- Email: [email protected]
Cultural Adaptation of the WHOQOL Questionnaire for Taiwan
BRIEF COMMUNICATION
Cultural Adaptation of the WHOQOL Questionnaire for Taiwan Grace Yao,1* Jung-Der Wang,2,3 Chih-Wen Chung2 This brief report describes cultural adaptation of the WHOQOL-100 questionnaire for Taiwan. The standard cultural adaptation procedure was used for questionnaire translation, response scale generation, and national items design. A field test was conducted on 1068 participants randomly selected from 17 hospitals all over Taiwan. Several psychometric criteria were used for selecting 12 out of the 20 national items previously proposed from three focus groups. In terms of reliability and validity, the psychometric properties of the culturally adapted WHOQOL questionnaire are good and the values are comparable with those in studies conducted in 15 countries. [J Formos Med Assoc 2007;106(7):592–597] Key Words: cultural adaptation, Taiwan, WHOQOL
In the last 20 years, health care providers and researchers have agreed that treatment intervention should be evaluated in part by its impact on healthrelated quality of life (QOL). Although people have been measuring QOL for many years, there is as yet little agreement on a definition.1,2 There is a need to conduct global studies on a wide range of disease groups in a wide range of cultures. Towards this end, the World Health Organization (WHO) has been developing a universal measure of QOL called the WHOQOL Questionnaire,3,4 which can be accepted and applied cross-culturally. The WHO began developing this questionnaire for generic use in 1991 and finished field tests in 1995. The WHOQOL questionnaire contains 24 facets organized into six broad domains: physical, psychological, level of independence, social relationships, environment, and spirituality/religion/ personal beliefs. Each facet contains four items. Four additional items measure “overall QOL and general health” (forming Facet-G). The final
version of the questionnaire (called the WHOQOL100) contains 100 culturally comparable items. However, each culture may add culture-specific questions, called national items, so that the questionnaire can also reflect cultural attributes. The purpose of this report is to document adaptation of the WHOQOL questionnaire for Taiwan. Specifically, the procedure for national item selection is presented.
Methods Questionnaire translation To adapt the WHOQOL questionnaire to Taiwan, the WHOQOL-Taiwan group composed of diverse backgrounds followed WHO recommendations for translation procedures of health status instruments.5,6 The Taiwan version is expected to be conceptually, semantically and technically equivalent to the standard form.7
©2007 Elsevier & Formosan Medical Association .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
Department of Psychology, College of Science, and 2Institute of Occupational Medicine and Industrial Hygiene, College of Public Health, National Taiwan University, and 3Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan. Received: April 24, 2006 Revised: December 28, 2006 Accepted: February 6, 2007
592
*Correspondence to: Dr Grace Yao, Department of Psychology, National Taiwan University, No. 1, Section 4, Roosevelt Road, Taipei 106, Taiwan. E-mail: [email protected]
J Formos Med Assoc | 2007 • Vol 106 • No 7
Cultural adaptation of the WHOQOL for Taiwan
Response scale generation The items in the standard questionnaire use four types of 5-point Likert response scales (i.e. capacity, frequency, intensity, and evaluation). According to WHOQOL studies, the anchor points should be universally applicable, but the three categories between the anchor points may need to be adjusted according to the culture.8 To obtain culturally appropriate equal-interval scale descriptors, we followed WHOQOL procedures3,5,8 to conduct a study on 156 participants who were randomly selected from 10 hospitals in Taiwan. As a result, three intermediate descriptors for each scale type were selected.9
Design of national items According to the documents of the WHOQOL,6,10 three focus groups consisting of health professionals, patients with a variety of diseases and health conditions, and patients’ caregivers met to discuss whether the ordinary six domains/24 facets/100 items were sufficient to describe their personal views with respect to QOL. Appropriate domains/ facets/items were proposed to compensate for insufficient areas, resulting in 20 proposed national items (available upon request). Of the 20 items, 11 were classified into two new facets: being respected/accepted (guanxi/mianzi) and eating/food. We hypothesized that the two new facets should be classified into the social relationships and environmental domains, respectively. The other nine were related to dependence on alternative medicine (mostly traditional Chinese medicine) and national health insurance (Taiwan’s nationally subsidized health care program), personal fortune, fulfillment of personal goals, support from a significant other, and maintaining a good relationship with relatives. These nine items were classified into existing facets and domains. Standard scoring system was applied. A higher score indicates a better QOL. Criteria for selection of national items According to the WHOQOL Group,5 the psychometric properties of new items should be comparable to the global data and to data collected in the J Formos Med Assoc | 2007 • Vol 106 • No 7
host country. That is, the psychometric properties of new items should be at least as good as existing items in the same domain and facet. We conducted several analyses to select appropriate national items from the proposed 20. The quantitative method and selection criteria were as follows: a. Items with means between 2 and 4 were selected. Items with larger variance have better discriminative power. b. A high exploratory factor analysis (EFA) factor loading on the proposed facet or domain indicates the item is classified correctly. An EFA was conducted on the items in each new facet and the existing items in the same domain to test whether the newly proposed facets could be extracted. An EFA was also conducted on the 24 ordinary facet scores plus the two new facet scores to see which domains and factors the two new facets belonged to. c. High correlation coefficients between items and their hypothesized domains/facets indicate that the items are classified correctly. d. Higher correlation coefficients between the items and the Facet-G score, the individual items in Facet-G, and the Total-QOL score indicates that the items make a larger contribution to predicting general QOL. e. Stepwise regression analysis was conducted to predict the Total-QOL score from items in the same facet. Better items should be included in the regression equation with larger standardized regression coefficients (beta) and higher predictive power (larger % of total variance explained). f. If the Cronbach’s alpha of a facet decreases significantly after deleting an item from the facet, then the item is important and should not be deleted. g. Independent t tests were conducted on the items to test for ability to distinguish healthy from unhealthy participants. h. Cluster analysis (CA) was used to investigate the similarity among items. Items with greater similarity are more likely to be clustered together. A good new item should not be hierarchically clustered too early with the original 593
G. Yao, et al
four items in the same facet, which would indicate too much overlap.11 However, the item should also not be clustered too late, which would indicate that the item is very different from those in the same facet. In this study, we conducted CA using average linkage. Both dendrogram and horizontal icicle plots of the items in the same facet were examined.
i. Multidimensional scaling (MDS) also identifies similarities among items. Items with greater similarity show smaller geometric distance. We first conducted one dimensional (1-D) nonmetric MDS to check the fit of the new item(s) in an existing facet. A model fit index (stress) of less than 0.10 indicated that the items were in the same dimension. A good item should be
Table 1. Demographic data* Unhealthy subjects
Healthy subjects
Total
385 (51.2) 362 (48.1) 5 (0.7)
145 (45.9) 168 (53.2) 3 (0.9)
530 (49.6) 530 (49.6) 8 (0.8)
42.19 ± 15.42 (17–89)
40.12 ± 13.21 (18–75)
41.57 ± 14.83 (17–89)
Education Illiterate Primary school Middle high school High school College Graduate Other Missing
8 (1.1) 111 (14.8) 103 (13.7) 243 (32.3) 243 (32.3) 30 (4.0) 3 (0.4) 11 (1.5)
1 (0.3) 18 (5.7) 37 (11.7) 109 (34.5) 126 (39.9) 19 (6.0) 1 (0.3) 5 (1.6)
9 (0.8) 129 (12.1) 140 (13.1) 352 (33.0) 369 (34.6) 49 (4.6) 4 (0.4) 16 (1.5)
Marital status Single Married/living together Divorced/separated Widowed Other Missing
180 (23.9) 493 (65.6) 29 (3.9) 19 (2.5) 4 (0.5) 27 (3.6)
76 (24.1) 220 (69.6) 14 (4.4) 4 (1.3) 0 (0.0) 2 (0.6)
256 (24.0) 713 (66.8) 43 (4.0) 23 (2.2) 4 (0.4) 29 (2.7)
Current health status† Very bad Bad Not bad/not good Good Very good Missing
63 (8.4) 135 (18.0) 296 (39.4) 205 (27.3) 35 (4.7) 18 (2.4)
3 (0.9) 15 (4.7) 95 (30.1) 160 (50.6) 41 (13.0) 2 (0.6)
66 (6.2) 150 (14.0) 391 (36.6) 365 (34.2) 76 (7.1) 20 (1.9)
Interview method Self-administered Interviewer-assisted Interviewer-administered Missing
636 (84.6) 65 (8.6) 36 (4.8) 15 (2.0)
306 (96.8) 9 (2.8) 0 (0.0) 1 (0.3)
942 (88.2) 74 (6.9) 36 (3.4) 16 (1.5)
Gender Male Female Missing Age (yr)
*All data are presented as n (%), except for age which is presented as mean ± standard deviation (range); †5-point Likert scale.
594
J Formos Med Assoc | 2007 • Vol 106 • No 7
J Formos Med Assoc | 2007 • Vol 106 • No 7
O O O O O O O O O O O O O O O O O O O O
Average
O O O X O O X X O O O O O O O O O O O O
SD
O O O O O O X X O O O O O O O O O O O X
EFA O O O O O O X X O O O O O O O O X O O X
I-D O O O O O O X X O O O O O O O O O O O X
C. I-F X O O O O O X X O O O O O O O O O O O X
I-G O O O O O O X O O O O O O O O O O O O X
G1 X O O O O O X X O O O O O O O O O O O X
G2 X O O O O O X X O O O O O O O O O O O X
G3 X O O O O O X X O O O O O O O O O O O O
G4 X X O O O O X X O O O O O O O O O O O X
I-Total QOL X X O O X O X X O X X O X O X O O X X O
Step reg O O O O O O X X O O O O O O O O X O O X
α↓ X O X X X O O O O O O X X O O O O O O O
DV O O O O O O O O O O O O O O O O O O O O
CA
2D/3D MDS centrality X O X O O O X X O O O O O O O O O O O O
1D MDS fit X† X† O O O O O O X† X† X† X† X† X† O O O O O X†
8/17 14/17 15/17 15/17 15/17 17/17 4/17 5/17 16/17 15/17 15/17 14/17 14/17 16/17 16/17 17/17 15/17 16/17 16/17 7/17
Passing rate
*National items selected for the Taiwan version; †represents achieving goodness of fit after deleting the item(s) in the facet. SD = standard deviation; EFA = exploratory factor analysis; I-D = item–domain correlation; C. I-F = corrected item–facet correlation; I-G = item–Facet G correlation; G1–G4 = item–Gi correlation (I = 1,2,3,4); I-Total QOL = item–overall quality of life correlation; Step reg = stepwise regression; a↓ = Cronbach’s a↓ after deleting the item; DV = discriminant validity; CA = cluster analysis; 1D = 1-dimensional; 2D/3D = 2-dimensional/3-dimensional; MDS = multidimensional scaling; O = pass; X = fail.
N1 N2* N3* N4 N5* N6* N7 N8 N9* N10* N11 N12 N13* N14* N15* N16* N17 N18* N19* N20
Item no.
Table 2. Summary of national item selection
Cultural adaptation of the WHOQOL for Taiwan
595
G. Yao, et al
geometrically in the center of the original four items in the same facet (centrality), which indicates that the item is spatially and conceptually appropriate with the items in the same facet.11,12 Multidimensional (2-D or 3-D) nonmetric MDS was also conducted to check the centrality of new items in an existing facet. j. When several items matched the described selection criteria, item selection was based on researchers’ subjective judgment. For example, researchers compared the conceptual and semantic meanings of items to select the most appropriate one.
plus the 20 proposed national items on 1068 participants randomly selected from 17 hospitals all over Taiwan. Participants included inpatients, outpatients, health professionals, and volunteers with a good mix of age, gender, social background, and heath status. Table 1 presents the demographic data. The proportions of the demographic data fit the WHOQOL protocol requirements.8
Results Selection of national items
Field test We conducted the field test using a questionnaire with the 100 standard WHOQOL items
Table 2 summarizes the results of the 17 criteria (the detailed results are available upon request). The last column is the passing rate. A higher rate
Table 3. Summary of the psychometric results of the WHOQOL-Taiwan (long) version, including the 12 national items Reliability Internal consistency1
Test–retest reliability2 (all p < 0.01)
Validity Content validity3 (all p < 0.01)
Discriminant validity4 Concurrent validity5 (all p < 0.01) Prediction validity6 Construct validity7
At facet level: 0.59–0.92 At domain level: 0.78–0.92 The whole questionnaire: 0.97 At item level: 0.36–0.78 At facet level: 0.68–0.85 At domain level: 0.75–0.91 The whole questionnaire: 0.86
Item & hypothesized facet: 0.57–0.91 Item & hypothesized domain: 0.41–0.85 Facet & hypothesized domain: 0.58–0.84 (excluding F24 = D6) Inter-domain: 0.22–0.68 Domains & Total QOL: 0.64–0.85 Significant differences were found between healthy and unhealthy subjects on most of the items, facets, domains, and Total QOL score Domain and the corresponding visual analog rating scale: 0.49–0.62 64.2% Facet-G score variance was explained by domain scores EFA: four factors (physical health, psychological, social, environmental) were extracted, 58.3% of the total variance was explained CFA: four-factor model is the most plausible model, CFI = 0.86
Analytical methods: 1Cronbach’s α at the facet and domain levels and the whole questionnaire. 2The correlations of the items/facets/ domains, and the whole questionnaire at 2–4 week intervals on 158 subjects. 3The correlations between the item and the hypothesized facet/domain, inter-domain, and facet–Total QOL score. 4t tests to distinguish between healthy and unhealthy subjects. 5Correlations between domains and their corresponding 100-point visual analog score. For example, psychological domain score was correlated with the 100-point visual analog item “In general, how satisfied are you with your psychological health”. 6Conducting multiple regression by using domain scores as the predictor variable and the Facet-G score as the criterion. 7EFA: principal factor analysis + promax rotation; CFA: using EQS software, three (one-factor, four-factor, and six-factor) models were analyzed. QOL = quality of life; EFA = exploratory factor analysis; CFA = confirmatory factor analysis.
596
J Formos Med Assoc | 2007 • Vol 106 • No 7
Cultural adaptation of the WHOQOL for Taiwan
indicates a better item. Twelve items were selected for the WHOQOL-Taiwan version. Four of these were selected for each of the two new facets, and the other four were for existing facets. The two new facets, “Being respected/accepted (guanxi/mianzi)” and “Eating/food” were classified into the social relationships and environmental domains, respectively.7 These results agreed with our hypothesis.
Psychometric properties of the WHOQOL-Taiwan version The methods for validating the psychologic properties of the questionnaire are briefly described in Table 3. Table 3 summarizes the results of the standard psychometric analyses for reliability and validity conducted on the WHOQOL-Taiwan version including the 12 new national items and two new facets. Its psychometric properties are good and the values are comparable to those in other studies.7,13,14 As with other studies, the original six domains can be further simplified into four domains: physical health, psychological, social relationships, and environment (see the EFA and CFA results). The WHOQOL-Taiwan version is a reliable and valid assessment instrument for measuring QOL in Taiwan.
Conclusion This brief report describes the cultural adaptation of the standard WHOQOL-100 questionnaire for Taiwan. We followed the standard adaptation procedure on questionnaire translation, response scale generation, and national items design. The results showed that the WHOQOL-Taiwan version is a reliable and valid QOL measurement for Taiwanese. Readers interested in this questionnaire can contact the first author of this report for detailed information.
Acknowledgments This study was supported by grants from the National Science Council (NSC 87-2312-B-002-001, J Formos Med Assoc | 2007 • Vol 106 • No 7
NSC 87-2413-H-002-021, NSC 88-2314-B-002-344, NSC 89-2312-B-002-001, NSC 94-2413-H-002018, NSC 95-2413-H-002-002, NHRI-EX949204PP).
References 1. Hunt SM. The problem of quality of life. Qual Life Res 1997;6:205–12. 2. Spitzer B. Quality of Life Assessments in Clinical Trials. New York: Raven Press, 1998. 3. The WHOQOL Group. The development of the World Health Organization Quality of Life assessment instrument (the WHOQOL). In: Orley J, Kuyen W (eds). Quality of Life Assessment: International Perspectives. Berlin: SpringerVerlag, 1994:41–57. 4. The WHOQOL Group. The World Health Organization Quality of Life assessment (WHOQOL): position paper from the World Health Organization. Soc Sci Med 1995;41: 1403–9. 5. World Health Organization. Resources for New WHOQOL Centers. Geneva: WHO (MNH/PSF/95.2), 1995. 6. World Health Organization. WHOQOL Protocol for New Centers. Geneva: WHO (MNH/PSF/94.4), 1994. 7. The WHOQOL-Taiwan Group. The User’s Manual of the Development of the WHOQOL-100 Taiwan Version, 2nd edition. Taipei: WHOQOL-Taiwan Group, 2005. [In Chinese] 8. World Health Organization. WHOQOL Study Protocol. Geneva: WHO (MNH/PSF/93.9), 1993. 9. Lin MR, Yao KP, Hwang JS, Wang JD. Scale descriptor selection for Taiwan-version of questionnaire of World Health Organization Quality of Life. Chin J Public Health (Taipei) 1999;18:262–70. [In Chinese] 10. World Health Organization. WHOQOL Focus Group Moderator Training. Geneva: WHO (MNH/PSF/92.9), 1992. 11. Skevington SM, Bradshaw J, Saxena S. Selecting national items for the WHOQOL: conceptual and psychometric considerations. Soc Sci Med 1999;48:473–87. 12. Kemmler G, Holzner B. Multidimensional scaling as a tool for uncovering the internal structure of quality of life instruments. Qual Life Res 1997;6:435. 13. Power M, Bullinger M, Harper A, The WHOQOL Group. The World Health Organization WHOQOL-100: tests of the universality of quality of life in 15 different cultural groups worldwide. Health Psychol 1999;18: 495–505. 14. The WHOQOL Group. The World Health Organization Quality of Life assessment (WHOQOL): development and general psychometric properties. Soc Sci Med 1998;46: 1569–85.
597
Cultural Adaptation of the WHOQOL Questionnaire for Taiwan Grace Yao,1* Jung-Der Wang,2,3 Chih-Wen Chung2 This brief report describes cultural adaptation of the WHOQOL-100 questionnaire for Taiwan. The standard cultural adaptation procedure was used for questionnaire translation, response scale generation, and national items design. A field test was conducted on 1068 participants randomly selected from 17 hospitals all over Taiwan. Several psychometric criteria were used for selecting 12 out of the 20 national items previously proposed from three focus groups. In terms of reliability and validity, the psychometric properties of the culturally adapted WHOQOL questionnaire are good and the values are comparable with those in studies conducted in 15 countries. [J Formos Med Assoc 2007;106(7):592–597] Key Words: cultural adaptation, Taiwan, WHOQOL
In the last 20 years, health care providers and researchers have agreed that treatment intervention should be evaluated in part by its impact on healthrelated quality of life (QOL). Although people have been measuring QOL for many years, there is as yet little agreement on a definition.1,2 There is a need to conduct global studies on a wide range of disease groups in a wide range of cultures. Towards this end, the World Health Organization (WHO) has been developing a universal measure of QOL called the WHOQOL Questionnaire,3,4 which can be accepted and applied cross-culturally. The WHO began developing this questionnaire for generic use in 1991 and finished field tests in 1995. The WHOQOL questionnaire contains 24 facets organized into six broad domains: physical, psychological, level of independence, social relationships, environment, and spirituality/religion/ personal beliefs. Each facet contains four items. Four additional items measure “overall QOL and general health” (forming Facet-G). The final
version of the questionnaire (called the WHOQOL100) contains 100 culturally comparable items. However, each culture may add culture-specific questions, called national items, so that the questionnaire can also reflect cultural attributes. The purpose of this report is to document adaptation of the WHOQOL questionnaire for Taiwan. Specifically, the procedure for national item selection is presented.
Methods Questionnaire translation To adapt the WHOQOL questionnaire to Taiwan, the WHOQOL-Taiwan group composed of diverse backgrounds followed WHO recommendations for translation procedures of health status instruments.5,6 The Taiwan version is expected to be conceptually, semantically and technically equivalent to the standard form.7
©2007 Elsevier & Formosan Medical Association .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
Department of Psychology, College of Science, and 2Institute of Occupational Medicine and Industrial Hygiene, College of Public Health, National Taiwan University, and 3Department of Internal Medicine, National Taiwan University Hospital, Taipei, Taiwan. Received: April 24, 2006 Revised: December 28, 2006 Accepted: February 6, 2007
592
*Correspondence to: Dr Grace Yao, Department of Psychology, National Taiwan University, No. 1, Section 4, Roosevelt Road, Taipei 106, Taiwan. E-mail: [email protected]
J Formos Med Assoc | 2007 • Vol 106 • No 7
Cultural adaptation of the WHOQOL for Taiwan
Response scale generation The items in the standard questionnaire use four types of 5-point Likert response scales (i.e. capacity, frequency, intensity, and evaluation). According to WHOQOL studies, the anchor points should be universally applicable, but the three categories between the anchor points may need to be adjusted according to the culture.8 To obtain culturally appropriate equal-interval scale descriptors, we followed WHOQOL procedures3,5,8 to conduct a study on 156 participants who were randomly selected from 10 hospitals in Taiwan. As a result, three intermediate descriptors for each scale type were selected.9
Design of national items According to the documents of the WHOQOL,6,10 three focus groups consisting of health professionals, patients with a variety of diseases and health conditions, and patients’ caregivers met to discuss whether the ordinary six domains/24 facets/100 items were sufficient to describe their personal views with respect to QOL. Appropriate domains/ facets/items were proposed to compensate for insufficient areas, resulting in 20 proposed national items (available upon request). Of the 20 items, 11 were classified into two new facets: being respected/accepted (guanxi/mianzi) and eating/food. We hypothesized that the two new facets should be classified into the social relationships and environmental domains, respectively. The other nine were related to dependence on alternative medicine (mostly traditional Chinese medicine) and national health insurance (Taiwan’s nationally subsidized health care program), personal fortune, fulfillment of personal goals, support from a significant other, and maintaining a good relationship with relatives. These nine items were classified into existing facets and domains. Standard scoring system was applied. A higher score indicates a better QOL. Criteria for selection of national items According to the WHOQOL Group,5 the psychometric properties of new items should be comparable to the global data and to data collected in the J Formos Med Assoc | 2007 • Vol 106 • No 7
host country. That is, the psychometric properties of new items should be at least as good as existing items in the same domain and facet. We conducted several analyses to select appropriate national items from the proposed 20. The quantitative method and selection criteria were as follows: a. Items with means between 2 and 4 were selected. Items with larger variance have better discriminative power. b. A high exploratory factor analysis (EFA) factor loading on the proposed facet or domain indicates the item is classified correctly. An EFA was conducted on the items in each new facet and the existing items in the same domain to test whether the newly proposed facets could be extracted. An EFA was also conducted on the 24 ordinary facet scores plus the two new facet scores to see which domains and factors the two new facets belonged to. c. High correlation coefficients between items and their hypothesized domains/facets indicate that the items are classified correctly. d. Higher correlation coefficients between the items and the Facet-G score, the individual items in Facet-G, and the Total-QOL score indicates that the items make a larger contribution to predicting general QOL. e. Stepwise regression analysis was conducted to predict the Total-QOL score from items in the same facet. Better items should be included in the regression equation with larger standardized regression coefficients (beta) and higher predictive power (larger % of total variance explained). f. If the Cronbach’s alpha of a facet decreases significantly after deleting an item from the facet, then the item is important and should not be deleted. g. Independent t tests were conducted on the items to test for ability to distinguish healthy from unhealthy participants. h. Cluster analysis (CA) was used to investigate the similarity among items. Items with greater similarity are more likely to be clustered together. A good new item should not be hierarchically clustered too early with the original 593
G. Yao, et al
four items in the same facet, which would indicate too much overlap.11 However, the item should also not be clustered too late, which would indicate that the item is very different from those in the same facet. In this study, we conducted CA using average linkage. Both dendrogram and horizontal icicle plots of the items in the same facet were examined.
i. Multidimensional scaling (MDS) also identifies similarities among items. Items with greater similarity show smaller geometric distance. We first conducted one dimensional (1-D) nonmetric MDS to check the fit of the new item(s) in an existing facet. A model fit index (stress) of less than 0.10 indicated that the items were in the same dimension. A good item should be
Table 1. Demographic data* Unhealthy subjects
Healthy subjects
Total
385 (51.2) 362 (48.1) 5 (0.7)
145 (45.9) 168 (53.2) 3 (0.9)
530 (49.6) 530 (49.6) 8 (0.8)
42.19 ± 15.42 (17–89)
40.12 ± 13.21 (18–75)
41.57 ± 14.83 (17–89)
Education Illiterate Primary school Middle high school High school College Graduate Other Missing
8 (1.1) 111 (14.8) 103 (13.7) 243 (32.3) 243 (32.3) 30 (4.0) 3 (0.4) 11 (1.5)
1 (0.3) 18 (5.7) 37 (11.7) 109 (34.5) 126 (39.9) 19 (6.0) 1 (0.3) 5 (1.6)
9 (0.8) 129 (12.1) 140 (13.1) 352 (33.0) 369 (34.6) 49 (4.6) 4 (0.4) 16 (1.5)
Marital status Single Married/living together Divorced/separated Widowed Other Missing
180 (23.9) 493 (65.6) 29 (3.9) 19 (2.5) 4 (0.5) 27 (3.6)
76 (24.1) 220 (69.6) 14 (4.4) 4 (1.3) 0 (0.0) 2 (0.6)
256 (24.0) 713 (66.8) 43 (4.0) 23 (2.2) 4 (0.4) 29 (2.7)
Current health status† Very bad Bad Not bad/not good Good Very good Missing
63 (8.4) 135 (18.0) 296 (39.4) 205 (27.3) 35 (4.7) 18 (2.4)
3 (0.9) 15 (4.7) 95 (30.1) 160 (50.6) 41 (13.0) 2 (0.6)
66 (6.2) 150 (14.0) 391 (36.6) 365 (34.2) 76 (7.1) 20 (1.9)
Interview method Self-administered Interviewer-assisted Interviewer-administered Missing
636 (84.6) 65 (8.6) 36 (4.8) 15 (2.0)
306 (96.8) 9 (2.8) 0 (0.0) 1 (0.3)
942 (88.2) 74 (6.9) 36 (3.4) 16 (1.5)
Gender Male Female Missing Age (yr)
*All data are presented as n (%), except for age which is presented as mean ± standard deviation (range); †5-point Likert scale.
594
J Formos Med Assoc | 2007 • Vol 106 • No 7
J Formos Med Assoc | 2007 • Vol 106 • No 7
O O O O O O O O O O O O O O O O O O O O
Average
O O O X O O X X O O O O O O O O O O O O
SD
O O O O O O X X O O O O O O O O O O O X
EFA O O O O O O X X O O O O O O O O X O O X
I-D O O O O O O X X O O O O O O O O O O O X
C. I-F X O O O O O X X O O O O O O O O O O O X
I-G O O O O O O X O O O O O O O O O O O O X
G1 X O O O O O X X O O O O O O O O O O O X
G2 X O O O O O X X O O O O O O O O O O O X
G3 X O O O O O X X O O O O O O O O O O O O
G4 X X O O O O X X O O O O O O O O O O O X
I-Total QOL X X O O X O X X O X X O X O X O O X X O
Step reg O O O O O O X X O O O O O O O O X O O X
α↓ X O X X X O O O O O O X X O O O O O O O
DV O O O O O O O O O O O O O O O O O O O O
CA
2D/3D MDS centrality X O X O O O X X O O O O O O O O O O O O
1D MDS fit X† X† O O O O O O X† X† X† X† X† X† O O O O O X†
8/17 14/17 15/17 15/17 15/17 17/17 4/17 5/17 16/17 15/17 15/17 14/17 14/17 16/17 16/17 17/17 15/17 16/17 16/17 7/17
Passing rate
*National items selected for the Taiwan version; †represents achieving goodness of fit after deleting the item(s) in the facet. SD = standard deviation; EFA = exploratory factor analysis; I-D = item–domain correlation; C. I-F = corrected item–facet correlation; I-G = item–Facet G correlation; G1–G4 = item–Gi correlation (I = 1,2,3,4); I-Total QOL = item–overall quality of life correlation; Step reg = stepwise regression; a↓ = Cronbach’s a↓ after deleting the item; DV = discriminant validity; CA = cluster analysis; 1D = 1-dimensional; 2D/3D = 2-dimensional/3-dimensional; MDS = multidimensional scaling; O = pass; X = fail.
N1 N2* N3* N4 N5* N6* N7 N8 N9* N10* N11 N12 N13* N14* N15* N16* N17 N18* N19* N20
Item no.
Table 2. Summary of national item selection
Cultural adaptation of the WHOQOL for Taiwan
595
G. Yao, et al
geometrically in the center of the original four items in the same facet (centrality), which indicates that the item is spatially and conceptually appropriate with the items in the same facet.11,12 Multidimensional (2-D or 3-D) nonmetric MDS was also conducted to check the centrality of new items in an existing facet. j. When several items matched the described selection criteria, item selection was based on researchers’ subjective judgment. For example, researchers compared the conceptual and semantic meanings of items to select the most appropriate one.
plus the 20 proposed national items on 1068 participants randomly selected from 17 hospitals all over Taiwan. Participants included inpatients, outpatients, health professionals, and volunteers with a good mix of age, gender, social background, and heath status. Table 1 presents the demographic data. The proportions of the demographic data fit the WHOQOL protocol requirements.8
Results Selection of national items
Field test We conducted the field test using a questionnaire with the 100 standard WHOQOL items
Table 2 summarizes the results of the 17 criteria (the detailed results are available upon request). The last column is the passing rate. A higher rate
Table 3. Summary of the psychometric results of the WHOQOL-Taiwan (long) version, including the 12 national items Reliability Internal consistency1
Test–retest reliability2 (all p < 0.01)
Validity Content validity3 (all p < 0.01)
Discriminant validity4 Concurrent validity5 (all p < 0.01) Prediction validity6 Construct validity7
At facet level: 0.59–0.92 At domain level: 0.78–0.92 The whole questionnaire: 0.97 At item level: 0.36–0.78 At facet level: 0.68–0.85 At domain level: 0.75–0.91 The whole questionnaire: 0.86
Item & hypothesized facet: 0.57–0.91 Item & hypothesized domain: 0.41–0.85 Facet & hypothesized domain: 0.58–0.84 (excluding F24 = D6) Inter-domain: 0.22–0.68 Domains & Total QOL: 0.64–0.85 Significant differences were found between healthy and unhealthy subjects on most of the items, facets, domains, and Total QOL score Domain and the corresponding visual analog rating scale: 0.49–0.62 64.2% Facet-G score variance was explained by domain scores EFA: four factors (physical health, psychological, social, environmental) were extracted, 58.3% of the total variance was explained CFA: four-factor model is the most plausible model, CFI = 0.86
Analytical methods: 1Cronbach’s α at the facet and domain levels and the whole questionnaire. 2The correlations of the items/facets/ domains, and the whole questionnaire at 2–4 week intervals on 158 subjects. 3The correlations between the item and the hypothesized facet/domain, inter-domain, and facet–Total QOL score. 4t tests to distinguish between healthy and unhealthy subjects. 5Correlations between domains and their corresponding 100-point visual analog score. For example, psychological domain score was correlated with the 100-point visual analog item “In general, how satisfied are you with your psychological health”. 6Conducting multiple regression by using domain scores as the predictor variable and the Facet-G score as the criterion. 7EFA: principal factor analysis + promax rotation; CFA: using EQS software, three (one-factor, four-factor, and six-factor) models were analyzed. QOL = quality of life; EFA = exploratory factor analysis; CFA = confirmatory factor analysis.
596
J Formos Med Assoc | 2007 • Vol 106 • No 7
Cultural adaptation of the WHOQOL for Taiwan
indicates a better item. Twelve items were selected for the WHOQOL-Taiwan version. Four of these were selected for each of the two new facets, and the other four were for existing facets. The two new facets, “Being respected/accepted (guanxi/mianzi)” and “Eating/food” were classified into the social relationships and environmental domains, respectively.7 These results agreed with our hypothesis.
Psychometric properties of the WHOQOL-Taiwan version The methods for validating the psychologic properties of the questionnaire are briefly described in Table 3. Table 3 summarizes the results of the standard psychometric analyses for reliability and validity conducted on the WHOQOL-Taiwan version including the 12 new national items and two new facets. Its psychometric properties are good and the values are comparable to those in other studies.7,13,14 As with other studies, the original six domains can be further simplified into four domains: physical health, psychological, social relationships, and environment (see the EFA and CFA results). The WHOQOL-Taiwan version is a reliable and valid assessment instrument for measuring QOL in Taiwan.
Conclusion This brief report describes the cultural adaptation of the standard WHOQOL-100 questionnaire for Taiwan. We followed the standard adaptation procedure on questionnaire translation, response scale generation, and national items design. The results showed that the WHOQOL-Taiwan version is a reliable and valid QOL measurement for Taiwanese. Readers interested in this questionnaire can contact the first author of this report for detailed information.
Acknowledgments This study was supported by grants from the National Science Council (NSC 87-2312-B-002-001, J Formos Med Assoc | 2007 • Vol 106 • No 7
NSC 87-2413-H-002-021, NSC 88-2314-B-002-344, NSC 89-2312-B-002-001, NSC 94-2413-H-002018, NSC 95-2413-H-002-002, NHRI-EX949204PP).
References 1. Hunt SM. The problem of quality of life. Qual Life Res 1997;6:205–12. 2. Spitzer B. Quality of Life Assessments in Clinical Trials. New York: Raven Press, 1998. 3. The WHOQOL Group. The development of the World Health Organization Quality of Life assessment instrument (the WHOQOL). In: Orley J, Kuyen W (eds). Quality of Life Assessment: International Perspectives. Berlin: SpringerVerlag, 1994:41–57. 4. The WHOQOL Group. The World Health Organization Quality of Life assessment (WHOQOL): position paper from the World Health Organization. Soc Sci Med 1995;41: 1403–9. 5. World Health Organization. Resources for New WHOQOL Centers. Geneva: WHO (MNH/PSF/95.2), 1995. 6. World Health Organization. WHOQOL Protocol for New Centers. Geneva: WHO (MNH/PSF/94.4), 1994. 7. The WHOQOL-Taiwan Group. The User’s Manual of the Development of the WHOQOL-100 Taiwan Version, 2nd edition. Taipei: WHOQOL-Taiwan Group, 2005. [In Chinese] 8. World Health Organization. WHOQOL Study Protocol. Geneva: WHO (MNH/PSF/93.9), 1993. 9. Lin MR, Yao KP, Hwang JS, Wang JD. Scale descriptor selection for Taiwan-version of questionnaire of World Health Organization Quality of Life. Chin J Public Health (Taipei) 1999;18:262–70. [In Chinese] 10. World Health Organization. WHOQOL Focus Group Moderator Training. Geneva: WHO (MNH/PSF/92.9), 1992. 11. Skevington SM, Bradshaw J, Saxena S. Selecting national items for the WHOQOL: conceptual and psychometric considerations. Soc Sci Med 1999;48:473–87. 12. Kemmler G, Holzner B. Multidimensional scaling as a tool for uncovering the internal structure of quality of life instruments. Qual Life Res 1997;6:435. 13. Power M, Bullinger M, Harper A, The WHOQOL Group. The World Health Organization WHOQOL-100: tests of the universality of quality of life in 15 different cultural groups worldwide. Health Psychol 1999;18: 495–505. 14. The WHOQOL Group. The World Health Organization Quality of Life assessment (WHOQOL): development and general psychometric properties. Soc Sci Med 1998;46: 1569–85.
597