BJOG: an International Journal of Obstetrics and Gynaecology DOI: 1 0 . 1 1 1 1 / j . 1 4 7 1 - 0 5 2 8 . 2 0 0 4 . 0 0 0 6 7 .

x
March 2004, Vol. 111, pp. 249 –257

Duloxetine versus placebo in the treatment of European and

Canadian women with stress urinary incontinence
Philip van Kerrebroecka, Paul Abramsb, Rainer Langec, Mark Slackd,
Jean-Jacques Wyndaelee, Ilker Yalcinf, Richard C. Bumpf,*,
for the Duloxetine Urinary Incontinence Study Group
Objective To assess the efficacy and safety of duloxetine in women with stress urinary incontinence.
Design Randomised double-blind, placebo-controlled clinical trial.
Setting Fort-six centres in seven European countries and Canada.
Population Four hundred and ninety-four women aged 24– 83 years identified as having predominant
symptoms of stress urinary incontinence using a clinical algorithm that was 100% predictive of urodynamic
stress urinary incontinence in a subgroup of 34 women.
Methods The case definition included a predominant symptom of stress urinary incontinence with a weekly
incontinence episode frequency
7, the absence of predominant symptoms of urge incontinence, normal
diurnal and nocturnal frequencies, a bladder capacity
400 mL and both a positive cough stress test and
positive stress pad test. Subjects completed two urinary diaries prior to randomisation and three diaries
during the active treatment phase of the study, each completed during the week prior to monthly visits.
Subjects also completed quality of life questionnaires at each visit. Safety was assessed by the evaluation of
treatment-emergent adverse events, discontinuation of treatment because of adverse events, serious adverse
events, vital sign measurements, electrocardiograms (ECG) and clinical laboratory tests.
Intervention After a two-week placebo lead-in, women received placebo or duloxetine 40 mg BD for
12 weeks.
Main outcome measures The percentage decrease in incontinence episode frequency and the change in the
Incontinence Quality of Life (I-QOL) questionnaire total score were prespecified as co-primary outcome
variables in the protocol.
Results Incontinence episode frequency decreased significantly with duloxetine compared with placebo
(median decrease of 50% vs 29%; P = 0.002) with comparable improvements in the more severely
incontinent subgroup (those experiencing at least 14 incontinence episodes per week at baseline; 56% vs
27% decreases; P < 0.001). The primary analysis of I-QOL scores did not reveal a significant difference
between treatment groups, due primarily to the carrying forward of low scores from patients who
discontinued treatment very early due to duloxetine-associated adverse events. The increase in I-QOL scores
was significantly greater for duloxetine than for placebo at each of the three postrandomisation visits after 4,
8, and 12 weeks of treatment. Discontinuation rates for adverse events were higher for duloxetine (22% vs
5%; P < 0.001) with nausea being the most common reason for discontinuation (5.3%). Nausea tended to be
mild to moderate, not progressive, and transient.
Conclusions The findings support duloxetine as a potential treatment for women with stress urinary
incontinence.

a
INTRODUCTION
Department of Urology, University Hospital Maastrecht,
Netherlands Stress urinary incontinence is defined by the Interna-
b
Bristol Urological Institute, Southmead Hospital, UK tional Continence Society as the complaint of involuntary
c
Department of Gynaecology, DRK-Krankenhaus, Alzey, leakage of urine on effort or exertion, or on sneezing or
Germany coughing.1 It occurs because a positive urethral pressure
d
Hinchingbrooke & Addenbrooke’s Hospitals, Cambridge, gradient over bladder pressure cannot be maintained during
UK abrupt increases in abdominal pressure. Stress urinary
e
Department of Urology, University of Antwerp, Belgium incontinence is the most common type of urinary inconti-
f
Lilly Research Laboratories, Indianapolis, Indiana, USA nence in women, with 78% of women presenting with the
* Correspondence: Dr R. C. Bump, Eli Lilly and Company Corporate symptom of stress urinary incontinence in either pure
Center, Indianapolis, Indiana 46285, USA. (49%) or mixed (29%) forms.2 Using year 2000 population
D RCOG 2004 BJOG: an International Journal of Obstetrics and Gynaecology www.blackwellpublishing.com/bjog
250 P. VAN KERREBROECK ET AL.
figures, approximately 18 million women over the age of
20 years have the symptom of pure stress urinary inconti-
nence while 29 million have stress urinary incontinence in
pure or mixed forms in the eight countries that were
included in the current study. Based on data showing that
stress urinary incontinence is bothersome and severe in
17% to 24% of incontinent women,3 it can be calculated
that between three and seven million of these women have
severe or bothersome stress urinary incontinence.
Currently, the most accepted forms of therapy for female
stress urinary incontinence are pelvic floor muscle training,
various types of behavioural interventions and continence
surgery. Only a small minority of women with bothersome
stress urinary incontinence undergo continence surgery. For
example, in the United States only 130,000 did so in 1997,4
fewer than 5% of American women with bothersome stress
urinary incontinence. An undetermined number of women
with stress urinary incontinence use various forms of pelvic
floor muscle training in an attempt to improve their
continence, however, they often fail to comply with a
pelvic floor muscle training program both acutely and
chronically. In one study, the majority of women (61%)
prescribed pelvic floor muscle training for stress urinary
incontinence were not compliant; 15% trained only once
weekly, 3% once monthly and 43% not at all.5 There are
currently no data to explain this relatively poor compliance,
however, it is possible that some women are discouraged by
the relatively slow response with pelvic floor muscle
training and thus fail to dedicate the 15 –20 weeks recom-
mended to determine their response.6 There is currently no
globally approved pharmacological agent for the treatment
of women with stress urinary incontinence.
Numerous animal studies have implicated serotonin
(5-HT) and norepinephrine in the neural control of lower
urinary tract function. In non-rodent species, serotonergic
agonists suppress parasympathetic activity and enhance
sympathetic and somatic activity in the lower urinary
tract,7 – 9 effects that promote urine storage by relaxing
the bladder and increasing outlet resistance. Noradrenergic
agonists and antagonists variably affect sympathetic and
somatic activity in the lower urinary tract, depending on the
adrenergic receptor subtype with which they interact.10 – 15
Duloxetine hydrochloride is a balanced and potent in-
hibitor of serotonin and norepinephrine reuptake, and is
being developed for the treatment both of stress urinary
incontinence and of major depression. Preclinical pharma-
cological data demonstrate a superiority of duloxetine with
respect to its relative affinity in binding to the norepineph-
rine and serotonin transport sites compared with venlafax-
ine, the only available selective dual reuptake inhibitor.16
Based upon the continence-promoting properties of
serotonin and norepinephrine, animal studies were con-
ducted with duloxetine to determine its effects on lower
urinary tract function. Duloxetine significantly increased
bladder capacity and sphincteric muscle activity in the cat
acetic acid bladder model, effects that were reversed by the
nonselective serotonergic antagonist methiothepin and by
prazosin and LY53857, alpha-1 adrenergic and 5-HT2 sero-
tonergic receptor antagonists, respectively.17 It has further
been demonstrated that the storage-promoting effects of
duloxetine on the lower urinary tract are unique to dual
serotonin and norepinephrine reuptake inhibition in a single
molecule, and are not duplicated by the administration of
single reuptake inhibitors alone or in combination.18
Duloxetine’s ability to increase rhabdosphincter contrac-
tility via stimulation of pudendal motor neuron alpha-1
adrenergic and 5-HT2 receptors, is believed to be the under-
lying mechanism responsible for its efficacy in women with
stress urinary incontinence reported recently from a phase
two randomised controlled trial.19 This manuscript reports
the results from a phase three trial performed in Western
Europe and Canada. The aim of this study was to further
characterise the efficacy and tolerability of duloxetine
hydrochloride as a potential pharmacological treatment
for female stress urinary incontinence.
METHODS
Women aged 18 years and older with a clinical diagnosis
of bothersome stress urinary incontinence of at least three
months duration were invited to participate in this double-
blind, placebo-controlled, randomised clinical trial con-
ducted at 46 study centres in Belgium, Canada, Denmark,
France, Germany, the Netherlands, Sweden and the United
Kingdom. The institutional review board for each site
approved the study and written informed consent was
obtained from all participants. To enrol, women must have
reported a predominant symptom of stress urinary inconti-
nence with seven or more incontinence episodes per week,
a diurnal urinary frequency of less than eight per day, a
nocturnal urinary frequency of two or less per day and the
absence of predominant symptoms of urge incontinence.
All patients underwent a supine bladder infusion. Those
who were unable to tolerate the filling to 400 mL or who
experienced a first sensation of bladder filling <100 mL
were excluded. After bladder filling, both a positive cough
stress test and a positive stress pad test were required for
inclusion. This clinical algorithm has been demonstrated
to predict urodynamic stress incontinence with an accu-
racy of 92%.19 In the current study, a subset of 34 patients
had filling phase multichannel cystometry at five centres,
all of whom had urodynamic stress incontinence con-
firmed according to International Continence Society
standards.1
After a two-week screening period and a two-week no-
drug lead-in period followed by a two-week single-blind,
placebo lead-in period, patients were randomised under
double-blind conditions to 12 weeks of treatment with
duloxetine 40 mg BD or placebo (Fig. 1). All patients
received two identical capsules twice daily and were seen
at four-week intervals throughout the treatment period.
D RCOG 2004 Br J Obstet Gynaecol 111, pp. 249 – 257
 DULOXETINE TREATMENT OF STRESS URINARY INCONTINENCE
Fig. 1. Study design and the timing of acquisition of urinary diary and quality of life measurements reported.
Patients were randomised using a computerised voice-
response system at a central location. A stratified random-
isation using baseline incontinence episode frequency (<14
or
14 episodes/week) was performed to prevent potential
imbalance in incontinence severity. Treatment assignments
were also balanced at each investigative site.
The percent change in incontinence episode frequency
from baseline to endpoint (calculated from patient com-
pleted daily paper diaries) was one of the two prespecified
primary outcome variables. The daily diaries were also
used to collect the number of voids, the time of voids and
the time study medication was taken. Two diaries were
completed prior to randomisation, one during the second
week of the no-drug lead-in and one during the second
week of the blinded placebo lead-in. Three diaries were
completed during the active treatment phase of the study,
each completed during the week prior to the three monthly
visits. Figure 1 provides an overview of the study design
and the timing of acquisition of diaries and other variables.
The second prespecified primary outcome measure of
efficacy was the change from baseline to endpoint in the
total score from the Incontinence Quality of Life (I-QOL)
questionnaire.20,21 This validated, disease-specific, 22-item
instrument was collected at each visit (Fig. 1) and evaluates
the effects of urinary incontinence in three domains (avoid-
ance and limiting behaviour, social embarrassment and
psychosocial impact). The I-QOL includes questions that
evaluate both the distress and impact of urinary inconti-
nence with a score of 100 being the best possible and zero
being the worst possible quality of life. It is one of the two
quality-of-life instruments that received the World Health
Organization’s Second International Consultation on In-
continence highest (‘highly recommended’) rating for con-
D RCOG 2004 Br J Obstet Gynaecol 111, pp. 249 – 257
251
dition-specific measures of incontinence impact in men and
women.22
Another validated quality-of-life measure, the Patient
Global Impression of Improvement (PGI-I) rating was
obtained at each postrandomisation visit and served as a
prespecified secondary efficacy outcome variable. It is a
single question, measuring patient’s self-perceived im-
provement in her urinary tract condition since she started
treatment.23 Finally, the validated Patient Global Impression
of Severity (PGI-S) scale23 was completed only at baseline.
All efficacy variables were analysed according to the
intention-to-treat principle using data from all randomised
subjects with at least one postrandomisation outcome
measure. Prematurely discontinuing subjects had their last
outcome measure carried forward only in the intention-to-
treat primary analysis. This analysis is the basis for all
principal efficacy conclusions. No efficacy outcome data
were collected after discontinuation for these subjects.
Because subjects who discontinued the trial prematurely
may have experienced changes in incontinence and quality
of life that could bias the results, efficacy was also eval-
uated by visit.
Safety was assessed by the evaluation of treatment-
emergent adverse events, discontinuation of treatment be-
cause of adverse events, serious adverse events, vital sign
measurements, electrocardiograms and clinical laboratory
tests. Adverse events were elicited during an interview
using nonprobing inquiry at each visit and were recorded
regardless of perceived causality. An event was considered
treatment-emergent if it occurred for the first time or
worsened during the double-blind treatment period. A
serious adverse event was defined according to the Inter-
national Consultation on Harmonization guidelines and
252 P. VAN KERREBROECK ET AL.

included any adverse associated with death, a threat to life,

hospitalisation, a severe or permanent disability, a congen-
ital anomaly or a newly diagnosed cancer or was consid-
ered significant for another reason by the investigator.
Patient compliance with the protocol was assessed at
each visit by counting unused medication and by monitor-
ing patient diaries.
The statistical analysis plan was specified a priori. The
percent decrease in incontinence episode frequency was
compared between duloxetine and placebo groups using
van Elteren’s test24 (a type of stratified Wilcoxon test) with
baseline incontinence severity as the stratification variable.
Percent change in incontinence episode frequency is
reported as a median because a few extreme outliers
(placebo þ729%, duloxetine þ2300%) substantially dis-
torted the means. Mean changes in I-QOL scores were
analysed using an ANCOVA model. The dependent vari-
able was the change in I-QOL score and the model included
terms for baseline scores, treatment, site and baseline
incontinence severity strata. The same ANCOVA model
was used to analyse changes in mean voiding intervals.
Categorical variables were analysed by using Pearson’s m2
test or Fisher’s exact test. Mean population differences for
laboratory, vital sign and electrocardiogram data were
analysed using ANOVA. These continuous data were also
assessed for clinically important outliers using prespecified
definitions.
Enrolment into the study was set to end when approx-
imately 440 patients (220 per treatment group) had been
Fig. 2. Flow diagram of the study. *Numbers in the diagram differ from
assigned to a treatment group. The sample size was deter- those in Tables 2 and 3 because some patients discontinuing at a visit still
mined to provide at least 80% power to detect a treatment completed diaries and I-QOL questionnaires. The ‘other’ category includes
difference of 20% in the median percent change in incon- investigator decision, patient decision and lost to follow up.
tinence episode frequency and of 3.5 points in the mean
change in I-QOL total score at a 0.05 Type I error level.
These assumptions were based upon observations from the in the duloxetine group. Figure 2 summarises the flow of
phase two trial19 and from the minimum important I-QOL patients through the study.
difference reported by Patrick et al.21 On average, patients in the placebo group took 94% of
Analyses were performed using SAS software, version their treatment doses compared with 82% of doses in the
8.1 (SAS Institute, Cary, North Carolina). A two-sided duloxetine group ( P < .001). This difference in compliance
alpha level of 0.05 was considered statistically significant was attributable to limited duloxetine consumption by
for treatment effects. subjects who discontinued from the trial early and was
not significant after the first postrandomisation visit.
Table 1 presents baseline demographic and incontinence
RESULTS data. Of the baseline variables, only age differed signifi-
cantly between the treatment groups with the placebo group
Four hundred and ninety-four women aged 24 to 83 being an average two years older than the duloxetine group.
years were randomised to receive duloxetine (247) or Overall, approximately two-thirds of the study population
placebo (247) between December 2000 and January 2002. considered their urinary tract function to be moderately or
Most patients (92%) completed at least one postrandomisa- severely abnormal and half averaged two or more inconti-
tion diary (86% duloxetine, 98% placebo), while 98% nence episodes every day.
completed at least one I-QOL questionnaire (97% dulox- The decrease in incontinence episode frequency was
etine, 99% placebo) and were included in the primary significantly greater in the duloxetine group than in the
analysis. In the placebo group, 92% of patients completed placebo group both in the primary analysis and in the by-
the 12-week study compared with 73% of those in the visit analysis (Table 2). Women who had more severe
duloxetine group ( P < 0.001); the difference was attribut- incontinence (those with 14 or more incontinence episodes
able to a higher discontinuation rate related to side effects per week) had a response similar to the entire study
D RCOG 2004 Br J Obstet Gynaecol 111, pp. 249 – 257
 DULOXETINE TREATMENT OF STRESS URINARY INCONTINENCE 253
a
Table 1. Baseline clinical characteristics of illness severity for all randomised patients. Data are mean (SD) unless otherwise indicated; BMI ¼ body mass
index; PFMT ¼ pelvic floor muscle training; IEF ¼ incontinence episode frequency; I-QOL ¼ Incontinence Quality of Life Questionnaire; PGI-S ¼ Patient
Global Impression of Severity question.

Duloxetine Placebo
b
Randomised N 247 247
Caucasian 244 (99%) 241 (98%)
Age, yearsc 52 (11) 54 (10)
BMI, kg/m2 28 (6) 27 (5)
Prior continence surgery 19 (7.7%) 19 (7.7%)
Perform PFMT 46 (18.8%) 48 (19.4%)
IEF per week (SD) [range] 17.3 (12.9) [0 – 81] 17.2 (12.4) [1 – 68]
Voiding interval, minutes 152 (39) 153 (39)
I-QOL score 66.7 (20.1) 64.5 (21.3)
Moderate or severely abnormal urinary tract function by PGI-S 166 (69.5%) 169 (71.0%)
14 or more IEF per week 125 (50.6%) 131 (53.0%)
a
Baseline is the last nonmissing visit score on or prior to randomisation.
b
Every randomised subject did not provide information for each variable; percentages are calculated using the number of responding patients as the
denominator.
c
P ¼ 0.01.

population (a 55.9% median reduction in incontinence
episodes with duloxetine compared with a 26.8% reduction
with placebo, P < 0.001).
Over half (51.9%) of the women in the duloxetine group
had a 50% to 100% decrease in their incontinence epi-
sode frequency compared with 33.5% of those in the pla-
cebo group ( P < 0.001). The superior improvements with
duloxetine were observed at the first postrandomisation
assessment visit (four weeks) and persisted throughout
the 12-week double-blind treatment period (Table 2). In
addition, patients in the duloxetine group also increased
their average voiding interval significantly compared with
those in the placebo group (15.0 vs 3.8 minutes, respec-
tively; P < 0.001).

Table 2. Incontinence episode frequency: primary analysis and by-visit results. CI ¼ confidence interval; IEF ¼ Incontinence episode frequency.

Treatment group (N)a Time point nb Median IEF Median percent 95% CI for median Pe
per week IEF change from baselinec percent change in IEFd

Primary analysisf
Placebo (247) Baseline 242 14.0
Endpoint 9.0
Change 3.0 29.3 36.8, 20.0
Duloxetine (247) Baseline 212 13.0
Endpoint 7.0
Change 6.0 50.0 57.1, 42.9 0.002

Data by visitg
Placebo (247) Visit 2 247 15.9
Visit 3 (randomisation) 246 14.0
Visit 4 — 4 weeks 237 11.0 20.0 27.1, 10.9
Visit 5 — 8 weeks 230 9.0 29.8 40.6, 23.8
Visit 6 — 12 weeks 225 9.1 28.6 36.4, 18.4
Duloxetine (247) Visit 2 246 15.0
Visit 3 (randomisation) 245 13.0
Visit 4 — 4 weeks 210 6.6 53.6 60.0, 46.9 <0.001
Visit 5 — 8 weeks 187 7.0 53.9 61.1, 46.2 <0.001
Visit 6 — 12 weeks 176 7.0 51.8 58.8, 42.9 0.002
a
N ¼ number randomised.
b
n ¼ number with diary data available for specified analysis or visit.
c
Approximate 95% confidence interval for the median percent change was obtained using nonparametric methods for order statistics.
d
Baseline is the last nonmissing visit score on or prior to randomization.
e
P values for duloxetine versus placebo obtained from van Elteren’s test for analysis of the percent changes in IEF where the stratification variable was
baseline IEF severity.
f
The prespecified primary analysis was based on the intention-to-treat principle using data from every randomised subject with at least one
postrandomisation measure. Prematurely discontinuing subjects had their last outcome measure carried forward in the primary analysis.
g
Analysis compares subjects who had IEF diary records at both the baseline and the respective postrandomisation visit.

D RCOG 2004 Br J Obstet Gynaecol 111, pp. 249 – 257

254 P. VAN KERREBROECK ET AL.

Table 3. Incontinence Quality of Life Questionnaire: primary analysis and by-visit results. CI ¼ confidence interval; I-QOL ¼ Incontinence Quality of Life
Questionnaire.

Treatment group (N)a Time point nb Mean I-QOL Mean change in I-QOL 95% CI for treatment Pe
score from baselinec difference in I-QOLd

Primary analysisf
Placebo (247) Baseline 245 64.4
Endpoint 68.5 4.1
Duloxetine (247) Baseline 240 66.6
Endpoint 72.2 5.5 0.5, 4.1 0.127

Data By Visitg
Placebo (247) Visit 2 245 61.1
Visit 3 (randomisation) 244 64.6
Visit 4 — 4 weeks 244 67.6 3.2
Visit 5 — 8 weeks 233 70.0 5.4
Visit 6 — 12 weeks 228 69.1 4.3
Duloxetine (247) Visit 2 242 62.9
Visit 3 (randomisation) 245 66.6
Visit 4 — 4 weeks 237 72.1 5.6 0.8, 4.6 0.006
Visit 5 — 8 weeks 192 75.8 8.6 1.6, 6.3 0.001
Visit 6 — 12 weeks 178 74.6 7.3 1.0, 6.2 0.008
a
N ¼ number randomised.
b
n ¼ number with diary data available for specified analysis or visit.
c
95% confidence interval for treatment difference (Dulox 80 mg minus placebo) was obtained by using LSMEANS from the ANCOVA model.
d
Baseline is the last nonmissing visit score on or prior to randomisation.
e
P values for duloxetine versus placebo obtained from the comparison of duloxetine 80 mg group to placebo using Type III Sums of Squares from an
ANCOVA model that includes change in I-QOL as dependent variable and baseline I-QOL score, treatment, severity strata and site as independent variables.
f
The prespecified primary analysis was based on the intention-to-treat principle using data from every randomised subject with at least one
postrandomisation measure. Prematurely discontinuing subjects had their last outcome measure carried forward in the primary analysis.
g
Analysis compares subjects who had I-QOL scores at both the baseline and the respective postrandomisation visit.

Changes in the I-QOL scores did not show a significant placebo group (81% vs 64%; P < 0.001). Table 4 lists all
difference between duloxetine and placebo in the primary side effects that occurred in at least 5% of patients on
analysis (Table 3). In contrast, there were significant duloxetine or that were significantly more common with
differences in these parameters when data were analysed duloxetine. Nausea was the most common side effect with
by visit (Table 3). The differences between the primary and duloxetine. Most (94%) nausea with duloxetine was
by-visit analysis for the I-QOL results are largely explained reported within four weeks of the start of therapy, 75%
by the carrying forward of low I-QOL scores from dulox- was mild to moderate in severity at onset, and in no
etine subjects who discontinued their medication very early instance did it increase in severity. The majority (57 of
due to side effects but who still completed their initial four- 69; 83%) of patients who developed nausea on duloxetine
week treatment visit I-QOL questionnaire. Of the 53 completed the study. Of these, 40% reported resolution of
duloxetine-treated subjects who discontinued at visit 4, 42 the nausea by one week while 75% reported resolution
(79%) did so due to adverse effects. Discontinuing subjects within one month. While dry mouth and constipation were
took a median of 10 doses of duloxetine (five days of treat-
ment) and most (47 or 89%) completed the visit 4 I-QOL
Table 4. Treatment-emergent adverse events occurring in at least 5% of
questionnaire. Their increases in I-QOL scores averaged patients randomised to duloxetine or occurring significantly more often
only 0.6 points. with duloxetine than with placebo. Values are expressed as n (%).
The analysis of PGI-I data revealed results similar to
Duloxetine Placebo P
those observed with the I-QOL analysis. Only slightly more
subjects in the duloxetine group than in the placebo group Nausea 69 (27.9) 16 (6.5) <0.001
considered their urinary tract condition to be ‘very much Dry mouth 48 (19.4) 6 (2.4) <0.001
better, much better, or a little better’ in the intention-to-treat Constipation 35 (14.2) 10 (4.0) <0.001
Fatigue 34 (13.8) 11 (4.5) <0.001
analysis (duloxetine 56.2%, placebo 48.2%; not significant).
Insomnia 31 (12.6) 3 (1.2) <0.001
However, significantly more subjects who completed the Dizziness 30 (12.1) 8 (3.2) <0.001
12-week PGI-I in the duloxetine group than in the placebo Headache 24 (9.7) 19 (7.7) 0.524
group considered their urinary tract condition to be im- Sweating increased 21 (8.5) 4 (1.6) <0.001
proved (duloxetine 64.8%, placebo 48.4%; P ¼ 0.008). Vomiting 16 (6.5) 5 (2.0) 0.024
Somnolence 10 (4.0) 0 0.002
Treatment-emergent adverse events were experienced by
Tremor 10 (4.0) 0 0.002
more patients in the duloxetine group compared with the
D RCOG 2004 Br J Obstet Gynaecol 111, pp. 249 – 257

Table 5. Discontinuations for adverse events in 1% or more of patients
randomised to duloxetine. Values are expressed as n (%).
Duloxetine Placebo
(N ¼ 247) (N ¼ 247)
For any adverse event 53 (21.5) 12 (4.9)
Nausea 13a (5.3) 2 (0.8)
Dizziness 9 (3.6) 1 (0.4)
Insomnia 5 (2.0) 1 (0.4)
Attention disturbance 3 (1.2) 1 (0.4)
a
One subject discontinued due to nausea that was a pre-existing condi-
tion that neither started nor worsened after she started taking duloxetine.
also common side effects, they tended to be mild to
moderate (90% and 86%, respectively) in the majority of
patients and rarely (one and two patients, respectively) led
to discontinuation from the study.
The discontinuation rate due to adverse events was sig-
nificantly higher for the duloxetine group compared with
the placebo group (21.5% vs 4.9%; P < .001). Table 5 lists
all adverse events that resulted in a 1% or higher discon-
tinuation rate for duloxetine.
There was a significant increase in mean heart rate
during treatment with duloxetine compared with placebo;
however, the less than three beats per minute increase with
duloxetine was within the normal range. In addition, outlier
analysis confirmed that there were no clinically relevant
effects of duloxetine on heart rate. There were no signif-
icant differences in the mean change for systolic or dia-
stolic blood pressure between the duloxetine and placebo
groups. Three patients (two placebo, one duloxetine) de-
veloped sustained hypertension (systolic or diastolic blood
pressure of at least 140 or 90, respectively, and increased at
least 10 mm from baseline for three consecutive visits). A
comprehensive analysis of corrected QT intervals revealed
no arrhythmogenic tendencies with duloxetine. There were
no deaths in the study. Two placebo-treated subjects and
two duloxetine-treated patients experienced serious adverse
events after randomisation. None resulted in a chronic
sequela.
DISCUSSION
In this study, duloxetine was noted to be significantly
more effective than placebo in women with stress urinary
incontinence across the spectrum of stress urinary inconti-
nence severity. With patients averaging two to three incon-
tinence episodes per day and some patients having as many
as 10 incontinence episodes per day (Table 1), the study
population included women who would be traditionally
considered candidates for continence surgery.
A 50% to 100% reduction in incontinence episodes was
seen in more than half of women treated with duloxetine.
This effect was noted within the first four weeks of
treatment and persisted through the three-month study,
D RCOG 2004 Br J Obstet Gynaecol 111, pp. 249 – 257
DULOXETINE TREATMENT OF STRESS URINARY INCONTINENCE 255
indicating a prompt and durable response. The reduction
in incontinence episode frequency with duloxetine for the
P half of the study population that averaged at least 14
incontinence episodes per week was slightly greater than
the reduction observed for the entire population, whereas
<0.001
0.007 the placebo response was somewhat lower in this more
0.02 severe subgroup.
0.22 The improvements with duloxetine were observed de-
0.62 spite a significant increase in the average voiding interval
for duloxetine compared with placebo. Less frequent void-
ing indicates that the improvements with duloxetine were
not the result of emptying the bladder more often, a
modification of behaviour that can result from the comple-
tion of diaries in the clinical trial setting.
These significant improvements in incontinence were
not reflected by improvements in quality of life as mea-
sured with the I-QOL and PGI-I instruments in the primary
analysis. These findings contrast with those of the phase
two study in the United States,19 a large phase three study
in the United States and Canada25 and another phase three
study conducted in Europe, South Africa, South America
and Australia26 that together randomised a total of 1419
women with stress urinary incontinence. In each of these
other studies, significant improvements in both quality of
life scales were seen with duloxetine compared with
placebo in the intention-to-treat analysis. In the largest of
these,25 it was demonstrated that these quality of life
improvements resulted from improved stress urinary incon-
tinence and were not the result of the treatment of unrec-
ognised symptoms of depression.
However, quality of life data for patients who completed
this study showed significant improvements with duloxe-
tine compared with placebo. The result is largely explained
by the fact that the results of the intention-to-treat analysis
were influenced by carrying forward the low I-QOL scores
of the duloxetine patients who discontinued very early due
to side effects. The data show that these patients took
medication for a very short period and discontinued before
they had a chance to experience benefit although they still
completed the I-QOL questionnaire.
Significantly more patients experienced treatment-emer-
gent adverse events with duloxetine than with placebo with
22% of duloxetine patients discontinuing therapy due to
side effects. Nausea was the most frequent side effect on
duloxetine with 5.3% of all duloxetine patients leaving the
study due to nausea. Nausea tended to have its onset very
early after starting duloxetine. It was mild or moderate in
most cases, did not worsen after its onset and resolved
within one week to one month of therapy in most cases.
Eighty-three percent of women who experienced duloxe-
tine-related nausea completed the study. Dry mouth and
constipation were not uncommon side effects, however,
these were seldom severe and rarely led to discontinuation
of therapy.
While alpha-adrenergic receptor agonists have been used
for the treatment of stress urinary incontinence, the major
256 P. VAN KERREBROECK ET AL.
concern with these agents is the inability to separate the
peripheral urethral smooth muscle stimulatory effects from
cardiovascular effects in this class of drugs.27 Duloxetine
has a predominant central mode of action through the sacral
cord pudendal motor nucleus to stimulate the urethral
rhabdosphincter17 rather than a peripheral mode of action
on smooth muscle. The slight increase in heart rate with
duloxetine provides evidence for a mild, clinically unim-
portant peripheral alpha-adrenergic action that was not
associated with any significant change in blood pressure
or clinically important changes in the electrocardiogram.
CONCLUSION
Our data are consistent with published data from North
and South America, Africa and Australia.19,25,26 However,
the improvements with duloxetine must be weighed against
a considerable discontinuation rate due to early side effects.
It is likely that more patients will continue duloxetine in
clinical practice if they are reassured that any early side
effects are unlikely to worsen or persist. Overall, the
efficacy and safety data suggest that an incontinent patient
and her physician should be able to assess the adequacy of
her response by comparing her improvement in inconti-
nence with any persisting side effects after four weeks of
duloxetine treatment.
In summary, these data support duloxetine as a pharma-
cological treatment for women with stress urinary inconti-
nence, thus increasing the range of options for women with
this problem.
Acknowledgements
The Duloxetine UI Study Group included the following
principal investigators and their staffs; investigators
received funding to conduct this study from the sponsor:
Paul Abrams, FRCS, MD, Southmead Hospital, Bristol,
UK; Karl Moller Bek, PhD, Skejby Sygehus, Aarhus N,
Denmark; BLH Bemelmans, MD, PhD, Radboud Depart-
ment of Urology, Nijmegen, Netherlands; Steven Bryniak,
MD, St. Joseph Hospital, Saint John, Canada; Hendrik
Cammu, MD, A. Z. – V. U. B., Brussels, Belgium; Pierre
Costa, MD, PhD, Hospital Gaston Doumergue, Nimes
Cedex, France; KPJ Delaere, MD, PhD, Atrium Medisch
Centrum, Heerlen, Netherlands; Marc Dhont, MD, PhD, U.
Z. Gent, Gent, Belgium; David Eiley, MD, Ultra-Med,
Pointe Claire, Canada; Per Ekstrom, MD, PhD, Hospital
Ystads, Ystad, Sweden; Christine Evans, FRCS, MD, Glan
Clwyd Hospital, Denbighshire, UK; Karel Everaert, MD,
PhD, U. Z. Gent, Gent, Belgium; Christian Falconer, MD,
PhD, Danderyds Hospital, Danderyd, Sweden; Aino Fianu-
Jonasson, MD, PhD, Huddinge University Hospital, Stock-
holm, Sweden; Robert Freeman, MD, DM, FRCOG,
Derriford Hospital, Derriford, UK; Francois Haab, MD,
Hospital Tenon, Paris, France; Tahseen Hasan, FRCS,
MD, Freeman Hospital, Newcastle, UK; Lars Henriksson,
MD, University Hospital MAS, Malmö, Sweden; Tim Hil-
lard, DM, FRCOG, Poole Hospital, Poole, UK; Sven Hun-
dertmark, MD, Ausbildungszentrum fuer Frauenheilkunde
Berlin, Berlin, Germany; Udo Jonas, MD, PhD, Medizini-
sche Hochsschule Hannover, Hannover, Germany; Preben
Kjolhede, MD, PhD, University Hospital, Linkoping, Swe-
den; Kenneth Krohn, Vrinnevijukhuset i Norrkoping, Norr-
koping, Sweden; Rainer Lange, MD, DRK-Krankenhaus,
Alzey, Germany; Paula Lowenadler, MD, Trelleborgs
Hospital, Trelleborg, Sweden; Hansjoerg Melchior, MD,
Staedtische Kliniken, Kassel, Germany; Birger Reinhold
Moller, MD, Odense Universitets Hospital, Odense, Den-
mark; Mahmoud Nachabe, MD, Greenfield Park, Canada;
Ole Nielsen, Grindsted, MD, Denmark; Allan Patrick, MD,
King Tower, Fredericton, Canada; Bo Pettersson, MD,
PhD, Countess Of Chester Hospital, Chester, UK; Jorgen
Praest, MD, Randers Central Sygehus, Randers, Denmark;
Volker Ragosch, MD, Ausbildungszentrum fuer Frauen-
heilkunde, Berlin, Germany; Olof Rugarn, MD, PhD,
Vrinnevi Hospital, Norrkoping, Sweden; Göran Samsioe,
MD, PhD, Kvinnohalsan, Lund, Sweden; JH Schagen van
Leeuwen, MD, PhD, St Antonius Ziekenhuis, Nieuwegein,
Netherlands; Erik Schick, MD, Maisonneuve Rosemount
Hospital, Montreal, Canada; Mark Slack, MD, Peter-
borough District Hospital, Peterborough, UK; Torsten
Sorensen, MD, Kolding Sygehus, Kolding, Denmark;
Stuart Stanton, FRCS, FRCOG, St. George Hospital,
London, UK; Tim Terry, FRCS, MS, Leicester General
Hospital, Leicester, UK; PHM van de Weijer, MD, PhD,
Gelre Ziekenhuizen-location Juliana, Apeldoorn, Nether-
lands; PEV van Kerrebroeck, MD, University Hospital
Maastricht, Maastricht, Netherlands; Kristiaan Vekemans,
MD, Medisch Centrum Practimed, Tessenderlo, Belgium;
HAM Vervest, MD, PhD, St. Elisabeth Ziekenhuis,
Tilburg, Netherlands; ME Vierhout, MD, PhD, University
Hospital Rotterdam, Rotterdam, Netherlands; Jean-Jacques
Wyndaele, MD, DSC, PhD, FEBU, FISCOSU, University
Antwerpen, Edegem, Belgium.
Conflict of interest statement
Dr Yalcin and Dr Bump are both full-time employees of
Eli Lilly and hold stock and stock options in the company.
Dr van Kerrebroeck, Dr Abrams, Dr Lange, Dr Slack and
Dr Wyndaele have served on advisory boards for and have
consulting agreements with Eli Lilly and were investigators
in this study that was funded by Eli Lilly and Boehringer
Ingelheim.
References
1. Abrams P, Cardozo L, Fall M, et al. The standardisation of terminol-
ogy of lower urinary tract function: report from the standardistion sub-
committee of the International Continence Society. Neurourol Urodyn
2002;21:167 – 178.
D RCOG 2004 Br J Obstet Gynaecol 111, pp. 249 – 257
 DULOXETINE TREATMENT OF STRESS URINARY INCONTINENCE
2. Hampel C, Wienhold D, Benken N, Eggersmann C, Thüroff JW.
Definition of overactive bladder and epidemiology of urinary incon-
tinence. Urology 1997;50(S6A):4 – 14.
3. Hannestad YS, Rortveit G, Sandvik H, Hunskaar S. A community-
based epidemiological survey of female urinary incontinence. The
Norwegian EPINCONT study. J Clin Epidemiol 2000;53:1150 – 1157.
4. Brown JS, Waetjen LE, Subak LL, Thome DH, Ven Den Eeden S,
Vittinghoff E. Pelvic organ prolapse surgery in the United States,
1997. Am J Obstet Gynecol 2002;186:712 – 716.
5. Lagro-Janssen T, Van Weel C. Long-term effect of treatment of female
incontinence in general practice. Br J Gen Pract 1998;48:1735 – 1738.
6. Wilson PD, Bø K, Hay-Smith J, et al. Conservative management
in women. In: Abrams P, Khoury S, Wein A, editors. Incontinence.
Plymouth, UK: Health Publication, 2002:573 – 624.
7. Espey MJ, Du H-J, Downie JW. Serotonergic modulation of spinal
ascending activity and sacral reflex activity evoked by pelvic nerve
stimulation in cats. Brain Res 1998;798:101 – 108.
8. Thor KB, Hisamitsu T, de Groat WC. Unmasking of a neonatal so-
matovesical reflex in adult cats by the serotonin autoreceptor agonist
5-methoxy-N,N-dimethyltryptamine. Brain Res Dev Brain Res 1990;
54:35 – 425.
9. Danuser H, Thor KB. Spinal 5-HT2 receptor-mediated facilitation of
pudendal nerve reflexes in the anaesthetized cat. Br J Pharmacol
1996;118:150 – 154.
10. Krier J, Thor KB, de Groat WC. Effects of clonidine on the lumbar
sympathetic pathways to the large intestine and urinary bladder of the
cat. Eur J Pharmacol 1979;59:47 – 53.
11. Pedersen E, Torring J, Klemar B. Effect of the alpha-adrenergic
blocking agent thymoxamine on the neurogenic bladder and urethra.
Acta Neurol Scand 1980;61:107 – 114.
12. Gajewski J, Downie JW, Awad SA. Experimental evidence for a cen-
tral nervous system site of action in the effect of alpha-adrenergic
blockers on the external urinary sphincter. J Urol 1984;132:403 – 409.
13. Downie JW, Bialik GJ. Evidence for a spinal site of action of cloni-
dine on somatic and viscerosomatic reflex activity evoked on the
pudendal nerve in cats. J Pharmacol Exp Ther 1988;246:352 – 358.
14. Downie JW, Espey MJ, Gajewski JB. Alpha 2-adrenoceptors not imi-
dazole receptors mediate depression of a sacral spinal reflex in the cat.
Eur J Pharmacol 1991;195:301 – 304.
15. Espey MJ, Downie JW, Fine A. Effect of 5-HT receptor and adreno-
ceptor antagonists on micturition in conscious cats. Eur J Pharmacol
1992;221:167 – 170.
16. Bymaster FP, Dreshfield-Ahmad LJ, Threlkeld PG, et al. Compara-
D RCOG 2004 Br J Obstet Gynaecol 111, pp. 249 – 257
257
tive affinity of duloxetine and venlafaxine for serotonin and norepi-
nephrine transporters in vitro and in vivo, human serotonin receptor
subtypes, and other neuronal receptors. Neuropsychopharmacology
2001;25:871 – 880.
17. Thor KB, Katofiasc MA. Effects of duloxetine, a combined serotonin
and norepinephrine reuptake inhibitor, on central neural control of
lower urinary tract function in the chloralose-anesthetized female
cat. J Pharmacol Exp Ther 1995;274:1014 – 1024.
18. Katofiasc MA, Nissen J, Audia JE, Thor KB. Comparison of the ef-
fects of serotonin selective, norepinephrine selective, and dual sero-
tonin and norepinephrine reuptake inhibitors on lower urinary tract
function in cats. Life Sci 2002;71:1227 – 1236.
19. Norton PA, Zinner NR, Yalcin I, Bump RC. Duloxetine versus pla-
cebo in the treatment of stress urinary incontinence. Am J Obstet
Gynecol 2002;187:40 – 48.
20. Wagner TH, Patrick DL, Bavendam TG, Martin ML, Buesching DP.
Quality of life of persons with urinary incontinence: development of a
new measure. Urology 1996;47:67 – 71.
21. Patrick DL, Martin ML, Bushnell DM, Yalcin I, Wagner TH, Bue-
sching DP. Quality of life of women with urinary incontinence:
further development of the incontinence quality of life instrument
(I-QOL). Urology 1999;53:71 – 76.
22. Donovan JL, Badia X, Corcos J, et al. Symptom and quality of life
assessment. In: Abrams P, Cardozo L, Khoury S, Wein A, editors.
Incontinence, 2nd edition. Plymouth, UK: Health Publication, 2002:
267 – 316.
23. Yalcin I, Bump RC. Validation of two global impression question-
naires for incontinence. Am J Obstet Gynecol 2003;189:98 – 101.
24. van Elteren PH. On the combination of independent two-sample tests
of Wilcoxon. Bull Int Stat Inst 1960;37:1 – 13.
25. Zinner NR, Dmochowski RR, Norton PA, Miklos JR, Yalcin I,
Bump RC. Duloxetine versus placebo in the treatment of North
American women with stress urinary incontinence. J Urol 2003;
170:1259 – 1263.
26. Millard RJ, Moore K, Rencken R, Yalcin I, Bump RC for the Dulox-
etine UI Study Group. Duloxetine versus placebo in the treatment of
stress urinary incontinence: a four continent randomized clinical trial.
Br J Urol Int. In press.
27. Andersson K-E, Appell R, Awad S, et al. Pharmacological treatment
of urinary incontinence. In: Abrams P, Khoury S, Wein A, editors.
Incontinence. Plymouth, UK: Health Publication, 2002:481 – 511.
Accepted 16 October 2003