QUALITY OF LIFE OF PERSONS WITH URINARY INCONTINENCE: DEVELOPMENT OF A NEW MEASURE* T.H. WAGNER,
ABSTRACT Objectives. Our objective was to develop a self-report quality of life measure specific to urinary incontinence (I-QOL) that could be used as an outcome measure
in clinical trials and in patient care centers. of 20 individuals with urinary incontinence. Refining the questionnaire was accomplished by structured interviews of 17 individuals with urinary incontinence. Testing the I-QOL’s psychometric properties involved two administrations (n = 62) along with measures of psychologic well-being and functional status. Results. The rigorous development process ensured that the measure was complete and understandable. The I-QOL proved to be internally consistent (alpha 0.95) and highly reproducible (r = 0.93; 18 days; SD 4). For discriminant validity, severity of incontinence (P < 0.000 1) and number of medical appointments in the past year to treat incontinence (P c 0.000 1) significantly predicted I-QOL scores, Convergent validity analyses confirmed our predictions that the I-QOL scores were more closely related to overall well-being than bodily pain. Conclusions. The I-QOL proved to be valid and reproducible as a self-administered measure for assessing quality of life of patients with urinary incontinence. UROLOGY8 47: 67-72, 1996.
Methods. The I-QOL was developed from interviews
rinary incontinence is characterized as the involuntary loss of urine sufficient enough to be a problem. ’ Recent steps have been taken to establish clinical practice guidelines because treatment options vary widely in terms of cost, side effects, and efficacy Treatment efficacy is routinely measured as a decrease in incontinence episodes or a reduction in the number of pads used; however, these outcome measures are limited in that they fail to take into account patients’ perceptions of their condition and treatment outcome. Clearly, patient-relevant outcome measures should be considered in the evaluation of alternative treatments, and this perspective is becoming more generally accepted by the urologic community. *This research was funded by a research contract from Eli Lilly and Company. D.P.B. is an employee of Eli Lilly and Company; T.H.W., M.L.M., D.L.P., and T.G.B. have no direct or indirect financial contracts with Eli Lilly and Company. Please contact D.P.B. for the most recent version of the I-QOL. T.H.W. is now a NIA predoctoral trainee at the University of California at Berkeley and can be contacted via e-mail: [email protected]
From the Cost and Outcomes Assessment Team, Department of Health Services, and the Department of Urology, University of Washington, Seattle, Washington; and Eli Lilly and Company, Lilly Research Laboratories, Indianapolis, Indiana Reprint requests: Donald L. Patrick, Ph.D., MSPH, Department of Health Services, The University of Washington, 146 N. Canal Street, Suite 310, Seattle, WA 98103 Submitted: May 15, 199.5, accepted (with revisions): July 21, 1995 UROLOGYa 47 (I), 1996
A number of generic and urinary incontinencespecific measures have been used to assessthe effects of incontinence on functioning and well-being. 2-7We asked incontinent patients to review the content of these instruments, and surprisingly most thought that none of the measures adequately covered their concerns. Participants said that too many of the questions dealt with symptoms and too few questions reflected their emotions. Based on these interviews, we decided to develop a quality of life measure specific to urinary incontinence (I-QOL). During the development of the I-QOL, emphasis was placed on making sure that it covered all of the important issues and that each item was easy to understand. The IQOL was designed to be applicable to all persons who experience chronic urinary incontinence. The measure was then tested for its validity and reproducibility
MATERIAL AND METHODS We followed three consecutive steps to develop and validate the I-QOL: (1) identification of QOL items specific to urinary incontinence; (2) development and refinement of the draft I-QOL questionnaire; and (3) assessment of the psychometric properties of the I-QOL. Participants were recruited from the University of Washington, Department of Urology, and through newspaper advertisements. Human subjects’ approval was obtained and participants signed an informed consent. Participants were screened over the telephone and were only allowed to participate in a single step. People were excluded if they were under age 18 years or if they experienced symptoms inconsistent with stress, urge, or mixed incontinence (such as
postvoid dribbling). Recruitment was stratified by gender, incontinence type, and incontinence severity to ensure a wide diversity of participants. Incontinence type (stress, urge, or mixed) was identified by collecting self-report symptom data as defined by Fultz and Herzog.8 Urge incontinence was defined as losing urine before the person can get to the bathroom. Stress incontinence was identified as losing urine when the person coughs, sneezes, runs, walks, jumps, or does some other specific activity. People were classified as having mixed incontinence if they reported having both stress and urge symptoms. Questions about undergarment use and altering physical or social life provided the primary measure of incontinence severity. We defined mild incontinence as those who said they do not wear protection or wear it only when needed and do not alter their social activities. Severe incontinence was defined as wearing heavy protection all the time and changing social activities. All others were categorized as having moderate incontinence. Additional estimates of severity included participant self-report and self-reported number of medical appointments to treat the incontinence.
IDENTIFICATION 0~ QUALITY~FLIFE
Fifteen women and 5 men were interviewed by trained quality-of-life researchers. Each interview started by asking the person, “How does urinary incontinence affect your quality of life?” Additional interview questions were derived from a review of relevant sociomedical literature and from the needs-based model of Hunt and McKenna.9 The tape-recorded interviews were analyzed to identify QOL items, which were viewed as statements representing an issue about urinary incontinence that the respondent said was important. Each item maintained, as closely as possible, the language used by the participants, and as a result no items were copied from existing questionnaires. The list of approximately 80 issues was reduced to 28 after removing items that were obviously redundant, related to treatment, or were not applicable to all persons with stress, urge, or mixed incontinence.
COGNITIVE DEBRIEFING OF THE QUALITYOF~JFE SPECIFIC TO INCONTINENCE Seventeen individuals with incontinence completed the 28-item measure and were interviewed about the measure’s response categories and format. Respondents were then queried about the relevance and importance of each item. The interviews indicated that all 28 items were consistently interpreted by each of the respondents, and we found no evidence to add or remove any items.
PSYCHOMETRIC TESTING OF THE QUALITYOFLIFE SPECIFIC TO INCONTINENCE The primary objective of this step was to test the psychometric properties of the 28-item I-QOL questionnaire. Sixty-eight persons were sent the I-QOL questionnaire through the mail. All of the I-QOL items included a four-point Likert response scale (very much, moderately, a little, not at all). Scoring the I-QOL questionnaire involved summing the responses into a single score. The sum score was then transformed to a O-100 scale to facilitate interpretation, with a higher number representing a better QOL.2 We reviewed data on each I-QOL item for the following characteristics: (1) ceiling effect in which more than 50% circled “Not at all”; (2) greater than 5% missing data; (3) indications that an item measures a different construct demonstrated by an item-to-total correlation less than 0.40; and (4) redundancy with another item as indicated by an interitem
correlation more than 0.70. Our goal was to remove items that adversely affect the ability of the I-QOL questionnaire to discriminate between different groups of people, as well as diminish its chances of detecting important changes that result from treatment. This information, along with the interview data, was used to remove items. For example, if a pair of items were highly correlated, we removed the item that appeared less important to participants. We were cautious, however, about removing too many items for fear of compromising the measure’s content validity. Content validity is the extent to which the domains of interest, which in this case are the participants’ concerns, are comprehensively sampled by the itemslo After removing items, the I-QOL was assessed for reliability (that is, internal consistency and reproducibility) and validity (convergent and discriminant). Internal consistency of an additive scale is determined by the correlations of items within the scale.” Alpha values range between 0 and 1, and the minimal acceptable level is 0.70.12 An acceptable alpha level indicates that the items perform well enough together to be a composite score. Reproducibility refers to the correlation between responses on the same items administered at different times.13 To assess reproducibility, all persons who returned the first questionnaire were sent the same questionnaire a second time, approximately 2 weeks later. We expected that this was a sufficient amount of time for respondents to forget their original answers, but not long enough for their QOL to change dramatically. The intraclass correlation coefficient (ICC) is the preferred method for estimating reproducibility.14 In general, a correlation of 0.70 is minimally acceptable, but a correlation above 0.80 is preferred.‘* In practical terms, a measure with low reproducibility will have trouble differentiating between meaningful change and random variation. Convergent validity involves comparing the I-QOL to logically related measures. If a priori predictions of association are accurate, then convergent validity is achieved. To assess convergent validity, the Short Form 36-Item Health Survey (SF-36)‘6 and Psychological General Well-Being schedule (PGWB)” were also completed by each participant. The SF36 is a generic functional status measure and the PGWB is a generic measure of psychological well-being. We expected the scores on the I-QOL to be more highly correlated with well-being than functional status. More specifically, we postulated that the I-QOL would be most closely associated with the overall well-being score on the PGWB and would be decreasingly correlated with the mental health and bodily pain dimensions of the SF-36. There are different methods for assessing discriminant validity. Because of its intuitive appeal, we evaluated the measure’s ability to discriminate between persons with mild, moderate, and severe incontinence. We anticipated that QOL scores would worsen as incontinence severity increased.
SAMPLE CHARACTERISTICS Sixty-two of the 68 participants (91%) returned the I-QOL questionnaire. We conducted a nonrespondent analysis on data collected during the initial telephone screen. Those who did not return the first questionnaire were not significantly different with respect to age, gender, incontinence type or severity. Sixty (97%) of the 62 persons sent a retest returned the questionnaire an average of 18 days later (SD 4).
Item * 1 I worry about wetting myself 2 I feel embarrassed talking about my incontinence with others 3 I have to watch how much I drink because of my incontinence 4 I worry about coughing or sneezing because of my incontinence 5 I have to be careful standing up after I’ve been sitting down because of my incontinence 6 I worry about where toilets are in new places 7 I feel depressed because of my incontinence 8 Because of my incontinence, I don’t feel free to leave my home for long periods of time 9 Having incontinence hurts my self-confidence 10 I feel frustrated because my incontinence prevents me from doing what I want 1 1 I worry about others smelling urine on me 12 Incontinence is always on my mind 13 It’s important for me to make frequent trips to the toilet 14 I avoid laughing because of my incontinence 15 I feel ashamed because of my incontinence 16 Because of my incontinence, it’s important to plan every detail in advance 17 I worry about my incontinence getting worse as I grow older 18 I have a hard time getting a good night of sleep because of my incontinence 19 I worry about being embarrassed or humiliated because of my incontinence 20 I avoid hugging others because of my incontinence 2 1 My incontinence makes me feel like I’m not a healthy person 22 My incontinence makes me feel helpless 23 I get less enjoyment out of life because of my incontinence 24 I worry about not being able to get to the toilet on time 25 I feel like I have no control over my bladder 26 I have to watch what I drink because of my incontinence 27 My incontinence limits my choice of clothing 28 I worry about having sex because of my incontinence *Items
Item-Total Correlation 0.68 0.33 0.49 0.40 0.53 0.65 0.67 0.72 0.85 0.81 0.58 0.78 0.62 0.57 0.75 0.71 0.59 0.72 0.78 0.58 0.72 0.79 0.82 0.66 0.75 0.71 0.79 0.58
and 20 were removed.
Women comprised 68% of the sample and the average age was 64 years (SD 14); 96% of the participants were white. Educational status and income were widely distributed; 39% had a high school education or less and the median income was between $15,000 and $24,999. Approximately 70% of the respondents reported having the condition for 5 years or less; 40% reported not seeking medical treatment for their incontinence in the past year. The sample consisted of 14 (25%) mildly, 21 (36%) moderately, and 23 (40%) severely incontinent individuals as determined by our classification algorithm; the breakdown by self-perceived severity was similar: 20 (33%) were mild, 17 (28%) were moderate, and 24 (39%) were severe. Separating the sample by type yielded 10% with stress incontinence, 7% with urge, and 84% with mixed. SCALE REDUCTION
Two items demonstrated ceiling effects, one item had more than 5% missing data, one item indicated a low item-to-total correlation, and nine pairs of items had correlations above 0.70 (Table I). In the end, we chose to remove six items. Although it had 12% missing data, the item about having sex because of incontinence was retained [email protected]
because of its importance to our interviewees. Further analyses indicated that missing data were related to persons without partners, and thus persons who skipped this question were assigned an item score of zero. After removing the six items, the I-QOL was assessed for validity and reliability RELIABILITY
Internal consistency, as measured by Cronbach’s alpha, was 0.95, indicating that the 22 items perform well enough together to be a composite score. Reproducibility as measured by the ICC was 0.93 after an average retest period of 18 days (SD 4). VALIDITY
For convergent validity, correlations between the I-QOL and domains of the psychologic wellbeing (PGWB) ranged between 0.45 and 0.62. Correlations between the I-QOL and functional status domains on the SF-36 ranged from 0.35 to 0.67 (Table II). Thus, the expectation that the IQOL would be more closely correlated to the PGWB than the SF-36 was not confirmed. Analyses did indicate, however, that the I-QOL scores were more closely related to well-being than bodily pain, as was predicted. 69
Convergent validity of quality of life specific to urinary incontinence II. Psychological General Well-Being schedule and Short Form 36-Item domain
PCWB Total PGWB Health worry concern Energy and vitality Depressed mood Tension and anxiety Positive well-being Behavioral and emotional KEY: I-QOL
oJfiJe spec~~c fo urinary
Discriminant of life specific to urinary
Characteristic Classification of disease severity (P < 0.000 11 Mild Moderate Severe Self-perceived disease severity (P < 0.000 1) Mild Moderate Severe Appointments for treatment in last year (P < 0.000 1) 0 l-2 3+ Total
Correlation with I-QOL 0.62 0.59 0.55 0.52 0.51 0.48 0.45 incontinence;
= S~OII Form
validity of the quality incontinence measure No.
Mean Scores (SD)
15 21 23
68.5 (2 1 .O) 47.7 (22.7) 25.7 (16.8)
20 15 24
70.3 (15.4) 39.4 (20.51 25.9 (17.8)
24 16 19
59.4 (20.2) 45.1 (25.5) 24.9 [20.8)
As expected, I-QOL scores were significantly worse as severity classification (mild, moderate, and severe), self-perceived severity (mild, moderate, and severe), and the number of medical appointments to treat incontinence in the past year (continuous variable segmented into tertiles) increased (F(2,56) = 21.0,34.8, 13.1, P
score with scores Correlation with I-QOL 0.67 0.62 0.53 0.53 0.53 0.52 0.43 0.35
MOS SF-36 Role, physical Social function Physical functioning Role, emotion Mental health General health Vitality Bodily pain 36.Ilrm
cant predictor of QOL: R2 = 0.66, F(1,55) = -3.9, P
We developed a highly reliable and valid selfadministered questionnaire to assess the QOL for persons with urinary incontinence. Most of the expectations about how the I-QOL would perform were statistically confirmed with a high degree of confidence. Multivariate regression verified that scores on the I-QOL were not significantly affected by demographic variables. This is reassuring, since the I-QOL was developed from a model presupposing each item to be meaningful to all persons with urinary incontinence. The ability of the I-QOL to discriminate between groups was compared with the SF-36 and PGWB. Results clearly demonstrated that the I-QOL was more sensitive for detecting differences between levels of self-perceived incontinence severity than either the PGWB or SF-36; this was also true for our classification of incontinence severity and the number of medical appointments in the last year. As might be expected, disease severity as determined by our classification system was highly correlated with participants’ self-reported severity (Spearman’s Y = 0.70). When these severity estimates disagreed, most people (60%) said they were less severe than the rating as determined by our severity classification. One explanation for this discrepancy is that people have positively biased self-perceptions of their situation, which Taylor and Brown” argue are normal, healthy, and promote psychological well-being. One limitation of this study was that we were not able to assess the relationship between the I-QOL and more clinically objective measures of severity, such as the perineal pad test. This limitation was recognized at the beginning of the study, but we thought it was more critical to recruit a wide range of subjects, including those not seeking medical care. More discriminant validity analyses will be [email protected]
47 (11, 1996
conducted in future clinical trials when such data will be available on each subject. A second limitation, which affects the generalizability of the findings, is that the sample does not typify all persons with incontinence. This was not feasible within the scope of the study, and this, it is hoped, will be addressed in future research. The results from this study are highly favorable and suggest that the I-QOL fills the current need for a QOL instrument that covers all of the important issues and is applicable to all persons with urinary incontinence. Although the I-QOL has already proven ready for use in clinical trials, there is ongoing research to translate and validate the measure in Europe. In upcoming controlled clinical trials, the ability of the I-QOL to detect minimally significant changes in QOL that result from treatment (that is, responsiveness) will be assessed. In addition, a study on the measure’s internal domain structure is also being conducted in the United States to determine if subscores are more clinically meaningful than a single, overall QOL score, which might lead to the formation of a subscale that would be easier to use on a routine clinical basis. ACKNOWLEDGMENT. To Paul Froese, Kelly Graham, Scott Ramsey, Lisa Smith Wagner, and Diane Wild for their assistance in developing the measure and helpful comments on this document.
REFERENCES 1. Urinary Incontinence Guideline Panel: Urinary incontinence in adults: clinical practice guideline. Rockville, MD: Agency for Health Care Policy and Research, U.S. Department of Health and Human Services, 1992. 2. Grimby A, Milsom I, Molander U, Wiklund I, and Ekelund P: The influence of urinary incontinence on the quality of life of elderly women. Age Ageing 22: 82-89, 1993. 3. Hunskaar S, and Vinsnes V: The quality of life in women with urinary incontinence as measured by the sickness impact profile. J Am Geriatr Sot 40: 976-977, 1991. 4. Yu LC: Incontinence stress index. Measuring psychological impact. J Gerontol Nurs 13: 18-25, 1987. 5. Wyman JF, Harkins SW, Choi SC, Taylor JR, and Fantl JA: Psychosocial impact of urinary incontinence in women. Obstet Gynecol 70: 378-381, 1987. 6. Cella D: Functional Assessment of Incontinence Therapy, version 2, Chicago, Rush Division of Psychosocial Oncology, 1994. 7. Shumaker SA, Wyman JF, Uebersax JS, McClish D, and Fantl JA: Health related quality of life measures for women with urinary incontinence: the incontinence impact questionnaire and the urogenic distress inventory. Qua1 Life Res 3: 291-306, 1994. 8. Fultz NH, and Herzog AR: Measuring urinary incontinence in surveys. Gerontologist 11: 708-713, 1991. 9. Hunt SM, and McKenna SP: The QLDS: a scale for the measurement of quality of life in depression. Health Policy 22: 307-319, 1992. 10. Guyatt GH, Feeny DH, and Patrick DL: Measuring healthrelated quality of life. Ann Intern Med 118: 622-629, 1993. 11. Cronbach LJ: Coefficient alpha and the internal structure of tests. Psychometrika 16: 297-334, 1951. 12. Nunnally JC: Psychometric Theory, vol. 2. New York: Basic Books, 1978, pp 229-246. [email protected]
13. Patrick DL, Wild DJ, Johnson ES, Wagner TH, and Martin ML: Cross-cultural validation of quality of life measures, in Orley J, and Kuyken W (Eds): @a&y ofI.ije Assessment: International Perspectives. Berlin, Springer-Verlag, 1994, pp 19-32. 14. Deyo RA, Diehr P, and Patrick DL: Reproducibility and responsiveness of health status measures: statistics and strategies for evaluation. Controlled Clin Trials 12: 142s-158s, 1991. 15. Aday LA: Designing and Conducting Health Surveys: A Comprehensive Guide. San Francisco: Jossey-Bass, 1991, pp 41-50. 16. Ware JE, and Sherbourne CD: The MOS 36-item shortform health survey (SF-36). I. Conceptual framework and item selection. Med Care 30: 473-483, 1992. 17. Dupuy HJ: The psychological general well-being (PGWB) index, in Wenger NK, Mattson ME, Furberg CD, and Elinson J (Eds): Assessment of Quality of Life in Clinical Trials of Cardiovascular Therapies. New York: Le Jacq, 1984, pp 170-183. 18. Taylor SE, and Brown JD: Positive illusions and wellbeing revisited: separating fact from fiction. Psycho1 Bull 116: 21-27, 1994. EDITORIAL COMMENTS The primary goal of health care is the improvement of patient outcomes. Accurate and reproducible methods of measuring outcomes have become one of medicine’s principal challenges. Implicit in this process is the integration of two models of health. One is the clinical or biomedical, which focuses on etiologies of disease, pathophysiology, and clinical outcomes. The other is from the social science disciplines where the focus is on functional status and quality of life. The former are concepts familiar to all physicians-the latter are becoming so as we attempt to involve patients in their own care and to understand better how they are affected by conditions or therapeutic interventions. Urinary incontinence can certainly have a profound impact on patients’ quality of life.‘,* Having a uniform, reproducible, and parsimonious instrument to measure this has several important values. First, it allows clinicians and researchers to evaluate comparability of individual patients as well as study populations. Second, it facilitates comparisons of results of studies performed at different sites and with different therapies. This is particularly important when evaluating conditions such as incontinence, where so much emphasis is placed on patients’ perceptions of their problem. Finally, it can serve as an excellent initiation to a dialogue between the patients and their physicians. It is gratifying that urologists utilizing psychometric and clinometric methodologies are increasingly involved in the development and validation of instruments such as the one presented here. As students of genitourinary disease entities and practitioners of their therapies, it is vital that we continue to play an integral role in this process. REFERENCES 1. Grimby A, Milsom I, Molander U, Wiklund I, and Ekelund P: The influence of urinary incontinence on the quality of life of elderly women. Age Ageing 22: 82-89, 1993. 2. Hunskaar S, and Vinsnes A: The quality of life in women with urinary incontinence as measured by the sickness impact profile. J Am Geriatr Sot 39: 378-382, 1991. Michael P. O’Leary, M.D. Harvard Medical School Brigham and Women’s Hospital 45 Francis Street ASBIlI-3 Boston, MA 02115 71