Supplementary Data for Health-Related Quality of Life After Stroke:Reliability and Validity

Supplementary data for“Health-related quality of life after stroke:reliability and validity of the Duke Health Profile for usein Viet Nam”

Methods

HRQoL Instruments

The Duke Health Profile (DHP) is a generic health-related quality of life (HRQoL) instrument. It is a brief self-report instrument with good acceptability.1, 2 The DHP has 17 questions that form 10 domains. Six of the 10 domains refer to health function (higher scores indicate better health status): physical health (5 items), mental health (5 items), social health (5 items), general health (combining the 15 items used for the physical, mental and social health dimensions to indicate overall well-being), perceived health (single item) and self-esteem (5 of the items used for the mental and social health domains). The remaining four domains refer to health dysfunction (higher scores indicate poorer health status): anxiety (6 of the items used for general health), depression (5 of the items used for physical and mental health), pain (single item used for physical health) and disability (single item). Each question has 3 possible responses scored as 0, 1 or 2. Responses to the constituent items in each domain are added and the mean of the raw scores is normalized to lie on scale of 0 to 100. In reporting the results of this study, the scores for anxiety, depression, pain and disability were subtracted from 100 so the scale ranged from 0 (poorest health) to 100 (best possible health) for each of the 10 dimensions. Prior to use, the DHP was translated and back-translated. This version has previously been validated among healthy adolescents in VietNam.3

The EuroQoL (EQ5D) instrument is a standardized measure of health status developed by the EuroQoL Group in order to provide a simple, generic measure of health for clinical and economic appraisal.45, 6 Applicable to a wide range of health conditions and treatments, it provides a simple descriptive profile and a single index value for health status. The version of the EQ5D used in this study was the EQ-5D-3L. It contains the EQ5D descriptive system and the EQ visual analogue scale (EQ-VAS). The descriptive system comprises 5 components: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. Each component has 3 response levels: no problems (0), some problems (1), and inability or extreme problems (2). The respondent is asked to indicate his/her health state by marking the box against the most appropriate statement in respect of each of the 5 components. The EQ VAS requires a respondent to record their self-rated health on a visual analogue scale with the endpoints labelled as “Best imaginable health state” and “Worst imaginable health state”.

Statistical Analysis

The ten dimensions of the DHP were coded and calculated according to the manual2. To calculate an EQ5D utility score, the components were weighted and summed using weights for the South Korean general population7 in the absence of a value set specific for the population of VietNam.

Inter-method comparisons were undertaken to assess convergent validity by calculating rank correlations between the DHP dimension scores and EQ5D component scores, overall score or EQ-VAS scores. Patterns of departure between the scores from each instrument were assessed by visual inspection of Bland Altman plots. For this purpose, the score from each instrument was standardized by subtracting the mean of its scores from the instrument and dividing by the standard deviation of all scores from that instrument. Linear regression methods were used to investigate whether the differences between standardized scores varied systematically with study factors such as sex, age and severity of stroke severity of disability at admission, severity of disability at 3 month follow up and type of stroke.

Results

Characteristics of the total cohort

Characteristics of the 108 participants (66 participants from the cohort study) in the validation study are summarized in Table 1. The remaining 315 subjects, in the cohort of 381,included 54.9% (173/315) males and with mean age at time of diagnosis of 58.3 (SD 12.8) years for males and 64.7 (SD 14.4) years for females.Ischemic stroke accounted for 76.5% (241/315) of their diagnoses, 41.5% (127/306) had severe impairment at admission, 73.2% (229/315) had severe disability at admission and 43.2% (134/313) had severe disability at three months. In these key respects, the participants in the validation study and non-participants were generally alike. Characteristics of 94 caregivers are summarized in Table S2.

Assessment of HRQOL made with EQ5D

The assessments of HRQoL made with EQ5D are also shown in Table S1. The overall scores were slightly greater (indicating better HRQoL) for males than for females, and decreased with age (r = –0.33), severity of disability at admission (r = –0.45), severity of impairment at admission (r = –0.45) and severity of disability at three months (r = –0.83). Contrary to the results for DHP, they were slightly greater for patients with ICH than for patients with IS. The scores on individual components mirrored the associations with the overall score. With higher scores reflecting poorer HRQoL, the component scores were slightly greater for females but none of the differences by sex reached statistical significance. Consistent with those for EQ5D, the EQ5D visual analogue scale (EQ-VAS) scores were slightly greater (indicating better health) for patients with ICH than for patients with IS (p = 0.72). The VAS scores were greater on average (p = 0.40) for men [65.7 (SD 17.9)] than for women [62.3 (SD 23.7)], and decreased with age (r = –0.06), and with severity of impairment at admission (r = –0.19), severity of disability at admission (r = –0.18) and at 3 months (r = –0.49).

Test and retest reliability of EQ5D

Table S4 and Table S5contains the results for the components of the EQ5D, its overall score, and the EQ-VAS. Consistent with the re-test measurements made using the DHP, the assessments of HRQoL using the EQ5D were generally higher on the second occasion. For the EQ5D utility score, the mean difference of 0.01 (SD 0.04) was significantly greater than zero (p = 0.02). The component score with the greatest mean difference was anxiety/depression (p = 0.11). After weighting, however, the greatest contributor to the mean difference in EQ5D overall scores was usual activities (p = 0.18), ahead of anxiety/depression (p = 0.36). The change in usual activities occurred because six (an unusually large number) of the most severely-affected patients reassessed their capacity to perform usual activities from 3 (“I am unable to perform my usual activities”) to 2 (“I have some problems with performing my usual activities”) on the second occasion. On each of the component items, and in respect of the overall score (p = 0.17) for which the mean differences were 0.16 (SD 0.50) for men and 0.04 (SD 0.03) for women, the disagreements were principally due to men upgrading their assessment of HRQoL on the second occasion. The test-retest differences were not influenced strongly by age (r = 0.03), type of stroke (r = 0.20), severity of impairment (r = 0.06), or severity of disability at admission (r = 0.11) and at 3 months (r = -0.04). Inspection of Bland-Altman plots of difference against average did not reveal systematic patterns in the differences or greater than expected numbers of outlying values. The greatest number of differences (n=35) occurred on re-measurement of anxiety. The intra-class correlation coefficients (ICC) ranged from 0.41 for anxiety to 0.89 for self-care. It was 0.99 for the overall score, in which the component scores are weighted and the penalty for the overwhelmingly most frequent change (moving from 1 = least severely affected to 2 = moderately affected, or conversely) is minor.

Consistent with the measurements made by the DHP and EQ5D, the EQ-VAS scores on the EQ5D instrument were generally greater on re-test but the mean difference was small (p = 0.58). The mean differences were almost identical for males [1.02 (SD 16.6)] and females [1.00 (SD 20.8)] and, whilst not influenced strongly by the following factors, decreased with age (r = –0.08), severity of impairment at admission (r= –0.12), severity of disability at admission (r = –0.08) and at three months (r = –0.02). Inspection of Bland-Altman plots of difference against average did not reveal systematic patterns in the differences or greater than expected numbers of outlying values. Reflecting the greater contribution of within-person differences, the ICC was moderate (0.56).

Table S2 about here

Patient-proxy reliability of EQ5D

Also shown in Table S6 and Table S7 are the mean proxy-patient differences for components of the EQ5D and its overall score, and for the EQ-VAS. The assessments of morbidity (p = 0.16), usual activity (p = 0.06) and pain (p = 0.37) using the EQ5D were greater on proxy assessment. For all components of EQ5D, the proxy-patient differences were more extreme when the proxy was a male than when the proxy was a female but these differences were not statistically significant. The proxy-patient differences in self-care (p = 0.01), usual activity (p = 0.01) and EQ5D overall score (p = 0.01) were associated with age of the patient. For most EQ5D components, the proxy-patient differences were greater for patients with ischemic stroke and for patients with severe impairment or disability at admission or severe disability after three months.

The proxy-patient differences on EQ5D were relatively smaller than those on the DHP, and without a consistent pattern but with proxies providing slightly lower HRQoL assessments overall and on the EQ-VAS. The stratified results indicate that, other than by the EQ-VAS, male proxies tend to over-estimate HRQoL relative to patients and/or relative to female proxies (data not shown), and that female proxies tend to underestimate HRQoL relative to patients and/or relative to male proxies (data not shown). The ICC for proxy-patient variation in overall EQ5D was lower, and the ICC for the anxiety component was higher, but otherwise the ICCs were generally similar to the corresponding values for patient test-retest.

Table S1.EQ5D domains, utility score and visual analogue scale (VAS)
Male (N=55) / Female (N=53)
% / (n/N) / % / (n/N)
Mobility
No problem / 45.5 / (25/55) / 34.0 / (18/53)
Some problem / 41.8 / (23/55) / 56.6 / (30/53)
Extreme problem / 12.7 / (7/55) / 9.4 / (5/53)
Self-care
No problem / 61.8 / (34/55) / 54.7 / (29/53)
Some problem / 9.1 / (5/55) / 22.6 / (12/53)
Extreme problem / 29.1 / (16/55) / 22.6 / (12/53)
Usual activity
No problem / 46.3 / (25/54) / 41.5 / (22/53)
Some problem / 25.9 / (14/54) / 26.4 / (14/53)
Extreme problem / 27.8 / (15/54) / 32.1 / (17/53)
Pain/Discomfort
No problem / 47.3 / (26/55) / 43.4 / (23/53)
Some problem / 50.9 / (28/55) / 49.1 / (26/53)
Extreme problem / 1.8 / (1/55) / 7.6 / (4/53)
Anxiety/Depression
No problem / 56.4 / (31/55) / 54.7 / (29/53)
Some problem / 40.0 / (22/55) / 37.7 / (20/53)
Extreme problem / 3.6 / (2/55) / 7.6 / (4/53)
Utility score: Mean (SD*) / 0.68 / (0.30) / 0.67 / (0.30)
VAS: Mean (SD*) / 65.7 / (17.9) / 62.3 / (23.7
* Standard Deviation
Table S2. Characteristics of caregivers who participated in the study
Characteristics / % / (n/N)
Age: Mean(SD) / 48.3 / (12.1)
Female sex / 77.0% / (67/87)
Relationship with patients
Spouse / 41.8% / (38/91)
Children / 41.8% / (38/91)
Siblings / 6.6% / (6/91)
Parents / 2.2% / (2/91)
Others / 7.6% / (7/91)

1

Table S3. Rank correlation of patient-proxy difference on assessments of DHP dimensions with study factors
The DHP dimensions
PH / MH / SH / Per / SE / Anx / Dep / AD / Pain / Dis / GH
Patient characteristics
Age / 0.17 / 0.03 / 0.17 / 0.15 / 0.25 / 0.03 / 0.11 / 0.13 / -0.02 / -0.15 / 0.22
Female sex / 0.15 / -0.04 / -0.08 / -0.03 / 0.02 / 0.10 / 0.06 / -0.12 / -0.12 / 0.00 / 0.07
Severe stroke at admission / -0.11 / 0.00 / -0.03 / -0.03 / -0.03 / -0.16 / -0.10 / -0.14 / -0.19 / -0.12 / -0.13
Severe disability at admission / -0.13 / -0.13 / -0.03 / 0.15 / -0.02 / -0.15 / -0.23 / -0.18 / -0.15 / -0.15 / -0.13
Severe disability at 3 months / -0.03 / -0.14 / 0.13 / 0.20 / 0.02 / -0.09 / -0.18 / -0.16 / -0.17 / -0.08 / -0.05
Proxy characteristics
Age / 0.15 / -0.03 / 0.05 / -0.05 / -0.02 / -0.02 / 0.03 / -0.04 / 0.06 / -0.04 / 0.05
Female sex / 0.01 / -0.01 / 0.01 / 0.06 / -0.19 / -0.11 / -0.16 / -0.20 / -0.03 / 0.09 / -0.13
PH: physical health, MH: mental health, SH: social health, Per: perceived health, SE: self-esteem, Anx: anxiety, Dep: depression, AD:anxiety-depression, Dis:disability, GH: general health.

1

Table S4: Test and retest reliability of measurements of HRQoL made one week apart of patient responses to the EQ5D
Test / Re-test / Difference
Mean / SD / Mean / SD / Mean / SD / ICC* / SEM† / MDD‡
Morbidity / 0.094 / (0.123) / 0.092 / (0.129) / -0.001 / (0.052) / 0.915 / 0.037 / 0.102
Self-care / 0.043 / (0.058) / 0.041 / (0.056) / -0.001 / (0.024) / 0.909 / 0.017 / 0.048
Usual activity / 0.076 / (0.089) / 0.068 / (0.084) / -0.007 / (0.052) / 0.818 / 0.037 / 0.103
Pain/Discomfort / 0.025 / (0.033) / 0.025 / (0.033) / 0 / (0.023) / 0.777 / 0.017 / 0.046
Anxiety/Depression / 0.026 / (0.038) / 0.027 / (0.037) / 0.004 / (0.042) / 0.433 / 0.030 / 0.083
EQ5D overall score / 0.676 / (0.299) / 0.686 / (0.304) / 0.010 / (0.04) / 0.989 / 0.031 / 0.084
VAS / 64.1 / 20.9 / 64.7 / (18.4) / 1.00 / (18.8) / 0.545 / 13.20 / 30.80

* ICC= Intra-class Correlation Coefficient, † SEM= Standard Error of Measurement, ‡ MDD= Minimal Detectable Difference

1

Table S5. Percent agreement and weighted Kappa for test-retest measurements made one week apart of patient responses to the EQ5D.
EQ5D components / Percent agreement / Weighted Kappa
Morbidity / 96.3% / 0.86
Self-esteem / 94.1% / 0.86
Usual activity / 91.6% / 0.80
Pain/discomfort / 93.8% / 0.65
Anxiety/Depression / 87.3% / 0.40

1

Table S6: Patient-proxy reliability of measurements of HRQoL using the EQ5D
Test / Re-test / Difference
Mean / SD / Mean / SD / Mean / SD / ICC* / SEM† / MDD‡
Morbidity / 0.100 / (0.130) / 0.094 / (0.132) / -0.006 / (0.068) / 0.845 / 0.049 / 0.134
Self-care / 0.459 / (0.596) / 0.047 / (0.059) / 0.014 / (0.031) / 0.854 / 0.022 / 0.060
Usual activity / 0.084 / (0.091) / 0.074 / (0.088) / -0.089 / (0.046) / 0.855 / 0.033 / 0.091
Pain/Discomfort / 0.027 / (0.035) / 0.026 / (0.037) / -0.001 / (0.034) / 0.513 / 0.024 / 0.067
Anxiety/Depression / 0.026 / (0.038) / 0.029 / (0.047) / 0.003 / (0.039) / 0.571 / 0.027 / 0.076
EQ5D overall score / 0.66 / (0.31) / 0.67 / (0.32) / 0.01 / (0.16) / 0.858 / 0.113 / 0.160
VAS / 63.0 / (20.5) / 62.7 / (18.8) / -1.40 / (18.8) / 0.539 / 13.25 / 36.74
* ICC= Intra-class Correlation Coefficient, † SEM= Standard Error of Measurement, ‡ MDD= Minimal Detectable Difference

1

Table S7. Percent agreement and weighted Kappa for measurements of patient responses and proxy responses to the EQ5D
EQ5D components / Percent agreement / Weighted Kappa
Morbidity / 95.4% / 0.84
Self-esteem / 93.3% / 0.85
Usual activity / 93.0% / 0.84
Pain/discomfort / 91.0% / 0.56
Anxiety/Depression / 88.6% / 0.51

1

References

1.Parkerson GR, Jr., Broadhead WE, Tse CK. The duke health profile. A 17-item measure of health and dysfunction. Med Care. 1990;28:1056-1072

2.Parkerson GR, Jr. User's guide for duke health measures. 2002.

3.Hanh VT, Guillemin F, Cong DD, Parkerson GR, Jr., Thu PB, Quynh PT, Briancon S. Health related quality of life of adolescents in vietnam: Cross-cultural adaptation and validation of the adolescent duke health profile. J Adolesc. 2005;28:127-146

4.Kranciukaite D, Rastenyte D. Measurement of quality of life in stroke patients. Medicina (Kaunas). 2006;42:709-716

5.Mauldin PD, Simpson KN, Palesch YY, Spilker JS, Hill MD, Khatri P, Broderick JP. Design of the economic evaluation for the interventional management of stroke (iii) trial. Int J Stroke. 2008;3:138-144

6.van Exel NJ, Scholte op Reimer WJ, Koopmanschap MA. Assessment of post-stroke quality of life in cost-effectiveness studies: The usefulness of the barthel index and the euroqol-5d. Qual Life Res. 2004;13:427-433

7.Lee YK, Nam HS, Chuang LH, Kim KY, Yang HK, Kwon IS, Kind P, Kweon SS, Kim YT. South korean time trade-off values for eq-5d health states: Modeling with observed values for 101 health states. Value Health. 2009;12:1187-1193

1