐2.0 [‐7.91, 3.91]
16.2 Duloxetine 40 mg daily versus 20 mg daily	1	67	Mean Difference (IV, Fixed, 95% CI)	‐3.80 [‐10.32, 2.72]
16.3 Duloxetine 40 mg daily versus 30 mg daily	1	43	Mean Difference (IV, Fixed, 95% CI)	‐1.80 [‐7.21, 3.61]
17 General health status (PGI‐I)	0	0	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
18 General Health Status ('Very much better', 'much better' or 'a little better' from PGI‐I)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
18.1 Duloxetine 80 mg daily versus 40 mg daily	1	277	Risk Ratio (M‐H, Fixed, 95% CI)	1.19 [0.89, 1.58]
18.2 Duloxetine 80 mg daily versus 20 mg daily	1	278	Risk Ratio (M‐H, Fixed, 95% CI)	1.42 [1.04, 1.94]
18.3 Duloxetine 40 mg daily 20 mg daily	1	275	Risk Ratio (M‐H, Fixed, 95% CI)	1.19 [0.86, 1.66]
19 Numbers experiencing adverse events	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
19.4 Duloxetine 80 mg daily versus 40 mg daily	1	277	Risk Ratio (M‐H, Fixed, 95% CI)	1.05 [0.83, 1.32]
19.5 Duloxetine 80 mg daily versus 20 mg daily	1	278	Risk Ratio (M‐H, Fixed, 95% CI)	1.16 [0.91, 1.48]
19.6 Duloxetine 40 mg daily 20 mg daily	1	275	Risk Ratio (M‐H, Fixed, 95% CI)	1.10 [0.86, 1.42]
20 Numbers withdrawing from treatment/trial arm	1		Risk Ratio (M‐H, Fixed, 95% CI)	Subtotals only
20.1 Duloxetine 80 mg daily versus 40 mg daily	1	277	Risk Ratio (M‐H, Fixed, 95% CI)	1.21 [0.67, 2.19]
20.2 Duloxetine 80 mg daily versus 20 mg daily	1	278	Risk Ratio (M‐H, Fixed, 95% CI)	1.59 [0.83, 3.05]
20.3 Duloxetine 40 mg daily 20 mg daily	1	275	Risk Ratio (M‐H, Fixed, 95% CI)	1.32 [0.67, 2.61]
21 Increase in voiding interval (minutes)			Other data	No numeric data
21.1 Duloxetine 80 mg daily versus 40 mg daily			Other data	No numeric data
21.2 Duloxetine 80 mg daily versus 20 mg daily			Other data	No numeric data
21.3 Duloxetine 40 mg daily versus 20 mg daily			Other data	No numeric data
22 Reduction in incontinence episodes per week ‐ IEF (quantification of symptoms)	1		Mean Difference (IV, Fixed, 95% CI)	Subtotals only
22.1 Duloxetine 30 mg daily versus 20 mg daily	1	60	Mean Difference (IV, Fixed, 95% CI)	‐5.1 [‐8.00, ‐0.20]
22.2 Duloxetine 40 mg daily versus 20 mg daily	1	67	Mean Difference (IV, Fixed, 95% CI)	‐6.30 [‐12.08, ‐0.52]
22.3 Duloxetine 40 mg daily versus 30 mg daily	1	59	Mean Difference (IV, Fixed, 95% CI)	‐1.20 [‐5.71, 3.31]
23 Numbers not cured after 1 year (objective)	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
24 Number not cured after 5 years (objective)	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
25 Socioeconomic measures	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
26 Other outcomes	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
28 Frequency of sensation of urgency (quantification of symptoms)	0	0	Odds Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

7.1. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 1 Numbers not cured during treatment (quantification of symptoms).

7.3. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 3 Numbers not improved during treatment (quantification of symptoms).

7.9. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 9 Number of incontinence episodes per 24hrs ‐ IEF (quantification of sysmptms).

Number of incontinence episodes per 24hrs ‐ IEF (quantification of sysmptms)
Study	Higher dose	p value (placebo)	Lower dose	p value (placebo)
Duloxetine 80 mg daily versus 40 mg daily
Norton 2002	‐58	0.04	‐59	0.02
Duloxetine 80 mg daily versus 20 mg daily
Norton 2002	‐58	0.04	‐44	0.6
Duloxetine 40 mg daily versus 20 mg daily
Norton 2002	‐59	0.02	‐44	0.6

7.10. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 10 Mean reduction in volume/weight of urine loss on pad test (quantification of symptoms ‐ SPT).

7.11. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 11 Mean reduction in volume/weight of urine loss on pad test (quantification of symptoms ‐ 24HPW).

7.12. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 12 Number of micturiations per 24hrs (quantification of symptoms).

Number of micturiations per 24hrs (quantification of symptoms)
Study	mict./day (higher)	p value	mict./day (lower)	p value
Duloxetine 80 mg daily versus 40 mg daily
Norton 2002	‐0.8	0.3	‐1.0	0.09
Duloxetine 80 mg daily versus 20 mg daily
Norton 2002	‐0.8	0.3	‐0.8	0.4
Duloxetine 40 mg daily versus 20 mg daily
Norton 2002	‐1.0	0.09	‐0.8	0.4

7.13. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 13 Numbers not cured (objective ‐ SPT).

7.14. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 14 Numbers not cured (objective ‐ CST).

7.15. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 15 Numbers not cured (objective ‐ Diary).

7.16. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 16 General health status (I‐QoL).

7.18. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 18 General Health Status ('Very much better', 'much better' or 'a little better' from PGI‐I).

7.19. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 19 Numbers experiencing adverse events.

7.20. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 20 Numbers withdrawing from treatment/trial arm.

7.21. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 21 Increase in voiding interval (minutes).

Increase in voiding interval (minutes)
Study	Higher dose	Lower dose	p value
Duloxetine 80 mg daily versus 40 mg daily
Norton 2002	18	18
Duloxetine 80 mg daily versus 20 mg daily
Norton 2002	18	16
Duloxetine 40 mg daily versus 20 mg daily
Norton 2002	18	16

7.22. Analysis.

Comparison 7 A higher dose of a SNRI is better than a lower dose, Outcome 22 Reduction in incontinence episodes per week ‐ IEF (quantification of symptoms).

Comparison 9. SNRI plus conservative non‐pharmacological therapy better than conservative non‐pharmacological therapy alone.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Number of incontinence episodes per 24hrs ‐ IEF (quantification of symptoms)			Other data	No numeric data
1.1 Duloxetine 20 mg daily			Other data	No numeric data
1.2 Duloxetine 40 mg daily			Other data	No numeric data
1.3 Duloxetine 80 mg daily			Other data	No numeric data
2 General health status (PGI‐I)	1	100	Odds Ratio (M‐H, Fixed, 95% CI)	1.28 [0.55, 2.96]
3 General health status (I‐QoL)	1	100	Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]

9.1. Analysis.

Comparison 9 SNRI plus conservative non‐pharmacological therapy better than conservative non‐pharmacological therapy alone, Outcome 1 Number of incontinence episodes per 24hrs ‐ IEF (quantification of symptoms).

Number of incontinence episodes per 24hrs ‐ IEF (quantification of symptoms)
Study	med.% Combined	med.% PFMT	p value
Duloxetine 80 mg daily
Ghoniem 2005	57	35	<0.05

9.2. Analysis.

Comparison 9 SNRI plus conservative non‐pharmacological therapy better than conservative non‐pharmacological therapy alone, Outcome 2 General health status (PGI‐I).

9.3. Analysis.

Comparison 9 SNRI plus conservative non‐pharmacological therapy better than conservative non‐pharmacological therapy alone, Outcome 3 General health status (I‐QoL).

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Cardozo 2004.

Methods	RCT (Double blind placebo‐controlled) Multi‐centre ITT
Participants	n= 109 Age range 33‐75 years Recruitment period May 2001 to October 2002 Study period: 8 weeks Baseline characteristics: Mean IEF/ week = 24.7 Mean I‐QOL = 53.6 PGI‐S = Moderately to severely abnormal bladder function was experienced by approximately 89% Prior continence surgery = 16.4% Patients on hormone replacement therapy = 47.3% Inclusion: Urodynamic stress incontinence, > or =14/week IEF, scheduled for continence surgery, Exclusion: Not specified
Interventions	I (52): placebo for 8 weeks II (46): duloxetine 40 mg twice daily for 4 weeks, then escalating to 60 mg twice daily for 4 weeks
Outcomes	Median decrease in IEF: I: 2.9 II: 7.1 ( P < 0.001) Decrease in IEF at 4 weeks: I: 26.3%, II: 54.7% (no data to extract exact numbers) Decrease in IEF at 8 weeks: I: 28.6%, II: 64% (no data to extract exact numbers) PGI‐I (very much better/much better): I: 4/52, II:17/51 PGI‐I (a little better or no change): I:42/52, II:31/51 PGI‐I (a little worse to very much worse): I: 6/52, II: 3/51 I‐QoL (mean +‐ SD): I: 2.4 +/‐ 9.4, II:10.6 +/‐ 19.1 Adverse events: I: 37/52, II: 43/46. Adverse events: Nausea, constipation, headache, dry mouth, fatigue, dizziness, insomnia, somnolence and vomiting. Serious adverse events: cardiovascular (not significantly different in both arms). Discontinued medication because of adverse events: I: 3/54, II: 18/55
Notes	Ref ID {19088} Authors found that 20% of women with USI were somewhat or not very much keen on continence surgery being on Duloxetine compared with 0% of those in the placebo arm. Three non responders at 80 mg daily became responders at 120 mg daily. Increasing the dose from 80 mg to 120 mg daily did not increase side effects.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Dmochowski 2003.

Methods	RCT (double blind, placebo controlled) ITT
Participants	n = 683 women Age range 22‐84 Baseline characteristics: Mean IEF/week(SD): I: 19(14.6) II: 18.2(14.3) Mean I‐QoL(SD): I:64.3(17.7) II: 62(20.2) [Mean IEF/ week = 18.2 ] [Mean I‐QOL score = 62 ] PGI‐S = Moderately to severely abnormal bladder function was experienced by approximately 68% Prior continence surgery = 12.2% Inclusion: Female 18 years or older, bothersome SUI for at least 3 months, predominant SUI with IEF at least 7 per week, absence of predominant UUI, daily frequency less than 8, nocturia less than 3, positive CST and SPT Exclusion: Unable to tolerate bladder filling to 400 ml, first sensation less than 100 ml on bladder filling, treatment with antidepressant medication
Interventions	I (339): placebo II (344): duloxetine(40 mg twice daily) Duration of treatment: 12 weeks Oral Tablets
Outcomes	Numbers not cured during treatment: I: 319/339, II: 308/344 Absolute decrease IEF/week: I: ‐3 II: ‐7 (P < 0.001) I‐QoL (mean +‐ SD): I: 6.8 (+/‐ 13.8), II: 11.1 (+/‐14.8) I‐QoL(mean score change): I: 134/339 (7.2) II: 213/344 (12.1) Adverse events: I: 170/339 II: 255/344. Nausea, fatigue, insomnia, dry mouth, constipation, somnolence, dizziness, headache, diarrhoea Withdrawal: I:14/339 II: 82/344. Nausea, fatigue, insomnia, somnolence, dizziness, blurred vision
Notes	Ref ID {16484} SNRI vs placebo No power calculation
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Ghoniem 2005.

Methods	RCT (double blind, controlled, multicentre, ITT)
Participants	n=201 women Age range = 29‐75 years Baseline characteristics: Median IEF/week (range): I: 18.9(10.3, 299.4) II: 22(13,140.9) III: 18.3(6.4,78.5) IV 19.4 (10,70.5) Mean I‐QoL(SD): I:64.9(17.1) II: 61.4(22.2) III: 59.8(20.6) IV:61.6 (22.3) Median pads per week (range) I:9.8(0,43.1) II:8.6(0,45.9) III:8.1(0,44.0) IV: 9.7(0,53.5) Mean PFM grade(SD) I:not done II:5.2(1.7) III:not done IV: 5.2(1.7) Prior continence surgery = I:6 II: 8 III: 4 IV: 4 Inclusion: Female 18 ‐ 75 years, urodynamic stress urinary continence, no detrussor overactivity within 6 months before entry or a positive cough stress test and normal micturiation freuency (less than 8 voids per day), predominant SUI at least 2 per day Exclusion: Advanced pelvic organ prolapse, acitve or recurrent UTI, continence surgery within 1 year, current device or pharmaceutical incontinence treatment, prior hip fracture or replacement and prior formal PFMT with a continence or physical therapist.
Interventions	I (47): placebo + imitation PFMT (no active treatment) II (50): PFMT + placebo (PFMT only) III (52): Duloxetine 40mg twice daily + imitation PFMT (Duloxetine only) IV (52): Duloxetine 40mg twice daily + PFMT (combined) Duration of treatment: 12 weeks Oral Tablets PFMT regimen: Qualified instructors gave subjects assigned to PFMT verbal instructions and manual feedback, during a digital pelvic examination, the subject was asked to contract the muscles she would to prevent urine or flatus loss and hold for 6‐8 seconds. The examiner confirmed the pelvic floor muscles were contracted without dominant contractions of the abdominal, gluteal or hip adductor muscles. Once proper contractions were confirmed, subjects received written instructions to perform 3 sets of 10 long (6 to 8 seconds) and 2 sets of 10 rapid (1 to 2 seconds) contractions 4 days weekly, for a total of 200 contractions weekly. Imitation PFMT: The subject was instructed to cross her legs at the ankles with her knees and hips flexed while sitting or supine and abduct the hips, holding the contractions for 6 to 8 seconds while the therapist palpated the hip abductors and abdominal muscles and confirmed the abductors were contracted without dominan contractions of abdominal muscles.
Outcomes	Median % change in pad use: I:‐10.5 II:‐24.8 III:‐35.3 IV:‐45.7 Median % change IEF/week: I: ‐28.9 II: ‐34.7 III: ‐56.9 IV: ‐57.4 I‐QoL (mean change): I: 4.8, II: 7.8, III:8.3, IV:13.1 PGI‐I (percentage change): I:42.2, II:65.3, III:54.0, IV:70.6 Adverse events: Group I and Group II: 58/97 Group III and Group IV: 85/104. (adverse events in decending order) Nausea, dizziness, dry mouth , constipation, insomnia, somnolence, asthenia. Withdrawal: Group I and Group II:8/97 Group III and Group IV: 28/104. Nausea, insomnia, dizziness, somnolence, asthenia.
Notes	SNRI vs placebo vs PFMT vs combination SNRI and PFMT Measures of cure and improvement not used, median % change in IEF used.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Kinchen 2005.

Methods	RCT (Double blind, placebo‐controlled) naturalistic study
Participants	n = 451 Mean age 53.5 years for placebo and 52.7 years for Duloxetine Inclusion: women > 18 years with > or = 1 IEF/ week with any type of incontinence Exclusion: Pregnancy, breast‐feeding, active UTI, participation in previous Duloxetine trial, arrhythmias, hypertension, liver disease, seizures, unstable cardiac illness. Duration of study: 36 weeks
Interventions	I (227): placebo II (224): duloxetine 80 mg daily (40 mg bd)
Outcomes	Mean change in I‐QoL, PGI‐I
Notes	The adjusted mean change in I‐QoL total score was greater than placebo, but not significant in LOCF analysis
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Manning 2005.

Methods	RCT (Double blind, randomised placebo‐controlled)
Participants	n = 617 women Age range 28‐86 Baseline characteristics: Overall mean IEF/week: 25/week 436(71%) had baseline IEF >14 Prior continence surgery = 12.2% Inclusion: Adult female with SUI diagnosed by urodynamics within previous 12 months or using a simple 2‐question instrument (S/UIQ). Subject had to have at least 7 incontinence episodes per week by diary and at least twice as many SUI episoded as UUI by S/UIQ. Exclusion: not specified
Interventions	I( 311): Placebo II (306): Duloxetine (dose not specified) Duration of study = 6 weeks
Outcomes	Median % change IEF/week = I:‐22.2, II:‐46.4 (p<0.001) PGI‐I(% change in "better" category) I: 44 II: 64(P < 0.01) Change in KHQ total score = I:‐2.6, II:‐9.2 (p<0.0001)
Notes	SNRI vs placebo IEF median % change described and KHQ used to measure QoL Changes in PGI were obtained from a graph manually
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Millard 2003.

Methods	RCT (double blind placebo controlled) ITT
Participants	n = 458 women Age = over 18 years Centres: 38 study centres Duration of study: 12 weeks Baseline characteristics: Median IEF/ week = 7.5 Prior continence surgery = 12.1% Inclusion: SUI => 3 months duration, => 7 Incontinence episodes/week, Diurnal frequency <9/day, Nocturnal frequency <3/ night, absence of predominant urge incontinence, positive stress pad test >2 g, positive cough‐stress test (visualisation of urine leakage with cough Exclusion: first sensation to void at 100 ml, no sensation at all during filling Patient compliance: each visit, by counting unused medication and monitoring diaries
Interventions	I (231): placebo II (227): duloxetine 80 mg daily (40 mg twice daily) orally Duration of treatment: 12 weeks
Outcomes	Numbered NOT cured: I: 217/231, II: 211/227 Numbers NOT improved: I:100/231, II: 92/227 I‐QoL: mean +/‐ SD: I:6.5 (+/‐ 17.3), II: 11.1 (+/‐ 16.4) Adverse events: I: 137/231, II: 173/227. Nausea, headache, Insomnia, Constipation, Dry mouth, Dizziness, Fatigue, somnolence, Anorexia, Vomiting, Increased sweating. Adverse events leading to discontinuation: I: 4/231, II: 39/227. Nausea, anxiety and Insomnia
Notes	Trial conducted in 4 continents "significant efficacy of Duloxetine in treating women with SUI. 54% median reduction in incontinence episodes with Duloxetine was associated with significant improvement in I‐QoL and PGI‐I"
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Mulcahy 1996.

Methods	RCT (double blind, placebo controlled)
Participants	n = 92 women Mean age 55 SUI: I(11) II(21), UUI: I(8) II(9), MUI: I(18) II(25) Inclusion: Daily incontinence over previous 4 weeks, evidence of incontinence from voiding/leakage diary, positive 1 hr SPT and/ or 24 hr PT during 2 weeks placebo lead‐ in Exclusion: not specified
Interventions	I (37): placebo II (55): duloxetine(20 mg daily) Duration of treatment: 3 weeks Oral tablets
Outcomes	Number NOT improved during treatment: I: 19/37 II: 13/55
Notes	Ref ID {4618} Abstract SSRI vs placebo Primary aim was to determine efficacy and safety of Duloxetine. Definition of improvement: =>30% reduction in urine loss on SPT and IEF Adverse events not specified, "no difference in adverse events between both arms of the study" Results in patients with MUI were equivocal and Duloxetine could not be distinguished from placebo by urodynamic evaluation in any incontinence type subgroup No means and no SD for any continuous data, only median % change in IEF results. Figures in abstract not labelled for proper interpretation and lacks minor gridline.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Norton 2002.

Methods	RCT (double blind, placebo controlled) ITT
Participants	n = 553 I: (138) mean age 50.2 (SD = 8.9), mean weight 78.7 (18), baseline frequency/24 hr 9.4 (2.5) II: (138) mean age 49.4 (7.3), mean weight 81.2 (21.8), baseline frequency/24 hrs 9.2 (2.6) III: (137): mean age 49.4 (8.0), mean weight 77.8 (16.5), baseline frequency/24 hrs 10.0 (2.6) IV: (140) mean age 49.3 8.8), mean weight 76.2 (14.5), baseline frequency/24 hrs 9.8 (2.5) Mean IEF/ week = 9.2‐10.0 Inclusion: Female with incontinence aged 18‐65 years, clinical SUI for at least 3 months, IEF at least 4 per week, predominantly SUI, 7 or fewer episodes of frequency per day, 2 or fewer episodes nocturia per night, CST positive, SPT more than 2 g per hour Exclusion: Previous continence/ prolapse operation, first sensation less than 100 ml on bladder filling, unable to tolerate bladder filling to 400 ml, no sensation on bladder filling to 400 ml
Interventions	I (138): placebo II (138): duloxetine(20 mg daily) III (137): duloxetine(20 mg twice daily) IV (140): Duloxetine (40 mg twice daily) Duration of treatment: 12 weeks Oral tablets
Outcomes	Numbers not cured during treatment (subjective) I: 6/138 II: 10/138 III: 14/137 IV: 17/140 Numbers not cured during treatment(objective‐SPT<2 g) I: 24/138 II: 28/138 III: 26/137 IV: 27/140 Numbers not cured during treatment(objective‐CST) I: 20/138 II: 26/138 III: 25/137 IV:26/140 Median % change in SPT (range) I: ‐30%(‐100‐2175%) II: ‐11%(‐100‐3240%) III: ‐43%(‐100‐5800%) IV: ‐29%(‐100‐12333%) No of micturitions/24 hrs I: ‐0.6 II: ‐1.0(P = 0.05) III: ‐1.2(P = 0.003) IV: ‐1.4(P = <0.001) Median % change of IEF/wk: I: ‐41% II: ‐54% (P = 0.06) III: ‐59%( P = 0.002 ) IV: ‐64%(P = <0.001) PGI‐I(% change in "very much better" and "much better" categories) I: 27 II: 31(P = 0.5) III: 37(P = 0.09) IV: 44( P = 0.005) Mean change in I‐QOL I: 5.8 II: 5.3( P = 0.6) III: 7.8(P = 0.16) IV: 9.3(P = 0.03) Adverse effects I: 61/138 II: 62/138 III: 68/137 IV: 73/140 Nausea, headache, somnolence, dizziness, menorrhagia, diarrhoea, constipation, dry mouth, insomnia, sinusitis, fatigue, nasopharyngitis, URTI Withdrawals I: 7/138 II: 13/138 III: 17/137 IV: 21/140
Notes	Ref ID {14640} SSRI vs placebo SSRI higher dose vs SSRI lower dose Study shows a significant increase in mean void interval with Duloxetine, although not in protocol. Also stratified for severity of incontinence and analysed for those with greater than 14 IEF/wk at baseline. Placebo response rate was a 40% reduction in IEF.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Van Kerrebroeck 2004.

Methods	RCT (double blind, placebo controlled) ITT
Participants	n = 494 women Age 24‐83 Baseline characteristics: Mean I‐QOL score = 66.7 PGI‐S = Moderately to severely abnormal bladder function in approximately 69.5% Prior continence surgery = 7.7% Currently using PFMT = 18.8% Inclusion: Over 18 years old, predominant SUI with IEF of at least 7 per week, daily frequency less than 8, nocturia twice or less per night, absence of predominant UUI, positive CST and SPT after bladder filling Exclusion: First sensation at less than 100 ml on bladder filling, unable to tolerate bladder filling to 400 ml
Interventions	I (247): placebo II (247): duloxetine (40 mg twice daily) Duration of treatment: 12 weeks Oral tablets
Outcomes	Median IEF/wk I: ‐3.0(95%CI ‐36.8, ‐20.0) II: ‐6.0( P = 0.002, 95%CI ‐57.1,‐42.9) PGI‐I (% feeling "very much better", "much better" or "a little better") I: 48.4% II: 64.8% (P = 0.008) Mean change in I‐QOL I: 4.1 II: 5.5 (P = 0.127 95%CI ‐0.5, 4.1) Adverse events I: 158/247 II: 200/247 Nausea, dizziness, insomnia, attention disturbance, dry mouth, constipation, fatigue, headache, sweating increased, vomiting, somnolence, tremor. Withdrawal I: 12/247 II: 53/247 (Nausea, dizziness, insomnia, attention disturbance)
Notes	Ref ID {16935} SSRI vs placebo Median IEF was used as there were a few extreme outliers. Also found that Duloxetine increased average voiding interval.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Zinner 1998.

Methods	RCT (Double blind , placebo‐controlled) ITT
Participants	n = 286 (SUI= 140, MUI= 146)
Interventions	I (34): placebo II (34): duloxetine 20 mg daily III (26): duloxetine 30 mg daily IV (33): duloxetine 40 mg daily Duration of study: 6 weeks
Outcomes	SPT ‐ mean (SD): I: ‐4.7 (15.5), II: ‐13.5 (26.2), III: ‐5.3 (16.3), IV: ‐12.2 (21.7) IEF (number/week) mean (SD): I: ‐5.9 (7.6), II: ‐13.9 (12.8), III: ‐8.8 (6.1), IV: ‐7.6 (11.3) 24HPW (g) mean (SD): I: ‐9.4 (43.3), II: ‐40.8 (65), III: ‐17.6 (49.2), IV: ‐19.6 (41.4) I‐QoL mean (SD): I: 2.6 (8.8), II: 12 (16), III: 10 (6.4), IV: 8.2 (10.8)
Notes	Ref ID {5696}
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

24HPW+ 24hours pad weight; bd = twice daily; CST = Cough stress test; Excl = Exclusion criteria; IEF = Incontinence episode frequency; IEF = Incontinence episode frequency; Incl = Inclusion criteria; IPMFT= Imitation Pelvic Muscle Floor Training; I‐QoL = Incontinence quality of life; ISI = Incontinence severity index; ITT = Intention to treat; MUI = Mixed (stress and urge)urinary incontinence; NS = Not stated; PGI‐S = Patient Global Impression of severity; PT = Pad test; PVR = Post void residual volume; RCT = Randomised Controlled Trial; SD = Standard deviation; SPT = Stress pad test; SUI = Stress urinary incontinence; UI = Urinary incontinence; UUI = Urge urinary incontinence.

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Hurley D	This is an abstract describing a meta‐analysis of 4 RCTs (references for all 4 are not specified) mainly describing the safety profile of the Duloxetine 40 mg vs placebo
Skinner 2004	Report described drug pharmacokinetics only.
Yalcin 1998	This study describes the validity of Incontinence Severity Index (ISI) only
Yalcin 2003	This is a study combining two other studies: Bump 2003 and Dmochowski 2003.
Yalcin 2004	This study reports on 4 other RCTs (Norton 2002, Dmochowski 2003, Van Kerrebroeck 2004 and Millard 2004) and assess the placebo effect on SUI only

Contributions of authors

Paramananthan Mariappan, Zoe Balantyne and Ammar Alhasso reviewed the publications and collated suitable data for entry into the tables of comparison. 
 Paramananthan Mariappan and Ammar Alhasso completed the analysis. 
 Paramananthan Mariappan wrote the Background, description of included studies, results and discussion. This was confirmed by Ammar Alhasso. James N'Dow ensured accuracy of data collection and monitored all aspects of this review. 
 Ammar Alhasso coordinated the project with the Cochrane Incontinence Group and also wrote to the authors of the reviewed publications for data clarification.

Declarations of interest

None known

Edited (no change to conclusions)