Quality of Life Research,
5,
pp. 443-449
Cross-cultural development of a quality of life measure for men with erection difficulties T. H. Wagner, *+ D. L. Patrick, P. S. Froese
S. R McKenna
and
Department of Health Services, School of Public Health and Community Medicine, University of Washington, Seattle, USA (T. H. Wagner, D. L. Patrick, P. S. Froese); Galen Research, Manchester, England (S. P. McKenna)
Erection difficulties have a profound effect on a man’s quality of life, however, the emotional consequences are often overlooked in quantitative research where most sex-related questionnaires focus on a man’s functional ability. Consequently, we developed a cross-cultural instrument to measure quality of life specific to male erection difficulties (QOL-MED). The items in the QOL-MED originated from interviewing forty men with erection difficulties in Seattle and Boston. Twelve men In the USA and 29 men in England helped us refine the instrument. Testing the QOL-MED’s psychometric properties involved two administrations over a twoweek period in the USA (n = 40) and the UK (n = 29). For dlscrlminant validity, we predicted quality of life would worsen with increased self-perceived severity of the condition. After controlling for years with erection difficulties in a linear regression model, we found a significant negative association between self-perceived severity and quality of life for men in the UK only (p
$I
research was supported from a research grant from Syntex,
To whom correspondence should be addressed at University of California, Berkeley, 1642 Francisco, Berkeley, CA 84703, USA; Phone: (510) 8481587; Fax: (510) 643-8614; Internet: thwagnerOgamet.berkeley.edu.
l
+ Mr. Wagner is now a doctoral student at the University of California at Berkeley, Berkeley, CA, USA.
@
1996 Rapid
Science
Publishers
Introduction Approximately 10% of healthy ‘young’ males have been reported to complain of erectile dysfunction.’ This number refers only to those whose experience is not an obvious result of some other injury or disease. Segraves ef al.’ found that in men with chronic diseases or disability, the prevalence of erectile dysfunction exceeds 30%. It has also been reported that approximately SO% of the male population experiences occasional erectile difficulty that does not require treatment.2 Differences in prevalence rates are also related to the difficulty in defining and diagnosing the condition. The US National Institutes of Health defined a male as sexuaJly functional when he is able to maintain an erection long enough to ‘permit satisfactory sexual intercourse.‘3 Clearly the use of the term ‘satisfactory’ implies subjective feeling about one’s ability. If treatment is desired, one can choose between a variety of drugs, surgeries and therapies, all of which have differing degrees of success and risks. The effectiveness of therapies is often determined by patient report. Previously used self-report outcome measures for erection difficulties have focussed primarily on functional status, which is the impairment resulting from the health condition’s effect on the physical, social and/or mental functioning. An example would be the reduced functional capacity of a man with erection difficulties to physically obtain an erection. Functional status measures are limited in that they do not incorporate what the person feels is important with respect to their condition. It is the subject’s evaluation of the effects of the condition and its treatment in relation to an individual’s goals, values, and expectations, which is unique to quality of life.’ In assessing the quality of life of men who suffer from erectile dysfunction, it is very difficult to assess
Quality
of Life Research
Vol5
1996
443
T. H.
Wagner et al.
the over-all effect it has on their emotional and personal life because, to some degree, the dysfunction may be the result of a pre-existing psychological problem. Nevertheless, it is clear that erectile problems produce a certain amount of stress and anxiety that were not experienced prior to the dysfunction. Clearly erection difficulties have a profound effect on a man’s quality of life; however, the emotional consequences are often overlooked in quantitative research because most sex-related questionnaires focus on functional ability Thus, the goal of this study was to develop a quality of life measure specific to men with erection difficulties (QOL-MED). The QOLMED was designed for use in controlled clinical trials, however, we expect that it may be useful for collecting important information in routine clinical care.
Methods Recruiting men for this study was challenging due to the highly sensitive nature of the condition. In the USA, participants were recruited from advertisements in the Se&e Times newspaper and from the Urology Clinic at Boston University. Participants in the UK were recruited from clinic populations in London and Leeds with the help of health care professionals. Human subjects approval was obtained and participants signed an informed consent. Men were excluded from participation if they were under the age of 18, had a sexual disorder other than problems with getting or keeping an erection (i.e., premature ejaculation), were currently enrolled in a Syntex sponsored clinical trial, or were not in a stable, heterosexual relationship. Inclusion and exclusion criteria were determined through participant self-report. The development and validation of the QOL-MED involved three successive steps: (1) identification of quality of life items relevant to erection difficulties; (2) development and refinement of the draft QOL-MED questionnaire and (3) assessment of the psychometric properties of the QOL-MED.
Identification of quality to erection difficulties
of life items relevant
The theoretical foundation for the measure was Hunt and McKenna’s5 needs-based model. The theory posits that an individual’s quality of life is dependent on his or her ability to meet basic and learned human needs. Effective treatment allows more needs to be met leading to an improvement in quality of life. Condition-specific quality of life instruments can be
444
Quality
of Life Research
Vol5
1996
constructed using this model by identifying those needs interfered with by the specific illness. To this end we conducted semi-structured interviews with twenty males with erection difficulties in both Seattle and Boston. We developed items (questions) from the interview transcripts, maintaining whenever possible the participants’ vernacular. Issues were excluded if they were not related to the condition, were related to treatment, or if they were not applicable to all men with erection difficulties. From this step, we produced a draft QOL-MED instrument.
Refinement
of the draft QOL-MED
questionnaire
The draft questionnaire was built from paraphrasing participants’ statements from the interviews into viable items. This process by itself, however, does not ensure that the items are intelligible, appropriate and meaningful. To address this, we recruited an additional 12 men in the USA and 29 men in the UK to review the draft questionnaire. This vital step reinforced 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 items6 Participant’s identified some vague phrases and two missing issues, and as a result a 27-item pilot QOL-MED questionnaire was produced.
Assessment of the psychometric of the QOL-MED
properties
Testing the psychometric properties of the QOL-MED required all participants to complete the postal survey and a two-week retest. Eligibility criteria for this step also required that the man’s partner be willing to complete a survey on how erection difficulties affect her quality of life. This criterion proved too restrictive in the UK, and thus UK men were allowed to enter without involving their partner. Results from the female partners will be reported elsewhere. At the first administration, the Psychological General Well-Being Schedule7 (PGWB) was completed along with the QOL-MED in the USA and UK. The Short Form 36-Item Health Survey’ (SF-36) was also used in the USA. At both sites, we collected data on perceived severity and duration of condition, as well as age and education level. For the retest, men completed the QOL-MED only. To maximize the QOL-MED’s ability to discriminate between different groups of people, as well as improve its chances of detecting important changes
QOLfir male erectile dysfuncton resulting from treatment, we reviewed each item for the following adverse characteristics: (1) ceiling effect in which the majority of the respondents circled Not at All’; (2) missing data greater than 5%; (3) an item-to-total correlation, which indicates an item measures a different construct and (4) an inter-item correlation of 0.70, which indicates redundancy with another item. After removing items, the QOL-MED was assessed for reliability (reproducibility), internal consistency and validity (i.e., convergent and discriminant). In the USA, the pilot QOL-MED instrument consisted of four-point Likert response scales that were scored as: 4 -‘Very much’, 3 -‘Moderately‘, 2 --‘A little’, and 1 -‘Not at all’. The response categories were changed slightly for the UK to maintain conceptual equivalence (‘Very much’, ‘Quite a lot’, ‘A little’, and ‘Not at all’). The items’ ordinal values were summed to calculate a total quality of life score, and then the score was transformed so that a zero represented the worst possible quality of life, and one hundred represented the best. Discriminant validity. We expected scores on the QOL-MED instrument to be negatively associated with the men’s self-perceived disease severity (mild, moderate and severe). This was tested using data collected during the first administration. Convergent validity. The Psychological General WellBeing Schedule (PGWB) was administered to all men in the USA and UK. The Medical Outcomes Study Short Form Health Survey (SF-36) was also administered in the USA. In general, we expected that correlations would be higher between the QOGMED and PGWB than between the QOL-MED and SF-36 because the PGWB is a measure of well-being while the SF-36 measures functional status (See Table 4). More specifically, we anticipated that the QOL-MED would be more highly correlated with depression, energy and anxiety than bodily pain or physical functioning scores. Reproducibility. Men were asked to complete the QOL-MED instrument at first administration and again 2 weeks later. The intraclass correlation coefficient (ICC) was calculated to assess the degree of reliability.’ Although a minimum correlation coefficient of 0.70 is necessary to claim an instrument is reliable,” it is optimal to have ICC correlations above 0.85, particularly when the instrument is intended for use in a clinical setting. Znternal consistency. Cmnbach’s coefficient a was used to assess internal consistency Cronbach’s a is used to analyze additive scales to determine if the items
within a scale are highly associated. For a composite score, it is optimal to have a levels above 0.90, although 0.70 is usually considered a minimally acceptable level.”
Results Sample
characteristics
In the USA field test, 40 of the 42 (95%) participants recruited returned the QOL-MED questionnaire. Of the 40 people sent a retest questionnaire, 37 (93%) returned it an average of 24 (SD = 9) days later, for a total response rate of 88%. Response rates were significantly worse in the UK. Of the 55 patients included in the study 30 (55.6%) were returned and of these, only 24 (80.0%) returned the second set of measures on average 19 (SD = 5) days later, for a total response rate of 44%. Within the UK, response rates varied considerably; participants recruited at Leeds had an overall response rate of 65%, while those from London had a response rate of 29%. Table 1 details the sample characteristics. American males experienced erection difficulties for a median of two years, while the median length in the UK was approximately four years. Men in the USA were much more unlikely to have visited the doctor in the past year
Scale reduction Three items in the USA demonstrated a ceiling effect, which indicated that more than 50% of the responses were ‘Not at All’ (Table 2). In the UK, eight items exhibited similar poor response distribution. The percentage of missing data was very small in both the USA and the UK. There was one item in the USA and four in the UK that had item-to-total correlations below 0.40. Thirteen pairs of items in the USA had inter-item correlations above 0.70, and twenty-five pairs in the UK had correlations above 0.70. Based on this information, we removed nine items from both the USA and UK measures, and then both versions of the instrument were assessed for validity, internal consistency and reproducibility.
Validity Discriminant The relationship
validity between the men’s perceived disease
Quality of Life Research
Vol5
1996
445
T. H.
Wagner
et al.
Table 1. Sample characteristics Characteristic
Category
%
n
Age
< 45 46-60 61+
30.0 40.0 30.0
12 16 12
23.3 36.7 40.0
7 11 12
Time with erection difficulties (yrs)
Cl l-2 3-5 >5
17.5 32.5 22.5 17.5
7 13 9 7
13.3 16.7 36.7 33.3
4 5 11 10
Medical appointments
0 1-3 4+
50.0 26.0 22.0
20 12 6
0.0 33.3 66.7
0 10 20
No problem Mild Moderate Severe
0.0 10.3 42.5 45.0
0 10 17 18
3.3 16.7 40.0 40.0
1 5 12 12
(per yr for ED)
Self-rated severity
severity and their quality of life is presented in Table 3. Among USA and UK males the mean scores decreased, although the relationship between perceived severity and quality-of-life scores was not statistically significant using one-way analysis of variance (ANOVA), F(2,35) = -0.59, p = 0.56; F(2,25) = -2.6, p = 0.09, respectively The relationship between sexual satisfaction in the last two weeks and quality of life was only tested in the USA; very few couples reported having intercourse so significant differences were not detected. The relationship between men’s perceived severity of illness and their quality of life was then assessed in a linear regression model. It was determined that the length of time having erection difficulties confounds the relationship between perceived severity and quality of life. After taking this variable into account, the trend was the same, the negative association between severity and quality of life reached statistical significance only in the UK, R2 = 0.26, F(1,26) = -2.9, p < 0.01. Included in the regression model were perceived severity and length of time with erection difficulties.
Convergent
validity
Table 4 presents the expected and observed Pearson’s correlations between the erection difficultly quality
446
Quality
of Life Research
Vol 5
1996
USA (n = 40) % n
UK (n P 30)
of life measure and the SF-36 and PGWB. In general, the expectations that the QOLMED would be more closely correlated to the PGWB than the SF-36 were not borne out in the United States sample.
Reproducibility The intraclass correlation coefficient (ICC) was 0.78 and 0.91 for the USA and UK male measures, respectively.
Internal
Consistency
Cronbach’s cxwas 0.94 and 0.96 for the USA and UK male instruments, respectively.
Discussion The QOLMED was originally developed for use in clinical trials. Respondents had no problems completing the instrument and thought that it was complete in coverage. While we believe it can be used in clinical trials, we also feel that this instrument could provide useful information in routine care. In general, the results are favourable, however the variability between the findings in the UK and USA
QOL fir male erectile dysfuncton Table
2. QOL-MED item reduction+ USA
Item list Celling effect 1 I feel frustrated because of my erection problem.
item-
r * 0.70
total
Ceiling effect
Item-
r > 0.70
total
6.7
0.64
6,W
26.7
0.79
7,16,17
0.63
92
30.0
0.76
12,13,15, 16,20
5.0
0.70
10
6.7
0.69
5,6
5.0
0.65
16.7
0.72
4,6,10
12.5
0.69
16.7
0.63
W5,
2.5
0.54
2 My erection problem makes me feel depressed.
12.5
0.66
3 I feel like less of a man because of my erection problem.
20.0
4 I have lost confidence in my sexual ability. 5 I worry that I won’t be able to get or keep an erection. 6 My erection problem is always on my mind.
UK
2
10,21
7 I feel that I have lost control over my erections.
5.0
0.44
10.0
0.55
6 I blame myself for my erection problem.
27.5
0.70
17,13
60.0
0.37
9 I feel angry because of my erection problem.
27.5
0.65
2,3,15
20.0
0.59
17
7.5
0.79
10.0
0.76
5,6,16
11 I have lost pleasure in sex because of my erection problem.
17.9
0.44
26.7
0.76
15,21
12 I am embarrassed
12.5
0.60
13,15
20.0
13 I worry about being humiliated because of my problem.
27.5
0.60
3,13
0.60
6,12
30.0
0.74
3,12
14 I try to avoid having sex.
25.0
0.53
15 I feel different from other men because of my erection problem.
27.5
0.60
16
40.0
0.56
36.7
0.61
3,16,16
16 I get less enjoyment out of life because of my erection problem.
17.5
0.60
23.3
0.65
2,3,10, 20-233
17 I feel guilty about my erection problem.
32.5
0.76
6,15
36.7
0.75
2,9
18 I am afraid to ‘make the first move’ towards sex.
22.5
0.59
14
36.7
0.66
15
19 I worry that my partner blames herself for my erection problem.
52.5
0.32
70.0
0.30
20 I worry about letting her down because of my erection problem.
10.0
0.59
21
10.3
0.60
3,16,21
21 I worry that I’m not satisfying her because of my erection problem.
5.0
0.56
20
10.3
0.61
6,16,20
22 I worry that we are growing apart because of my erection problem.
25.0
0.50
53.3
0.63
23 I worry that she is looking for someone else because of my problem.
65.0
0.56
75.9
0.36
24 I feel that she blames me for my erection problem.
40.0
0.62
69.0
0.42
25 I worry that she thinks I don’t want her because of my erection problem.
40.0
0.51
69.0
0.46
26 I have trouble talking to her about my problem.
35.9
0.53
55.2
0.39
27 My erection problem interferes with my daily activities.
66.7
0.59
56.6
0.64
10 I worry about the future of my sex life.
about my problem.
4
9,12,17
16
16
Items 8,13,19, 22-27 were removed + Missing data did not exceed 3.3% for any item.
Quality
of Life Research
Vol 5
1996
447
T. H. Wagner et al.
Table 3. Discriminant validity by self-perceived severity Measure Mild
Mean scores (SD) Moderate Severe
p value Total
US Version (n = 38)
50.9 (34.9)
40.7 (23.8)
36.9 (20.9)
40.0 (23.3)
0.56
UK Version (n = 28)
62.6 (33.8)
49.0 (21.9)
34.4 (21.7)
45.2 (25.5)
0.09
Table 4. Convergent validity of the QOL-MED measure Correlations Predicted*
l
USA male (n = 35)
UK male (n = 29)
PGWB Positive well-being Health worry and concern Depressed mood Behavioural and emotional control Energy Tension and anxiety Total PGWB
0.20-0.40 > 0.40 > 0.40 > 0.40 > 0.40 0.20-0.40 > 0.40 0.20-0.40
0.15 0.25 0.36 0.38 0.19 0.27 0.30+
0.37+ 0.39 0.49+ 0.31 -0.26 0.51+ 0.52
SF-36 Physical functioning Social functioning Role physical Role emotional Mental health Vitality and energy Bodily pain General health
c 0.20 0.20-0.40 c 0.20 0.20-0.40 0.20-0.40 0.20-0.40 < 0.20 0.20-0.40
0.13+ 0.36+ 0.22 0.12 0.29+ 0.14 0.35 0.17
NA NA NA NA NA NA NA NA
Predictions were made prior to analyzing the data to determine if the measure ‘behaves’ in the expected manner
+ Predictions matched observed values
was not expected. We believe many of the discrepancies can be traced to disparities between the USA and UK samples. The samples reflect convenience; nonetheless, it appears that the men recruited in the USA and UK were perhaps quite different. AII of the men in the UK had visited the doctor at least once in the last year for this problem. Approximately half of the American respondents had not seen a doctor in the past year; in fact, some men mentioned that we were thefirst people they had ever talked to about this problem. Although we have limited data from one urologist to whom we referred patients, we believe some of the men in the USA saw this as an opportunity to learn more about their problem and
448
Quality
of Life
Research Vol5
1996
seek medical advice. Thus, one of the assumptions underlying reproducibility, which is subject stability between measurements, may have been compromised. In other words, the changes between test and retest might be related to meaningful quality of life changes. A second concern has to do with the convergent validity results, which indicate how the new measure performs in relation to already validated measures with similar and dissimilar constructs. It was not expected that in the USA the QOL-MED would be moderately related to the bodily pain and social functioning domains of the SF-36. Similarly, it was not expected that in the UK the QOL-MED would be negatively correlated with energy and vitality.
QOLfir male erectile dysfincton Again, one possible explanation is the difference in samples. Men in the USA were recruited primarily through advertisements in the local newspapers; recmiting through medical clinics did not prove fruitful. But men in the UK were recruited through clinicians. Therefore, it is not surprising that men in the UK had a significantly greater number of medical appointments for treating erection difficulties. It is felt that these differences are probably connected to the variation in convergent validity results, yet this remains unanswered. We were also concerned by the low response rate in the UK. While this remains unexplained, men in the UK were less comfortable talking about sexual matters. Perhaps they felt uncomfortable answering such personally sensitive questions. The response rate was higher in Leeds than London. In the former case, a majority of the participants had the opportunity to meet the researcher (SPM). Another point worth elucidating is the process of eliminating items. Without a doubt, item reduction is a complicated process that directly affects validity, internal consistency and reproducibility. We were very cautious about the possibility of removing an item that was important to men with erection clifficulties. Consequently, as the measure is used and we learn more about it, it may be possible for additional items to be removed for easier use on a routine clinical basis. Generally the qualitative comments about the QOL-MED were very positive and the psychometric results for reproducibility and internal consistency were good in the USA and UK, as was discriminant validity in the UK. We will he conducting additional analyses on data currently being gathered in European validation studies. The QOL-MED is an important step towards gaining a better understanding of how erection difficulties affect quality of life.
Riley, Dr Kevan Wylie, Mr Ivan Rothman, Mr Paul Abel, Mr Marc Laniardo and Mr Terry Payton for their help recruiting participants and technical guidance. In addition, Robert Hoop and Jennie Best provided helpful comments on this manuscript. Please contact Jennie Best at Roche Bioscience, Palo Alto, CA for a copy of the QOL-MED.
References 1. !&gravesRT,SchoenbergHW, Ivanoff J.Serum testosterone and prolactin levels in erectiIe dysfunction. J Sex Marital
Ther 1983; 9(l): 19-26. 2. Mason DR. ErectiIe dysfunctions: assessment and care. Nurse Pratt 1989; 14(12): 2334. 3. Consensus development conference statement. National Institutes of Health. Impotence. December 7-9,1992. Int J Impt Res1993; S(4):181-284. 4. WHOQOL Group. Study protocol for the World Health Organization project to develop a quality of life instrument. Qua1 Life Res 1993; 2: 153-159. 5. Hunt S, McKenna SP.The QLDS: a scalefor the measurement of quality of life in depression. Health Policy 1992; 22: 307-319. 6. Guyatt GH, Feeny DH, Patrick DL. Measuring healthrelated quality of life. Annals of Internal Medicine 1993; 118: 622-629. 7. Dupuy HJ. The psychologicaI general well-being (PGWE) index. In: Wenger NR, Mattson ME, Furberg CD, Ebnson J, eck. Assessment of quality of life in clinical trials ofcardiovasculartherapies.New York Le Jacq, 1984: 170-183. 8. Ware JE, Sherboume CD. The MO!? 36-item short-form health survey (SF-36).I. Conceptual framework and item selection. Med Care 1992; 30(6): 473-483. 9. Deyo RA, Diehr P, Patrick DL. Reproducibility and responsivenessof health status measures. Statistics and strategiesfor evaluation. Contml Clin Trials1991; 12: 142s158s. 10. NnnnaEy JC Psychometrictheory,2nd ed. New York: Basic Books, 1978: 229-247.
Acknowledgments We would like to thank Dr Richard Berger, Dr Alan
&xeived 10 February 1996; accepted 16 March 1996)
Quality of L# Research Vol5. 1996
449