Pergamon, J Clin Epidemiol Vol. 48, No. 3, pp. 339-343, 1995. Copyright © 1995 Elsevier Science Ltd. Printed in Great Britain. 0895-4356(94)00147-2
The sensitivity for stress incontinence was 0.66 (95% confidence interval ±0.08), specificity 0.88 (±0.06). The corresponding values for urge incontinence were 0.56 (±0.15) and 0.96 (±0.03), and for mixed incontinence 0.84 (±0.10) and 0.66 (±0.07). Using these indices of validity as corrective measures for the diagnostic distribution reported in the epidemiological survey, the percentage of stress incontinence increased from 51% to 77%, while mixed incontinence was reduced from 39% to 11%. Pure urge incontinence increased from 10% to 12%.
Mixed incontinence will be overreported
in epidemiological surveys. Correction for validity indicates
that a larger majority than hitherto reported may have pure stress
incontinence.
Key words: Urinary incontinence, Women, Urodynamics, Sensitivity and specificity, Prevalence studies, Validity of results
Treatment options differ considerably between urge and stress incontinence, and the distinction is of interest both clinically and epidemiologically. However, the validity of the diagnostic criteriae used in epidemiological surveys remains uncertain.
The present study was designed
to validate simple diagnostic questions concerning female urinary
incontinence, for the subsequent implementation in an epidemiological
survey. The sensitivity and specificity of the diagnostic questions
were to be used as corrective measures for the diagnostic distribution
reported in the epidemiological survey.
The first part of the study was conducted during the years 1988-1992 at the out-patient clinic of the Department of Gynecology and Obstetrics, Trondheim University Hospital. Included were 250 consecutive women referred from the primary health care because of urinary incontinence. Before proceeding to other diagnostic procedures all patients were interviewed in a standard fashion by a nurse using a structured questionnaire designed for the study.
A positive answer to the following question was presumed to be an indication of stress incontinence: "Do you lose urine during sudden physical exertion, lifting, coughing or sneezing?" Urge incontinence was presumed to be indicated by a positive answer to the question: "Do you experience such a strong and sudden urge to void that you leak before reaching the toilet?" A positive answer to both questions was registered as mixed incontinence.
Severity of leakage was estimated by a validated severity index [9]. Typically, slight incontinence denotes leakage of drops a few times a month, moderate incontinence daily leakage of drops, and severe incontinence larger amounts at least once a week.
Afterwards, the women were interviewed and examined by a gynecologist who was not informed about the answers given in the previous interview. All 250 women went through the following diagnostic procedures: An inquiry concerning lower urinary tract function, frequency/volume charts filled in at home for 48 hours, urinalysis, and a gynecological examination, including a stress provocation test with full bladder. Flowmetry was recorded during spontaneous micturition, after which residual urine volume was measured. A medium-fill water urethrocystometry was performed with the patient in the semi-supine position. Finally, resting and stress urethral pressure profiles were measured.
In the urodynamic evaluation genuine stress incontinence was diagnosed when, in the absence of a detrusor contraction, involuntary loss of urine occurred as a result of vigorous coughing. Urge incontinence was diagnosed when loss of urine occurred as a result of involuntary detrusor contractions during the filling phase (detrusor instability). Methods, definitions and units conform to the standard recommended by the International Continence Society [10]. The final diagnosis made by the gynecologist was considered "gold standard" for the anamnestic classification.
The second part of the study
was conducted during the spring of 1992. The National Register
provided names and addresses for the 2 366 female inhabitants
(
20 years) registered in the rural community of Rissa, Norway.
All received an anonymous postal questionnaire, with a reminder
one month later. Among the 1 820 (77%) who answered, 535 (29.4%)
reported urinary incontinence. The age distribution among responders
did not differ from non-responders.
Any frequency or amount of leakage was considered "incontinence". Based on similar diagnostic questions, a classification of stress, urge, and mixed incontinence was made.
Since the survey was anonymous,
it is not known how many of the responders were also included
in the previous validation study. Considering the size of the
population of Rissa and the rest of the county, it may be assumed
that seven or eight women were included in both materials.
The parameters of diagnostic validity were subsequently applied to the results of the epidemiological survey. In addition to sensitivity (Se) and specificity (Sp), the following sums were known:
True positives (a) + false positives (b) = x
False negatives (c) + true negatives
(d) = y
The number of true negatives was calculated by the formula:
False negatives: c = y - d
True positives: 
Finally, the adjusted ("true")
number of each diagnosis (a+c) was calculated.
Among the 250 patients at the outpatient clinic three did not complete the urodynamic evaluation, five were too young, five could not be classified by the diagnostic questions, and one was neither classifiable nor old enough. Among the 535 women identified as incontinent in the epidemiological survey nineteen were not classifiable and seven were of unknown age. Hence, the final materials consisted of 236 women at the outpatient clinic and 509 women in the epidemiological survey.
For 64% of the patients the gynecologist accepted the diagnosis suggested by the urodynamic evaluation, for 36% the diagnosis was changed because history or clinical findings conflicted with the urodynamic diagnosis. The most frequent deviation (made for 23% of the patients) was a diagnosis of urge incontinence (or mixed incontinence) although no detrusor instability had been demonstrated during the urodynamic evaluation.
The anamnestic classification, final diagnoses, and indices of diagnostic value are shown in Tables 2 and 3. When sensitivity and specificity were calculated in different groups of severity, no systematic or significant difference could be detected between the groups (Table 4).
In the epidemiological survey
the percentage of stress incontinence increased after the adjustment,
while the percentage of mixed incontinence decreased. The percentage
of urge incontinence was little affected by the procedure (Table
5).
The first part of the study was conducted among a selected sample of incontinent women. Since patients in the general population have milder problems [9], sensitivity of a questionnaire in that situation may be lower. For the same reason, the specificity may be higher. However, our analyses of different severity categories could not confirm this theory (Table 4), and we conclude that differences in severity do not seem to invalidate the transfer of sensitivity and specificity from clinic to community. However, it is still possible that increased awareness of the problem among clinic patients may result in increased sensitivity.
Basically, sensitivity and specificity are independent of prevalence, and cannot be used to "correct" the overall reported prevalence of incontinence in the population. Thus, the adjusted percentages presented in this study are only valid for the diagnostic distribution among those already defined incontinent. The prevalence of different anamnestic diagnoses did not differ substantially between the two materials.
Our main finding is that the majority of women who report mixed incontinence probably will be given a diagnosis of stress incontinence if examined more closely. The reason for this is the low specificity (high number of false positives) of the diagnostic questions for mixed incontinence. The high specificity for urge incontinence explains why this diagnosis is little affected by the adjustment procedure. A highly specific test is best for ruling in disease. It is possible to increase the specificity for stress incontinence by diagnosing it only in the absense of urge incontinence (and vice versa), but this will also reduce the sensitivity [11,12].
Our non-adjusted results from the epidemiological survey are in agreement with comparable studies [18]. The exception is the study by Sommer et al. who reported a very high percentage of urge incontinence [3]. Their result may be biased since all participants were asked to fill in a frequency chart for 72 hours before returning the questionnaire. This may have selected individuals who found the procedure meaningful.
No epidemiological study has
been published in which a urodynamic evaluation has been applied
to a truly unselected incontinent population. Diokno et al. reported
cystometric characteristics of continent and incontinent noninstitutionalized
elderly (
60 years), and found uninhibited detrusor contractions in only
12% of the incontinent women [13]. Fantl et al. did a urodynamic
evaluation of 145 community-dwelling older women (
55 years) who entered a clinical trial of behavioral treatment.
Excluding the unclassified, 69% were diagnosed as stress, 13%
urge, and 18% mixed incontinence [14]. Thus, in materials less
selected than most urodynamic studies, the diagnostic distribution
compares well with our adjusted results. Ideally, external validation
of our survey results could have been done by clinical and urodynamic
evaluation of the responders, but since the majority of them were
not bothered by their incontinence, this was never considered
feasible [9].
Nevertheless, our adjusted results
should be interpreted with caution. Interviews and postal questionnaires
may have different sensitivity and specificity. Minor differences
in sensitivity and specificity would bring about large deviations
in the adjusted numbers. This is illustrated by the fact that
the adjusted total number of diagnoses (
20 years) do not exactly match the numbers reached by simply adding
the two subgroups (20 - 50 years and
50 years). Further studies are obviously needed to clarify the
true diagnostic distribution of incontinence in the general population.
Adding to this uncertainty comes the question about the validity of the gold standard used [15,16]. A urodynamic test performed at a single point in time cannot retrospectively confirm a patient's past history of urge incontinence with any reliability [16]. Doubts about the validity of single diagnostic tests were the main reason why we chose the final diagnosis made by the gynecologist as our gold standard, even if this entailed an element of subjectivity.
Bearing these reservations in mind, some clinical implications may be drawn from this study. Anamnestic questions may be misleading with regard to the true diagnosis. As far as stress incontinence is concerned, it is not acceptable in a preoperative setting to have a large number of false positives. Irreversible and possibly harmful surgical procedures should not be based solely on responses to questionnaires or interviews [12].
Treatment options in general practice, however, are less risky. Pelvic floor exercises will not be harmful if applied to a patient with detrusor instability. Thus, in a setting where history is the most important diagnostic tool, sensitive questions should be preferred, and lower specificity accepted. Our results indicate that pure urge incontinence is very rare among premenopausal women. Probably, all women reporting urinary incontinence to the general practitioner should be taught pelvic floor exercises.
The most frequent error, however,
will be a diagnosis of mixed incontinence when, in fact, the correct
diagnosis is pure stress incontinence. These women will be at
risk of receiving useless medical treatment. The treatment options
are not irreversible, however, and usually not harmful. In fact,
both general practitioners and gynecologists occasionally prescribe
anticholinergic medication to patients with isolated stress incontinence
[17,18].
Due to low specificity of the diagnostic questions, mixed incontinence will be overreported in epidemiological surveys. Correction for validity indicates that a larger majority than hitherto reported may have pure stress incontinence.
Artikkel kan hentes i Word-format her