Health-Related Quality of Life in Patients With Different Diseases Measured With the EQ-5D-5L: A Systematic Review

1As a quantitative indicator of health-related quality of life (HRQoL), the health utility reflects people's preference for a given health state. The health utility is measured on a scale from zero to one, where zero represents death and one represents full health. The worse the perception of the health status is, the lower the utility value. It can be a negative value when a health state is perceived as being worse than death. There are several preference-based measurement tools for health utility, such as the EuroQol 5 dimensions (EQ-5D) family of instruments, the Short Form-6 Dimensions (SF-6D), and the Health Utilities Index (HUI). Health utility can be used as quality-of-life weight to calculate QALYs in cost-utility analysis (CUA). Thus, health utility plays an important role not only in the measurement of HRQoL but also in health economics evaluations.

2The EQ-5D, developed by the European Quality of Life Group (EuroQol Group), is currently one of the most widely used questionnaires in HRQoL research. The original version of the EQ-5D was introduced in 1990 and contains five dimensions: Mobility, Self-Care, Usual Activities, Pain/Discomfort, and Anxiety/Depression. For each dimension, there were three levels to describe the severity, namely, have no problems, have some problems, and have extreme problems, which could describe 243 different health states. However, there may be some issues when using the EQ-5D-3L to detect small changes in mild conditions, and the EQ-5D-3L had obvious ceiling effects. Therefore, in 2005, the EuroQol Group developed a new version of the EQ-5D based on the same five dimensions but with five rather than three severity levels (EQ-5D-5L); this instrument could detect 3,125 unique health states. Published studies have shown that compared with the EQ-5D-3L, the EQ-5D-5L was significantly more sensitive, with reduced ceiling effects.

3To derive health utility from the responses on the EQ-5D instruments, country-specific value sets need to be estimated. Since 2016, more than 20 countries and regions have published standard EQ-5D-5L value sets (Europe: 9; Asia: 9; Americas: 3; Africa: 1). In 2012, before any standard EQ-5D-5L value set was established, van Hout et al. developed a crosswalk project to map the EQ-5D-5L to the EQ-5D-3L, enabling researchers to obtain a crosswalk value set for the EQ-5D-5L based on published EQ-5D-3L standard value sets. Besides that, the psychometric properties of the EQ-5D-5L have been validated in both general and disease populations.

4In recent years, with the availability of the EQ-5D-5L value sets, an increasing number of studies have used the EQ-5D-5L to measure the HRQoL of patients with different diseases and perform economic evaluations to support health decision-making. At present, a comprehensive review of these studies is lacking. For HRQoL measured with EQ-5D-5L, cross-sectional studies mainly focus on the current health status of the patients while randomized controlled trials (RCTs) pay attention to the effects of different interventions on health outcomes. This study focuses on the use of the EQ-5D-5L to explore the variation in health utility in patients in different conditions, provide information to perform CUAs, and inform health policies.

5This systematic review and meta-analysis was performed in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The protocol was registered on PROSPERO with ID CRD42020158694 (https://www.crd.york.ac.uk/PROSPERO/). Literature searches were conducted in Medline via Ovid, Embase via Ovid, The Cochrane Library, and Web of Science from January 2012 to October 2019 with combinations of the following search terms: “quality of life,” “QoL,” “HRQoL,” “HRQL,” “EQ-5D,” “EQ-5D-5L,” “five level,” “EuroQol,” “five dimensions,” “randomized controlled trial,” “RCT,” and “diseases” (details in Supplementary Table 1).

6According to the selection criteria, all studies were original cross-sectional studies reporting EQ-5D-5L utilities for any specific disease with or without comorbidities and using country-specific value sets or the crosswalk method (mapping from EQ-5D-3L). Due to the lack of EQ-5D-5L standard value sets in many countries, the crosswalk method is the most important value set to calculate utility measured by EQ-5D-5L. In addition, the crosswalk method is recommended by the National Institute for Health and Care Excellence (NICE) to perform CUA when EQ-5D-5L is used to measure health outcomes in England. Therefore, it is useful and necessary to include these articles in this review. Studies reported that multiple utility values using value sets from different countries in the same published article were also included. The language of publication was limited to English. This review excluded reviews, protocols, or abstracts; studies focused on the general population; longitudinal studies or effects evaluation studies of different interventions; studies that reported only synthetic utilities of multiple diseases, non-EQ-5D-5L utilities, or no utilities; and studies unrelated to HRQoL.

7After removing duplicates, title and abstract screening was conducted by two authors independently. Following the application of the selection criteria, all eligible studies with full-texts were read, and the relevant references were checked manually. Two researchers independently collected the data using a predesigned data extraction table, including author, publication year, country or region, sample size, disease type, mean age, health utility, EQ-5D VAS score, proportions with problems in the five dimensions, value set, and administration method (i.e., face-to-face, telephone survey). When there was any discrepancy between the two researchers, it was resolved by discussion.

8Quality assessment was conducted with the 11-item cross-sectional research checklist developed by the Agency for Healthcare Research and Quality (AHRQ). According to the description in the study and the AHRQ checklist, the reviewer selects one of three options (“Yes,” “No,” and “Unclear”) for each item. “Yes” was assigned one point, while “No” or “Unclear” was assigned zero points. The quality level of each study was determined by summing all the item scores. For each assessed study, 0–3 points indicated low quality, 4–7 points indicated moderate quality, and 8–11 points indicated high quality.

9This review involved the analysis of the range of mean health utility values of the overall sample (or subgroups when there is no overall utility value reported) among different studies and value sets used in each study for a specific disease with or without comorbidities. In addition, this study reports the ranges in mean EQ-VAS scores and responses on each dimension of the EQ-5D-5L.

10Meta-analysis was performed to synthesize utility data when three or more studies reported utility values and standard error/deviation for a specific disease. For any study that reported multiple utility values for the same sample using different EQ-5D-5L value sets, the average value or the utility calculated by using a local country-specific value set was applied in meta-analysis. Heterogeneity was assessed with the I2 statistic. Random-effect (DerSimonian–Laird estimator method) and fixed-effect (inverse variance method) models were both used to calculate the pooled utility for a specific disease. Sensitivity analysis was conducted by removing EQ-5D-5L utility values derived from crosswalk value sets. All analyses were performed with R (version 4.0.5).

11A total of 9,500 articles were identified from the four databases, and four additional studies were obtained from the manual search. After eliminating duplicates, 6,409 documents were screened to assess eligibility, of which 98 articles were finally included in qualitative analyses and 35 studies were included in meta-analyses (Figure 1). Those 98 articles involved 98,085 patients. The included studies were published between January 2006 and March 2018 (Table 1). Except for three studies that only included male patients and one study that only included female patients, the rest of the studies included patients of both sexes. Twenty studies did not report the mode of administration. Of the remaining 78 studies, 47.4% involved the face-to-face administration of the survey, 47.4% involved self-administered surveys, and 5.2% involved telephone surveys. The AHRQ checklist scores ranged from four to nine points, the median was six points, and the mode was five points (details in Supplementary Table 2). There were no low-quality studies; 87 studies and 11 studies were of moderate and high quality, respectively. The data about the distributions of EQ-5D-5L are summarized in Supplementary Table 3.

12In this review, health utility values derived from the EQ-5D-5L were reported for 50 different diseases. Among these, diabetes mellitus, neoplasms, multiple sclerosis, cardiovascular disease, chronic obstructive pneumonia disease (COPD), human immunodeficiency virus (HIV) infection, chronic kidney disease, and fracture were reported in three or more studies and meta-analyses were performed for these diseases (fracture was not included in meta-analysis, because only two of the studies reported standard error/deviation). The sensitivity analysis results (remove all the utility values derived from the crosswalk value set) are presented in Supplementary Figure 1.

13For patients with diabetes mellitus (Table 2), 12 studies reported health utility values ranging from 0.31 to 0.99. The Chinese standard EQ-5D-5L value set and Crosswalk UK value set were used to derive the utility values in the studies that reported the highest value and lowest value, respectively. The former focused on diabetes patients without diabetic retinopathy with a mean disease duration of 10.3 years and a mean age of 67.9 years, while the latter involved patients with severe comorbidities on hemodialysis, with a mean age of 60.3 years. Additionally, Lamu et al. used eight country value sets (England, the Netherlands, Spain, Canada, Uruguay, China, Japan, and Korea) to analyze 924 diabetic patients from six countries. The results showed that the utility value calculated with the Uruguay value set was the highest at 0.880, while the lowest, 0.735, was derived with the value set from the Netherlands. The EQ-5D VAS scores were reported to range from 50.9 to 72.6 in six studies. Among the five dimensions of the EQ-5D-5L, pain/discomfort was the dimension with the most reported problems. The prevalence of diabetes comorbidities ranged from 55 to 100%, which was one of the most important factors negatively affecting the HRQoL of patients.

14The meta-analytic utility estimate of diabetes mellitus was 0.83 (95% confidence interval (CI) = 0.77–0.90, heterogeneity I2 = 100%, P = 0.00) using the random-effect model, and it was 0.93 (95% CI = 0.93–0.93) using the fixed-effect model. The results are presented in Figure 2A.

15Seven studies reported health utility values for cancer patients ranging from 0.62 to 0.90. The highest utility value was in early-stage prostate cancer patients using the crosswalk UK value set, while the lowest value was in colorectal cancer patients, 49.7% of whom had stage III–IV disease, applying the China value set. The EQ-5D VAS scores ranged from 56.2 to 77.5 in two studies. The decrease in health utility in cancer patients was mainly due to problems related to the pain/discomfort dimension of the EQ-5D-5L. As the cancer progressed, the health utility value decreased.

16The pooled utility value of cancer patients was 0.75 (95% CI = 0.68–0.82, heterogeneity I2 = 96%, P <0.01) using the random-effect model, and it was 0.80 (95% CI = 0.78–0.81) using the fixed-effect model (Figure 2B).

17The health utility ranged from 0.31 to 0.78 for multiple sclerosis patients in six studies. The upper and lower utility values were generated with the crosswalk France value set and the crosswalk UK value set, respectively. The study with the highest value reported a shorter disease duration (9 vs. 15 years) than the study with the lowest utility value. In addition, the former had a higher proportion of relapsing–remitting multiple sclerosis patients than the latter (71.5 vs. 52.8%). EQ-5D VAS scores ranged from 58.3 to 78.0 in five studies. Pain/discomfort and usual activities were the dimensions with the most reported problems among multiple sclerosis patients.

18The meta-analytic utility estimate of multiple sclerosis patients was 0.56 (95% CI = 0.47–0.66, heterogeneity I2 = 99%, P <0.01) using the random-effect model, and it was 0.67 (95% CI = 0.66–0.68) using the fixed-effect model (Figure 2C).

19For cardiovascular disease patients, the health utility values ranged from 0.56 to 0.85 in eight studies. The lowest value was derived from the Chinese value set, while the study with the highest value did not report the value set used. In the study with the highest utility value, all patients were evaluated 4 years after cardiac arrest, and the proportion of men was 80%. In the study with the lowest value, the patients had atrial fibrillation; 43% of them were men, and 23% had diabetes mellitus. Berg et al. compared utility values among nine subgroups of patients with different cardiovascular diseases. Among these subgroups, heart transplant patients had the highest value, which was 0.82, while arrhythmia patients had the lowest value, which was 0.70. The EQ-5D VAS scores ranged from 61.4 to 77.8 in six studies. Anxiety/depression and pain/discomfort were the dimensions with the most reported problems among cardiovascular disease patients.

20The pooled utility value of cardiovascular disease patients was 0.77 (95% CI = 0.75–0.79, heterogeneity I2 = 99%, P <0.01) using the random-effect model, and it was 0.76 (95% CI = 0.75–0.76) using the fixed-effect model (Figure 2D).

21For patients with COPD, the health utility values ranged from 0.68 to 0.79 in four studies. The crosswalk US value set and UK standard EQ-5D-5L value set were used in the studies that reported the highest utility value and the lowest value, respectively. The mean age of COPD patients in the study reporting the lowest utility was 70.4 years, and the mean predicted forced expiratory volume in 1 s (FEV1) was 49.8%. Meanwhile, the patients in the study with the highest value had a younger mean age (68.5 years old) and a better predicted FEV1. The EQ-5D VAS scores ranged from 60.5 to 70.6 in four studies. Mobility was the dimension with the most problems affecting the HRQoL of COPD patients based on EQ-5D-5L. In addition, as the predicted FEV1 decreased, the health utility value in COPD patients decreased.

22The synthesized utility value of COPD patients was 0.75 (95% CI = 0.71–0.80, heterogeneity I2 = 96%, P <0.01) using the random-effect model, and it was 0.76 (95% CI = 0.75–0.77) using the fixed-effect model (Figure 2E).

23The health utility values of patients infected with HIV ranged from 0.65 to 0.90 in four studies, and both extreme values were derived with a crosswalk value set [Thailand and Spain]. The study with the highest utility value involved patients in relatively good condition and without any comorbidities, while the study, with the lowest value focused on patients who had symptomatic HIV infections. The EQ-5D VAS scores ranged from 68.8 to 88.6 in four studies. The decrease in utility in HIV-infected patients was mainly due to problems related to the anxiety/depression dimension of the EQ-5D-5L.

24The pooled utility value of patients infected with HIV was 0.84 (95% CI = 0.80–0.88, heterogeneity I2 = 83%, P <0.01) using the random-effect model, and it was 0.81 (95% CI = 0.80–0.82) using the fixed-effect model (Figure 2F).

25For chronic kidney disease patients, the health utility values ranged from 0.37 to 0.89 in three studies. The Japan value set and crosswalk UK value set were used to calculate the highest utility value and the lowest value, respectively. The mean age of chronic kidney disease patients in the study reporting the highest value was 49.8 years old, and all of them had received kidney transplants, while those in the study reporting the lowest value were 59.4 years old, and 33.7% of them had been on dialysis for 4 years or longer. One study reported that the EQ-5D VAS score was 59.4. Among the five dimensions, self-care was the dimension with the most reported problems among chronic kidney disease patients.

26The meta-analytic utility estimate of chronic kidney disease was 0.70 (95% CI = 0.48–0.92, heterogeneity I2 = 99%, P <0.01) using the random-effect model, and it was 0.76 (95% CI = 0.74–0.78) using the fixed-effect model (Figure 2G).

27The health utility values of patients with fractures ranged from 0.56 to 0.88 in the three studies. However, neither of the studies that reported the maximum and minimum values described the value sets used. The patients in the study reporting the highest value had midshaft clavicular fractures and a much younger mean age (44.5 vs. 73.5 years old) than the osteoporotic vertebral compression fracture patients in the study reporting the lowest value. Two studies reported EQ-5D VAS scores of 80.3 and 77.2. No information was available for the dimensions that contributed the most to the HRQoL of fracture patients.

28For Prader–Willi syndrome, hypertension, ulcerative colitis, ankylosing spondylitis, psoriasis, actinic keratosis, Parkinson's disease, overactive bladder, hereditary angioedema, spinal cord injury, schizophrenia, hemophilia, asthma, and hepatitis, only two studies reported the health utility values for patients with each disease. For the remaining 29 diseases, the HRQoL and utility values were only reported by one study each. Patients with adolescent idiopathic scoliosis had the highest utility value of 0.93, while children with Morquio A syndrome, who must use wheelchairs, had the lowest value of −0.18.

29Furthermore, two studies compared utility values calculated with different country-specific value sets in the same sample. For patients with psoriasis living in central South China, value sets for Japan, China, and the UK were used separately to obtain the EQ-5D-5L utility values, and the results were 0.86, 0.90, and 0.90, respectively. van Dongen-Leunis et al. used two EQ-5D-5L country-specific value sets to calculate the health utility of acute leukemia patients, and the value derived from the Dutch value set (0.81) was lower than that derived from the UK value set (0.85). The rest of the studies all used a single value set. Compared with other dimensions, pain/discomfort was the dimension with the most problems reported by patients in most of the studies.

30In this study, we reviewed the health utility values in patients with different diseases according to the EQ-5D-5L in cross-sectional surveys. We found that the EQ-5D-5L has been widely applied in populations with specific diseases, including various chronic non-communicable diseases, such as diabetes mellitus, neoplasms, multiple sclerosis, and cardiovascular disease, and infectious diseases, such as HIV and Dengue fever. The health utility values for a specific disease measured by the EQ-5D-5L differed based on patient characteristics, survey location, the use of country-specific value sets, and other factors. Meta-analyses were performed to synthesized utility data of any specific disease reported in three or more studies.

31Health utility measures the preference of people for a given health state and reflects their status with regard to quality of life. Sex is one of the factors that affect health utilities. There are differences in the perception of health status between males and females, and in most of the included studies that reported sex-specific utilities, men had better HRQoL as measured by the EQ-5D-5L than women. For instance, the utility value was 0.80 for men with COPD and 0.69 for women with COPD, and the proportion of men who reported having problems on all five dimensions was lower than the proportion of women. In addition, health utility values decreased as the age of patients increased due to the deterioration of physical function and reduced disease tolerance. Among patients with COPD, for example, the utility value for patients under 65 years of age (0.77) was lower than that for patients who were 65 years old and older (0.79).

32In general, the severity of disease is reflected by the magnitude of the health utility value. The variation in values measured by EQ-5D-5L for the same disease under different conditions reflects its discriminative ability. As the disease progresses, the utility value decreases. Alvarado-Bolaños et al. used Hoehn and Yahr staging to categorize Parkinson's disease patients into groups with mild, moderate, and severe disease, and the utility values were 0.77, 0.65, and 0.47, respectively. In addition, the number of comorbidities and the different types of comorbidities substantially affect the HRQoL of patents. Patients who have comorbidities usually report a lower utility value than those without comorbidities. Van Duin et al. reported that the utility value was 0.90 in patients with HIV infections who did not have any comorbidities; however, it was reduced to 0.84 when patients had comorbid diseases. In Al-Jabi's study, for hypertension patients with one, two, and three or more comorbidities, the utility values were 0.81, 0.73, and 0.66, respectively.

33Various living environments result in different lifestyles, which may influence HRQoL and health utility. Zyoud et al. reported that among patients with end-stage renal disease in Palestine, those living in villages had a higher mean utility value than those living in cities (0.44 vs. 0.29). In another study, among patients with cardiovascular disease, the utility value was a little bit higher for those living in urban Vietnam than those in rural areas (0.82 vs. 0.81).

34To calculate health utility, the target patients' responses to the EQ-5D-5L and a country-specific value set are needed. The health preferences of patients living in different countries are affected by their social environment, living standards, and health system. Therefore, the EQ-5D-5L value sets estimated based on residents' preferences for health states vary across countries or regions. Different results can be observed in the same sample when various country value sets are used to calculate health utility values. In the same sample of patients with acute leukemia, van Dongen-Leunis et al. reported that the value obtained with the Dutch value set was higher than that obtained with the UK value set. In countries where the EQ-5D-5L utility value set has been estimated, it is more appropriate to use the local value set. Before any standard country-specific EQ-5D-5L value set was published, the crosswalk method developed by van Hout et al. in 2012 was an alternative means of calculating health utility measured by EQ-5D-5L. For cost-utility analyses performed in England, the NICE recommends the use of the crosswalk method to obtain EQ-5D-5L utility values and calculate quality-adjusted life-years (QALYs) because there are some concerns about the current standard value set published by Devlin et al.. In this review, a crosswalk value set was used in half of the studies to calculate utility values due to the lack of a local standard EQ-5D-5L value set when the survey was conducted. Therefore, the crosswalk value set is still important for researchers to calculate health utility.

35The heterogeneity of health utility derived from different studies for any specific disease is significant. Although, this may lead to some issues of the direct comparison among these studies, the trend of variation and the influence factors of health utility can be observed. In addition, to perform CUA, different sources of health utilities are need to be identified and applied in the model. The summarization and review of health utility for different diseases are helpful and useful.

36There are some limitations of this study. Among the 50 different diseases analyzed in this review, nearly half of them were only discussed in one study each. The included studies were limited to those published in English. In addition, some of the studies did not describe the value set used. This review focused on health utility measured by the EQ-5D-5L in cross-sectional studies, and the comparison of different utility-based instruments (i.e., SF-6D, HUI) in populations with specific diseases needs further exploration.

37A deeper understanding of the HRQoL and health utility of patients with different diseases facilitates the provision of a more appropriate range of services for disease management and treatment. In addition, health utility is used for HRQoL weighting when calculating QALYs. QALY is used as the outcome measure in CUA and plays an important role in health technology assessments. The summarization of health utility from various sources provides information to perform CUA which could inform health decision making and the reasonable allocation of health resources.

38The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

39TZ, HG, and AM designed this study protocol. HG, AM, and TZ conceived the literature strategies. LW and MR reviewed the title/abstract independently. TZ and LW performed the original study review. TZ, HG, and YZ extracted and analyzed the data from included studies. TZ and MR assessed the methodological quality with AHRQ checklists. TZ and YZ contributed to the writing of the manuscript. All the authors approved the final version of this systematic review.

40The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.