If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Correspondence: Takeru Shiroiwa, MPH, PhD, Center for Outcomes Research and Economic Evaluation for Health (C2H), National Institute of Public Health, 2-3-6 Minami, Wako, Saitama 351-0197, Japan.
Population norms of preference-based measures (PBMs) have been established in some countries and are important for cost-effectiveness analysis and interpretation of PBM values. No comparative data exist for population norms with regard to the EQ-5D-5L and Health Utilities Index Mark 3.
•
We established the Japanese population norms of the EQ-5D-5L and Health Utilities Index Mark 3 using data from a large sample. Age, sex, household income, and education level had a significant influence on health state index values. Disutility associated with diseases and symptoms was also estimated, revealing that Parkinson disease, dementia, stroke, and depression were associated with a large disutility.
•
These findings can contribute to a more reliable analysis of economic evaluations and may help clarify the characteristics of the 2 PBMs, which in turn will lead to an appropriate interpretation of cost-effectiveness analysis to facilitate healthcare decision making.
Abstract
Objectives
This study aimed to establish the Japanese population norms of the EQ-5D-5L and Health Utilities Index Mark 3 (HUI3) and estimate the disutility associated with diseases and symptoms.
Methods
We performed a door-to-door survey of the general population by random sampling. The planned sample size was 10 000 residents (age ≥16 years) of 334 districts in Japan. In addition to the EQ-5D-5L and HUI3 questionnaires, questions regarding demographic factors and self-reported main diseases and symptoms were asked. The EQ-5D-5L and HUI3 responses were converted to index values on the basis of Japanese value sets. Summary values by age and sex were calculated to obtain Japanese normative values. A multiple linear model was used to examine relationships between these values and diseases and symptoms.
Results
We collected 10 183 responses from 334 districts. The mean EQ-5D-5L index values were 0.821 (male) and 0.774 (female) in the age group of 80 to 89 years, which were lower compared with 0.978 (male) and 0.967 (female) in the age group of 16 to 19 years. Similar trends were observed for the HUI3 values. Age, sex, household income, and education level had a significant influence on the values of both instruments. When measured with the EQ-5D-5L, Parkinson disease, dementia, and stroke were associated with the largest disutility (>0.2), and the disutility for depression was approximately 0.18. In contrast, the HUI3 disutility values for Parkinson disease and dementia were approximately 0.4.
Conclusions
This study established the Japanese population norms of the EQ-5D-5L and HUI3, which can be used in healthcare decision making and contribute to a more reliable analysis of economic evaluations.
Preference-based measures (PBMs), such as the EQ-5D, Health Utilities Index Mark 3 (HUI3), and Short-Form 6 Dimensions (SF-6D), are used to calculate quality-adjusted life-years (QALYs) in cost-effectiveness analysis. Responses to these questionnaires are converted to quality-of-life values (referred to as utilities, index values, or quality-of-life scores) anchored to 0 (death) and 1 (full health) on the basis of predetermined value sets. Therefore, the same health state assessed with the same PBM may result in different values across countries and regions.
In some countries, the population norms of PBMs have been established, which help interpret PBM data and can be used as values without diseases or symptoms in cost-effectiveness analysis. The population norms of the EQ-5D-5L have been reported in Germany,
Health-related quality of life measured with the EQ-5D-5L: estimation of normative index values based on a representative German population sample and value set.
Quality of life of the Indonesian general population: test–retest reliability and population norms of the EQ-5D-5L and WHOQOL-BREF [published correction appears in PLoS One. 2018;13(8):e0203091].
Jiang R, Janssen MFB, Pickard AS. US population norms for the EQ-5D-5L and comparison of norms from face-to-face and online samples. 2021;30(3):803-816.
Health Utilities Index Mark 3 scores for major chronic conditions: population norms for Canada based on the 2013 and 2014 Canadian Community Health Survey.
One way to collect general population norm data is to include PBMs in national surveys for government statistics. Nevertheless, some countries, including Japan, do not permit this practice. Therefore, we conducted an original survey to collect data necessary for constructing the population norms of the EQ-5D-5L and HUI3 in Japan. This study had the 3 major objectives. First, we aimed to establish the Japanese population norms of the EQ-5D-5L and HUI3. Normative data of the HUI3 in the general Japanese population are still lacking. As for the EQ-5D-5L, our group has reported population norms in 2013,
but these values need to be updated. The sample size was limited in the previous survey, and with this large sample survey we expect to improve the generalizability of the results. Second, we set out to simultaneously collect data on respondents’ self-reported diseases and symptoms to estimate the disutilities associated with various diseases and symptoms. These data can be useful for QALY calculations in economic evaluations. The final objective was to clarify the characteristics of the EQ-5D-5L and HUI3 by comparing their normative values. In fact, this is the first study to compare EQ-5D-5L norms with those of the HUI3.
Methods
Instruments
In this cross-sectional study, we measured respondents’ health state with the EQ-5D-5L and HUI3. The EQ-5D-5L
comprises 5 dimensions (“mobility,” “self-care,” “usual activities,” “pain･discomfort,” and “anxiety･depression”), with each dimension having 5 levels. The EQ-5D-5L was introduced to improve the sensitivity of the original EQ-5D (EQ-5D-3L), which has 5 dimensions with 3 levels. The Japanese value set for the EQ-5D-5L, which has been developed on the basis of the societal preferences of the general population,
consists of 8 attributes (“vision,” “hearing,” “speech,” “ambulation,” “dexterity,” “emotion,” “cognition,” and “pain”), which are mapped from answers to the HUI questionnaire comprising 15 items. The HUI was developed on the basis of the multiattribute utility theory, and a multiplicative utility function is used to convert responses to scores. The Japanese HUI3 multiattribute utility function has recently been developed.
Development of a multiplicative, multi-attribute utility function and eight single-attribute utility functions for the health utilities index mark 3 in Japan.
Japanese is the official language of Japan and is almost exclusively spoken in the country. Therefore, only Japanese versions of the EQ-5D-5L and HUI3 were used in this survey.
We also included in our survey some questions from the National Livelihood Survey conducted annually by the Japanese Ministry of Health, Labour, and Welfare. These questions ask respondents whether they have any diseases for which they consult a doctor and whether they have any subjective symptoms. Those who answer “yes” must then select their main diseases and symptoms from among a list of 40 symptoms (eg, fever, sluggishness, sleeplessness) and diseases (eg, diabetes, obesity, hyperlipidemia).
Sampling Method
For the present survey, a total of 10 000 respondents (age ≥16 years) from 334 districts in Japan were selected by random sampling. The 334 districts (cho-me in Japanese) were selected using the following methods. First, the number of districts in 8 regions (Hokkaido, Tohoku, Kanto, Chubu, Kinki, Chugoku, Shikoku, and Kyushu) was determined in proportion to the population of each region. Then, for each region, the number of districts belonging to each stratum (i.e., prefecture [47 prefectures in Japan] × category of size of municipality [approximately 1700 municipalities in Japan]) was calculated on the basis of the population of the stratum. The surveyed districts were randomly selected according to the allocated number of districts in each stratum. Respondents were also randomly sampled from each selected district, stratified by sex and age. Therefore, for each district, the planned sample size was 1 (men aged 16–19 years and women aged 16–19 years) and 2 (men aged 20–29 years to 80–89 years and women aged 20–29 years to 80–89 years), totaling 30 respondents. Individuals in hospitals and nursing homes were excluded.
The Basic Resident Register was used to select respondents living in each district in a random manner. In Japan, each municipality has its own Basic Resident Register data, which include information, such as name, sex, address, and date of birth. We obtained permission from each municipality to use these data in this study.
A door-to-door survey was conducted between November 2019 and January 2020 (before the coronavirus disease outbreak in Japan). Investigators visited the registered addresses and distributed the questionnaire forms. If informed consent was obtained, they then collected the forms after a few days and checked for missing data. These visits continued until the planned number of responses was collected for each district. The investigators obtained informed consent from all respondents. This sampling method was the same as that used in our previous survey (conducted between January 2013 and March 2013),
except that the sample size (1000 vs 10 000) and the number of districts (100 vs 334) were larger.
Statistical Analysis
Summary index values were calculated for background factors. Responses to the EQ-5D-5L and HUI3 were converted to index values using Japanese value sets.
Development of a multiplicative, multi-attribute utility function and eight single-attribute utility functions for the health utilities index mark 3 in Japan.
The percentage of individuals who reported no problem or any problem was similarly computed. The interclass correlation coefficient (ICC) was used to compare the EQ-5D-5L and HUI3 values.
A multiple linear model was used to examine relationships between the EQ-5D-5L and HUI3 values and sociodemographic factors. Age, sex, region of residence, household income, employment status, education level, and marital status were included in the model. Next, to estimate the size of the disutility associated with diseases and symptoms, a model that included age category, sex, and main disease or symptom was constructed to calculate estimated regression coefficients. Considering that the utility distribution was censored at 1, the Tobit model was also applied.
Spearman correlation coefficients were used to address the problem of multicollinearity between independent variables.
All statistical analyses were performed using SAS 9.4, SAS Institute Inc., Cary, NC, USA. This study was approved by the ethics committee of the National Institute of Public Health.
Results
Response Rate and Demographic Factors
We collected 10 183 responses from 334 districts. The overall response rate was 34.1%, and the response rates in younger age groups tended to be lower than those in older age groups (16–19 years, 33.4%; 20–29 years, 26.5%; 30–39 years, 31.8%; 40–49 years, 32.4%; 50–59 years, 33.2%; 60–69 years, 38.2%; 70–79 years, 40.4%; and 80–89 years, 39.6%). The region with the highest response rate was Kyushu (45.4%) and that with the lowest response rate was Kanto (29.3%). Generally, response rates in larger cities tended to be lower.
Respondent demographic factors are shown in Table 1. The proportion of the national population in each region as of October 2019 as estimated by the Japanese government was 4.2% in Hokkaido, 6.9% in Tohoku, 34.4% in Kanto, 16.8% in Chubu, 17.7% in Kinki, 5.8% in Chugoku, 2.9% in Shikoku, and 11.3% in Kyushu. The sampled respondents were representative of the entire Japanese population.
Because we sampled respondents principally by the same number among the age and sex categories (except for age 16–19 years), it was natural that the characteristics of demographic factors were inconsistent with those of the entire Japanese population. When comparing the data shown in Table 1 with those of the Japanese general population, the percentage of young individuals in our sample was larger than that in the entire Japanese population. According to the National Livelihood Survey by the Ministry of Health, Labour, and Welfare, the average household income in 2018 was ¥5.5 million ($52 000; $1 = ¥106, as of September 2020), with a median of ¥4.3 million ($41 000). The median household income among our respondents ranged from ¥4 million to ¥6 million. According to the 2019 Labour Force Survey, part-time workers accounted for 22.6% of all workers (23.8% in our sample), 19.9% of Japanese individuals graduated from a university or graduate school in 2010, and the government predicted that 26.6% and 17.8% of Japanese men and women, respectively, were unmarried at the age of 50 years (known as “lifetime unmarried people”) in 2020.
The characteristics of our study population did not differ largely from these figures.
Population Norms of the EQ-5D-5L and HUI3
The EQ-5D-5L and HUI3 summary index values, stratified by categories of age and sex, are shown in Table 2. Both values tended to be higher in younger individuals, and the percentage of respondents in full health had the same tendency. In the age group of 16 to 19 years, the mean EQ-5D-5L values were 0.978 (male) and 0.967 (female), and the mean HUI3 values were 0.892 (male) and 0.888 (female). The higher the age, the lower the values: in the age group of 80 to 89 years, the mean EQ-5D-5L values were 0.821 (male) and 0.774 (female), and the mean HUI3 values were 0.714 (male) and 0.679 (female). EuroQol–visual analog scale scores and responses to each item are presented in the Appendix in Supplemental Materials found at https://doi.org/10.1016/j.jval.2021.03.010.
Table 2Population norms of EQ-5D-5L and HUI3.
PBM
Age, y
Sex
N
Mean
SD
Median
% of full health
EQ-5D-5L
16-19
Male
362
0.978
0.059
1.000
85.9
Female
381
0.967
0.073
1.000
80.6
20-29
Male
637
0.951
0.091
1.000
71.1
Female
642
0.953
0.088
1.000
72.1
30-39
Male
633
0.953
0.091
1.000
71.6
Female
632
0.944
0.104
1.000
69.2
40-49
Male
640
0.947
0.094
1.000
68.0
Female
652
0.945
0.090
1.000
65.6
50-59
Male
615
0.931
0.109
1.000
60.2
Female
687
0.925
0.106
1.000
56.0
60-69
Male
682
0.930
0.120
1.000
60.9
Female
688
0.927
0.104
1.000
55.7
70-79
Male
723
0.889
0.154
0.895
46.8
Female
710
0.876
0.157
0.895
42.3
80-89
Male
765
0.821
0.203
0.889
36.2
Female
734
0.774
0.220
0.831
26.8
HUI3
16-19
Male
356
0.892
0.151
0.940
27.9
Female
369
0.888
0.157
0.940
23.6
20-29
Male
619
0.855
0.187
0.912
22.6
Female
632
0.873
0.158
0.930
19.2
30-39
Male
617
0.856
0.175
0.912
17.9
Female
617
0.875
0.157
0.930
17.6
40-49
Male
629
0.871
0.158
0.930
19.7
Female
640
0.884
0.147
0.931
19.9
50-59
Male
594
0.844
0.166
0.902
9.6
Female
663
0.863
0.153
0.912
10.3
60-69
Male
663
0.837
0.185
0.902
8.1
Female
669
0.870
0.143
0.912
10.0
70-79
Male
664
0.807
0.199
0.875
6.1
Female
669
0.818
0.194
0.875
8.5
80-89
Male
674
0.714
0.254
0.813
4.3
Female
645
0.679
0.271
0.738
5.6
HUI3 indicates Health Utilities Index Mark 3; PBM, preference-based measure.
The EQ-5D-5L values were higher than HUI3 values in all age and sex categories. The percentage of respondents in full health was also higher when measured with the EQ-5D-5L. For example, 85.9% (male) and 80.6% (female) of the respondents aged 16 to 19 years reported a full health state on the EQ-5D-5L compared with 27.9% (male) and 23.6% (female) on the HUI3. When the linear regression model without an intercept was applied using the EQ-5D-5L and HUI3 values of the same respondent, the following estimated equation was obtained: “EQ-5D-5L” = 1.07 × “HUI3.” The HUI3 values were lower than the EQ-5D-5L values in 83.2% of our respondents. The ICC between the HUI3 and EQ-5D-5L values was 0.53.
As shown in Figure 1, the EQ-5D-5L values were higher than the HUI3 values in all age and sex categories. The EQ-5D-5L values obtained in the present survey were comparable to those obtained in our previous survey in 2013. The largest difference between the present and previous values was observed in the age group of 60 to 69 years; nevertheless, the difference was only 0.024, which is smaller than the minimal clinically important difference for the EQ-5D-5L.
Estimation of an instrument-defined minimally important difference in EQ-5D-5L index scores based on scoring algorithms derived using the EQ-VT version 2 valuation protocols.
Relationships Between EQ-5D-5L and HUI3 Values and Demographic Factors
Table 3 shows the relationships between the EQ-5D-5L and HUI3 values and demographic factors. Model 1 included only age category, sex, household income, and education level, all of which had an influence on the EQ-5D-5L values in the previous survey. Model 2 included all demographic factors. Because the absolute values of Spearman correlation coefficients among demographic factors were low (<0.3), the influence of multicollinearity was not considered in this study. The results of analyses using both models revealed age and sex to be statistically significant factors. The effect size of sex was small, but the direction of the effect was different between the 2 instruments. Household income was also a significant factor for both instruments. Individuals with a higher education level tended to have a better health state. Region of residence did not have a significant influence on health state, and no difference by area was observed. The results obtained using the Tobit model are shown in the Appendix in Supplemental Materials found at https://doi.org/10.1016/j.jval.2021.03.010.
Table 3Relationship between values by both PBM and demographic factor.
Variable
Model 1
Model 2
EQ-5D-5L
HUI3
EQ-5D-5L
HUI3
Coefficient
P-value
Coefficient
P-value
Coefficient
P-value
Coefficient
P-value
Intercept
0.906
<.001
0.709
<.001
0.948
<.001
0.791
<.001
Age, y
16-19
—
—
—
—
—
—
—
—
20-29
−0.027
.023
−0.007
.703
−0.036
.003
−0.021
.227
30-39
−0.033
.010
−0.006
.728
−0.050
<.001
−0.043
.0207
40-49
−0.039
.002
−0.002
.899
−0.057
<.001
−0.042
.0242
50-59
−0.056
<.001
−0.029
.118
−0.073
<.001
−0.069
<.001
60-69
−0.047
<.001
−0.011
.554
−0.056
<.001
−0.044
.020
70-79
−0.080
<.001
−0.029
.118
−0.072
<.001
−0.046
.019
80-89
−0.148
<.001
−0.108
<.001
−0.123
<.001
−0.108
<.001
Sex
Male
—
—
—
—
—
—
—
—
Female
−0.008
.013
0.023
<.001
−0.011
.001
0.018
.001
Household income (¥10 000)
<100
—
—
—
—
—
—
—
—
100-200
0.017
.0417
0.058
<.001
0.012
.141
0.050
<.001
200-400
0.039
<.001
0.084
<.001
0.027
<.001
0.065
<.001
400-600
0.050
<.001
0.099
<.001
0.032
<.001
0.071
<.001
600-1000
0.055
<.001
0.117
<.001
0.034
<.001
0.085
<.001
1000-1500
0.057
<.001
0.136
<.001
0.036
<.001
0.104
<.001
1500-2000
0.056
<.001
0.135
<.001
0.034
.009
0.101
<.001
>2000
0.071
<.001
0.136
<.001
0.047
.003
0.103
<.001
Education level
Elementary or junior high
—
—
—
—
—
—
—
—
High school
0.026
<.001
0.041
.0412
0.022
<.001
0.037
<.001
College
0.030
<.001
0.048
.0479
0.027
<.001
0.045
<.001
University or graduate
0.036
<.001
0.065
.0653
0.033
<.001
0.062
<.001
Other
0.012
.297
0.059
.0585
0.037
.005
0.091
<.001
Region of residence
Hokkaido
—
—
—
—
Tohoku
0.008
.349
−0.001
.915
Kanto
0.003
.734
0.003
.757
Chubu
0.005
.505
−0.001
.916
Kinki
0.002
.773
0.004
.745
Chugoku
−0.007
.465
−0.014
.301
Shikoku
0.004
.736
0.003
.864
Kyushu
0.014
.093
0.023
.047
Employment status
Full-time worker
—
—
—
—
Part-time worker
0.006
.254
0.006
.384
Self-employed or manager
−0.002
.763
−0.003
.723
Homemaker
−0.008
.223
−0.010
.291
Retiree
−0.064
<.001
−0.070
<.001
Other
−0.033
.002
−0.036
.004
Marital status
Married
—
—
—
—
Single
−0.016
.001
−0.050
<.001
Divorced/bereaved
−0.017
.001
−0.022
.002
R2
0.143
0.168
0.109
0.132
Coefficient >0.05 and significant
Coefficient <0.05 and significant
HUI3 indicates Health Utilities Index Mark 3; PBM, preference-based measure.
Table 4 shows the relationships between the EQ-5D-5L and HUI3 values and the respondents’ self-reported main diseases. For the EQ-5D-5L, Parkinson disease, dementia, and stroke were associated with the largest disutility (>0.2), and depression was associated with a disutility of approximately 0.18. For the HUI3, Parkinson disease and dementia were associated with the largest disutility (approximately 0.4). Respondents with stroke and depression had estimated disutilities of >0.2. The pattern of disutility was similar between the EQ-5D-5L and HUI3, but differences in values were considerable for some diseases (eg, dementia and ear disease). Some diseases did not cause a significant decrease in the EQ-5D-5L and HUI3 values (eg, thyroid disease, hyperlipidemia, hypertension, asthma, tooth diseases, and gout).
Table 4Disutility associated with disease.
Variable
N
EQ-5D-5L
HUI3
Coefficient
P-value
Coefficient
P-value
Intercept
0.989
<.001
0.897
<.001
Age, y
16-19
743
—
—
—
—
20-29
1279
−0.018
.001
−0.023
.006
30-39
1265
−0.019
.001
−0.018
.025
40-49
1292
−0.018
.001
−0.004
.670
50-59
1302
−0.028
<.001
−0.021
.013
60-69
1370
−0.021
<.001
−0.013
.121
70-79
1433
−0.057
<.001
−0.042
<.001
80-89
1499
−0.129
<.001
−0.145
<.001
Sex
Male
5057
—
—
—
—
Female
5126
−0.011
<.001
0.011
.003
Main disease
No disease
5693
—
—
—
—
Diabetes
341
−0.046
<.001
−0.055
<.001
Obesity
8
−0.034
.417
0.019
.780
Hyperlipidemia
135
−0.002
.816
−0.004
.782
Thyroid disease
45
−0.006
.722
−0.046
.097
Depression
140
−0.184
<.001
−0.282
<.001
Dementia
27
−0.222
<.001
−0.426
<.001
Parkinson disease
15
−0.352
<.001
−0.421
<.001
Other neuropathic diseases
49
−0.211
<.001
−0.232
<.001
Eye diseases
158
−0.049
<.001
−0.101
<.001
Ear diseases
33
−0.033
.114
−0.108
.001
Hypertension
838
−0.005
.275
−0.006
.388
Stroke
92
−0.265
<.001
−0.293
<.001
Angina or myocardial infarction
131
−0.073
<.001
−0.081
<.001
Other cardiovascular diseases
142
−0.054
<.001
−0.068
<.001
Cold
18
−0.083
.004
−0.040
.368
Allergic rhinitis
68
−0.027
.069
−0.049
.029
COPD
11
−0.114
.002
−0.075
.161
Asthma
85
−0.025
.0603
−0.035
.086
Other respiratory diseases
72
−0.121
<.001
−0.116
<.001
Gastroduodenal diseases
90
−0.053
<.001
−0.040
.040
Liver or gallbladder diseases
53
−0.067
<.001
−0.072
.004
Other gastrointestinal diseases
87
−0.087
<.001
−0.085
<.001
Tooth diseases
214
−0.013
.129
−0.021
.091
Atopic dermatitis
78
−0.031
.025
−0.049
.014
Other dermatologic diseases
91
−0.043
.001
−0.037
.050
Gout
37
−0.012
.556
−0.022
.460
Rheumatoid arthritis
63
−0.103
<.001
−0.096
<.001
Arthrosis
194
−0.157
<.001
−0.101
<.001
Shoulder pain
108
−0.051
<.001
−0.043
.014
Back pain
323
−0.117
<.001
−0.100
<.001
Osteoporosis
58
−0.034
.034
−0.054
.025
Kidney diseases
83
−0.079
<.001
−0.122
<.001
Benign prostatic hyperplasia
89
−0.029
.024
−0.054
.008
Menopausal problem
7
−0.048
.286
−0.007
.925
Bone fracture
45
−0.124
<.001
−0.125
<.001
Injury without bone fracture and burn injury
38
−0.134
<.001
−0.075
.009
Anemia or blood diseases
24
−0.079
.001
−0.121
.001
Malignant neoplasm
101
−0.084
<.001
−0.103
<.001
Pregnant or postpartum disorders
13
−0.058
.082
0.019
.695
Infertility
8
−0.077
.071
−0.118
.059
Other
273
−0.080
<.001
−0.101
<.001
Unknown
5
−0.093
.082
−0.053
.550
R2
0.276
0.180
Coefficient >0.1 and significant
0.05< Coefficient <0.1 and significant
Coefficient <0.05 and significant
COPD indicates chronic obstructive pulmonary disease; HUI3, Health Utilities Index Mark 3.
Table 5 shows the relationships between the EQ-5D-5L and HUI3 values and the respondents’ self-reported main symptoms. Almost all symptoms listed in Table 5 were associated with disutilities significantly larger than 0, with effect sizes exceeding 0.05. Symptoms that had the most significant influence on disutilities included “limb motion problems” (0.300), “chest pain” (0.212), and “acraturesis” (0.211) for the EQ-5D-5L and “limb motion problems” (0.328), “forgetfulness” (0.301), and “hearing impairment” (0.268) for the HUI3.
We conducted a large random sample survey to establish the Japanese population norms of the EQ-5D-5L and HUI3. We also constructed a catalog of disutility values by disease and symptom for use in QALY calculations. For example, the EQ-5D-5L value of women aged 50 to 59 years with depression can be calculated thus: 0.989 (intercept) – 0.028 (age) – 0.011 (sex) – 0.184 (disease) = 0.766. The updated values were comparable to and consistent with those of our previous survey in 2013, suggesting that our survey method is reproducible and reliable. The established population norms are based on responses from the general population living at home (those who were hospitalized or living in nursing homes were excluded) and are thus considered “community-based” values that do not reflect acute or severe health states. In addition, to estimate the size of the disutility, we only used the main disease or symptom reported by the respondents. Interactions between multiple diseases were not considered.
In the present study, the ICC between the EQ-5D-5L and HUI3 values was 0.53. In our previous survey, the ICCs were 0.802 between the EQ-5D-3L and EQ-5D-5L, 0.249 between the EQ-5D-3L and SF-6D, and 0.234 between the EQ-5D-5L and SF-6D.
The ICC between the EQ-5D-5L and HUI3 was larger than that between the EQ-5D-5L and SF-6D, suggesting that the EQ-5D-5L index values are more comparable to those of HUI3 than to those of SF-6D. Nevertheless, it should be noted that the EQ-5D-5L values were generally higher than the HUI3 values (an average of approximately 7% higher).
Although the EQ-5D-5L dimensions (“mobility,” “self-care,” “usual activities,” “pain･discomfort,” and “anxiety･depression”) and HUI3 attributes (“ambulation,” “dexterity,” “emotion,” and “pain”) share similar items, the HUI3 is more sensitive to the influence of dementia and ear/eye disease (which includes the related items “vision,” “hearing,” and “cognition”). In fact, the symptoms “forgetfulness” (0.301) and “hearing impairment” (0.268) have large disutility coefficients. Only the HUI3 has the “speech” item (speech-related diseases and symptoms were excluded in the present survey). Owing to these differences (forgetfulness and hearing impairment), either the EQ-5D-5L and HUI3 are sensitive to certain diseases and symptoms. The mixed use of measurements using both instruments may thus result in an arbitrary analysis, which could be misleading with regard to favorable analysis conclusions.
With regard to the relationships of the 2 instruments with demographic factors, education level and household income had a significant influence. For the EQ-5D-5L, similar trends have been reported in Spain,
In addition, the EQ-5D-5L index values are slightly higher in men than in women, regardless of country. The effect size of sex is approximately 0.01 in Japan, and similar values have been reported in other countries (0.019 in Trinidad and Tobago,
). Although this sex-based difference is commonly observed in many countries, the underlying reason is unclear.
This study has a limitation. Although our survey was based on a rigid random sampling method, the response rate was lower than that of the previous survey. The lower response rate may be due to changes in Japan’s social environment in the past decade. Japanese individuals now pay more attention to security and personal information than before because their importance has been stressed by the government and police from the perspective of preventing crime. Thus, people are more reluctant to participate in surveys, especially when a survey involves unfamiliar researchers visiting their homes and asking questions regarding their personal information. Indeed, some respondents suspected our survey to be fraudulent, and in 1 case, the police were called. In addition, younger individuals have increasingly variable lifestyles today, and it is difficult to hand out questionnaires in person (because researchers cannot be certain when they are home). Under these circumstances, the nonresponse rate for national surveys, to which Japanese citizens are required by law to respond, has increased greatly from 1.7% in 2000 to 13.1% in 2015. This issue represents a common problem in the current Japanese government statistics. Nonetheless, we believe that our sampling method resulted in no serious bias in terms of background factors. In a future study, nevertheless, the sampling method must be reevaluated to ensure a sufficiently high response rate for updating normative values.
Conclusion
We established the Japanese population norms of the EQ-5D-5L and HUI3 on the basis of a large random sample survey. In addition, disutility by disease and symptom was also estimated. Our findings can be used in healthcare decision making and may contribute to a more reliable analysis of economic evaluations and clarification of the characteristics of the EQ-5D-5L and HUI3, which will help in selecting suitable PBMs and offering appropriate interpretations.
Article and Author Information
Author Contributions:Concept and design: Shiroiwa, Noto, Fukuda
Acquisition of data: Noto
Analysis and interpretation of data: Shiroiwa, Noto, Fukuda
Drafting of the manuscript: Shiroiwa
Critical revision of the paper for important intellectual content: Noto, Fukuda
Statistical analysis: Shiroiwa
Supervision: Fukuda
Conflict of Interest Disclosures: The authors reported no conflicts of interest.
Funding/Support: This work was supported by the research budget of the National Institute of Public Health.
Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Health-related quality of life measured with the EQ-5D-5L: estimation of normative index values based on a representative German population sample and value set.
Quality of life of the Indonesian general population: test–retest reliability and population norms of the EQ-5D-5L and WHOQOL-BREF [published correction appears in PLoS One. 2018;13(8):e0203091].
Jiang R, Janssen MFB, Pickard AS. US population norms for the EQ-5D-5L and comparison of norms from face-to-face and online samples. 2021;30(3):803-816.
Health Utilities Index Mark 3 scores for major chronic conditions: population norms for Canada based on the 2013 and 2014 Canadian Community Health Survey.
Development of a multiplicative, multi-attribute utility function and eight single-attribute utility functions for the health utilities index mark 3 in Japan.
Estimation of an instrument-defined minimally important difference in EQ-5D-5L index scores based on scoring algorithms derived using the EQ-VT version 2 valuation protocols.
00171-6&id=gr1.jpg){kind=link}