Vol. 44 No. 6 December 2012 Journal of Pain and Symptom Management 923

Transcription

1 Vol. 44 No. 6 December 2012 Journal of Pain and Symptom Management 923 Brief Methodological Report Measurement Properties of the Chinese Version of the Kidney Disease Quality of Life-Short Form (KDQOL-SFÔ) in End-Stage Renal Disease Patients With Poor Prognosis in Singapore Yin Bun Cheung, PhD, CStat, Ying Ying Seow, MD, Li Min Qu, MSN, and Alethea Chung Pheng Yee, MD Singapore Clinical Research Institute (Y.B.C.), Department of General Medicine, Khoo Teck Puat Hospital (Y.Y.S.), and National Cancer Centre (L.M.Q., A.C.P.Y.), Singapore, Republic of Singapore Abstract Context. The Kidney Disease Quality of Life (KDQOLÔ) instrument is widely used to assess care of end-stage renal disease (ESRD) patients. Objectives. This study aimed to demonstrate the measurement properties of the Chinese (simplified characters) version of the KDQOL-Short FormÔ (KDQOL-SFÔ) in ethnic Chinese ESRD patients in Singapore. Methods. Seventy-eight new ESRD patients, initially conservatively managed, were interviewed and followed-up; 31 later commenced dialysis. Quality of life (with the KDQOL-SF), Karnofsky Performance Status, and estimated glomerular filtration rate (egfr) were assessed at 0, 3, 6, 9, 12, 18, and 24 months. Nonparametric statistics were used to assess correlation and changes. Results. The patterns of descriptive summary and floor and ceiling effects were similar to those reported in the original English version. Most of the kidney disease-targeted scales and the generic scales of the KDQOL-SF demonstrated 1) criterion validity in terms of correlation with objectively measured and/or subjectively rated variables, 2) convergent/divergent validity in terms of correlation patterns between kidney disease-targeted and generic scales, and 3) responsiveness to change in terms of difference before and after dialysis, except for Work Status. Internal consistency reliability was satisfactory, but that for the Work Status scale was too low (alpha ¼ 0.32) and for two other kidney disease-targeted scales, Sexual Function and Dialysis Staff Encouragement, was too high (1.0). Conclusion. Most of the scales in the Chinese version of the KDQOL-SF were valid and achieved internal consistency reliability, except for the Work Status scale. Also, the internal consistency reliability of two disease-targeted scales was too high, suggesting room for reduction of some items to reduce burden on Address correspondence to: Ying Ying Seow, MD, Department of General Medicine, Khoo Teck Puat Hospital, 90 Yishun Central, Singapore Ó 2012 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved , Republic of Singapore. yingseow@ gmail.com Accepted for publication: December 31, /$ - see front matter doi: /j.jpainsymman

2 924 Cheung et al. Vol. 44 No. 6 December 2012 respondents. J Pain Symptom Manage 2012;44:923e932. Ó 2012 U.S. Cancer Pain Relief Committee. Published by Elsevier Inc. All rights reserved. Key Words Validation, KDQOL, quality of life, Chinese Introduction Chronic illnesses and their treatments, including end-stage renal disease (ESRD) and dialysis, have great impact on the functioning and well-being of patients. Dialysis is both life-saving and life-altering. It changes a person s lifestyle completely and hence has great impact on his/her physical, mental, and social wellbeing. The Singapore Renal Registry in 2008 showed that the incidence of new ESRD patients in Singapore was 1212, with 894 (73.8%) commencing on renal replacement therapy (RRT). 1 The Kidney Disease Quality of Life (KDQOLÔ) instrument was developed for assessing the quality of life of individuals with kidney disease and on RRT. 2 The KDQOL-Short Form (KDQOL-SFÔ) is a version of the KDQOL that includes the Short-Form 36 Health Survey (SF-36 Ò ) generic core with other questions pertaining to specific dimensions of chronic kidney disease. It has been used in ESRD patients both on dialysis and on those opting for conservative measures. 3 This is an important tool for understanding ESRD patients experiences and has been used in cross-continent comparisons. 4 Poor KDQOL scores also have been shown to be a powerful and independent predictor of hospitalization and death for dialysis patients. 5 ESRD patients who are very elderly and those with multiple comorbidities have a very poor prognosis, with an expected life expectancy of less than two years. 6,7 Quality of life as an outcome would hence be of even more importance as there is a limited duration of life. Furthermore, these are also patients who are expected to benefit most from palliative care. The World Health Organization definition of palliative care includes the impeccable assessment of physical, psychological, social, and spiritual aspects of life. 8 This highlights the importance of quality-of-life assessment in this group of patients. Four Asian language versions of the KDQOL- SF have been reported in the literature. Three articles reported the translation, adaptation, and validation of a Filipino, 9 a Japanese, 10 and a Korean version 11 of the KDQOL-SF. A dissertation in Hong Kong reported the translation and adaptation of the instrument to Hong Kong Chinese followed by a pilot test on 43 ESRD patients. 12 Hong Kong and Taiwan use the traditional Chinese characters, whereas mainland China and Singapore use simplified Chinese characters. The use of words also can vary between places. 13 A cross-sectional study in Singapore assessed the validity and internal consistency of the KDQOL-SF. 14 However, in that study, over 99% of the 980 interviews used the English version and the analysis pooled all the interviews. The U.S. Food and Drug Administration recommends that translated versions of patient-reported outcomes should be assessed for their measurement properties. 15 Globally, the Chinese language has the largest number of native speakers and this is projected to continue. 16 Among Singapore residents aged 75 years or older who are literate, 57% are literate in Chinese only. 17 Drawing on data from a longitudinal study on the health-related quality of life and survival of ESRD patients, this report aims to assess the measurement properties of the Chinese version (simplified characters) of the KDQOL-SF in these patients. Methods Study Design and Participants This analysis is part of a larger study on the quality of life, survival, and role of palliative care in ESRD, which received ethics approval by the SingHealth Centralized Institutional Review Board. The study population came from renal wards and outpatient clinics of a tertiary hospital. The Singapore General Hospital (SGH) is the largest health care institution

3 Vol. 44 No. 6 December 2012 Validating the Chinese KDQOL-SF Ô in Singapore 925 in Singapore, caring for 45% of new ESRD patients in (unpublished SGH data, HL Choong, 2011). The inclusion criteria were estimated glomerular filtration rate (egfr) (Cockcroft Gault equation) 8e12 ml/minute, either 75 years old or older, or with Charlson Comorbidity Index (CCI) of 8 or greater, and spoke English and/or Chinese. (For the present analysis of the Chinese version of the KDQOL-SF, only the Chinese speakers were included.) Initiation of dialysis is usually recommended when egfr is about 10 ml/minute. 18 The exclusion criterion was cognitive impairment. Demographic and medical history data were collected at baseline. KDQOL-SF scores, Karnofsky Performance Status (KPS) scores, and laboratory and symptom data were assessed at enrollment and at 3, 6, 9, 12, 18, and 24 months or until death. The same study nurse conducted the KPS rating and interviews for all patients throughout the study period. Sample size was originally planned for the purpose of comparing poor prognosis ESRD patients deciding to undergo dialysis vs. those opting for conservative management. The present analysis only analyzed the 78 Chinese-speaking participants. This subgroup sample size has 77% power (5% Type I error rate) to detect a correlation of 0.3 between a scale score and a validation criterion variable (PASS Software, NCSS, Kaysville, UT). Measures The KDQOL-SF (version 1.2) comprises kidney disease-targeted domains and generic quality-of-life domains. The disease-targeted scales include Symptoms/Problems (12 items), Effects of Kidney Disease (eight items), Burden of Kidney Disease (four items), Work Status (two items), Cognitive Function (three items), Quality of Social Interaction (three items), Sexual Function (two items), Sleep (four items), Social Support (two items), Dialysis Staff Encouragement (two items), and Patient Satisfaction with Care (one item). The generic scales are the same as the RAND 36- Item Health Survey, whose items are identical to those of the SF-36 but the scoring algorithm for two of the eight scales was different. 19 The eight scales include Physical Function (10 items), Role-Physical (four items), Pain (two items), General Health (five items), Emotional Well-Being (five items), Role-Emotional (three items), Social Function (two items), and Energy/Fatigue (four items). The Chinese version (simplified characters) of the KDQOL-SF was translated and backtranslated by the KDQOL Working Group, RAND, and MAPI Institute 19 (personal communication with MAPI). The KDQOL-SF allows interviewer administration. The manual for use and scoring provides detailed instructions on interview administration. Because of the low literacy in older Chinese-speaking people in Singapore 17 and poor health status of the patients, this study used interviewer administration. Version 1.2 differs from the latest version (1.3) only in a screening question preceding the sexual functioning items. 19 The items and scales were scored according to the manual. The KPS score is a strong predictor of patients quality of life. 20 It ranges from 0 (dead) to 100 (normal), with scores #40 indicating dependency, 50e60 requiring assistance, and $70 self-caring. The RAND 36-Item Health Survey also includes a selfrated change of health in general compared with one year ago, measured on a five-point scale ( much better, somewhat better, about the same, somewhat worse, or much worse ). This question is not part of any of the eight generic scales and is designed to use individually. egfr was calculated, based on measured serum creatinine, at each visit. These three measures were used as objective and subjective criteria in the evaluation of the scales validity. Switching from conservative management to RRT is a major change that impacts on patient experience both positively (e.g., symptom control) and negatively (e.g., financial and other burdens). We used this as a criterion to evaluate responsiveness to change. Patient demographic and clinical data also were collected at the first interview, most of which are self-explanatory (Table 1). The CCI is an age-modified score of number and severity of comorbidities (including cardiac, pulmonary, hepatic, diabetes, etc.) proven to prognosticate for mortality in ESRD. 7 Scores of three or less indicate that a patient has very few comorbidities, whereas a score of eightormoreindicatesthepresenceofmultiple and/or advanced stage(s) of various medical condition(s).

4 926 Cheung et al. Vol. 44 No. 6 December 2012 Table 1 Patient Characteristics at Baseline (n ¼ 78) Characteristics Statistics or Category Value or n (%) Age Median 76 Range (52, 98) Gender Female 35 (44.9) Male 43 (55.1) Marital status Single 3 (3.9) Married 49 (62.8) Separated 3 (3.9) Widow 23 (29.5) Number of residents Living alone 5 (6.4) in household With one or two 33 (42.3) people With three or more 38 (48.7) people Unknown 2 (2.6) Primary carer Spouse 30 (38.5) Children 8 (10.3) Domestic helper 17 (21.8) Others 23 (29.5) Education None 28 (35.9) Primary school 37 (47.4) Secondary school 13 (16.7) or above Employment Unemployed 45 (57.7) Retired 28 (35.9) Self-employed 5 (6.4) Karnofsky Median 60 performance score Range (30, 90) CCI score Median 5 Range (2, 9) egfr (ml/minute) Median 10 Range (8.0, 11.4) Primary renal disease Diabetes mellitus 51 (65.4) Glomerulonephritis 3 (3.9) Hypertension/renal 19 (24.4) artery stenosis Obstructive uropathy 1 (1.3) Stone 4 (5.1) egfr ¼ estimated glomerular filtration rate; CCI ¼ Charlson Comorbidity Index. Statistical Analysis The participants were interviewed one to seven times during the study period, depending on whether they survived through the whole duration. Standard statistical analysis assumes independency between data points. In particular, nonparametric methods, which do not involve distribution assumptions, are not designed for correlated data. As such, except for analysis of responsiveness to change, we used only one measure per patient. By default, the baseline measure was used. However, the Dialysis Staff Encouragement and Patient Satisfaction scales were applicable only to patients who were on dialysis. In this study, the participants began with conservative management and fewer than half of them switched to dialysis during the study and, therefore, ever answered these questions. Furthermore, previous studies of the Singaporean population had found very high nonresponse rates to qualityof-life questions related to sexual life. 21,22 The present study also had the same problem of a high nonresponse rate to the Sexual Function scale. Use of the baseline data only would lead to incomplete analysis of the KDQOL-SF in this Chinese population. As such, we searched for the earliest postbaseline assessment that maximized the number of valid responses to the dialysis/sexual items, if it existed. Then, the earliest assessment that included the smallest number of item nonresponses was included in the present analysis instead of the baseline measure. Ceilingeffect(%atmaximumscore)and floor effect (% at minimum score) were tabulated. Kane recommended 30% as the cutoff for acceptable ceiling and floor effects in outcome measures. 23 The Shapiro-Wilk method was used to test for normality of the scale scores. The KPS score, self-rated change in health and egfr were used to examine criterion validity, using the Spearman s rank correlation coefficient (rho). Both KDQOL-SF and self-rated change in health were patient reported. To overcome the possible interpretation problem that any associations between the self-reported measures were spurious, the other two measures (KPS score and egfr) served to countercheck. The Spearman s rho was used to examine the association between the kidney disease-targeted scales and the generic scales to assess convergent/divergent validity. Internal consistency reliability was estimated using Cronbach s alpha. Responsiveness to change was assessed by comparing the quality-of-life scores before and after starting RRT among patients who changed from conservative management to dialysis, using the Wilcoxon signed-rank test. Results A total of 78 Chinese-speaking patients were recruited for this study from September 2007 to March All of them were ethnic Chinese. Four patients dropped out of the study after participating up to the visit at six months.

5 Vol. 44 No. 6 December 2012 Validating the Chinese KDQOL-SF Ô in Singapore 927 Demographic and Clinical Characteristics Patient characteristics are shown in Table 1. The primary renal disease was mainly diabetes (65.4%) and hypertension (24.4%). At recruitment, the median KPS score was 60, with a range from 30 to 90, and 41% were in the range able to perform self-care (score 70 or above). The median CCI was 5 and the median egfr was 10 ml/minute. Descriptive Summary of Quality of Life At baseline, none of the 78 participants filled in the scale items on Dialysis Staff Encouragement and Patient Satisfaction with Care, as all participants were initially on conservative treatment. At baseline, only 11 participants responded to the Sexual Function items. However, the numbers of participants who ever responded to these three scales during the study period were 30, 31, and 15, respectively. The 78 assessments in the main analysis were mainly (58%) from the baseline assessment; 31% were from the visits between three to 12 months and the rest from the visits after 12 months. Table 2 describes the distribution of the KDQOL-SF scale scores. The overall pattern is similar to that presented in the KDQOL-SF manual. Among the kidney disease-targeted scales, the mean scores on Burden of Kidney Disease and Work Status were lower than the rest. Apart from the Burden of Kidney Disease, Work Status, and Sexual Function scales, the scales showed little floor effect. More scales showed a ceiling effect, especially the Cognitive Function and Sexual Function scales. Among the RAND 36-Item Health Survey scales, again similar to the patterns displayed in the KDQOL-SF manual, the Role-Physical and Role-Emotional scales had both substantial floor and ceiling effects, whereas the Pain and Social Function scales showed substantial ceiling effects. Only three of the diseasetargeted scales (Sleep, Dialysis Staff Encouragement, and Patient Satisfaction) and four of the generic scales (Role-Physical, General Health, Role-Emotional, and Energy/Fatigue) were not rejected for normal distribution assumption (each P > 0.05). Criterion Validity Table 3 shows the Spearman s rank correlation coefficient (rho) between the scales and KPS score, self-rated change in general health, and egfr. Most scales correlated significantly with at least one of the three criteria, except Work Status, Social Support, and Pain. The KPS scores were most strongly correlated with the Effects of Kidney Disease (rho ¼ 0.56), Physical Functioning (rho ¼ 0.87), and Table 2 Descriptive Summary of KDQOL-SF Scale Scores KDQoL-SF Parameters n Mean SD Skewness Kurtosis % Floor % Ceiling Cronbach s Alpha Kidney disease-targeted scales Symptom/problems Effects of kidney disease Burden of kidney disease , Work status Cognitive function Quality of social interaction Sexual function Sleep Social support Dialysis staff encouragement Overall health NA Patient satisfaction NA RAND 36-Item Health Survey scales Physical functioning Role-physical Pain General health Emotional well-being Role-emotional Social function Energy/fatigue KDQOL-SF ¼ Kidney Disease Quality of Life-Short Form.

6 928 Cheung et al. Vol. 44 No. 6 December 2012 Table 3 Spearman s Rank Correlation Coefficients Among KDQOL-SF Scales and KPS and Self- Rated Change in Health KDQOL-SF Parameters KPS Score Self-Rated Change in Health egfr Kidney disease-targeted scales Symptom/problems a 0.22 b Effects of kidney disease 0.56 c 0.25 a 0.29 b Burden of kidney disease b Work status Cognitive function 0.41 c 0.31 b 0.07 Quality of social 0.32 b 0.30 b 0.01 interaction Sexual function c 0.56 a Sleep 0.28 a Social support Dialysis staff a encouragement Patient satisfaction a RAND 36-Item Health Survey scales Physical functioning 0.87 c Role-physical 0.26 a 0.23 a 0.22 a Pain General health a 0.25 a Emotional well-being 0.31 b 0.30 b 0.24 a Role-emotional 0.30 b 0.28 a 0.12 Social function 0.29 a b Energy/fatigue 0.62 c 0.23 a 0.27 a KDQOL-SF ¼ Kidney Disease Quality of Life-Short Form; KPS ¼ Karnofsky Performance Status; egfr ¼ estimated glomerular filtration rate. a P < b P < c P < Energy/Fatigue (rho ¼ 0.62) scales. Self-rated change in health was most strongly correlated with Sexual Function (rho ¼ 0.77). egfr was most strongly correlated with Sexual Function (rho ¼ 0.56) and also strongly correlated with the scales that relate to health care: Dialysis Staff Encouragement (rho ¼ 0.39) and Patient Satisfaction with Care (rho ¼ 0.37). All the KDQOL-SF scales that were found to be associated with self-rated change in health also were found to be associated with either the KPS score, or egfr, or both. Convergent/Divergent Validity Table 4 shows the Spearman s rank correlation coefficients between the kidney diseasetargeted scales and the RAND 36-Item Health Survey generic scales. The disease-targeted scales Symptoms/Problems, Effect of Kidney Disease, and Sleep were associated with 8/8, 7/8, and 7/8 of the generic scales, respectively. Dialysis Staff Encouragement and Patient Satisfaction with Care had almost no association with the generic scales. The rest appeared to show convergent/divergent validity. For example, Sexual Function was more correlated with Role-Physical (rho ¼ 0.57) than others, and Quality of Social Interaction was more correlated with Emotional Well-Being (rho ¼ 0.54; P < 0.001) than with others. Internal Consistency Reliability The Cronbach s alpha values also are presented in Table 2. The two-item Work Status scale had the lowest alpha (0.32). The two-item Sexual Function and Dialysis Staff Encouragement scales both had a Cronbach s alpha of 1.00 after Table 4 Spearman s Rank Correlation Coefficients Between Kidney Disease-Targeted Scales and RAND 36-Item Health Survey Generic Scales Generic Scales Disease-Targeted Scales Physical Functioning Role- Physical Pain General Health Emotion Well-Being Role- Emotional Social Function Energy/ Fatigue Symptoms/problems 0.22 a 0.35 b 0.47 c 0.39 c 0.42 c 0.36 b 0.34 b 0.51 c Effects of kidney disease 0.58 c 0.44 c c 0.59 c 0.46 c 0.51 c 0.64 c Burden of kidney disease c 0.37 c b 0.19 Work status a a Cognitive function 0.34 b 0.24 a a 0.39 c c Quality of social 0.31 b c 0.30 b c interaction Sexual function a a a Sleep 0.35 b 0.36 b c 0.50 c 0.35 b 0.31 b 0.50 c Social support b 0.30 b a Dialysis staff a encouragement Patient satisfaction a P < b P < c P <

7 Vol. 44 No. 6 December 2012 Validating the Chinese KDQOL-SF Ô in Singapore 929 rounding to two decimal points. The correlation between the two items in each of the scales appeared to be too strong. The patients responded to the two items as if they concerned exactly the same experience. The other scales showed Cronbach s alpha values from 0.78 to the distributions of changes suggest an improvement after starting RRT, although only Role-Physical (P ¼ 0.013), General Health (P ¼ 0.050), and Energy/Fatigue (P ¼ 0.021) reached conventional levels of statistical significance. Responsiveness to Change During the study period, 31 of the participants switched from conservative management to dialysis. Three of them did not live until the next scheduled visit for quality-of-life assessment. Therefore, 28 participants were involved in the responsiveness to change analysis. Table 5 shows the median and first and third quartiles of changes in scale scores from the assessment before to the assessment after starting RRT. Starting RRT was associated with improvement in Symptoms/Problems (P < 0.001), Cognitive Function (P ¼ 0.047), and Sleep (P ¼ 0.044). However, the start of RRT also was associated with a decline in quality of life as measured by the Burden of Kidney Disease scale (P < 0.001). Among the generic scales, apart from Physical Functioning, all Table 5 Changes in KDQOL-SF Scale Scores in Relation to Switching from Conservative Treatment to Renal Replacement Therapy a KDQOL-SF Parameters First Third Quartile Median Quartile P-value Kidney disease-targeted scales Symptom/problems <0.001 Effects of kidney disease Burden of kidney <0.001 disease Work status Cognitive function Quality of social interaction Sexual function Sleep Social support RAND 36-Item Health Survey scales Physical functioning Role-physical Pain General health Emotional well-being Role-emotional Social function Energy/fatigue a Positive values denote improvement in health-related quality of life after change to renal replacement therapy. Pre- and postchange scores tested by the Wilcoxon signed-rank test. Discussion The KDQOL-SF has the advantage of containing both kidney disease-specific concerns and generic quality-of-life domains. It has the potential of being a comprehensive instrument for use in the examination of kidney disease patients. This is the first study that examined the measurement properties of the Chinese version (simplified characters) of the KDQOL-SF. A limitation of the present study of ESRD patients with poor prognoses is that there is no directly relevant data available to examine the representativeness of the sample. However, we note that the participants KPS scores (41% with a score of 70 or above) was similar to that in patients in a similar age range in the Singapore Renal Registry: among patients 60 years old or older, 41% were rated normal/mild effort in carrying out daily activities. 1 Furthermore, the study sample was roughly similar to the ESRD patients in the Singapore Renal Registry in terms of major demographic and clinical characteristics, for example, diabetes mellitus being the most common primary disease. 1 Therefore, we postulate that the findings on the measurement properties of the KDQOL-SF can be generalized to the ESRD patients with poor prognoses in Singapore. A second limitation of the study is that the sample size is rather small. It was estimated to have sufficient power to established criterion and convergent validity in terms of strength of correlation. It also met the minimum sample size (n ¼ 75) recommended by the KDQOL-SF developers for field testing of translated versions. 24 However, the sample size for the assessment of responsiveness to change was smaller (n ¼ 28). A third limitation was that the study was not planned to assess test-retest reliability. In this analysis of data from a longitudinal study, the KDQOL-SF was administered three months apart. This is not suitable for measuring test-retest reliability,

8 930 Cheung et al. Vol. 44 No. 6 December 2012 which usually requires retesting at about one to two weeks after the first test. 9e11 Future longitudinal studies with larger sample sizes to confirm the finding on responsiveness and assess test-retest reliability are warranted. Similar to many other patient-reported outcome measures, the KDQOL-SF allows interviewer administration. Caution is needed in the interpretation of the Dialysis Staff Encouragement and Patient Satisfaction scales because the accuracy could possibly be affected by the mode of administration. In the present study, the interviewer was a study nurse not involved in the health care delivery. The descriptive findings also did not show high means or large ceiling effects on these two scales. Hence, we speculate that, in this particular study, administration method might not be a problem. Users and readers should pay attention to this potential problem nonetheless. Most of the scales demonstrated criterion validity, internal consistency reliability, and responsiveness to change. We used three different clinical, laboratory, and self-rated variables as criteria for the assessment of validity because the KDQOL-SF is multifaceted. Most of the KDQOL-SF scales correlated with at least one of these markers. (The exceptions include Work Status, Social Support, and Pain, which will be discussed in subsequent paragraphs.) We did not perform any P-value adjustment for multiple comparisons, which is a highly debatable practice. 25,26 For example, Rothman 26 pointed out that the concept of adjustment for multiplicity is based on the untenable universal null hypothesis that all observed associations only reflect random variation. The starting of RRTwas clearly associated with an improvement in the Symptoms/Problems scale but a decline in the Burden of Kidney Disease scale. This agrees with expectation, given the nature of the therapy. Five of the other 15 scales showed a statistically significant improvement in response to RRT. Two of the scales (Dialysis Staff Encouragement and Patient Satisfaction with Care) cannot be studied as they were not applicable before starting dialysis. The minimum clinically important differences (MCID) of the disease-targeted scales have not been previously charted, but the MCID in the SF-36 scales are typically in the range of three to five points. 27 Table 5 suggests that the generic core scales of the KDQOL-SF did show a range of clinically important responses to the start of RRT. None of the participants in this study received a salary. The Work Status scale had a very low Cronbach s alpha, showed little association with quality-of-life predictor variables, and did not respond to the start of RRT. For such a patient group, the inclusion of the Work Status scale did not provide useful information. The measurement properties of the Work Status scale in a working population are not yet known for the Chinese version. Similar to previous studies of cancer patients in Singapore, 20,21 most participants did not respond to the questions on Sexual Function. Among those who did respond to the two items on Sexual Function and the two items on Dialysis Staff Encouragement, the Cronbach s alpha was This suggests redundancy. Although the two items in each of the scales are conceptually distinct, patients answered them identically. Although this redundancy does not directly harm the measurement property, it appeared that only one of the two items is needed. The Pain scale score (mean 84; 51% at ceiling) in this sample was clearly better than that in the original study of the English version (mean 60; 20% at ceiling). 2,19 This may be a result of the phenomenon that Asian ESRD patients tend to have better health and functioning than their Western counterparts. 4,28 Nevertheless, the Pain scale was correlated with the Symptoms/ Problems scale (rho ¼ 0.47; P < 0.001), giving some support to its validity. Despite the Social Function scale being responsive to changes in clinical and laboratory variables, the Social Support scale was not. Note that the Social Function questions ask about whether the disease interferes with a patient s social life, whereas the Social Support questions ask about whether a patient is satisfied with the social support s/he received. There is no reason to expect that Social Support should associate with clinical or laboratory variables or Social Function. Looking at the correlation between Social Support and Emotional Well-Being (rho ¼ 0.33; P ¼ 0.010) and Role-Emotional scales (rho ¼ 0.30; P ¼ 0.010), and its lack of correlation with physical aspects such as Physical Functioning and Role- Physical scales, the Social Support scale appears to show convergent validity and divergent validity.

9 Vol. 44 No. 6 December 2012 Validating the Chinese KDQOL-SF Ô in Singapore 931 In conclusion, the present study has shown that most of the scales in the Chinese version of the KDQOL-SF administered by a trained research nurse were valid and achieved internal consistency, with the exception of the Work Status scale. Further investigation on test-retest reliability is warranted. In addition, further validation studies using more specialized measures on aspects of social interaction and a larger sample size for responsiveness are desirable. Disclosures and Acknowledgments This study was supported by a SingHealth Research Grant. The authors declare no conflicts of interest. References 1. Choong HL. Seventh report of the Singapore Renal Registry 2007/2008. Singapore: Health Promotion Board, Hays RD, Kallich JD, Mapes DL, Coons SJ, Carter WB. Development of the Kidney Disease Quality of Life (KDQOL TM ) instrument. Qual Life Res 1994;3:329e Gil Cunqueiro JM, Garcia Cortes MJ, Foronda J, et al. Health-related quality of life in elderly patients in haemodialysis. Nefrologia 2003;23:528e Fukuhara S, Lopes AA, Bragg-Gresham JL, et al. Worldwide Dialysis Outcomes and Practice Patterns Study. Health-related quality of life among dialysis patients on three continents: the Dialysis Outcomes and Practice Patterns Study. Kidney Int 2003;64: 1903e Mapes DL, Lopes AA, Satayathum S, et al. Health-related quality of life as a predictor of mortality and hospitalization: the Dialysis Outcomes and Practice Patterns Study (DOPPS). Kidney Int 2003;64:339e Munshi SK, Vijayakumar N, Taub NA, et al. Outcome of renal replacement therapy in the very elderly. Nephrol Dial Transplant 2001;16:128e Di Iorio B, Cillo N, Cirillo M, et al. Charlson Comorbidity Index is a predictor of outcomes in incident haemodialysis patients and correlate with phase angle and hospitalization. Int J Artif Organs 2004;27:330e Sepulveda C, Marlin A, Yoshida T, et al. Palliative care: the World Health Organization s global perspective. J Pain Symptom Manage 2002;24:91e Bataclan RP, Dial MA. Cultural adaptation and validation of the Filipino version of Kidney Disease Quality of Life-Short Form (KDQOL-SF version 1. 3). Nephrology (Carlton) 2009;14:663e Green J, Fukuhara S, Shinzato T, et al. Translation, cultural adaptation and initial reliability and multitrait testing of the Kidney Disease Quality of Life instrument for use in Japan. Qual Life Res 2001;10:93e Park HJ, Kim S, Yong JS, et al. Reliability and validity of the Korean version of Kidney Disease Quality of Life instrument (KDQOL-SF). Tohoku J Exp Med 2007;211:321e Li CM. Development of the Hong Kong Chinese version of the Kidney Disease Quality of Life Short Form (KDQOL-SF) Available from Accessed October 21, Cheung YB, Thumboo J, Goh C, et al. The equivalence and difference between English and Chinese versions of two major cancer-specific, health-related quality of life questionnaires. Cancer 2004;101:2874e Joshi V, Mooppil N, Lim J. Validation of the Kidney Disease Quality of Life-Short Form: a crosssectional study of a dialysis-targeted health measure in Singapore. BMC Nephrol 2010;11:36e U.S. Food and Drug Administration. Guidance for industry. Patient-reported outcome measures: Use in medical product development to support labeling claims. Silver Spring, MD: FDA, Available from ComplianceRegulatoryInformation/Guidances/UCM pdf. Accessed October 21, Graddol D. The future of language. Science 2004;303:1329e Singapore Department of Statistics. Census of population Statistical Release 1. Singapore: Department of Statistics, European best practice guidelines for haemodialysis (Part 1). Nephrol Dial Transplant 2002;17: 10e Hays RD, Kallich JD, Mapes DL, et al. Kidney Disease Quality of Life Short Form (KDQOL-SFTM), version 1.3: A manual for use and scoring. Washington, DC: RAND, Available from org/pubs/papers/2006/p7994.pdf. Accessed October 21, Cheung YB, Goh C, Wong LC, et al. Quick_FLIC: validation of a short questionnaire for assessing quality of life of cancer patients. Br J Cancer 2004;90: 1747e Cheung YB, Daniel R, Ng GY. Response and non-response to a quality-of-life question on sexual life: a case study of the simple mean imputation method. Qual Life Res 2006;15:1493e Cheung YB, Thumboo J, Gao F, et al. Mapping the English and Chinese versions of the Functional Assessment of Cancer TherapydGeneral to the EQ- 5D utility index. Value Health 2009;12:371e376.

10 932 Cheung et al. Vol. 44 No. 6 December Kane RL. Outcomes measures. In: Kane RL, ed. Understanding health care outcomes research. Gaithersburg, MD: Aspen, 1997:17e KDQOL Working Group. Translation Available from kdqol_translation_info.html. Accessed December 12, Schulz KF, Grimes D. Multiplicity in randomised trials I: endpoints and treatments. Lancet 2005;365: 1591e Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology 1990;1:43e Samsa G, Edelman D, Rothman M, et al. Determining clinically important differences in health status measures: a general approach with illustration to the Health Utilities Index Mark II. Pharmacoeconomics 1999;15:141e Lee A, Challinor S, Low YS, et al. Poster presentation at 16th Annual Scientific Meeting of Singapore General Hospital, April 27e28, 2007.