The Incontinence Impact Questionnaire and the Urogenital

Distress Inventory: A revisit of their validity in women

without a urodynamic diagnosis
Marie-Andre Harvey, MD,a Betsy Kristjansson, MA,b David Griffith, MD,c and Eboo Versi, MD, PhDd
Ottawa, Ontario, Canada; Swindon, United Kingdom; and Peapack, NJ

OBJECTIVES: We sought to assess validity of the Incontinence Impact Questionnaire (IIQ) and the Urogeni-
tal Distress Inventory (UDI) (long and short forms) in incontinent women previously urodynamically undiag-
nosed.
STUDY DESIGN: Post hoc analysis of responses to the IIQ and UDI questionnaires were obtained from a
trial on a urethral device in community-dwelling incontinent women. Internal consistency and validity were
evaluated against the 1-hour pad test.
RESULTS: Internal consistency (Cronbachs alpha) for the long forms was high for the IIQ and moderately
high for the UDI, good for the IIQ-short, but poor for the UDI-short. Correlations with 1h pad test were low
and non-significant for both versions. Correlation of the short with the long forms was high.
CONCLUSION: In the community-dwelling population, without a urodynamic diagnosis, neither long nor
short versions of the questionnaires correlate with the severity of the urinary incontinence as shown by the
pad test. The validity of the current questionnaires in women without urodynamic diagnosis is questionable.
(Am J Obstet Gynecol 2001;185:25-31.)

Key words: Quality of life, urinary incontinence, psychometrics, urodynamic studies

Urinary incontinence is increasingly seen in the physi-
cians office, because of both its high prevalence (20% to
30% of middle-aged and 30% to 50% of elderly women1)
and the growing expectations for relief by women af-
fected by it. More and more the psychosocial impact of
urinary incontinence becomes an important aspect, and
several disease-specific tools have been designed to mea-
sure quality of life.2-4 A number of careful steps must be
followed in questionnaire development (Table I).5
The Incontinence Impact Questionnaire (IIQ) was de-
signed by Shumaker et al2 to assess the impact of urinary
incontinence on activities and emotions in women.
Thirty self-administered questions cover four domains:
physical activity, social relationships, travel, and emo-
tional health. Each question has a 4-point response scale;
patients are asked to rate the extent to which their uri-
nary incontinence affects their daily functioning (0, not
From the Department of Obstetrics and Gynecologya and the Department
of Epidemiology and Community Medicine,b University of Ottawa, the
Department of Obstetrics and Gynaecology, Princess Margaret Hospital,c
and Pharmacia Corp.
Presented at the Twenty-first Annual Meeting of the American UroGyne-
cologic Society, Hilton Head, SC, October 26-28, 2000.
Reprint requests: Marie-Andre Harvey, MD, FRCSc, University of Ot-
tawa, The Ottawa HospitalGeneral Campus, 501 Smyth Rd, Box 802,
Ottawa, Ontario, Canada K1H 8L6; E-mail: maharvey@ottawahospi-
tal.on.ca.
Copyright 2001 by Mosby, Inc.
0002-9378/2001 $35.00 + 0 6/6/116369
doi:10.1067/mob.2001.116369
at all; 1, slightly; 2, moderately; 3, greatly) (see Appendix
1). No time anchor is included in their formulation of the
questions. The Urogenital Distress Inventory (UDI),
which is meant to complement the IIQ, was developed at
the same time by the same group to assess the degree to
which symptoms associated with incontinence are trou-
bling. It consists of 19 questions covering 3 domains:
symptoms related to stress urinary incontinence, detrusor
overactivity, and bladder outlet obstruction. The re-
sponse scale is the same as the IIQ (Appendix 2).
Shumaker et al2 assessed the validity, reproducibility,
and sensitivity to change of the full-length IIQ and UDI in
a population of community-dwelling women recruited in
an uncontrolled clinical study. All patients had to fulfill
urodynamic criteria of genuine stress incontinence or de-
trusor instability. They reported a low but significant cor-
relation with incontinence severity on the 1-hour pad test
(r = 0.27). The authors concluded from this and other
psychometric tests that the questionnaires were reliable
and valid.
Short forms of the IIQ and UDI were developed with
the original data2 on the full-length questionnaire. Re-
gression analyses suggested that a 7- to 8-item question-
naire would accurately predict the IIQ long from total
score, and a 6-item form would predict the UDI long
form. Items constituting the short forms (the IIQ-7 and
the UDI-66) are shown in bold in Appendixes 1 and 2.
The short forms were validated through correlations

25
26 Harvey et al July 2001
Am J Obstet Gynecol

Table I. Steps of development of a scale (adapted from Table II. Demographic and baseline data of the
Streiner & Norman) population studied at baseline (without the use
of the device)
Devising the items
Evaluate the existing scales Standard
Develop new items (focus groups, interviews with patients, N Mean deviation Range
clinical observation, expert opinion)
Evaluate content validity (are all the domains under investiga- Age 154 55 11.8 29-86
tion covered?) Vaginal parity 155 2.2 1.7 0-9
Evaluate face validity (do the items appear on surface to be 1-h pad test (gm) 136 26.1 45.2 0-203.6
measuring what they actually are?) IIQ score* 149 28 19.1 0-85
Choose a response scale IIQ-7 150 1.1 0.7 0-3
Select the items UDI score* 150 16.5 7 4-42
Interpretability (eliminate ambiguous or incomprehensible UDI-6 150 1.4 0.5 0.3-2.8
items)
Pilot testing
*The long form scores were obtained by simple addition of the
Frequency of endorsement
individual items scores (see text).
Discrimination capacity
Internal consistency testing
Coefficient alpha
Reliability testing (on several different groups) tice or from the community to participate in the 4-week
Intraobserver reliability trial. Advertisements in local newspapers and radio sta-
Interobserver reliability tions were used to recruit participants. Patients under-
Test-retest reliability
Validity testing (on several different groups) went a standardized history (including grading on a scale
Criterion (correlation with a gold standard) of 4 [none, mild, moderate, and severe] different urinary
Construct validity (convergent, divergent and concurrent) symptoms) and a physical examination at the time of re-
(when no gold standard exists)
Sensitivity to change testing cruitment. Inclusion and exclusion criteria are reported
elsewhere.10 All patients completed the long forms of the
IIQ and UDI and underwent a standardized 1-hour pad
with long-form scores and with clinical data. Abrams et test at baseline. Subjects then wore the device for 1
al7 report a significant correlation with the 1h pad test. month. At the end of the month, they were asked to re-
Independent investigators have carried out additional peat the tests while wearing the device. Local institutional
validation studies, also in women selected on the basis of review board approval was obtained at the time of the
formal urodynamic testing.8,9 They have found that vari- original data collection.
ables pertaining to frequency and volume of loss could Scores on the long forms of the IIQ and UDI were cal-
predict the score of the IIQ-7 and UDI-6.6 In addition, culated in 2 fashions: with the method recommended by
positive responses to subscale questions (obstruction, the scale authors, in which a score was generated for each
stress incontinence or overactivity) are correlated with subscale and all subscales were weighted equally, and in a
the corresponding urodynamic (UDS) diagnosis.9 simple additive fashion because this is more convenient.
The weight of accumulated evidence suggests that the Correlation between the 2 methods was obtained by use
IIQ and UDI, both short and long forms, are valid for use of the Pearsons correlation coefficient. The assessment
with women who have a priori urodynamic diagnosis of of validity and internal consistency were done for each
urinary incontinence. However, neither the short forms method of scoring.
nor the long forms have ever been validated in women Internal consistency of the long forms was assessed with
without such diagnosis. The validity of a measurement is Cronbachs alpha. Validity was assessed through correla-
dependent on the population being measured.5 There- tion with the 1h pad test (Spearmans rho) and receiver-
fore, if a scale is to be used on a different population, psy- operator curve (ROC) analyses against the pad test,
chometric properties must be reestablished. The purpose which was dichotomized as positive (>2 g) or negative
of this study is to assess the reliability and validity of the (2 g). Criterion validation was also done by correlating
IIQ and UDI (long and short) in community-dwelling the scales score with the symptoms patients reported at
women with no a priori UDS diagnosis. In addition, we baseline. Sensitivity to change was assessed by use of the
wanted to determine whether a simpler method of scor- Wilcoxon signed rank test. Confirmation that a change
ing the long forms (simple addition) would yield similar was present was assessed by use of the Wilcoxon signed
results. rank test for comparing the pad test results (with versus
without the device).
Methods Scores for the short forms were extracted from the
Data from a prospective before/after clinical trial on long form questionnaires, and an average of the items re-
the efficacy of an external urethral device were used in sponded to were calculated, according to scale develop-
the validation study.10 Women were enrolled either ers instruction. Reliability of the short form was studied
through referral from an academic urogynecology prac- by use of inter-item correlation, item-total statistics, and
Volume 185, Number 1
Am J Obstet Gynecol
Table III. Reliability and validity of the long forms of the IIQ and UDI
IIQ
Internal consistency (Cronbachs alpha) 0.95
Convergent validity
Pad test (area under the curve of ROC) 0.543
Pad test (Spearmans correlation) 0.12 (NS)
Sensitivity to change
Change in score Z = 6.12 (P < .001) Z = 6.04 (P < .001) Z = 5.96 (P < .001)
Correlation of change in score with 0.10 (NS)
change in pad test result
Correlation with long form N/A
coefficient alpha. For the inter-item correlation, items
should correlate with each other, but a very high in-
teritem correlation is indicative of redundancy. In the lat-
ter two techniques, the analyses evaluate the correlation
of each individual item with the overall score of the test
calculated without that item. This represents a method
for checking the homogeneity of the scale.5 Two methods
were used: simple correlation and item-total statistic with
Cronbachs alpha. The first method consists of calculat-
ing the correlation of the individual item, with the scale
total omitting that item, with Pearsons correlation. In
the second method, an alpha for the total score is calcu-
lated with individual items deleted one at the time. If the
alpha increases significantly when a specific item is left
out, then its exclusion would increase the homogeneity
of the test. To assess the appropriateness of choice of
items on the short form, we studied item-total correlation
for the long form.
Statistical significance was set at a P value of .05. All
analyses were performed with SPSS for Windows 9.0
(SPSS Inc, Chicago, Ill).
Results
One hundred fifty-five women were enrolled in the
study. Five women dropped out after enrollment; thus
baseline data are available for 150 women. Validity and
reliability studies were performed on this group. Fifty-two
women dropped out before follow-up, leaving 98 patients
for whom end-of-study data are available. Evaluation of
sensitivity to change was performed on those 98 patients.
Reasons for dropping out, as well as the comparison be-
tween included/excluded subjects, are given in the origi-
nal report.10 They did not differ from women who com-
pleted the trial with regard to age, body mass index
(BMI), vaginal parity, or urinary symptoms expressed at
onset. Demographic characteristics of the 150 women
with baseline data are shown in Table II.
When compared, the two methods of calculating the
long form scores were highly correlated (0.91 for the UDI
and 0.99 for the IIQ). In addition, the method of scoring
had little impact on validity and reliability evaluations;
there was little difference in results for the two scoring
methods. Therefore for the sake of simplicity and clarity,
Harvey et al 27
UDI IIQ-7 UDI-6
0.76 0.84 0.52
0.513 0.58 0.526
0.07 (NS) 0.18 (P = .048) 0.07 (NS)
Z = 5.83 (P < .001)
0.001 (NS) 0.12 (NS) 0.12 (NS)
N/A 0.95 (P < .001) 0.87 (P < .001)
only the simple addition score will be used in the analy-
ses presented herein.
Long forms. Results for the reliability study (Cron-
bachs alpha), convergent validity and sensitivity to
change are presented in Table III. Coefficient alpha for
the IIQ was high, whereas that for the UDI was good.
ROC analysis showed that scores on the questionnaire
were largely unrelated to incontinence, as measured by
the pad test: the area under the curve was only 0.543 and
0.513, for the IIQ and the UDI respectively (Fig 1). An
area under the curve of 0.50 indicates a 50% probability
of correct classification. Thus this test performed little
better than chance alone in classifying a patient as incon-
tinent (as defined by a positive pad test result). No corre-
lation between the pad test result and the questionnaires
was found as shown in Table III. When correlations were
performed against the clinical symptoms expressed by
the patients, only the UDI showed consistent, albeit weak,
correlations (Table IV).
The null hypothesis that the tests without and with the
device were the same was tested. The pad test result was
significantly different (Z= 5.539, P < .01) with and with-
out the device, indicating that it was sensitive to change.
The change in questionnaire score was also significantly
different while the patient was wearing the device, indi-
cating that it too was sensitive to change (Table III).
Short forms
Reliability studies
INCONTINENCE IMPACT QUESTIONNAIRE. Items had low to
moderate correlation with each other (range 0.24 to
0.66). Item total correlation revealed moderate to good
correlation between individual items and the short form
score (range 0.47 to 0.66). The Cronbachs alpha for in-
ternal consistency is shown in Table III; it was moderately
high. Inter-item correlations on the long form revealed
that the 7 items that correlated best with the total score of
the long form were questions 1, 3, 6, 8, 12, 20, and 29.
UROGENITAL DISTRESS INVENTORY Items in this scale
showed moderate correlation of the items of the short
form to each other (range 0.04 to 0.57). Item analysis re-
vealed a weak and inverse correlation for item 4 (0.09)
and poor correlation between other items (range 0.23 to
0.44) and the short form score. The Cronbachs alpha for
28 Harvey et al July 2001
Am J Obstet Gynecol

Table IV. Correlation of the IIQ and UDI, long and

short, with symptoms expressed at baseline question-
naire

IIQ UDI IIQ-7 UDI-6

SUI 0.06 0.19* 0.03 0.08

UI 0.29* 0.53* 0.24* 0.48*
Daily pad use 0.13 0.18* NS 0.12
Urgency 0.33* 0.54* 0.31* 0.47*
Nocturia 0.18 0.44* NS 0.40*
VD
Strain to void 0.21* 0.27* 0.20* 0.29*
Incomplete emptying 0.12 0.26* 0.07* 0.22*
Hesitancy 0.08 0.25* 0.10 0.25*
Poor stream 0.04 0.16 0.03 0.18*

*Indicates a correlation with P < .05.

a weak correlation was noted between the BMI and the

Fig 1. ROC examining relation between pad test result (as di-
chotomized, positive or negative) and questionnaire score (IIQ
long form).
internal consistency was low at 0.53. Inter-item correla-
tion performed on the long form revealed that the 6
items that correlated best with the total score of the long
form were questions 1, 3, 4, 6, 13, and 14. However, when
we attempted to validate these modified short versions
against either the pad test results or the change in pad
test measurement, validity was not better (data not
shown).
Validity. Convergent validity results for the short forms
are presented in Table III. The correlation between the
short and long forms of the IIQ and UDI were both high.
There was little correlation between the short forms of
the IIQ and UDI and the weight of the pad test. The ROC
curve analysis revealed that the questionnaires do not
have any discriminant power in predicting an abnormal
pad test result: AUC were quite low, at 0.58 and 0.53, re-
spectively. Validity against clinical symptoms, as in the
case of the long forms, showed significant correlation, al-
though weak to moderate, only for the UDI-6 (Table IV).
Sensitivity to change. The scale scores were significantly
different with and without the device, indicating that they
were also sensitive to change (Table III). The change in
score value was then correlated with the change in pad
test by use of Spearmans correlation. No correlation be-
tween the change in the pad test score and the change in
test score was found.
Role of patients characteristics. The role of BMI, vagi-
nal parity, and age as predictors of the questionnaire
scores were sought. Nonparametric correlations (Spear-
mans) were used. A weak inverse correlation (r = 0.16, P
< .05) was found between age and the IIQ and IIQ-7, sig-
nifying that as age increased, scores were lower. Similarly,
IIQ (r = 0.17, P < .05) and the UDI-6 (r = 0.21, P < .05);
scores on the questionnaires were higher for people with
higher BMI. No correlation was found with vaginal parity.
Discussion
We have found that the IIQ and the UDI, either long or
short, do not appear to be valid in a population in whom
no prior urodynamic diagnosis has been established,
when validity is assessed with the criterion of 1h pad test.
The lack of validity is most likely due to differences be-
tween our population and the population in whom the
questionnaires were originally evaluated. It is important
that, before application in a new population, a question-
naire instrument be validated in that new setting. The
quality of life questionnaires studied are frequently dis-
tributed to women before any urodynamic evaluation in
both clinical and research settings. Our analysis demon-
strates that, in patients recruited from the community on
the basis of subjective complaints of urinary incontinence
and without an established urodynamic diagnosis, these
questionnaires are invalid as markers of urinary inconti-
nence severity. A population consisting of women with an
established UDS diagnosis may differ in terms of patho-
physiological condition and severity of incontinence
from the population of all patients with the complaint of
urinary incontinence. Moreover, urodynamic studies may
have false-negative results. The relative insensitivity of
standard multichannel urodynamic studies in detecting
detrusor instability has been highlighted in many ambu-
latory-monitoring studies performed on subjects with
symptoms.11-13 In addition, authors agree that symptoms
are not a good predictor for the UDS diagnosis of stress
incontinence14 or mixed or urge incontinence.15
The original trial had been designed to assess the ef-
fectiveness of an external urethral device as close as
possible to what would happen in general practice.
Nonetheless, our population probably differ somewhat
from that of the general population because subjects
Volume 185, Number 1
Am J Obstet Gynecol
participating in trials are often not representative of the
general population (eg, they may be healthier and more
educated).
Another explanation for our findings is that the 1-hour
pad test is not a good gold standard of severity to assess im-
pact of incontinence on quality of life. Pad tests may help
to measure the severity of incontinence. However, the as-
sumption that the more severe the incontinence, the
greater the effect on the quality of live is debatable. A pa-
tient with urgency/frequency and occasional urge inconti-
nence may have a decreased quality of life as measured by
the IIQ or the UDI, but the pad test result will be negative.
Similarly, a patient may have mild stress incontinence dur-
ing a specific activity (eg, playing tennis), and it may be
perceived as very problematic. Furthermore, two patients
with the same severity of incontinence might be affected
in a different manner by their condition. A better indica-
tor of quality of life might be the frequency of micturition,
perceived leakage quantity, number of incontinence
episodes/24 hours, as suggested by Robinson et al8 in a
telephone interview of women who had recently visited
their primary care physician for any reason. Thus valida-
tion of the questionnaires versus a voiding diary may yield
different results.
The 1-hour pad test was standardized in 1988 by the In-
ternational Continence Society. Recommendations were
issued for its use in clinical trials when quantitative
analysis of the urine loss is required for assessment and
comparison of treatments of different types of urinary
incontinence.16 However, its use in the evaluation of in-
continence is questionable because it has not proven to
be very accurate in detecting the presence or the severity
incontinence, unless the bladder volume is fixed at 200 to
300 mL or 50% to 75% of bladder capacity. This could ex-
plain why we did not find a relationship between the pad
test and the scales.
We also found that the long- and short-form UDI corre-
lated well with actual symptoms of which the patients com-
plained, which is reassuring, because this questionnaire is
meant to probe specifically the presence and bothersome-
ness of urinary symptoms. However, the IIQ did not corre-
late well with symptoms experienced by the patients,
meaning that the impact on the life activities of the incon-
tinence did not correlate with the importance the patient
was attributing to her symptoms.
In the analyses evaluating which items should constitute
a short version, we have arrived at divergent results from
the scale developers for the IIQ-7. In fact, we would retain
items 3, 12, and 29 instead of 5, 15, and 27. However, psy-
chometric properties were not improved.
The original IIQ and UDI were not designed to discrim-
inate between urge and stress incontinence, and the vali-
dation study was not performed according to UDS diag-
nosis. It is therefore possible that such discrimination
would have helped determine for which of the two groups
Harvey et al 29
(urge or stress) this questionnaire might have been more
valid. However, the design of our original trial did not in-
clude UDS diagnosis, preventing us from performing such
an analysis.
Our sample size was small, according to that routinely
used in psychometrics studies, where typically 10 subjects
per item are used. The IIQ includes 30 items, and as such,
a sample of 300 subjects would ideally be required.
Nonetheless, our sample size was similar to that of the orig-
inal scale developers (162 subjects).
Although we agree in principle that all subscales are of
equal importance and therefore warrant weight adjust-
ment as a result of the unequal number of questions in
each subscales, we have not found an alteration in the vali-
dation or in the sensitivity to change when performed with
either methods (ie, weighted score versus simple addition
score, data not shown). We are therefore challenging the
scoring method of the questionnaires. The original report
recommended generating a mean score for each subscale
and transforming it to attribute equal weight to all the
subscales; this would prevent giving higher weight to sub-
scales with more questions. Calculated this way, scores
range from 0 to 400 for the IIQ and from 0 to 300 for the
UDI. This method of scoring is cumbersome. We have
concluded that simple addition shows similar results.
Authorities in the field of health measurement scales sug-
gest that in scales with fewer than 20 or 40 items, weighting
may have some effect, supporting the position of the
scale developers. However, if the items are relatively homo-
geneous (as in these scales), the effect of weighting may
be minimal.
Finally, a note of caution is appropriate regarding the
development of short forms of scales. Ideally, patients
would independently and at different times complete the
long and the short versions of a questionnaire. This would
allow a more robust validation of such scales. However,
neither in this study nor in the original development was
this done, perhaps accounting for the limited correlation
found.
Further studies are necessary to assess the role of the IIQ
and UDI in the clinical setting. These tools remain valid in
the research setting where patients are enrolled on the
basis of UDS criteria and on whom an intervention is
being evaluated. The ideal interval between preinterven-
tion and postintervention application of a questionnaire
was not evaluated in our setting, and we therefore cannot
suggest an answer. Nonetheless, their validity is not sup-
ported for diagnosis, or for the assessment of intervention,
in the clinical setting in the absence of urodynamic evalu-
ation. This study highlights the importance of using a sur-
vey only in the population for whom it was validated.
We acknowledge Ian McDowell, PhD, from the Depart-
ment of Clinical Epidemiology and Community Medicine,
University of Ottawa, for critical review of this manuscript.
30 Harvey et al July 2001
Am J Obstet Gynecol

REFERENCES
1. Hunskaar S, Arnold EP, Burgio K, Diokno AC, Herzog AR, Mal-
lett VT. Epidemiology and natural history of urinary inconti-
nence. In: Abrams P, Khoury S, Wein A, editors. Incontinence,
1st International Consultation on Incontinence, Monaco 1998.
Plymouth: Health Publication Ltd; 1999. p. 204-5.
2. Shumaker SA, Wyman JF, Uebersax JS, McClish D, Fantl JA, for
the Continence Program in Women (CPW) Research Group.
Health-related quality of life measures for women with urinary
incontinence: the Incontinence Impact Questionnaire and the
Urogenital Distress Inventory. Qual Life Res 1994;3:291-306.
3. Lee PS, Reid DW, Saltmarche A, Linton L. Measuring the psy-
chosocial impact of urinary incontinence: the York Incontinence
Perceptions Scale (YIPS). J Am Geriatr Soc 1995;43:1275-8.
4. Kelleher CJ, Cardozo LD, Khullar V, Salvatore S. A new ques-
tionnaire to assess the quality of life of urinary incontinence
women. Br J Obstet Gynaecol 1997;104:1374-9.
5. Streiner DL, Norman GR. Health measurement scales: a practi-
cal guide to their development and use. 2nd ed. Oxford (UK):
Oxford University Press; 1995.
6. Uebersax JS, Wyman JF Shumaker SA, McClish DK, Fantl JA.
Short forms to assess life quality and symptom distress for uri-
nary incontinence in women: The Incontinence Impact Ques-
tionnaire and the Urogenital Distress Inventory. Continence
Program form Women Research Group. Neurourol Urodyn
1995;14:131-9.
7. Abrams P, Blaivas JG, Stanton SL, Andersen JT. The standardisa-
tion of terminology of lower urinary tract function recom-
mended by the International Continence Society. Br J Obstet Gy-
naecol 1990;97(Suppl 6):1-16.
8. Robinson D, Pearce KF, Preisser JS, Dugan E, Suggs PK, Cohen
SJ. Relationship between patient reports of urinary inconti-
nence symptoms and quality of life measures. Obstet Gynecol
1998;91:224-8.
9. Lemack GE, Zimmem PE. Predictability of urodynamic findings
based on the Urogenital Distress Inventory-6 questionnaire.
Urology 1999;54:461-6.
10. Versi E, Griffiths DJ, Harvey MA. A new external urethral occlu-
sive device for female urinary incontinence. Obstet Gynecol
1998;92:286-91.
11. Van Waalwijk Van Doom ESC, Remmers A, Janknegt RA. Extra-
mural ambulatory urodynamic monitoring during natural filling
and normal daily activities: evaluation of 100 patients. J Urol
1991;146:124-31.
12. Webb RJ, Ramsden PD, Neal DE. Ambulatory monitoring and
electronic measurement of urinary leakage in the diagnosis of
detmsor instability and incontinence. Br J Urol 1991;68:148-52.
13. Van Waalwijk Van Doom ESC, Janknegt RA. A telemetric ambu-
latory method to investigate the lower urinary tract function.
Neurourol Urodyn 1993;3:59.
14. Harvey MA, Versi E. Predictive value of clinical evaluation of
stress urinary incontinence: a summary of the published litera-
ture. Int Urogynecol J (in press).
15. Jenseen JK, Nielsen FR Jr, Ostergard DR. The role of patient his-
tory in the diagnosis of urinary incontinence. Obstet Gynecol
1994;83:904-10.
16. Artibani W, Andersen JT, Ostergard DR, et al. Imaging and other
investigations. In: Abrams P, Khoury S, Wein A, editors. Inconti-
nence, 1st International Consultation on Incontinence, Monaco
1998. Plymouth: Health Publication Ltd; 1999. p. 436.

Appendix 1. Incontinence Impact Questionnaire

Not at all Slightly Moderately Greatly

Some women find that accidental urine loss or prolapse may affect their activities, relationships, and feelings. The questions refer to
areas in your life that may have been influenced or changed by your problem. For each question, check the response that best describes
how much your activities, relationships, and feelings are being affected by urine leakage or prolapse.

Has urine leakage and/or prolapse affected your:

Ability to do household chores (cooking, housecleaning, laundry)?
Ability to do usual maintenance or repair work done in home or yard?
Shopping activities?
Hobbies and pastime activities?
Physical recreational activities such as walking, swimming, or other exercise?
Entertainment activities such as going to a movie or concert?
Ability to travel by bus or car for distances less than 20 minutes away from home?
Ability to travel by bus or car for distances greater than 20 minutes away from home?
Going to places if you are not sure about available restrooms?
Going on vacations?
Church or temple attendance?
Volunteer activities?
Employment (work) outside the home?
Having friends visit you in your home?
Participating in social activities outside your home?
Relationship with friends?
Relationship with family excluding husband/companion?
Ability to have sexual relations?
Way you dress?
Emotional health?
Physical health?
Sleep?
Does fear of odor restrict your activities?
Does fear of embarrassment restrict your activities?
In addition, does your problem cause you to experience any of the following:
Nervousness or anxiety?
Fear?
Frustration?
Anger?
Depression?
Embarrassment?

Items in bold indicate components of the short version.

Volume 185, Number 1 Harvey et al 31
Am J Obstet Gynecol

Appendix 2. Urogenital Distress Inventory

The next set of questions deals specifically with your accidental urine loss and/or prolapse.
The following symptoms have been described by women who experience accidental urine loss and/or prolapse. Please indicate
which symptoms you are now experiencing, and how bothersome they are for you. Be sure to answer all items.

Not at all Slightly Moderately Greatly

1. Do you experience frequent urination? If yes, how much does it bother you?
2. Do you experience a strong feeling of urgency to empty your bladder? If yes, how much
does it bother you?
3. Do you experience urine leakage related to the feeling of urgency? If yes, how much
does it bother you?
4. Do you experience urine leakage related to physical activity, coughing or sneezing? If yes,
how much does it bother you?
5. Do you experience general urine leakage not related to urgency or activity? If yes, how
much does it bother you?
6. Do you experience small amounts of urine leakage (that is, drops)? If yes, how much
does it bother you?
7. Do you experience large amounts of urine leakage? If yes, how much does it bother you?
8. Do you experience night time urination? If yes, how much does it bother you?
9. Do you experience bed wetting? If yes, how much does it bother you?
10. Do you experience difficulty emptying your bladder? If yes, how much does it bother you?
11. Do you experience a feeling of incomplete bladder emptying? If yes, how much does it
bother you?
12. Do you experience lower abdominal pressure? If yes, how much does it bother you?
13. Do you experience pain when urinating? If yes, how much does it bother you?
14. Do you experience pain in the lower abdominal or genital area? If yes, how much does it
bother you?
15. Do you experience heaviness or dullness in the pelvic area? If yes, how much does it
bother you?
16. Do you experience a feeling of bulging or protrusion in the vaginal area? If yes, how
much does it bother you?
17. Do you experience pelvic discomfort when standing or physically exerting yourself? If yes,
how much does it bother you?
18. Do you have to push on the vaginal walls to have a bowel movement? If yes, how much
does it bother you?

Items in bold indicate components of the short version.

CORRECTION
In the article "Subtotal hysterectomy in modern gynecology: a decision analysis
(Scott JR, Sharp HT, Dobson MK, Norton PA, Warner HR. Am J Obstet Gynecol
1997;176:1186-92), there was an error in Table II. The cervical cancer death rate was
entered as 0.01 instead of 0.001, which changed the good days lost from 91.75 to 9.18.
Recalculation of good days lost for each surgical procedure gives 25.24 days for total
hysterectomy and 32.25 for subtotal hysterectomy.