|Year : 2020 | Volume
| Issue : 1 | Page : 25-32
Psychometric validation of the psoriasis disability index questionnaire (Translated Bengali Version): A cross-sectional study
Anaitulah Ahmad Mir1, Abhijit Chattopadhyay1, Jaytilak Pramanick1, Ashish Gautam2, Showkat Ahmad Mir3, Munmun Koley4, Subhranil Saha4
1 Department of Homoeopathic Materia Medica, National Institute of Homoeopathy, Kolkata, West Bengal, India
2 Department of Case Taking and Repertory, National Institute of Homoeopathy, Kolkata, West Bengal, India
3 Private Practice, Malpora Bomai Sopore, Baramulla, Jammu and Kashmir, India
4 Independent Researchers, Kolkata, West Bengal, India
|Date of Submission||13-Jul-2019|
|Date of Acceptance||09-Aug-2019|
|Date of Web Publication||27-Mar-2020|
Dr. Anaitulah Ahmad Mir
Department of Homoeopathic Materia Medica, National Institute of Homoeopathy, Block GE, Sector III, Salt Lake, Kolkata - 700 106, West Bengal
Source of Support: None, Conflict of Interest: None
Introduction: The Psoriasis Disability Index (PDI) is a pre-validated, self-administered questionnaire assessing quality of life in adults suffering from psoriasis. The English version contains 15 items measuring five components – activities, work/school performance, personal, leisure, and treatment. Purpose: We aimed to translate the PDI into Bengali and examine its psychometric properties. Methods: The PDI-Bengali version (PDI-B) was produced by standardized forward-backward translations. A cross-sectional study was conducted to gather responses by consecutive sampling. Reliability was tested using internal consistency and test-retest reliability analyses, concurrent validity by comparing with Psoriasis Area and Severity Index (PASI) and Dermatological Life Quality Index scores, while construct validity by exploratory factor analysis (n= 83) using principal component analysis (varimax rotation). Subsequently, confirmatory factor analysis (CFA; n = 85) was performed to verify the a priori scales by the goodness of fit model. Results: The internal consistency (Cronbach's alpha) and the intra-class correlation coefficients were 0.877 and 0.882, respectively. Acceptable values of internal consistency, test-retest reliability, and concurrent validity were found. All the items loaded above the pre-specified value of 0.4. Factor analyses using Varimax identified three components (social activities, concern, and personal activities) explaining 56.8% of the variation. The Kaiser-Meyer-Olkin was 0.836 and Bartlett's test of sphericity was P < 0.001. The goodness of fit of CFA model was excellent. Conclusion: The PDI-B, consisting of 15 items and framed within three components, is a valid and reliable questionnaire, but measured different dimensions from the English version.
Keywords: Confirmatory factor analysis, principal component analysis, psoriasis disability index, reliability, validity
|How to cite this article:|
Mir AA, Chattopadhyay A, Pramanick J, Gautam A, Mir SA, Koley M, Saha S. Psychometric validation of the psoriasis disability index questionnaire (Translated Bengali Version): A cross-sectional study. J Dermatol Dermatol Surg 2020;24:25-32
|How to cite this URL:|
Mir AA, Chattopadhyay A, Pramanick J, Gautam A, Mir SA, Koley M, Saha S. Psychometric validation of the psoriasis disability index questionnaire (Translated Bengali Version): A cross-sectional study. J Dermatol Dermatol Surg [serial online] 2020 [cited 2020 Jul 6];24:25-32. Available from: http://www.jddsjournal.org/text.asp?2020/24/1/25/281419
| Introduction|| |
In spite of considerable advancements in therapies, psoriasis reduces patients' quality of life (QoL). Patients with psoriasis experience physical discomfort, impaired emotional functioning, negative body image, absenteeism at workplace, and daily activity impairments. Even mild psoriasis affects patients' daily lives. The chronic and recurrent nature of the disease necessitates life-long treatment, which leads to a significant economic burden on patients and the health-care system. Increasing severity of psoriasis appears to correlate closely with increased severity of depression and a higher frequency of suicidal ideation., However, disease severity, as measured by instruments such as the Psoriasis Area and Severity Index (PASI), is not the sole factor determining the burden of illness because relatively minor psoriasis located on visible parts of the body may also have a detrimental effect on QoL.
The original Psoriasis Disability Index (PDI) is a 15item, validated, self-administered questionnaire designed to quantify the functional disability in the aspects of daily activities, employment, personal relationships, leisure, and treatment effects in adults with psoriasis over previous 4 weeks. The score ranges from 0 to 45 with higher scores reflecting greater QoL impairment. The PDI demonstrates the effects of interventions, both treatments and in health service research. The PDI has been translated into many languages and has been used in many countries.
We aimed to evaluate whether the Bengali version of the PDI-Bengali questionnaire (PDI-B) is a psychometrically sound tool to measure the impact of psoriasis and to examine cross-cultural adaptation considering linguistic equivalence.
| Materials and Methods|| |
This cross-sectional, observational, noninterventional study consisted of standardized translation, face validation by pilot testing, and field testing and psychometric evaluation of the PDI-B version.
It was conducted for 3 months, from April 2017 to June 2017 at National Institute of Homoeopathy, Kolkata, under Ministry of AYUSH, Government of India. Ethics approval for this study was obtained from the Institutional Ethics Committee (Ref. No. 5-023/NIH/PG/Ethical Comm. 2009/Vol. III/1964 [A/S]; dated March 27, 2017) prior to initiation.
- Forward translation: An expert committee was constructed, consisting of research methodologists, dermatologists, and linguistic experts. First, two independent native Bengali speakers translated the English version of PDI into the target language Bengali (T1 and T2). One of the translators was a dermatologist and therefore aware of the concepts that were being measured with the PDI, and the other translators were a language specialist with no medical background
- Synthesis of T1 and T2 into T1,2: The two translators then agree on a new consensus version of the translation (T1,2) that was verified by the expert committee supervising the project
- Back translation: For the back translation from Bengali into English, two English language translators (BT1 and BT2; one dermatologist and one linguistic expert) were required. Though born in India, both had been residing in the United States for over 20 years. Being blinded to the original English version, they both independently translated T1,2 back into English
- Committee review: The committee reviewed all the translations (T1 and T2, T1,2, B1 and B2) and the written report comparing the back translations with the forward translation T1,2. Based on those translations, the prefinal version was developed
- Face validation: The prefinal version of the questionnaire was tested on randomly (simple random sampling) chosen 10 patients visiting outpatient clinics of the hospital. Each completed the questionnaire and was then asked the meaning of each questionnaire item as well as whether or not they had problems with the questionnaire layout, content, clarity, language, instructions, or response scales. Any difficulties were noted and included in the final report. A detailed report written by the interviewing person, including proposed changes of the prefinal version based on the results of the face validity test, was then submitted to the expert committee
- Committee appraisal: The final version of the PDI-B was developed by the committee based on the results of the face validity testing and the written report. Questionnaire translation stages are depicted in [Figure 1]. The final version of the PDI-Bcan is found as a supplementary file.
Field testing and validation
Content validity of the PDI questionnaire was previously evaluated of the original English version and was therefore not tested in this study. The additional testing was done to evaluate construct validity.
Aged 18–45 years, diagnosed with psoriasis, patient's ability to read Bengali and written consent to participate. Patients undergoing any therapy were included after a washout period of 1 month.
Patients with psychiatric diseases, pregnant women, and lactating mothers, cases suffering from uncontrolled systemic illness or life-threatening infections or any vital organ failure, self-reported immune-compromised state, and substance abuse and/or dependence.
Of 118 eligible patients approached, 83 (response rate 70.3%) returned the questionnaire and these responses were subjected to exploratory factor analysis (EFA), and 85 further responses were subjected to confirmatory factor analysis (CFA).
Recommendations for adequate sample size to conduct factor analysis are between 50 and 250 with most authors recommending at least 100 individuals. We calculated the sample size using the subject to item ratio. A minimum of 75 participants were needed (15 items × 5 = 75) based on Gorsuch's formula for which a subject to item ratio 5:1 was used. By taking into account of a 10% attrition rate, the sample size was increased to 83 participants.
Consecutive sampling was adopted. Patients suffering from psoriasis who attended the outpatient clinics on the day of the data collection were consecutively approached and invited to participate in the study.
All the participants were provided with the self-administered questionnaire along with patient information sheets in local vernacular Bengali and informed consents were obtained. To ensure anonymized protection of the patient's privacy, all the patient identifiable information was concealed. Another section in the questionnaire sought information regarding patients' sociodemographic features. Completed questionnaires were put inside opaque envelopes and sealed at the study site. Thirty randomly chosen participants were selected for retest visits at approximately 2–3 weeks' interval to fill the same questionnaire again. All data were extracted in a specially designed Microsoft Excel master chart that was subjected to statistical analysis.
Analyses were performed using IBM® Statistical Package for Social Sciences (SPSS)® software, version 20.0 and SPSS Amos® version 20.0 (IBM Corp., Armonk, NY, USA). First, sampling adequacy was assessed using the Kaiser-Meyer-Olkin (KMO) measure value, and appropriateness of data was conducted using the Bartlett's test of sphericity. The KMO value of more than 0.50 with a significant Bartlett's test of sphericity with a P < 0.05 was considered suitable for factor analysis. Then, EFA using principal component analysis (PCA) with varimax rotation (Eigenvalue more than 1) was conducted to examine the PDI-B dimensionality and construct validity, i.e., the number and type of subscales in the instrument. Only factors with values of ≥0.40 were considered. Items were planned to be excluded if they revealed weak loadings (failing to load above 0.39 on any component) and showing general loadings of 0.40 on more than one component. Next, the reliability of the PDI-B was assessed using internal inconsistency and test-retest reliability analyses. To represent high internal consistencies, Cronbach's alpha of 0.5–0.7 and average item-total correlation in a moderate range between >0.3 and >0.9 were considered as reliable. Cronbach's alpha value of >0.9 was considered as excellent, while correlation near 0 indicated no meaningful construct. Intraclass correlation coefficient (ICC) values of >0.7 indicated that PDI-B was stable over time, values between 0.4 and 0.7 indicated fair reliability, while values of <0.4 indicated poor reliability. Paired t-tests were used to evaluate whether there was statistically significant change in scores on the PDI between the test-retest evaluations. The inter-item correlations between domains (item discriminant validity) and the overall PDI-B (internal item convergence) was assessed using correlation statistics. Correlation value of 0.4 or higher was considered adequate to support the internal consistency of the instrument. Concurrent validity was examined using Pearson's r statistics comparing the total PDI scores with simultaneously measured PASI scores and Dermatological Life Quality Index (DLQI) scores. Cohen's guidelines for the interpretation of correlation coefficients were used with a correlation of 0.10 being small, correlation of 0.30 moderate, and a correlation of 0.50 being large. Finally, a CFA model was developed to verify the a priori scales as suggested by EFA. The goodness of fit of the CFA models were evaluated utilizing the following multiple fit indices: comparative fit index (CFI), normed fit index (NFI), Tucker Lewis Index (TLI), root mean square error of approximation (RMSEA), standardized root mean square residual (SRMR), Bayesian information criterion (BIC), and Hoelter index. The recommendations for cutoff values indicating a good model fit are CFI/TLI ≥0.95, RMSEA ≤0.6, and SRMR ≤0.8., All the statistical tests were identified a priori, and no data imputation was performed for missing data. All statistical tests were two-tailed and were conducted with type I error probability fixed at 0.05.
| Results|| |
Sociodemographic characteristics of the representative sample were presented in terms of six variables – age, sex, body mass index, residence, educational status, and family income status [Table 1].
The mean and median PDI-B item values ranged from 1.1 to 1.9 and from 1.0 to 2.0 respectively, except item number 14 that revealed lower mean and median values of 0.5 and 0, respectively, than other items. Most of the items appeared to be negatively skewed and platykurtic. Few of the items (e.g. item 14) showed floor effect as high as 67.5% and high ceiling effect of 39.8% (e.g. item 6). [Table 2]
|Table 2: Descriptive statistics of 15-item Psoriasis Disability Index-Bengali questionnaire, grouped into the original five subscales (n=83)|
Click here to view
Exploratory factor analysis
The sample size of 83 seemed adequate for factor analysis as the average communalities after extraction was 0.568, above the preferred cutoff of 0.5. The KMO measure of sampling adequacy was 0.836 (Chi-square: 472.687, df = 105, P < 0.001), much greater than the minimum Kaiser criterion of 0.5, indicating the adequacy of the sample and compactness of correlation patterns. A significant Bartlett's test of sphericity also indicated that the R-matrix was not an identity matrix. We performed extraction using the principal component method for determining how many factors best explained the observed covariance matrix within the data set. The scree plot revealed very high eigenvalue for the first three components, and thereafter, the curve began to tail off gradually before the final plateau was reached [Figure 2]. The factor component matrix also supported the scree plot by representing information from initial unrotated solution and extracting three components explaining 56.8% of the total variance [Table 3]. The display of coefficients was sorted by size. Factor loadings were similar to regression weights (or slopes) and indicated the strength of the association between the variables and the factors. Next, the correlation matrix was scanned for values >0.9 for identifying multicollinearity and singularity. The determinant of the correlation matrix was 0.002. Thus, multicollinearity was not a problem. All items correlated fairly well, and none of the correlation coefficients were particularly large; there was no need to consider the elimination of any item at this stage. We expected that the factors were not correlated, and we selected an orthogonal rotation method (i.e., varimax, with Kaiser normalization). The rotated component matrix was a matrix of factor loadings for each variable onto each factor. The absolute values less than a specified value of 0.4 were suppressed, ensuring that factor loadings within ± 0.4 were not displayed in the output. After conducting factor rotation, the items were looked for that loaded onto the same factor. Three subcomponents of the main construct were identified and named as below [Table 4]:
|Table 4: Principal component analysis with rotated component matrix - factor loadings revealing three component structures|
Click here to view
- Items 11, 13, 2, 4, 10, 3, and 8: “Social activities”
- Items 6, 1, 12, 7, 15, and 5: “Concern”
- Items 14 and 9: “Personal activities.”
The Cronbach's alpha value for the overall PDI-B was 0.877, and ICC was 0.882. The Cronbach's alpha values for the three subscales were 0.853, 0.789, and 0.407 indicating acceptable to good reliability [Table 5]. The correlations between the overall and PDI-B subscales were generally higher than correlations between subscales [Table 6].
|Table 5: Internal consistency of the Psoriasis Disability Index-Bengali questionnaire (n=83)|
Click here to view
|Table 6: Correlations matrix between the Psoriasis Disability Index-Bengali components and the overall score (n=83)|
Click here to view
The scores on the PDI-B subscales were largely stable with insignificant mean score differences [Table 7].
|Table 7: Test-retest reliability analysis of the Psoriasis Disability Index-Bengali questionnaire (n=30)|
Click here to view
The PDI-B total score had a moderate to large correlation with the PASI (PDI-B: 22.0 ± 10.7 vs. PASI: 7.8 ± 7.3, r = 0.402, P < 0.01) and DLQI (PDI-B: 22.0 ± 10.7 vs. DLQI: 14.2 ± 7.7, r = 0.719, P < 0.01) scores, thus ensuring acceptable concurrent validity.
Confirmatory factor analysis
The indices of CFA that confirmed model fit were: CFI = 0.971, NFI = 0.789, TLI = 0.965, RMSEA = 0.038, SRMR = 0.081, BIC = 244.046, and Hoelter index (at α 0.05) = 95, thus indicating an excellent model fit and three distinct components [Figure 3].
| Discussion|| |
PDI is a validated English questionnaire consisting of 15 questions framed within five components and assessing QoL in patients suffering from psoriasis; but, until now, no validated Bengali version of the questionnaire is available. The English questionnaire underwent standardized forward-backward translation to produce the PDI-Bengali version. In contrast with the original five subscales English version, EFA using PCA of the PDI-B identified three components and the overall model goodness of fit was further confirmed by CFA. Thus, PDI-B was valid and reliable with adequately high Cronbach's alpha, ICC coefficients, concurrent validity, and test-retest reliability within acceptable limits.
One of the major strengths of this study was to apply EFA and CFA on two different samples as recommended.,, The overall and individual subscales PDI-B scores were comparable to other studies. Usually, floor and ceiling effects between 1% and 15% are considered optimal. However, the inclusion of responses on this basis would have eliminated majority of the questionnaire items; hence, we refrained from prespecifying any cutoff values in this aspect. Thus, there were considerable floor and ceiling effects in this study that might have affected the outcomes. Unlike other validation studies, there was no control (normal/healthy) group; hence, the assessment of item discriminant validity was not possible. Besides, the responsiveness of the questionnaire was not assessed because the treatment offered by the study site was homeopathy exclusively, and that was not an accepted standard treatment for psoriasis. The internal consistency for overall PDI-B is high and comparable to other PDI versions. However, the “personal activities” component showed a fairly low Cronbach's alpha. This was also reflected on the lower correlation between subscales [Table 6]. Alpha depends on the number of items and covariance between items. A score below 0.70 suggests that the items within the tool may not be measuring the same underlying construct, and poorly correlated items need to be deleted. However, as the number of items was not too many, correlation coefficients with the “social activities” and overall score were higher than 0.30, and retaining all the items revealed a fair fit in the CFA model; this, we decided not to eliminate any items. Alpha has very strict assumptions, including unidimensionality, uncorrelated errors, and identical covariance between the items (tau equivalence). In most of the cases, these assumptions are violated, and thus over- or underestimates the true reliability. Thus, alpha may not be the best choice for measuring reliability. The probable alternative may be Guttman's lambda or McDonald's omega, which are not based on tau-equivalence. The inter-item correlations among the subscales were satisfactorily high. Our sample size was similar to the original PDI development and validation study and other translations, but we were unable to confirm the postulated 5-component structure model in the factorial examination; rather our analysis identified three components only, thus implying limited usefulness for routine assessment of changes in psoriasis disability and QoL in West Bengal, probably because of possible redundancies. Further psychometric tests of PDI-B are warranted, especially with regard to validity and sensitivity. Conventionally, a loading of an absolute value >0.3 is considered with a sample size of 300. With a sample size of 83 only, a higher loading was chosen to be significant. Sixty percent (9/15) of the items had a strong factor loadings of 0.60 and above. Second, the PDI-B was administered to the patients who were able to read and understand the Bengali language. Therefore, the findings of this study could only be generalized to Bengali population only. Finally, the three-component model had an excellent fit in CFA. Thus, there is a need to translate and validate this questionnaire into other Indian languages and on larger sample to give better utilization in a multiethnic Indian population. Other limitations include the consecutive sampling method used in this study, which may be vulnerable to sampling bias.
The validated PDI-B is an important patient-reported instrument to measure psoriasis symptom severity and QOL impact. The responsiveness, or sensitivity to change, of the PDI-B to measure symptoms and life-impact of treatments needs to be tested in future investigations. Finally, in order to confirm the PDI-B can measure the impact of clinical treatment, the final step in this instrument's development will be to define a minimally important difference of change in this scale that reflects a clinically meaningful difference. Further validation using a larger sample size and more specific Rasch analyses is warranted to check the goodness of fit and confirm whether the sequence of the questionnaire requires any readjustment in future.
| Conclusion|| |
The PDI-B contains 15 items which are framed within 3-component model. It is a reasonably valid and reliable tool which can be used to measure the QoL in Bengali patients suffering from psoriasis. However, further analysis is recommended to strengthen the validity of the PDI-B.
We would like to acknowledge the institutional heads for allowing us to conduct the study. We are also grateful to the patients for their participation.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
de Korte J, Sprangers MA, Mombers FM, Bos JD. Quality of life in patients with psoriasis: A systematic literature review. J Investig Dermatol Symp Proc 2004;9:140-7.
Stern RS, Nijsten T, Feldman SR, Margolis DJ, Rolstad T. Psoriasis is common, carries a substantial burden even when not extensive, and is associated with widespread treatment dissatisfaction. J Investig Dermatol Symp Proc 2004;9:136-9.
Akay A, Pekcanlar A, Bozdag KE, Altintas L, Karaman A. Assessment of depression in subjects with psoriasis vulgaris and lichen planus. J Eur Acad Dermatol Venereol 2002;16:347-52.
Gupta MA, Schork NJ, Gupta AK, Kirkby S, Ellis CN. Suicidal ideation in psoriasis. Int J Dermatol 1993;32:188-90.
Heydendael VM, de Borgie CA, Spuls PI, Bossuyt PM, Bos JD, de Rie MA, et al
. The burden of psoriasis is not determined by disease severity only. J Investig Dermatol Symp Proc 2004;9:131-5.
Finlay AY, Coles EC. The effect of severe psoriasis on the quality of life of 369 patients. Br J Dermatol 1995;132:236-44.
Lewis VJ, Finlay AY. Two decades experience of the psoriasis disability index. Dermatology 2005;210:261-8.
Preacher KJ, MacCallum RC. Exploratory factor analysis in behavior genetics research: Factor recovery with small sample sizes. Behav Genet 2002;32:153-61.
Gorsuch RL. Factor analysis. 2nd
ed. Hillsdale: Lawrence Erlbaum Associates; 1983.
Bollen KA. A new incremental fit index for general structural equation models. Sociol Methods Res 1989;17:303-16.
Sitzia J. How valid and reliable are patient satisfaction data? An analysis of 195 studies. Int J Qual Health Care 1999;11:319-28.
Streiner DL, Norman GR. Health Measurement Scales: A Practical Guide to their Development and Use. Oxford: Oxford University Press; 2008.
Cicchetti DV. Guidelines, criteria, and rules of thumb for evaluating normed and standardized assessment instruments in psychology. Psychol Assess 1994;6:284-90.
Cohen J. Statistical Power Analysis for the Behavorial Sciences. 2nd
ed. Hillsdale, NJ: Lawrence Erlbaum Associates; 1977.
Hu L, Bentler P. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Struct Equ Modeling 1999;6:1-55.
Brown TA. Confirmatory Factor Analysis for Applied Research. New York: Guilford Press; 2006.
Costello A, Osborne J. Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Pract Assess Res Evaluation 2005;10. Available at: https://pareonline.net/pdf/v10n7.pdf
. [Last accessed on 2018 Dec 12].
Henson R, Roberts J. Use of exploratory factor analysis in published research: Common errors and some comment on improved practice. Educ Psychol Meas 2006;66:393-416.
Worthington R, Whittaker T. Scale development research: A content analysis and recommendations for best practices. Couns Psychol 2006;34:806-38.
McHorney CA, Tarlov AR. Individual-patient monitoring in clinical practice: Are available health status surveys adequate? Qual Life Res 1995;4:293-307.
Dunn TJ, Baguley T, Brunsden V. from alpha to omega: A practical solution to the pervasive problem of internal consistency estimation. Br J Psychol. 2014;105:399-412.
Dunn TJ, Baguley T, Brunsden V. From alpha to omega: A practical solution to the pervasive problem of internal consistency estimation. Br J Psychol 2014;105:399-412.
[Figure 1], [Figure 2], [Figure 3]
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5], [Table 6], [Table 7]