- Email: [email protected]
A pilot quality-of-life instrument for pruritus
A pilot quality-of-life instrument for pruritus Nisha S. Desai, MD,a Gabriele B. Poindexter, MD,b Yvette Miller Monthrope, MD,c Sandra E. Bendeck, MD,d Robert A. Swerlick, MD,e,f and Suephy C. Chen, MD, MSe,f,g Saint Louis, Missouri; Chapel Hill, North Carolina; Toronto, Ontario, Canada; Dallas, Texas; and Atlanta, Georgia Objective: We sought to develop a validated, reliable pruritus-specific quality-of-life (QOL) instrument, ItchyQoL. Methods: From 21 in-depth interviews with patients with pruritus, we developed 22 pruritus-specific items, and hypothesized 3 major constructs that explain the way pruritus affects patients’ QOL: symptoms, functional limitations, and emotions. We developed two versions of the pruritus QOL instrument, which assess for level of bother or frequency using items from the interviews and from generic skin QOL instruments, Skindex-16 (bother) and Skindex-29 (frequency). The instrument was tested for validity, reliability, and responsiveness. The frequency version was subsequently applied clinically to further evaluate its face validity. Results: A total of 89 patients with dermatologic conditions participated in the validation phase and 101 patients participated in the clinical application phase of the study. Construct validity was demonstrated by principal axes factor analyses and by demonstrating that differences in symptoms, functioning, and emotion differed among the varying levels of self-reported pruritus severity more than would be expected by chance (P \ .05 by analysis of variance). The instrument demonstrated reliability with internal consistency (Cronbach a: frequency 0.72-0.93 and bother 0.78-0.81) and reproducibility (intraclass correlation coefficient: frequency 0.91 and bother 0.84-0.87). The instrument suggested preliminary responsiveness for patients with improved disease for both frequency and bother items with both overall scores and the majority of the subscales scored demonstrating significant changes. Discriminant validity was shown by comparing differences in and the number of insensitive items between the pruritus-specific QOL instrument and the generic Skindex instruments. Limitations: Lack of generalizability and potential selection bias are limitations. Conclusions: This study represents, to our knowledge, the first attempt at a pruritus-specific QOL instrument that is reliable, valid, and responsive. ( J Am Acad Dermatol 2008;59:234-44.)
ruritus is the chief symptom in numerous dermatologic conditions,1-4 and a common symptom in systemic disorders,5-7 infectious diseases,8,9 and neoplastic diseases.9,10 The prevalence of pruritus in the general population is not well
P
documented but has been reported to be around 10%.11,12 It is the most common dermatologic problem in nursing homes along with xerosis.13 Not only is pruritus a widespread phenomenon, it also can have a significant quality-of-life (QOL) impact in
From Washington University, Saint Louisa; University of North Carolinab; University of Torontoc; University of Texas Southwesternd; Department of Dermatology, Emory University School of Medicine, Atlantae; and Division of Dermatologyf and Department of Health Services Research and Development,g Atlanta Veterans Administration Medical Center. Supported in part from an American Skin Association (ASA) Health Services Research Grant, the ASA David Martin Carter Research Scholar Award, and an Emory Skin Disease Research Center Pilot and Feasibility grant (No. P30AR42687) from the National Institute on Arthritis and Musculoskeletal and Skin Disease (NIAMS), National Institutes of Health (NIH). Dr Chen was supported in part by a Mentored Patient Oriented Career Development Award (No. K23AR02185-01A1) from NIAMS, NIH.
Conflicts of interest: None declared. Presented as abstracts at the 66th Annual Meeting of the Society for Investigative Dermatology, Miami, Florida, May 2005, and the 63rd Annual Meeting of the American Academy of Dermatology Annual Meeting, New Orleans, Louisiana, February 2005. Accepted for publication April 4, 2008. Reprint requests: Suephy C. Chen, MD, MS, Department of Dermatology, Emory University School of Medicine, 101 Woodruff Circle; Atlanta, GA 30322. E-mail: [email protected]. Published online June 12, 2008. 0190-9622/$34.00 ª 2008 by the American Academy of Dermatology, Inc. doi:10.1016/j.jaad.2008.04.006
234
J AM ACAD DERMATOL
Desai et al 235
VOLUME 59, NUMBER 2
Abbreviations used: ANOVA: analysis of variance ICC: intraclass correlation coefficient QOL: quality of life
patients’ lives. In a recent French study, more than 40% of patients with chronic pruritus described their disease as burdensome.12 Current treatments for pruritus are sparse, mostly consisting of antihistamines, antileukotrienes, and immunosuppressives, which are of varying efficacy. Researchers are challenged to develop better therapies, but pruritus, like pain, is a stimulus that cannot be measured directly. Rather, pruritus is measured either by observing the amount of scratching by the patient or by asking the patient to rate pruritus on some sort of severity scale, such as the visual analog scale or a numeric scale. However, neither the amount of scratching nor the visual analog/numeric scale measures the impact of pruritus on patients’ QOL. The extent to which a disease impacts patients’ QOL can be quantified by health status instruments. Health status instruments can either be generic or condition specific. Generic instruments capture broad-based QOL issues and, therefore, may be used to directly compare QOL across disparate disease states. However, generic instruments generally exclude QOL aspects that may be significant for a specific condition. Condition-specific instruments, by contrast, focus on QOL aspects pertinent to a particular condition. As a result, they have the advantage of being shorter, more appropriate, and more sensitive to subtle, specific QOL issues related to the disease in question. Currently there are no validated pruritus-specific QOL instruments. Although generic and skin-specific QOL instruments are available, they do not directly assess the issues specific to pruritus and, thus, will not adequately capture itch-specific QOL impact. In this study, we developed a pruritus-specific QOL instrument, called ItchyQoL, based on concerns and issues pertinent to patients with pruritus. We tested the instrument for validity, reliability, and responsiveness and we report preliminary results. In addition, we tested the instrument in a second sample of patients with pruritus to determine its measurement properties in the clinical setting.
METHODS All patients were consented and recruited from Emory University Dermatology Clinics, Atlanta, Ga; Grady Memorial Hospital, Atlanta, Ga; or the Veterans
Affairs Medical Center of Atlanta, Ga. We obtained approval for the protocol and consent process from the respective institution’s research committee and by the university institutional review board. Item development We conducted in-depth interviews with adult patients with pruritus of any cause to elicit any and all ways that pruritus affected their lives. Interviews were conducted until no new items were obtained. Based on all patient responses and their frequency, we composed pruritus-specific items. Items were developed into questions that addressed frequency, level of bother, or both. We conceptualized 3 major constructs from these items that explain the way pruritus affects patients’ QOL: symptoms, functioning limitations, and emotions. We administered the pruritus-specific items in addition to the 29-item version of Skindex14 and Skindex-1615 to comprehensively assess the constructs patients found relevant. Skindex-29 addresses QOL from a frequency perspective whereas Skindex-16 addresses level of bother. For all items, we used the same format as for the Skindex instruments, but changing the wording ‘‘skin condition’’ to ‘‘itchy skin condition.’’ All items inquired about the last 7 days. The responses for frequency type questions were ranked on a 5-point scale with designations of: never, rarely, sometimes, often, and all the time. A similar scale was developed for bother type questions: not bothered, little bothered, somewhat bothered, very bothered, and severely bothered. The final survey also contained demographic, pruritus severity, cause, and medication use questions. After the development of the initial instrument, it was administered to 5 adults with no medical training recruited from the general population for pilot testing, to test for clarity and comprehension. Based on the responses, we made the appropriate modifications to the instrument. Sample population, measures, and data collection All adult patients with active pruritus in any area of the body within the past 7 days were eligible to participate. Patients were recruited from our university dermatology clinic by referral from their dermatologist, or by telephone by contacting patients in our pruritus database. The database consists of patients seen at the university dermatology clinic identified by International Classification of Diseases, Ninth Revision code for pruritus. Interviewers administered either the bother or frequency version of the instrument to patients in a pseudorandom fashion. For each interviewer, the first patient was
236 Desai et al
J AM ACAD DERMATOL AUGUST 2008
randomly given one of the two versions and then the interviewer alternated administering the two versions of the survey. Patients answered all questions at baseline, at 72 hours (allowable range: 3-7 days), and at 2 months (allowable range: 2-3 months). The demographic questions were omitted in the latter two time points. Because the 72-hour data are used to test for reliability and because pruritus can change within a 72-hour time frame, patients were asked to remember their itch from their baseline time point. Patients were asked to self-rate their pruritus at the 2to 3-month time point as improved, worsened, or not changed based on the question, ‘‘How has your itchy skin condition changed since the last time you filled out this survey?’’ To further determine face validity, the frequency version of the final instrument was administered in the clinical setting. Adult patients from all study centers who had been pruritic in the past 7 days were asked to fill out the pruritus-specific QOL instrument, at the baseline time point only. Analyses All statistical analyses were performed with software (SPSS 12.0 for Windows, 12.0 edition, SPSS Inc, Chicago, Ill). Statistical significance was defined as P less than or equal to .05. Scoring Results for the instrument consisted of 3 subscale (symptom, function, and emotion) scores, based on the hypothesized constructs and an overall score. The subscale scores consist of the average of the responses to the items in a given subscale. For example, the subscale score for symptom was the average of all responses to the items pertinent to symptoms. The overall score consists of the average of the responses to all the items. The responses to the frequency items were scored on a 1 (never) to 5 (all the time) scale. Similarly, the responses to bother items ranged from 1 (not bothered) to 5 (severely bothered). Item reduction To create the most parsimonious evaluative instruments, we eliminated irrelevant and insensitive questions. We defined insensitive items as those where greater than 50% of patients responded with an extreme value (never/not bothered or all the time/severely bothered), an accepted cut-off threshold used in the literature.16,17 Psychometric evaluation The instrument was tested for reliability, responsiveness, and validity. We evaluated reliability with
internal consistency by Cronbach a coefficient and with reproducibility (test-retest reliability) by intraclass correlation coefficient (ICC). We tested for responsiveness by applying the paired t test to the baseline and 2-month answers for 3 groups: those who reported improvement, no change, or worsening of their pruritic condition. Validity was confirmed by examining face, content, construct, and discriminant validity. We ensured face and content validity of the instrument by conducting in-depth interviews with patients with pruritus. We also tested face validity in the clinical setting by administering the frequency version of the final instrument to a second sample of patients with pruritus. We compared subscale scores across demographic information, itch severity and frequency, diagnosis, and disease duration. One-way analysis of variance (ANOVA) was used for these comparisons. Construct validity was tested in two ways. We hypothesized that the items for bother and frequency would cluster into 3 factors that could be labeled as: symptoms, functioning, and emotions. This hypothesis was tested by using principle axis factor analysis followed by promax rotation for bother items, and varimax rotation for frequency items. We retained only those factors with eigenvalues greater than 2 and by application of the scree test.18 We identified the factor onto which items loaded by selecting the largest coefficient of that item among all the retained factors. Each factor was labeled by the predominant trait of the heavily loaded items. The regression factor scores were compared with the unweighted hypothesized subscale scores by Pearson correlation coefficients. The second way we tested construct validity was by comparing the subscale scores with the self-reported pruritus severity and pruritus frequency using one-way ANOVA. We hypothesized that the subscale scores would correlate with pruritus severity and pruritus frequency. Discriminant validity is the extent to which one instrument measures a certain health characteristic better than another instrument. A questionnaire that asks questions more pertinent to the patient’s actual condition would eliminate the floor effect (when data cannot take on a value lower than some particular number set by the instrument19), spare irrelevant questions, and improve the sensitivity of the questionnaire itself. Because Skindex-29 and Skindex-16 are QOL instruments specific to skin rather than pruritus, we hypothesized that they are not as sensitive as the test instrument in measuring QOL issues specific to pruritus. Discriminant validity was tested by comparing the floor effect of the final pruritusspecific QOL instrument with Skindex 29 and Skindex 16. We evaluated floor effect by quantifying
Desai et al 237
J AM ACAD DERMATOL VOLUME 59, NUMBER 2
the number of insensitive items (ie, those items where [50% of participants chose ‘‘never’’ as their response).
Table I. Demographics and pruritus characteristics
RESULTS
Age, y, mean (SD) Sex Male Race/ethnicity White/Caucasian Black/African American Other Education level # Grade 8 Any high school Any college Any postgraduate Total yearly household income, $ \10,000 10,000-25,000 25,000-50,000 50,000-75,000 75,000-100,000 [100,000 Itch frequency All of the time Most of the time Some of the time Once in awhile Never Itch severity Worst ever Very severe Average Itch occasionally Itch duration [3 mo 3-6 mo 6 mo-1 y 1-5 y 5-10 y [10 y When is itching the worst? All day Morning Afternoon Evening At night [1 of the above Diagnosis Urticaria/hives Dermatitis/eczema Other Idiopathic itching
Demographics We invited 21 patients to participate in face-toface in-depth interviews. After designing the initial pruritus-specific QOL instrument from these interviews, we contacted 187 patients for the validation phase of the study. In all, 61 patients were approached in clinic and 147 patients were contacted by telephone. Nearly all of the patients recruited from clinic (96%, N = 59), and approximately 16% (N = 30) of the invited patients by telephone agreed to participate. There was no significant difference in the socioeconomic profile (age, sex, race, education, or income levels) between patients recruited in clinic or by telephone. The majority of the patients contacted by telephone declined because they were no longer pruritic (22%, N = 32) or were not interested in the study (20%, N = 30); we were unable to contact 37% of patients (N = 55). In our cohort of 89 patients, 50 patients were administered the frequency version of the instrument and 39 patients were administered the bother version. Sixteen patients only completed the baseline survey and could not be contacted to complete the 72-hour or 2-month time points. A total of 23 patients completed the baseline and 72-hour time points only. Three patients did not complete the 72-hour time point, but were able to complete the 2-month time point. The overall mean age was 58.1 years (SD: 16.6), 39% were male, 74% were Caucasian, 21% were African American, 5% were of another race, and 75% had attended college or beyond (Table I). For the clinical application phase of the study, we visited 29 general dermatology clinics at a universitybased center, Department of Veterans Affairs hospital, or local inner-city hospital. Of 663 patients scheduled to arrive, 508 patients kept their appointment. Because of limited manpower and the amount of time required to administer the survey, we were able to approach 320 patients, and of those 115 patients (36%) were pruritic. In all, 104 of the patients with pruritus completed the frequency version of the instrument. This population had a mean age of 55.9 years; 46% were male, 69% were Caucasian, 23% were African American, and 5% were of another race. The majority (76%) of the patients had attended college or beyond (Table I). Disease characteristics Of the 89 patients who completed the instrument development and validation portion of the study, the
Instrument validation (N = 89)
Instrument application (N = 104)
58.1 (16.7)
55.9 (16.4)
39.3%
45.6%
74.2% 21.3% 4.5%
68.9% 22.6% 4.7%
3.4% 21.3% 51.7% 23.6%
2.9% 21.2% 51.9% 24.0%
7.2% 2.6% 32.5% 21.7% 12.0% 22.9%
10.9% 17.4% 23.9% 27.2% 9.8% 10.9%
25.8% 37.1% 30.3% 6.7% 0%
16.2% 27.6% 41.0% 15.2% 0%
10.1% 24.7% 47.2% 28.0%
4.8% 26.0% 39.4% 29.8%
4.5% 3.4% 16.9% 40.4% 15.7% 19.1%
18.4% 12.6% 15.5% 18.4% 7.8% 27.2%
32.6% 3.4% 3.4% 10.1% 22.5% 28.1%
32.7% 5.1% 6.1% 25.5% 27.6% 3.1%
10.1% 32.6% 25.8% 31.5%
6.9% 19.8% 14.9% 56.4%
majority of patients were able to attribute their pruritus to a specific cause (Table I) but 32% had idiopathic itching. The majority of patients rated the
238 Desai et al
J AM ACAD DERMATOL AUGUST 2008
severity of their pruritus as more than just occasionally with 10% rating as worst ever. Most patients had been itching longer than 1 year (75%). Of the 104 patients who completed the clinical application portion of the study, the majority (56%) had idiopathic itching. Like the previous cohort, most patients rated the severity as more than just occasionally and the majority (53%) had been itching for more than 1 year (Table I). Item development, analysis, and reduction We interviewed a total of 21 patients, and no new responses were obtained after the 15th patient. We developed 27 pruritus-specific items from the indepth patient interview session, grouped into the following constructs: 6 symptom, 10 functional limitation, and 11 emotion (Table II). Eighteen of these items are measured with both frequency and bother. The remaining 9 items pertained to emotion and were measured by frequency only. Twelve of the items developed in the interviews were already present in Skindex-29 of which 8 were also analogous to Skindex-16 items. Therefore, 15 new items were created and are unique to the instrument (Table II). Both frequency and bother versions were formulated to query about the preceding 7 days. Of the 27 items, the patients demonstrated that their pruritus-related QOL was most affected by ‘‘needing to scratch’’ with a frequency score (SD) of 4.08 (0.80), and a bother score of 3.81 (1.03). QOL was least affected by ‘‘limiting the types of food you can eat’’ with a frequency score of 1.58 (1.06) and a bother score of 1.50 (0.99). Mean scores for all items are listed in Table II. Several items proved to be relatively insensitive, in which more than 50% of the patients answered ‘‘never or not bothered’’ to these items, and were subsequently eliminated. A total of 5 frequency items and 3 bother items were found insensitive. Of these, 3 of the items were frequency and bother versions of the items: ‘‘limiting the types of food you can eat,’’ ‘‘worry about having a contagious disease,’’ and ‘‘self-esteem changing.’’ The rest of the eliminated items were the frequency version of pruritus-specific items. The final version of the pruritus-specific QOL instrument consists of the 22 sensitive items. The frequency portion consists of 22 items, and the bother portion consists of 15 items. Feasibility Note that although several of the Skindex items were not included in the pruritus-specific QOL instrument, they were administered in the study for us to compare the test instrument with Skindex. The average duration to answer all 44 questions of the frequency instrument (test instrument 1 additional
Skindex-29 items) was 10.0 minutes (SD: 4.36 minutes). The average duration to all 36 questions of the bother instrument was 9.36 minutes (SD: 4.00 minutes). Reliability The overall pruritus-specific QOL instrument score demonstrated internal consistency reliability with Cronbach a ranging from 0.89 for bother and 0.92 for frequency (Table III). Systematically deleting one item from the analysis did not significantly raise the coefficients. The overall score also demonstrated reproducibility with ICC ranging from 0.87 for bother and 0.92 for frequency. Similar findings were observed for each of the 3 subscales (Table III). Internal consistency reliability was demonstrated with Cronbach a coefficient ranging from 0.76 to 0.81 for bother items, and from 0.72 to 0.93 for frequency items. Systematically deleting one item from the analysis did not significantly raise the coefficients. Reproducibility was demonstrated with ICC ranging from 0.84 to 0.87 for bother items, and 0.91 for frequency items. Responsiveness result Patients were divided into 3 categories (better, worse, and no change) based on their responses to the question, ‘‘How has your itchy skin condition changed since the first interview?’’ For the group that answered the bother-type questions, 13 patients reported improvement, 4 reported worsening, and 6 reported no change. For the group that answered the frequency-type questions, 16 patients reported improvement, 8 reported worsening, and 8 reported no change. Using the pruritus-specific overall and subscale scores (Table IV), we found improvement in the QOL scores of those patients who reported improvement in their pruritus; the difference was statistically significant (P # .01, using paired t test) for the overall score and all subscale scores in the bother and frequency versions except for bother symptom and function where there was a trend toward statistical significance (P = .06-.07). Future work with larger numbers of patients will elucidate this trend. We found no significant differences in all subscale scores of those who reported no change. We also did not find significant difference in all subscale scores of those who reported worsening of their pruritus; however, our number of patients was very small in these groups. Construct validity Principle axis factor analysis with varimax rotation was performed on the frequency items, and 3 factors were retained according to the criteria outlined in the
Desai et al 239
J AM ACAD DERMATOL VOLUME 59, NUMBER 2
Table II. Instrument items with hypothesized constructs and mean scores Item
Hypothesized construct
1. 2. 3. 4. 5. 6.
Sy Sz Sz S S S
2.35 2.82 2.94 2.66 4.08 2.98
(1.17) (1.24) (1.38) (1.48) (0.80) (1.46)
2.28 2.72 2.67 2.40 3.81 2.78
(1.37) (1.38) (1.39) (1.58) (1.03) (1.47)
F Fz Fz Fz Fy Fy F F F F
3.48 2.92 2.73 1.80 3.28 1.97 2.71 1.58 2.74 3.71
(1.24) (1.35) (1.40) (1.19) (1.22) (1.33) (1.17) (1.06) (1.53) (1.49)
2.98 2.81 2.75 1.75 3.18 1.93 2.60 1.50 2.51 2.55
(1.46) (1.36) (1.45) (1.17) (1.41) (1.40) (1.28) (0.99) (1.48) (1.53)
Ez Ez E Ey Ez E E
4.04 2.89 3.35 2.88 2.57 1.89 2.43
(1.11) (1.53) (1.26) (1.28) (1.36) (1.37) (1.36)
3.82 (1.27) 2.80 (1.58) N/A N/A N/A N/A N/A
E E E E
3.74 2.80 2.13 2.17
(1.21) (1.43) (1.28) (1.33)
N/A N/A N/A N/A
My itchy skin condition bleeds My skin hurts because of my itchy skin condition My itchy skin condition burns or stings I get scars from my itchy skin condition I need to scratch my itchy skin condition Temperature/seasonal changes aggravate my itchy skin condition 7. I spend a lot of money treating my itchy skin condition 8. My itchy skin condition makes it hard to work or do what I enjoy 9. My itchy skin condition affects my interaction with others 10. I tend to stay home because I don’t want to scratch in public§ 11. My itchy skin condition affects how well I sleep 12. My itchy skin condition interferes with my sex life§ 13. My itchy skin often makes it difficult to concentrate 14. My itchy skin condition limits the types of foods I can eat§ 15. My itchy skin condition limits the types of clothes I can wear 16. My itchy skin condition forces me to buy special soaps, detergents, and lotions 17. I am frustrated by my itchy skin condition 18. I am embarrassed by my itchy skin condition 19. My itchy skin condition drives me crazy/nuts 20. My itchy skin condition makes me feel angry or irritable 21. My itchy skin condition makes me feel depressed or sad 22. I worry I have a contagious disease§ 23. I worry about what other people think about me because of my itchy skin condition 24. I worry that the itching will last forever 25. I feel self-conscious because of my itchy skin condition 26. My personality has changed because of my itchy skin condition 27. My self-esteem has changed because of my itchy skin condition§
Frequency, mean (SD) score*
Bother, mean (SD) score*
Subscale scores of S, F, and E are calculated by average of item scores that pertain to that particular subscale. Frequency items are scored on scale of 1-5 (1 = never, 2 = rarely, 3 = sometimes, 4 = often, 5 = all the time). Bother items are scored on scale of 1-5 (1 = not bothered, 2 = little bothered, 3 = somewhat bothered, 4 = very bothered, 5 = severely bothered). E, Emotions; F, functioning; N/A, not applicable; S, symptoms. *Mean scores are derived from 89-person instrument validation population. y Also in Skindex-29. z Also in Skindex-29 and Skindex-16. § Insensitive items removed from final version of instrument.
‘‘Methods’’ section. From the items that loaded most heavily on factor 1, we could see that the predominant trait was emotions; factor 2 was clearly functioning, and factor 3 was symptoms. The Pearson correlation coefficient comparing the regression factor scores and the unweighted hypothesized subscale scores were significant (P\.01) and ranged from 0.52 to 0.77 (Table III). For the bother items, 3 factors were also retained after principle axis factor analysis with promax rotation. The predominant trait or construct loaded most heavily on factor 1 was functioning, factor 2 was emotions, and factor 3 was symptoms. The Pearson correlation coefficient for the bother items ranged from 0.86 to 0.93 (P \.01) (Table III).
In the clinical application cohort (n = 104) construct validity was confirmed by comparing the subscale scores with self-reported pruritus severity and frequency at the baseline time point. The severity levels were answered on a 5-point scale: worst ever, very severe, average, itch occasionally, and no itching at all. Pruritus frequency levels were measured on a similar scale (1-5): all of the time, most of the time, some of the time, once in awhile, and never. We found greater differences in symptom and emotion scores among the different levels of pruritus severity than would be expected by chance in both frequency and bother items (P \ .05, by ANOVA) (Table V). We also found significant differences in
240 Desai et al
J AM ACAD DERMATOL AUGUST 2008
Table III. Psychometric test results Psychometric test
Overall
Cronbach a coefficient
F: B: F: B:
Intraclass correlation coefficient Pearson correlation coefficient comparing RF scores and unweighted hypothesized subscale scores
0.92 0.89 0.92 0.87
Symptoms
Functioning
Emotions
F: 0.72 B: 0.78 F: 0.90 B: 0.84 FRF1: 0.20 FRF2: 0.37* FRF3: 0.52* BRF1: 0.75* BRF2: 0.88* BRF3: 0.86*
F: 0.82 B: 0.86 F: 0.92 B: 0.89 FRF1: 0.50* FRF2: 0.64* FRF3: 0.16 BRF1: 0.93* BRF2: 0.81* BRF3: 0.61*
F: 0.91 B: 0.76 F: 0.93 B: 0.84 FRF1: 0.77* FRF2: 0.37* FRF3: 0.32 BRF1: 0.88* BRF2: 0.91* BRF3: 0.70*
B, Bother item; F, frequency item; RF, regression factor. *P \ .01.
Table IV. Responsiveness scores Pruritus condition
ItchyQoL Worse (F: n = 8, B: n = 4)
Subscale
Symptoms Functioning Emotions Overall
ItchyQoL Same (F: n = 8, B: n = 6)
Symptoms Functioning Emotions Overall
ItchyQoL Better (F: n = 16, B: n = 13)
Symptoms Functioning Emotions Overall
Baseline, mean score 6 SD
2 mo, Mean score 6 SD
P value
F: 3.17 6 0.75 B: 3.23 6 0.55 F: 2.68 6 0.61 B: 2.48 6 0.78 F: 2.57 6 0.88 B: 2.75 6 1.41 F: 2.85 6 .56 B: 2.82 6 .68 F: 2.87 6 0.88 B: 2.54 6 0.97 F: 2.71 6 0.71 B: 2.35 6 0.99 F: 2.70 6 1.15 B: 2.78 6 1.30 F: 2.75 6 .89 B: 2.48 6 .97 F: 2.95 6 0.75 B: 2.65 6 0.88 F: 3.29 6 0.78 B: 2.82 6 0.84 F: 2.88 6 1.04 B: 3.12 6 1.14 F: 3.03 6 .78 B: 2.77 6 .79
F: 3.40 6 0.55 B: 3.17 6 0.71 F: 3.13 6 0.81 B: 2.86 6 1.11 F: 3.17 6 1.23 B: 3.88 6 1.31 F: 3.22 6 .73 B: 3.12 6 .83 F: 3.30 6 1.39 B: 2.71 6 0.97 F: 2.29 6 0.49 B: 2.07 6 0.48 F: 2.38 6 0.96 B: 2.93 6 1.59 F: 2.55 6 .57 B: 2.32 6 .75 F: 2.25 6 0.97 B: 2.35 6 1.08 F: 2.63 6 1.19 B: 2.33 6 1.21 F: 2.27 6 1.07 B: 2.39 6 1.04 F: 2.38 6 1.00 B: 2.34 6 1.10
F: .38 B: .63 F: .28 B: .39 F: .11 B: .19 F: .280 B: .377 F: .50 B: .22 F: .15 B: .39 F: .06 B: 1.00 F: .844 B: .199 F: \.01 B: .06 F: \.01 B: .07 F: \.01 B: \.01 F: \.01 B: .016
B, Bother item; F, frequency item.
symptom, functioning, and emotion scores among different levels of pruritus frequency in the frequency items, and between symptom and functioning scores in the bother items (P \ .05, by ANOVA) (Table V). Discriminant validity We demonstrated the discriminant validity of the pruritus-specific QOL instrument as compared with
Skindex-29 and Skindex-16 in patients by ascertaining that the test instrument had fewer insensitive items. We used the responses from the patients in both the instrument validation and application phases of the study. For frequency items, 1 of 22 of the test instrument items (4.5%) had at least 50% of the patients answering ‘‘never,’’ as compared with 9 of 29 (31.0%) Skindex-29 items. For bother items, none of the 15 test instrument items, and 2 of 16
Desai et al 241
J AM ACAD DERMATOL VOLUME 59, NUMBER 2
Table V. Face and construct validity Instrument validation Frequency, mean scores (N = 89)
Itch severity Worst ever Very severe Average Itch occasionally P value Itch frequency All of the time Most of the time Some of the time Once in awhile P value
S 3.68 3.26 2.84 2.51 .001 S 3.34 3.05 2.74 2.14 .003
F 3.27 3.32 3.00 2.84 .291 F 3.48 3.08 2.90 2.33 .011
Instrument application
Bother, mean scores (N = 89)
E 3.83 3.21 2.70 2.92 .015 E 3.45 2.99 2.66 2.55 .035
S 3.77 3.01 2.67 2.19 \.001 S 3.13 2.92 2.52 1.83 .006
F 3.16 3.01 2.66 2.48 .200 F 3.31 2.77 2.47 1.98 .003
Frequency, mean scores (N = 104)
E 4.22 3.59 3.02 3.16 .045 E 3.74 3.30 2.96 3.25 .207
S 3.38 3.24 2.78 2.25 \.001 S 3.57 3.06 2.48 2.19 \.001
F 3.00 2.96 2.13 1.62 \.001 F 3.29 2.53 1.94 1.44 \.001
E 3.60 2.93 2.20 1.86 \.001 E 3.03 2.76 2.02 1.78 \.001
E, Emotions; F, functioning; S, symptoms.
Table VI. Clinical application of ItchyQoL results Average subscale scores (SD)
Sex Male (n = 47) Female (n = 57) P value (ANOVA) Diagnoses Urticaria (n = 8) Dermatitis (n = 20) Other (n = 16) Idiopathic pruritus (n = 57) P value (ANOVA)
Symptoms
Functioning
Emotions
2.51 (0.73) 2.99 (0.81) .006
1.89 (0.79) 2.55 (1.21) .003
1.98 (0.85) 2.65 (1.23) .003
3.05 3.05 2.86 2.61 .149
2.89 2.46 2.42 1.99 .040
2.56 2.55 2.47 2.15 .361
(0.72) (0.79) (0.83) (0.81)
(1.22) (1.26) (1.01) (0.94)
(1.17) (1.18) (1.27) (0.96)
ANOVA, Analysis of variance.
(12.5%) Skindex-16 items, had at least 50% of the patients answering ‘‘not bothered.’’ Application of the pruritus-specific QoL instrument in the clinical setting When we applied the test instrument in the clinical application cohort of patients (n = 104), we observed several interesting findings (Table VI). First, although we found no differences in symptom, functioning, or emotion subscale scores among age, race, education, income, or disease duration, we did find that women had higher (P \ .01) average subscale scores than men despite no difference in itch severity or itch frequency between the two groups. Second, we detected a difference in functional QOL impact between the different diagnoses, with urticaria having the highest overall QOL impact (P = .04). On post hoc analysis, we found that those given a diagnosis of dermatitis had a higher symptomatic impact than those with idiopathic itching (P = .04).
To explore the sex QOL differences further, we compared sex distribution of the specific causes listed in the survey. We found that women reported higher percentages of all the diagnoses as compared with men (urticaria: 3.5% vs 1.4%, eczema: 7.6% vs 4.8%, other: 5.3% vs 4.1%). On the other hand, men were more apt to report that their origin has yet to be elucidated (20.4% vs 15.2%).
DISCUSSION This study represents a pilot QOL instrument specifically for patients with pruritus that has demonstrated preliminary reliability, responsiveness, and validity. We demonstrated reliability of the pruritusspecific QOL instrument with high internal consistency reliability and reproducibility. This ensures that the instrument will yield consistent results for a given patient in similar conditions and is relatively free from random error. We verified preliminary data for responsiveness by demonstrating that individual
242 Desai et al
J AM ACAD DERMATOL AUGUST 2008
construct and total scores improved in those patients who reported improvement in their pruritus. Scores did not significantly change in patients who reported that their pruritus worsened. However, subscale scores did increase, hinting at potentially greater QOL impact with worsening pruritus in the frequency and bother items in all constructs except for symptom impact for bother items (Table IV). We believe that with a larger sample size, this trend of increasing patient scores with worsening pruritus will become significant. An instrument may be reliable and responsive, but not valid. We ensured face and content validity by deriving the items from in-depth interviews with patients with pruritus. We confirmed construct validity by finding that the subscale scores correlated with self-reported severity and frequency in the expected direction and by demonstrating that our 3 hypothesized constructs (symptoms, emotions, and functioning) correlated with the factor analysis of the data. We also demonstrated that the pruritus-specific QOL instrument was more sensitive to QOL issues as compared with Skindex-29 and Skindex-16. Lastly, the test instrument is a feasible measure of pruritus to use in many settings. The combined frequency or bother instrument (test instrument 1 Skindex) took, on average (SD), 9.37 to 10.0 (4.004.36) minutes to complete; the 27-item pruritus instrument only took 6.71 (2.90) minutes to complete. In clinical application we found that, after item reduction, the 22-item frequency pruritus instrument took, on average, 3.11 minutes to complete. We believe that a 5- to 10-minute instrument is manageable for both patients and clinicians. The clinical application data that we obtained revealed many interesting findings about how pruritus affects people’s lives. The fact that urticaria has the highest overall QOL impact speaks to the need to develop therapies for chronic urticaria. The differences found in sex (women with more QOL impact than men) are also thought provoking. Looking at the data carefully, we see that patients with idiopathic pruritus reported less of a functional impact than the other causes (Table VI); because men reported more idiopathic cause than women, this may be the reason behind the lower functional impact in men compared with women. Similarly, we found that dermatitis has more symptomatic impact than idiopathic cause; because women reported dermatitis more commonly than men, this may explain why women reported more symptoms than men. Although the emotional impact was not significantly different comparing idiopathic versus known cause, it may be that men were on average less emotionally impacted by their pruritus because they
still had hope that an origin would be discovered and, thus, their pruritus would be treatable. Thus, the sex differences that we observed in QOL impact may be a result of the differences in causes behind the pruritus. This interesting finding begs to be elucidated with an epidemiologic study. Limitations The results of this study are limited by several factors. First, the responsiveness of the pruritusspecific QOL instrument has only been validated in patients whose disease either improved or did not change from baseline. This may limit the use of the instrument in investigational studies until further data are published regarding the instrument’s responsiveness in patients with worsening disease. Second, there is potential for selection bias in the instrument development phase of the study, as the majority of the patients recruited were from an academic institution. These patients may have had pruritus refractory to standard therapies, and a higher educational background than patients from the inner city or rural clinics. Thirdly, we administered the test instrument by telephone in a subset of patients. Although we do not anticipate that face-toface or self-administration would make a significant difference, future work is necessary to compare the administration by telephone versus in person. Fourth, a small sample of patients for this study also serves as a limitation. A larger study population is needed for confidence in the ability to generalize our results, and for confidence in the analyses such as the confirmatory factor analyses and responsiveness analyses. Fifth, we tested the reliability of the test instrument by comparing baseline and 72-hour answers, where the 72-hour time point asked the patient to recall their baseline pruritus. To test reliability, the second time point is chosen long enough from baseline so that the patient cannot recall the answers for their baseline survey, but not so long such that their clinical severity would change. We thought that 72 hours was long enough for patients to forget their exact answers to the test instrument, and that for the majority of patients, their pruritus would not change significantly. However, given that a small number of patients may have changed the severity of their pruritus, we asked the patients to recall their baseline pruritus. By doing so, we may have introduced an element of testing recall. We do not think that recall is an issue for a significant proportion of our population, but acknowledge this potential limitation. Lastly, we assumed that each item carries equal weight when attributing components to each axis (symptom, function, and emotion). Currently this is
J AM ACAD DERMATOL
Desai et al 243
VOLUME 59, NUMBER 2
the standard for developing this type of QOL instrument; however, the next step is to allow the patients to weigh the importance of the items in the survey. Additional studies are warranted for the pruritusspecific QOL instrument to be used with confidence in a general population. Validation needs to take place with a larger subset of less-educated patients from nonacademic institutions. Responsiveness needs to be verified in patients with worsening disease in a larger population of patients. Further work needs to be performed in validating other means of administrating the test instrument (face-toface and self-administration) and to apply other methodologies to incorporate patient weighing of importance of items. Potential uses of the pruritus-specific QOL instrument The current form of the pruritus-specific QOL instrument represents a promising patient-reported outcome measure of disease burden as a result of pruritus and might be practical for use in a clinical or research setting. The instrument may be used as a tool for clinicians who treat patients who have conditions in which pruritus is the major symptom. The clinician can use either the frequency or bother version of the instrument to help their patients with pruritus in many ways. First, the profile of the instrument can be analyzed to determine which aspect of the disease most impacts the patient in relation to symptoms, functional impairment, and emotions. This can differ greatly among individuals and, thus, the instrument can help to personalize the treatment and advice offered. Next, changes in QOL can be a useful guide in therapeutic intervention. Administering the survey at different time points during the course of therapy could be useful in mapping a patient’s response to that therapy. Having an instrument that demonstrates an improvement, no change, or worsening in QOL can greatly affect both the patient-doctor relationship and patient compliance with treatment. Lastly, clinicians can use the impact on QOL data to petition managed care and insurance companies when coverage is denied for therapies to treat pruritus. In the research setting, the pruritus-specific QOL instrument may be useful as an end point for epidemiologic and burden-of-disease studies. It may also be useful as a secondary or coprimary end point for clinical trials testing new therapies. The Food and Drug Administration has recently provided guidance regarding patient-reported outcomes,20 thus emphasizing the importance of well-developed QOL instruments. The development strategy of the test instrument has complied with these guidelines and
bridges the gap in the research arena for a pruritusspecific QOL patient-reported outcome. Although Skindex-29 and -16 are both well-validated and reliable QOL instruments (ICC: 0.88-0.92 and 0.88-0.90, respectively, and Cronbach a: 0.870.96 and 0.86-0.93, respectively), they are skin-specific instruments and were not developed to address the issues specific to pruritus. In the final version of the instrument, the pruritus-specific QOL instrument contains 15/22 unique items. Only 7 of the items were common to Skindex-16 or Skindex-29. Thus the Skindex instruments have a different role from that of the pruritus-specific QOL instrument. The Skindex instruments should be used with patients with pruritus for those researchers with an invested interest in comparing pruritic versus nonpruritic conditions. The pruritus-specific QOL instrument should be used when investigators are interested in a crosssectional picture of the impact of pruritus, a comparison across pruritic conditions, or determining whether patients are responding to a specific therapy. REFERENCES 1. Yosopovitch G, Goon A, Wee J, Chan YH, Goh CL. The prevalence and clinical characteristics of pruritus among patients with extensive psoriasis. Br J Dermatol 2000;143:969-73. 2. Koblenzer CS. Itching and the atopic skin. J Allergy Clin Immunol 1999;104(Suppl):S109-13. 3. Yosipovitch G, Ansari N, Goon A, Chan YH, Goh CL. Clinical characteristics of pruritus in chronic idiopathic urticaria. Br J Dermatol 2002;147:32-6. 4. Bernhard J. Pruritus in skin diseases. In: Bernhard JD, editor. Itch: mechanisms and management of pruritus. New York: McGraw-Hill; 1994. pp. 37-68. 5. Etter L, Myers SA. Pruritus in systemic disease: mechanisms and management. Dermatol Clin 2002;20:459-72. 6. Zucker I, Yosipovitch G, David M, Gafter U, Boner G. Prevalence and characterization of uremic pruritus in patients undergoing hemodialysis: uremic pruritus is still a major problem for patients with end-stage renal disease. J Am Acad Dermatol 2003;49:842-6. 7. Leuschner U. Primary biliary cirrhosis: presentation and diagnosis. Clin Liver Dis 2003;7:741-58. 8. Shapiro R, Samorodin C, Hood A. Pruritus as a presenting sign of acquired immunodeficiency syndrome. J Am Acad Dermatol 1987;16:1115-7. 9. Zirwas MJ, Seraly MP. Pruritus of unknown origin: a retrospective study. J Am Acad Dermatol 2001;45:892-6. 10. Braverman IM. Skin manifestations of internal malignancy. Clin Geriatr Med 2002;18:1-19. 11. Ogunbiyi AO, Daramola OO, Alese OO. Prevalence of skin diseases in Ibadan, Nigeria. Int J Dermatol 2004;43:31-6. 12. Wolkenstein P, Grob JJ, Bastuji-Garin S, Ruszczynski S, Roujeau JC, Revuz J. French people and skin disease: results of a survey using a representative sample. Arch Dermatol 2003;139:1614-9. 13. Norman RA. Xerosis and pruritus in the elderly: recognition and management. Dermatol Ther 2003;16:254-9. 14. Chren MM, Lasek RJ, Flocke SA, Zyzanski SJ. Improved discriminative and evaluative capability of a refined version of Skindex, a quality-of-life instrument for patients with skin diseases. Arch Dermatol 1997;133:1433-40.
244 Desai et al
J AM ACAD DERMATOL AUGUST 2008
15. Chren MM, Lasek RJ, Sahay AP, Sands LP. Measurement properties of Skindex-16: a brief quality-of-life measure for patients with skin diseases. J Cutan Med Surg 2001;5:105-10. 16. Chen SC, Yeung J, Chren MM. Scalpdex: a quality of life instrument for scalp dermatitis. Arch Dermatol 2002;138: 803-7. 17. Nicholson K, Abramova L, Chren MM, Yeung J, Chon SY, Chen SC. A pilot quality-of-life instrument for acne rosacea. J Am Acad Dermatol 2007;57:213-21.
18. Dillon W, Goldstein M. Multivariate analysis: methods and applications. New York: John Wiley and Sons; 1984. 19. Spilker B, editor. Quality of life and pharmacoeconomics in clinical trials. 2nd ed. New York: Lippincott-Raven; 1996. p. 43. 20. US Food and Drug Administration. Patient-reported outcome measures: use in medical product development to support labeling claims. February 2006. Available from: URL: http://www. fda.gov/CDER/GUIDANCE/5460dft.pdf. Accessed December 4, 2006.
ruritus is the chief symptom in numerous dermatologic conditions,1-4 and a common symptom in systemic disorders,5-7 infectious diseases,8,9 and neoplastic diseases.9,10 The prevalence of pruritus in the general population is not well
P
documented but has been reported to be around 10%.11,12 It is the most common dermatologic problem in nursing homes along with xerosis.13 Not only is pruritus a widespread phenomenon, it also can have a significant quality-of-life (QOL) impact in
From Washington University, Saint Louisa; University of North Carolinab; University of Torontoc; University of Texas Southwesternd; Department of Dermatology, Emory University School of Medicine, Atlantae; and Division of Dermatologyf and Department of Health Services Research and Development,g Atlanta Veterans Administration Medical Center. Supported in part from an American Skin Association (ASA) Health Services Research Grant, the ASA David Martin Carter Research Scholar Award, and an Emory Skin Disease Research Center Pilot and Feasibility grant (No. P30AR42687) from the National Institute on Arthritis and Musculoskeletal and Skin Disease (NIAMS), National Institutes of Health (NIH). Dr Chen was supported in part by a Mentored Patient Oriented Career Development Award (No. K23AR02185-01A1) from NIAMS, NIH.
Conflicts of interest: None declared. Presented as abstracts at the 66th Annual Meeting of the Society for Investigative Dermatology, Miami, Florida, May 2005, and the 63rd Annual Meeting of the American Academy of Dermatology Annual Meeting, New Orleans, Louisiana, February 2005. Accepted for publication April 4, 2008. Reprint requests: Suephy C. Chen, MD, MS, Department of Dermatology, Emory University School of Medicine, 101 Woodruff Circle; Atlanta, GA 30322. E-mail: [email protected]. Published online June 12, 2008. 0190-9622/$34.00 ª 2008 by the American Academy of Dermatology, Inc. doi:10.1016/j.jaad.2008.04.006
234
J AM ACAD DERMATOL
Desai et al 235
VOLUME 59, NUMBER 2
Abbreviations used: ANOVA: analysis of variance ICC: intraclass correlation coefficient QOL: quality of life
patients’ lives. In a recent French study, more than 40% of patients with chronic pruritus described their disease as burdensome.12 Current treatments for pruritus are sparse, mostly consisting of antihistamines, antileukotrienes, and immunosuppressives, which are of varying efficacy. Researchers are challenged to develop better therapies, but pruritus, like pain, is a stimulus that cannot be measured directly. Rather, pruritus is measured either by observing the amount of scratching by the patient or by asking the patient to rate pruritus on some sort of severity scale, such as the visual analog scale or a numeric scale. However, neither the amount of scratching nor the visual analog/numeric scale measures the impact of pruritus on patients’ QOL. The extent to which a disease impacts patients’ QOL can be quantified by health status instruments. Health status instruments can either be generic or condition specific. Generic instruments capture broad-based QOL issues and, therefore, may be used to directly compare QOL across disparate disease states. However, generic instruments generally exclude QOL aspects that may be significant for a specific condition. Condition-specific instruments, by contrast, focus on QOL aspects pertinent to a particular condition. As a result, they have the advantage of being shorter, more appropriate, and more sensitive to subtle, specific QOL issues related to the disease in question. Currently there are no validated pruritus-specific QOL instruments. Although generic and skin-specific QOL instruments are available, they do not directly assess the issues specific to pruritus and, thus, will not adequately capture itch-specific QOL impact. In this study, we developed a pruritus-specific QOL instrument, called ItchyQoL, based on concerns and issues pertinent to patients with pruritus. We tested the instrument for validity, reliability, and responsiveness and we report preliminary results. In addition, we tested the instrument in a second sample of patients with pruritus to determine its measurement properties in the clinical setting.
METHODS All patients were consented and recruited from Emory University Dermatology Clinics, Atlanta, Ga; Grady Memorial Hospital, Atlanta, Ga; or the Veterans
Affairs Medical Center of Atlanta, Ga. We obtained approval for the protocol and consent process from the respective institution’s research committee and by the university institutional review board. Item development We conducted in-depth interviews with adult patients with pruritus of any cause to elicit any and all ways that pruritus affected their lives. Interviews were conducted until no new items were obtained. Based on all patient responses and their frequency, we composed pruritus-specific items. Items were developed into questions that addressed frequency, level of bother, or both. We conceptualized 3 major constructs from these items that explain the way pruritus affects patients’ QOL: symptoms, functioning limitations, and emotions. We administered the pruritus-specific items in addition to the 29-item version of Skindex14 and Skindex-1615 to comprehensively assess the constructs patients found relevant. Skindex-29 addresses QOL from a frequency perspective whereas Skindex-16 addresses level of bother. For all items, we used the same format as for the Skindex instruments, but changing the wording ‘‘skin condition’’ to ‘‘itchy skin condition.’’ All items inquired about the last 7 days. The responses for frequency type questions were ranked on a 5-point scale with designations of: never, rarely, sometimes, often, and all the time. A similar scale was developed for bother type questions: not bothered, little bothered, somewhat bothered, very bothered, and severely bothered. The final survey also contained demographic, pruritus severity, cause, and medication use questions. After the development of the initial instrument, it was administered to 5 adults with no medical training recruited from the general population for pilot testing, to test for clarity and comprehension. Based on the responses, we made the appropriate modifications to the instrument. Sample population, measures, and data collection All adult patients with active pruritus in any area of the body within the past 7 days were eligible to participate. Patients were recruited from our university dermatology clinic by referral from their dermatologist, or by telephone by contacting patients in our pruritus database. The database consists of patients seen at the university dermatology clinic identified by International Classification of Diseases, Ninth Revision code for pruritus. Interviewers administered either the bother or frequency version of the instrument to patients in a pseudorandom fashion. For each interviewer, the first patient was
236 Desai et al
J AM ACAD DERMATOL AUGUST 2008
randomly given one of the two versions and then the interviewer alternated administering the two versions of the survey. Patients answered all questions at baseline, at 72 hours (allowable range: 3-7 days), and at 2 months (allowable range: 2-3 months). The demographic questions were omitted in the latter two time points. Because the 72-hour data are used to test for reliability and because pruritus can change within a 72-hour time frame, patients were asked to remember their itch from their baseline time point. Patients were asked to self-rate their pruritus at the 2to 3-month time point as improved, worsened, or not changed based on the question, ‘‘How has your itchy skin condition changed since the last time you filled out this survey?’’ To further determine face validity, the frequency version of the final instrument was administered in the clinical setting. Adult patients from all study centers who had been pruritic in the past 7 days were asked to fill out the pruritus-specific QOL instrument, at the baseline time point only. Analyses All statistical analyses were performed with software (SPSS 12.0 for Windows, 12.0 edition, SPSS Inc, Chicago, Ill). Statistical significance was defined as P less than or equal to .05. Scoring Results for the instrument consisted of 3 subscale (symptom, function, and emotion) scores, based on the hypothesized constructs and an overall score. The subscale scores consist of the average of the responses to the items in a given subscale. For example, the subscale score for symptom was the average of all responses to the items pertinent to symptoms. The overall score consists of the average of the responses to all the items. The responses to the frequency items were scored on a 1 (never) to 5 (all the time) scale. Similarly, the responses to bother items ranged from 1 (not bothered) to 5 (severely bothered). Item reduction To create the most parsimonious evaluative instruments, we eliminated irrelevant and insensitive questions. We defined insensitive items as those where greater than 50% of patients responded with an extreme value (never/not bothered or all the time/severely bothered), an accepted cut-off threshold used in the literature.16,17 Psychometric evaluation The instrument was tested for reliability, responsiveness, and validity. We evaluated reliability with
internal consistency by Cronbach a coefficient and with reproducibility (test-retest reliability) by intraclass correlation coefficient (ICC). We tested for responsiveness by applying the paired t test to the baseline and 2-month answers for 3 groups: those who reported improvement, no change, or worsening of their pruritic condition. Validity was confirmed by examining face, content, construct, and discriminant validity. We ensured face and content validity of the instrument by conducting in-depth interviews with patients with pruritus. We also tested face validity in the clinical setting by administering the frequency version of the final instrument to a second sample of patients with pruritus. We compared subscale scores across demographic information, itch severity and frequency, diagnosis, and disease duration. One-way analysis of variance (ANOVA) was used for these comparisons. Construct validity was tested in two ways. We hypothesized that the items for bother and frequency would cluster into 3 factors that could be labeled as: symptoms, functioning, and emotions. This hypothesis was tested by using principle axis factor analysis followed by promax rotation for bother items, and varimax rotation for frequency items. We retained only those factors with eigenvalues greater than 2 and by application of the scree test.18 We identified the factor onto which items loaded by selecting the largest coefficient of that item among all the retained factors. Each factor was labeled by the predominant trait of the heavily loaded items. The regression factor scores were compared with the unweighted hypothesized subscale scores by Pearson correlation coefficients. The second way we tested construct validity was by comparing the subscale scores with the self-reported pruritus severity and pruritus frequency using one-way ANOVA. We hypothesized that the subscale scores would correlate with pruritus severity and pruritus frequency. Discriminant validity is the extent to which one instrument measures a certain health characteristic better than another instrument. A questionnaire that asks questions more pertinent to the patient’s actual condition would eliminate the floor effect (when data cannot take on a value lower than some particular number set by the instrument19), spare irrelevant questions, and improve the sensitivity of the questionnaire itself. Because Skindex-29 and Skindex-16 are QOL instruments specific to skin rather than pruritus, we hypothesized that they are not as sensitive as the test instrument in measuring QOL issues specific to pruritus. Discriminant validity was tested by comparing the floor effect of the final pruritusspecific QOL instrument with Skindex 29 and Skindex 16. We evaluated floor effect by quantifying
Desai et al 237
J AM ACAD DERMATOL VOLUME 59, NUMBER 2
the number of insensitive items (ie, those items where [50% of participants chose ‘‘never’’ as their response).
Table I. Demographics and pruritus characteristics
RESULTS
Age, y, mean (SD) Sex Male Race/ethnicity White/Caucasian Black/African American Other Education level # Grade 8 Any high school Any college Any postgraduate Total yearly household income, $ \10,000 10,000-25,000 25,000-50,000 50,000-75,000 75,000-100,000 [100,000 Itch frequency All of the time Most of the time Some of the time Once in awhile Never Itch severity Worst ever Very severe Average Itch occasionally Itch duration [3 mo 3-6 mo 6 mo-1 y 1-5 y 5-10 y [10 y When is itching the worst? All day Morning Afternoon Evening At night [1 of the above Diagnosis Urticaria/hives Dermatitis/eczema Other Idiopathic itching
Demographics We invited 21 patients to participate in face-toface in-depth interviews. After designing the initial pruritus-specific QOL instrument from these interviews, we contacted 187 patients for the validation phase of the study. In all, 61 patients were approached in clinic and 147 patients were contacted by telephone. Nearly all of the patients recruited from clinic (96%, N = 59), and approximately 16% (N = 30) of the invited patients by telephone agreed to participate. There was no significant difference in the socioeconomic profile (age, sex, race, education, or income levels) between patients recruited in clinic or by telephone. The majority of the patients contacted by telephone declined because they were no longer pruritic (22%, N = 32) or were not interested in the study (20%, N = 30); we were unable to contact 37% of patients (N = 55). In our cohort of 89 patients, 50 patients were administered the frequency version of the instrument and 39 patients were administered the bother version. Sixteen patients only completed the baseline survey and could not be contacted to complete the 72-hour or 2-month time points. A total of 23 patients completed the baseline and 72-hour time points only. Three patients did not complete the 72-hour time point, but were able to complete the 2-month time point. The overall mean age was 58.1 years (SD: 16.6), 39% were male, 74% were Caucasian, 21% were African American, 5% were of another race, and 75% had attended college or beyond (Table I). For the clinical application phase of the study, we visited 29 general dermatology clinics at a universitybased center, Department of Veterans Affairs hospital, or local inner-city hospital. Of 663 patients scheduled to arrive, 508 patients kept their appointment. Because of limited manpower and the amount of time required to administer the survey, we were able to approach 320 patients, and of those 115 patients (36%) were pruritic. In all, 104 of the patients with pruritus completed the frequency version of the instrument. This population had a mean age of 55.9 years; 46% were male, 69% were Caucasian, 23% were African American, and 5% were of another race. The majority (76%) of the patients had attended college or beyond (Table I). Disease characteristics Of the 89 patients who completed the instrument development and validation portion of the study, the
Instrument validation (N = 89)
Instrument application (N = 104)
58.1 (16.7)
55.9 (16.4)
39.3%
45.6%
74.2% 21.3% 4.5%
68.9% 22.6% 4.7%
3.4% 21.3% 51.7% 23.6%
2.9% 21.2% 51.9% 24.0%
7.2% 2.6% 32.5% 21.7% 12.0% 22.9%
10.9% 17.4% 23.9% 27.2% 9.8% 10.9%
25.8% 37.1% 30.3% 6.7% 0%
16.2% 27.6% 41.0% 15.2% 0%
10.1% 24.7% 47.2% 28.0%
4.8% 26.0% 39.4% 29.8%
4.5% 3.4% 16.9% 40.4% 15.7% 19.1%
18.4% 12.6% 15.5% 18.4% 7.8% 27.2%
32.6% 3.4% 3.4% 10.1% 22.5% 28.1%
32.7% 5.1% 6.1% 25.5% 27.6% 3.1%
10.1% 32.6% 25.8% 31.5%
6.9% 19.8% 14.9% 56.4%
majority of patients were able to attribute their pruritus to a specific cause (Table I) but 32% had idiopathic itching. The majority of patients rated the
238 Desai et al
J AM ACAD DERMATOL AUGUST 2008
severity of their pruritus as more than just occasionally with 10% rating as worst ever. Most patients had been itching longer than 1 year (75%). Of the 104 patients who completed the clinical application portion of the study, the majority (56%) had idiopathic itching. Like the previous cohort, most patients rated the severity as more than just occasionally and the majority (53%) had been itching for more than 1 year (Table I). Item development, analysis, and reduction We interviewed a total of 21 patients, and no new responses were obtained after the 15th patient. We developed 27 pruritus-specific items from the indepth patient interview session, grouped into the following constructs: 6 symptom, 10 functional limitation, and 11 emotion (Table II). Eighteen of these items are measured with both frequency and bother. The remaining 9 items pertained to emotion and were measured by frequency only. Twelve of the items developed in the interviews were already present in Skindex-29 of which 8 were also analogous to Skindex-16 items. Therefore, 15 new items were created and are unique to the instrument (Table II). Both frequency and bother versions were formulated to query about the preceding 7 days. Of the 27 items, the patients demonstrated that their pruritus-related QOL was most affected by ‘‘needing to scratch’’ with a frequency score (SD) of 4.08 (0.80), and a bother score of 3.81 (1.03). QOL was least affected by ‘‘limiting the types of food you can eat’’ with a frequency score of 1.58 (1.06) and a bother score of 1.50 (0.99). Mean scores for all items are listed in Table II. Several items proved to be relatively insensitive, in which more than 50% of the patients answered ‘‘never or not bothered’’ to these items, and were subsequently eliminated. A total of 5 frequency items and 3 bother items were found insensitive. Of these, 3 of the items were frequency and bother versions of the items: ‘‘limiting the types of food you can eat,’’ ‘‘worry about having a contagious disease,’’ and ‘‘self-esteem changing.’’ The rest of the eliminated items were the frequency version of pruritus-specific items. The final version of the pruritus-specific QOL instrument consists of the 22 sensitive items. The frequency portion consists of 22 items, and the bother portion consists of 15 items. Feasibility Note that although several of the Skindex items were not included in the pruritus-specific QOL instrument, they were administered in the study for us to compare the test instrument with Skindex. The average duration to answer all 44 questions of the frequency instrument (test instrument 1 additional
Skindex-29 items) was 10.0 minutes (SD: 4.36 minutes). The average duration to all 36 questions of the bother instrument was 9.36 minutes (SD: 4.00 minutes). Reliability The overall pruritus-specific QOL instrument score demonstrated internal consistency reliability with Cronbach a ranging from 0.89 for bother and 0.92 for frequency (Table III). Systematically deleting one item from the analysis did not significantly raise the coefficients. The overall score also demonstrated reproducibility with ICC ranging from 0.87 for bother and 0.92 for frequency. Similar findings were observed for each of the 3 subscales (Table III). Internal consistency reliability was demonstrated with Cronbach a coefficient ranging from 0.76 to 0.81 for bother items, and from 0.72 to 0.93 for frequency items. Systematically deleting one item from the analysis did not significantly raise the coefficients. Reproducibility was demonstrated with ICC ranging from 0.84 to 0.87 for bother items, and 0.91 for frequency items. Responsiveness result Patients were divided into 3 categories (better, worse, and no change) based on their responses to the question, ‘‘How has your itchy skin condition changed since the first interview?’’ For the group that answered the bother-type questions, 13 patients reported improvement, 4 reported worsening, and 6 reported no change. For the group that answered the frequency-type questions, 16 patients reported improvement, 8 reported worsening, and 8 reported no change. Using the pruritus-specific overall and subscale scores (Table IV), we found improvement in the QOL scores of those patients who reported improvement in their pruritus; the difference was statistically significant (P # .01, using paired t test) for the overall score and all subscale scores in the bother and frequency versions except for bother symptom and function where there was a trend toward statistical significance (P = .06-.07). Future work with larger numbers of patients will elucidate this trend. We found no significant differences in all subscale scores of those who reported no change. We also did not find significant difference in all subscale scores of those who reported worsening of their pruritus; however, our number of patients was very small in these groups. Construct validity Principle axis factor analysis with varimax rotation was performed on the frequency items, and 3 factors were retained according to the criteria outlined in the
Desai et al 239
J AM ACAD DERMATOL VOLUME 59, NUMBER 2
Table II. Instrument items with hypothesized constructs and mean scores Item
Hypothesized construct
1. 2. 3. 4. 5. 6.
Sy Sz Sz S S S
2.35 2.82 2.94 2.66 4.08 2.98
(1.17) (1.24) (1.38) (1.48) (0.80) (1.46)
2.28 2.72 2.67 2.40 3.81 2.78
(1.37) (1.38) (1.39) (1.58) (1.03) (1.47)
F Fz Fz Fz Fy Fy F F F F
3.48 2.92 2.73 1.80 3.28 1.97 2.71 1.58 2.74 3.71
(1.24) (1.35) (1.40) (1.19) (1.22) (1.33) (1.17) (1.06) (1.53) (1.49)
2.98 2.81 2.75 1.75 3.18 1.93 2.60 1.50 2.51 2.55
(1.46) (1.36) (1.45) (1.17) (1.41) (1.40) (1.28) (0.99) (1.48) (1.53)
Ez Ez E Ey Ez E E
4.04 2.89 3.35 2.88 2.57 1.89 2.43
(1.11) (1.53) (1.26) (1.28) (1.36) (1.37) (1.36)
3.82 (1.27) 2.80 (1.58) N/A N/A N/A N/A N/A
E E E E
3.74 2.80 2.13 2.17
(1.21) (1.43) (1.28) (1.33)
N/A N/A N/A N/A
My itchy skin condition bleeds My skin hurts because of my itchy skin condition My itchy skin condition burns or stings I get scars from my itchy skin condition I need to scratch my itchy skin condition Temperature/seasonal changes aggravate my itchy skin condition 7. I spend a lot of money treating my itchy skin condition 8. My itchy skin condition makes it hard to work or do what I enjoy 9. My itchy skin condition affects my interaction with others 10. I tend to stay home because I don’t want to scratch in public§ 11. My itchy skin condition affects how well I sleep 12. My itchy skin condition interferes with my sex life§ 13. My itchy skin often makes it difficult to concentrate 14. My itchy skin condition limits the types of foods I can eat§ 15. My itchy skin condition limits the types of clothes I can wear 16. My itchy skin condition forces me to buy special soaps, detergents, and lotions 17. I am frustrated by my itchy skin condition 18. I am embarrassed by my itchy skin condition 19. My itchy skin condition drives me crazy/nuts 20. My itchy skin condition makes me feel angry or irritable 21. My itchy skin condition makes me feel depressed or sad 22. I worry I have a contagious disease§ 23. I worry about what other people think about me because of my itchy skin condition 24. I worry that the itching will last forever 25. I feel self-conscious because of my itchy skin condition 26. My personality has changed because of my itchy skin condition 27. My self-esteem has changed because of my itchy skin condition§
Frequency, mean (SD) score*
Bother, mean (SD) score*
Subscale scores of S, F, and E are calculated by average of item scores that pertain to that particular subscale. Frequency items are scored on scale of 1-5 (1 = never, 2 = rarely, 3 = sometimes, 4 = often, 5 = all the time). Bother items are scored on scale of 1-5 (1 = not bothered, 2 = little bothered, 3 = somewhat bothered, 4 = very bothered, 5 = severely bothered). E, Emotions; F, functioning; N/A, not applicable; S, symptoms. *Mean scores are derived from 89-person instrument validation population. y Also in Skindex-29. z Also in Skindex-29 and Skindex-16. § Insensitive items removed from final version of instrument.
‘‘Methods’’ section. From the items that loaded most heavily on factor 1, we could see that the predominant trait was emotions; factor 2 was clearly functioning, and factor 3 was symptoms. The Pearson correlation coefficient comparing the regression factor scores and the unweighted hypothesized subscale scores were significant (P\.01) and ranged from 0.52 to 0.77 (Table III). For the bother items, 3 factors were also retained after principle axis factor analysis with promax rotation. The predominant trait or construct loaded most heavily on factor 1 was functioning, factor 2 was emotions, and factor 3 was symptoms. The Pearson correlation coefficient for the bother items ranged from 0.86 to 0.93 (P \.01) (Table III).
In the clinical application cohort (n = 104) construct validity was confirmed by comparing the subscale scores with self-reported pruritus severity and frequency at the baseline time point. The severity levels were answered on a 5-point scale: worst ever, very severe, average, itch occasionally, and no itching at all. Pruritus frequency levels were measured on a similar scale (1-5): all of the time, most of the time, some of the time, once in awhile, and never. We found greater differences in symptom and emotion scores among the different levels of pruritus severity than would be expected by chance in both frequency and bother items (P \ .05, by ANOVA) (Table V). We also found significant differences in
240 Desai et al
J AM ACAD DERMATOL AUGUST 2008
Table III. Psychometric test results Psychometric test
Overall
Cronbach a coefficient
F: B: F: B:
Intraclass correlation coefficient Pearson correlation coefficient comparing RF scores and unweighted hypothesized subscale scores
0.92 0.89 0.92 0.87
Symptoms
Functioning
Emotions
F: 0.72 B: 0.78 F: 0.90 B: 0.84 FRF1: 0.20 FRF2: 0.37* FRF3: 0.52* BRF1: 0.75* BRF2: 0.88* BRF3: 0.86*
F: 0.82 B: 0.86 F: 0.92 B: 0.89 FRF1: 0.50* FRF2: 0.64* FRF3: 0.16 BRF1: 0.93* BRF2: 0.81* BRF3: 0.61*
F: 0.91 B: 0.76 F: 0.93 B: 0.84 FRF1: 0.77* FRF2: 0.37* FRF3: 0.32 BRF1: 0.88* BRF2: 0.91* BRF3: 0.70*
B, Bother item; F, frequency item; RF, regression factor. *P \ .01.
Table IV. Responsiveness scores Pruritus condition
ItchyQoL Worse (F: n = 8, B: n = 4)
Subscale
Symptoms Functioning Emotions Overall
ItchyQoL Same (F: n = 8, B: n = 6)
Symptoms Functioning Emotions Overall
ItchyQoL Better (F: n = 16, B: n = 13)
Symptoms Functioning Emotions Overall
Baseline, mean score 6 SD
2 mo, Mean score 6 SD
P value
F: 3.17 6 0.75 B: 3.23 6 0.55 F: 2.68 6 0.61 B: 2.48 6 0.78 F: 2.57 6 0.88 B: 2.75 6 1.41 F: 2.85 6 .56 B: 2.82 6 .68 F: 2.87 6 0.88 B: 2.54 6 0.97 F: 2.71 6 0.71 B: 2.35 6 0.99 F: 2.70 6 1.15 B: 2.78 6 1.30 F: 2.75 6 .89 B: 2.48 6 .97 F: 2.95 6 0.75 B: 2.65 6 0.88 F: 3.29 6 0.78 B: 2.82 6 0.84 F: 2.88 6 1.04 B: 3.12 6 1.14 F: 3.03 6 .78 B: 2.77 6 .79
F: 3.40 6 0.55 B: 3.17 6 0.71 F: 3.13 6 0.81 B: 2.86 6 1.11 F: 3.17 6 1.23 B: 3.88 6 1.31 F: 3.22 6 .73 B: 3.12 6 .83 F: 3.30 6 1.39 B: 2.71 6 0.97 F: 2.29 6 0.49 B: 2.07 6 0.48 F: 2.38 6 0.96 B: 2.93 6 1.59 F: 2.55 6 .57 B: 2.32 6 .75 F: 2.25 6 0.97 B: 2.35 6 1.08 F: 2.63 6 1.19 B: 2.33 6 1.21 F: 2.27 6 1.07 B: 2.39 6 1.04 F: 2.38 6 1.00 B: 2.34 6 1.10
F: .38 B: .63 F: .28 B: .39 F: .11 B: .19 F: .280 B: .377 F: .50 B: .22 F: .15 B: .39 F: .06 B: 1.00 F: .844 B: .199 F: \.01 B: .06 F: \.01 B: .07 F: \.01 B: \.01 F: \.01 B: .016
B, Bother item; F, frequency item.
symptom, functioning, and emotion scores among different levels of pruritus frequency in the frequency items, and between symptom and functioning scores in the bother items (P \ .05, by ANOVA) (Table V). Discriminant validity We demonstrated the discriminant validity of the pruritus-specific QOL instrument as compared with
Skindex-29 and Skindex-16 in patients by ascertaining that the test instrument had fewer insensitive items. We used the responses from the patients in both the instrument validation and application phases of the study. For frequency items, 1 of 22 of the test instrument items (4.5%) had at least 50% of the patients answering ‘‘never,’’ as compared with 9 of 29 (31.0%) Skindex-29 items. For bother items, none of the 15 test instrument items, and 2 of 16
Desai et al 241
J AM ACAD DERMATOL VOLUME 59, NUMBER 2
Table V. Face and construct validity Instrument validation Frequency, mean scores (N = 89)
Itch severity Worst ever Very severe Average Itch occasionally P value Itch frequency All of the time Most of the time Some of the time Once in awhile P value
S 3.68 3.26 2.84 2.51 .001 S 3.34 3.05 2.74 2.14 .003
F 3.27 3.32 3.00 2.84 .291 F 3.48 3.08 2.90 2.33 .011
Instrument application
Bother, mean scores (N = 89)
E 3.83 3.21 2.70 2.92 .015 E 3.45 2.99 2.66 2.55 .035
S 3.77 3.01 2.67 2.19 \.001 S 3.13 2.92 2.52 1.83 .006
F 3.16 3.01 2.66 2.48 .200 F 3.31 2.77 2.47 1.98 .003
Frequency, mean scores (N = 104)
E 4.22 3.59 3.02 3.16 .045 E 3.74 3.30 2.96 3.25 .207
S 3.38 3.24 2.78 2.25 \.001 S 3.57 3.06 2.48 2.19 \.001
F 3.00 2.96 2.13 1.62 \.001 F 3.29 2.53 1.94 1.44 \.001
E 3.60 2.93 2.20 1.86 \.001 E 3.03 2.76 2.02 1.78 \.001
E, Emotions; F, functioning; S, symptoms.
Table VI. Clinical application of ItchyQoL results Average subscale scores (SD)
Sex Male (n = 47) Female (n = 57) P value (ANOVA) Diagnoses Urticaria (n = 8) Dermatitis (n = 20) Other (n = 16) Idiopathic pruritus (n = 57) P value (ANOVA)
Symptoms
Functioning
Emotions
2.51 (0.73) 2.99 (0.81) .006
1.89 (0.79) 2.55 (1.21) .003
1.98 (0.85) 2.65 (1.23) .003
3.05 3.05 2.86 2.61 .149
2.89 2.46 2.42 1.99 .040
2.56 2.55 2.47 2.15 .361
(0.72) (0.79) (0.83) (0.81)
(1.22) (1.26) (1.01) (0.94)
(1.17) (1.18) (1.27) (0.96)
ANOVA, Analysis of variance.
(12.5%) Skindex-16 items, had at least 50% of the patients answering ‘‘not bothered.’’ Application of the pruritus-specific QoL instrument in the clinical setting When we applied the test instrument in the clinical application cohort of patients (n = 104), we observed several interesting findings (Table VI). First, although we found no differences in symptom, functioning, or emotion subscale scores among age, race, education, income, or disease duration, we did find that women had higher (P \ .01) average subscale scores than men despite no difference in itch severity or itch frequency between the two groups. Second, we detected a difference in functional QOL impact between the different diagnoses, with urticaria having the highest overall QOL impact (P = .04). On post hoc analysis, we found that those given a diagnosis of dermatitis had a higher symptomatic impact than those with idiopathic itching (P = .04).
To explore the sex QOL differences further, we compared sex distribution of the specific causes listed in the survey. We found that women reported higher percentages of all the diagnoses as compared with men (urticaria: 3.5% vs 1.4%, eczema: 7.6% vs 4.8%, other: 5.3% vs 4.1%). On the other hand, men were more apt to report that their origin has yet to be elucidated (20.4% vs 15.2%).
DISCUSSION This study represents a pilot QOL instrument specifically for patients with pruritus that has demonstrated preliminary reliability, responsiveness, and validity. We demonstrated reliability of the pruritusspecific QOL instrument with high internal consistency reliability and reproducibility. This ensures that the instrument will yield consistent results for a given patient in similar conditions and is relatively free from random error. We verified preliminary data for responsiveness by demonstrating that individual
242 Desai et al
J AM ACAD DERMATOL AUGUST 2008
construct and total scores improved in those patients who reported improvement in their pruritus. Scores did not significantly change in patients who reported that their pruritus worsened. However, subscale scores did increase, hinting at potentially greater QOL impact with worsening pruritus in the frequency and bother items in all constructs except for symptom impact for bother items (Table IV). We believe that with a larger sample size, this trend of increasing patient scores with worsening pruritus will become significant. An instrument may be reliable and responsive, but not valid. We ensured face and content validity by deriving the items from in-depth interviews with patients with pruritus. We confirmed construct validity by finding that the subscale scores correlated with self-reported severity and frequency in the expected direction and by demonstrating that our 3 hypothesized constructs (symptoms, emotions, and functioning) correlated with the factor analysis of the data. We also demonstrated that the pruritus-specific QOL instrument was more sensitive to QOL issues as compared with Skindex-29 and Skindex-16. Lastly, the test instrument is a feasible measure of pruritus to use in many settings. The combined frequency or bother instrument (test instrument 1 Skindex) took, on average (SD), 9.37 to 10.0 (4.004.36) minutes to complete; the 27-item pruritus instrument only took 6.71 (2.90) minutes to complete. In clinical application we found that, after item reduction, the 22-item frequency pruritus instrument took, on average, 3.11 minutes to complete. We believe that a 5- to 10-minute instrument is manageable for both patients and clinicians. The clinical application data that we obtained revealed many interesting findings about how pruritus affects people’s lives. The fact that urticaria has the highest overall QOL impact speaks to the need to develop therapies for chronic urticaria. The differences found in sex (women with more QOL impact than men) are also thought provoking. Looking at the data carefully, we see that patients with idiopathic pruritus reported less of a functional impact than the other causes (Table VI); because men reported more idiopathic cause than women, this may be the reason behind the lower functional impact in men compared with women. Similarly, we found that dermatitis has more symptomatic impact than idiopathic cause; because women reported dermatitis more commonly than men, this may explain why women reported more symptoms than men. Although the emotional impact was not significantly different comparing idiopathic versus known cause, it may be that men were on average less emotionally impacted by their pruritus because they
still had hope that an origin would be discovered and, thus, their pruritus would be treatable. Thus, the sex differences that we observed in QOL impact may be a result of the differences in causes behind the pruritus. This interesting finding begs to be elucidated with an epidemiologic study. Limitations The results of this study are limited by several factors. First, the responsiveness of the pruritusspecific QOL instrument has only been validated in patients whose disease either improved or did not change from baseline. This may limit the use of the instrument in investigational studies until further data are published regarding the instrument’s responsiveness in patients with worsening disease. Second, there is potential for selection bias in the instrument development phase of the study, as the majority of the patients recruited were from an academic institution. These patients may have had pruritus refractory to standard therapies, and a higher educational background than patients from the inner city or rural clinics. Thirdly, we administered the test instrument by telephone in a subset of patients. Although we do not anticipate that face-toface or self-administration would make a significant difference, future work is necessary to compare the administration by telephone versus in person. Fourth, a small sample of patients for this study also serves as a limitation. A larger study population is needed for confidence in the ability to generalize our results, and for confidence in the analyses such as the confirmatory factor analyses and responsiveness analyses. Fifth, we tested the reliability of the test instrument by comparing baseline and 72-hour answers, where the 72-hour time point asked the patient to recall their baseline pruritus. To test reliability, the second time point is chosen long enough from baseline so that the patient cannot recall the answers for their baseline survey, but not so long such that their clinical severity would change. We thought that 72 hours was long enough for patients to forget their exact answers to the test instrument, and that for the majority of patients, their pruritus would not change significantly. However, given that a small number of patients may have changed the severity of their pruritus, we asked the patients to recall their baseline pruritus. By doing so, we may have introduced an element of testing recall. We do not think that recall is an issue for a significant proportion of our population, but acknowledge this potential limitation. Lastly, we assumed that each item carries equal weight when attributing components to each axis (symptom, function, and emotion). Currently this is
J AM ACAD DERMATOL
Desai et al 243
VOLUME 59, NUMBER 2
the standard for developing this type of QOL instrument; however, the next step is to allow the patients to weigh the importance of the items in the survey. Additional studies are warranted for the pruritusspecific QOL instrument to be used with confidence in a general population. Validation needs to take place with a larger subset of less-educated patients from nonacademic institutions. Responsiveness needs to be verified in patients with worsening disease in a larger population of patients. Further work needs to be performed in validating other means of administrating the test instrument (face-toface and self-administration) and to apply other methodologies to incorporate patient weighing of importance of items. Potential uses of the pruritus-specific QOL instrument The current form of the pruritus-specific QOL instrument represents a promising patient-reported outcome measure of disease burden as a result of pruritus and might be practical for use in a clinical or research setting. The instrument may be used as a tool for clinicians who treat patients who have conditions in which pruritus is the major symptom. The clinician can use either the frequency or bother version of the instrument to help their patients with pruritus in many ways. First, the profile of the instrument can be analyzed to determine which aspect of the disease most impacts the patient in relation to symptoms, functional impairment, and emotions. This can differ greatly among individuals and, thus, the instrument can help to personalize the treatment and advice offered. Next, changes in QOL can be a useful guide in therapeutic intervention. Administering the survey at different time points during the course of therapy could be useful in mapping a patient’s response to that therapy. Having an instrument that demonstrates an improvement, no change, or worsening in QOL can greatly affect both the patient-doctor relationship and patient compliance with treatment. Lastly, clinicians can use the impact on QOL data to petition managed care and insurance companies when coverage is denied for therapies to treat pruritus. In the research setting, the pruritus-specific QOL instrument may be useful as an end point for epidemiologic and burden-of-disease studies. It may also be useful as a secondary or coprimary end point for clinical trials testing new therapies. The Food and Drug Administration has recently provided guidance regarding patient-reported outcomes,20 thus emphasizing the importance of well-developed QOL instruments. The development strategy of the test instrument has complied with these guidelines and
bridges the gap in the research arena for a pruritusspecific QOL patient-reported outcome. Although Skindex-29 and -16 are both well-validated and reliable QOL instruments (ICC: 0.88-0.92 and 0.88-0.90, respectively, and Cronbach a: 0.870.96 and 0.86-0.93, respectively), they are skin-specific instruments and were not developed to address the issues specific to pruritus. In the final version of the instrument, the pruritus-specific QOL instrument contains 15/22 unique items. Only 7 of the items were common to Skindex-16 or Skindex-29. Thus the Skindex instruments have a different role from that of the pruritus-specific QOL instrument. The Skindex instruments should be used with patients with pruritus for those researchers with an invested interest in comparing pruritic versus nonpruritic conditions. The pruritus-specific QOL instrument should be used when investigators are interested in a crosssectional picture of the impact of pruritus, a comparison across pruritic conditions, or determining whether patients are responding to a specific therapy. REFERENCES 1. Yosopovitch G, Goon A, Wee J, Chan YH, Goh CL. The prevalence and clinical characteristics of pruritus among patients with extensive psoriasis. Br J Dermatol 2000;143:969-73. 2. Koblenzer CS. Itching and the atopic skin. J Allergy Clin Immunol 1999;104(Suppl):S109-13. 3. Yosipovitch G, Ansari N, Goon A, Chan YH, Goh CL. Clinical characteristics of pruritus in chronic idiopathic urticaria. Br J Dermatol 2002;147:32-6. 4. Bernhard J. Pruritus in skin diseases. In: Bernhard JD, editor. Itch: mechanisms and management of pruritus. New York: McGraw-Hill; 1994. pp. 37-68. 5. Etter L, Myers SA. Pruritus in systemic disease: mechanisms and management. Dermatol Clin 2002;20:459-72. 6. Zucker I, Yosipovitch G, David M, Gafter U, Boner G. Prevalence and characterization of uremic pruritus in patients undergoing hemodialysis: uremic pruritus is still a major problem for patients with end-stage renal disease. J Am Acad Dermatol 2003;49:842-6. 7. Leuschner U. Primary biliary cirrhosis: presentation and diagnosis. Clin Liver Dis 2003;7:741-58. 8. Shapiro R, Samorodin C, Hood A. Pruritus as a presenting sign of acquired immunodeficiency syndrome. J Am Acad Dermatol 1987;16:1115-7. 9. Zirwas MJ, Seraly MP. Pruritus of unknown origin: a retrospective study. J Am Acad Dermatol 2001;45:892-6. 10. Braverman IM. Skin manifestations of internal malignancy. Clin Geriatr Med 2002;18:1-19. 11. Ogunbiyi AO, Daramola OO, Alese OO. Prevalence of skin diseases in Ibadan, Nigeria. Int J Dermatol 2004;43:31-6. 12. Wolkenstein P, Grob JJ, Bastuji-Garin S, Ruszczynski S, Roujeau JC, Revuz J. French people and skin disease: results of a survey using a representative sample. Arch Dermatol 2003;139:1614-9. 13. Norman RA. Xerosis and pruritus in the elderly: recognition and management. Dermatol Ther 2003;16:254-9. 14. Chren MM, Lasek RJ, Flocke SA, Zyzanski SJ. Improved discriminative and evaluative capability of a refined version of Skindex, a quality-of-life instrument for patients with skin diseases. Arch Dermatol 1997;133:1433-40.
244 Desai et al
J AM ACAD DERMATOL AUGUST 2008
15. Chren MM, Lasek RJ, Sahay AP, Sands LP. Measurement properties of Skindex-16: a brief quality-of-life measure for patients with skin diseases. J Cutan Med Surg 2001;5:105-10. 16. Chen SC, Yeung J, Chren MM. Scalpdex: a quality of life instrument for scalp dermatitis. Arch Dermatol 2002;138: 803-7. 17. Nicholson K, Abramova L, Chren MM, Yeung J, Chon SY, Chen SC. A pilot quality-of-life instrument for acne rosacea. J Am Acad Dermatol 2007;57:213-21.
18. Dillon W, Goldstein M. Multivariate analysis: methods and applications. New York: John Wiley and Sons; 1984. 19. Spilker B, editor. Quality of life and pharmacoeconomics in clinical trials. 2nd ed. New York: Lippincott-Raven; 1996. p. 43. 20. US Food and Drug Administration. Patient-reported outcome measures: use in medical product development to support labeling claims. February 2006. Available from: URL: http://www. fda.gov/CDER/GUIDANCE/5460dft.pdf. Accessed December 4, 2006.