Henry H. Bailey 1,2,3
1Department of Economics, St. Augustine Campus, The University of the West Indies
2 HEU, Centre for Health Economics, St. Augustine Campus, The University of the West Indies
3Arthur Lok Jack Global School of Business, The University of the West Indies
Dr. Henry Bailey
Department of Economics,
St. Augustine Campus,
The University of the West Indies,
Trinidad and Tobago.
E-mail: [email protected]
Copyright: This is an open-access article under the terms of the Creative Commons Attribution License which permits use, distribution, and reproduction in any medium, provided the original work is properly cited.
©2021 The Authors. Caribbean Medical Journal published by Trinidad & Tobago Medical Association
Quality of life measures are increasingly being used in clinical practice and in research. Internationally EQ-5D is the most commonly used generic health related quality of life instrument. The EQ-5D system comprises five dimensions: mobility, self-care, ability to conduct usual activities, pain/discomfort and anxiety/depression. This study sought to provide a set of population norms for Trinidad and Tobago to facilitate clinical and other researchers who wish to use the EQ-5D-3L value set that was recently created for Trinidad and Tobago. Population norms are baseline values against which the EQ-5D health status of patients or patient groups can be compared.
Self-reported health data using the EQ-5D-3L questionnaire were taken from two previously undertaken surveys of representative samples of the Trinidad and Tobago population. EQ-5D-3L health states, mean EQ VAS (visual analogue scale) scores and mean EQ-5D-3L index values were obtained for demographic groups.
Data were obtained for 590 respondents. The mean EQ VAS value was 82.3 and the mean EQ-5D-3L index value was 0.931. The EQ-5D dimensions with the highest rates of reported problems were pain/discomfort (29%) and anxiety/depression (14%). Self-reported health was observed to decline as age progresses and men reported higher levels of health than women.
The population norms in this study can be used as base-line values for key demographic groups in clinical practice and in clinico-economic research using EQ-5D-3L.
Health-related quality of life (HRQOL) instruments aim to capture the respondent’s (or patient’s) perspective on how they are affected by their state of health. Such instruments are increasingly being used in clinical practice and in research. One generic HRQOL instrument is the EQ-5D-3L health classification system developed by the EuroQol Group. This is a measure of health status for clinical use and economic analysis. The EQ-5D instrument comprises five dimensions: mobility, ability to perform self-care (bathing, dressing etc.), ability to perform usual activities (work, study, leisure etc.), pain/discomfort and anxiety/depression. Each dimension has 3 levels in the 3L version of the instrument (with level 1 being no problems, level 2 being some/moderate problems, and level 3 being extreme problems). A respondent completes the EQ-5D-3L questionnaire by indicating their health status on each dimension (using the appropriate level) and by completing the EQ VAS (visual analogue scale). This is a 0-100 scale on which 0 and 100 represent the worst and best levels of health imaginable to the respondent.
The respondent’s EQ-5D-3L state is coded by the level of problems reported by the respondent in the order: mobility, self-care, usual activities, pain/discomfort, anxiety/depression. For example, EQ-5D-3L state 21321 would indicate that the respondent has reported:
2: some problems walking (mobility)
1: no problems bathing/dressing (self-care)
3: extreme problems performing usual activities,
2: moderate pain/discomfort and
1: not anxious or depressed.
With five dimensions in three levels, the EQ-5D-3L instrument allows 35=243 possible states. An index value can be calculated for each EQ-5D-3L state. This gives a measure of how good or bad a community believes the state is. For example, EQ-5D-3L state 11121 (moderate pain with no problems on the other 4 dimensions) might be preferred over state 21121 (some problems with mobility and moderate pain with no problems on the other 3 dimensions). In this example state 21121 would have a lower index value than 11121. Index values can be estimated for the full set of states for a country or region, and this has been done for Trinidad and Tobago.1 These values are obtained in valuation studies in which a representative sample of citizens perform guided valuation tasks following a valuation protocol on a subset of EQ-5D states. Regression methods are then used to obtain values for all 243 states based on the valuation subset. The resulting index values are used as Quality of Life adjustments in Cost per QALY analysis to include the preferences of the population in the economic evaluation of health interventions.
The EQ-5D-3L system therefore provides three measures for a respondent: a health state, the EQ VAS value (i.e. the respondent’s subjective value of his/her own health) and the EQ-5D-3L index value which gives the relative value that society places on the individual’s state of health on a scale anchored at 1 (full health) and 0 (dead). These states and values can be used in economic evaluation of health interventions, and they can also be used to track patients, or patient groups through the course of an illness or treatment. For example, this can be done for a cohort of patients by tracking the relative frequency of problems reported on the 5 dimensions, or by tracking mean EQ VAS and/or index values. In order to use EQ-5D in this way, clinicians and analysts often need a set of population norms. These are the ‘base-line’ EQ-5D states, index values and EQ VAS scores of the normal population, adjusted for important socioeconomic factors (usually age and gender) such that the values used for comparison are drawn from a reference group that is demographically similar to the disease or treatment group.
Internationally, the EQ-5D health classification system is the most commonly used measure of health-related quality of life and QALY adjustment values. For over 10 years, EQ-5D has been the instrument recommended by the National Institutes for Health and Clinical Excellence in the United Kingdom for evaluations of health interventions.2 Most developed countries have EQ-5D index value sets.3 The list of countries in which EQ-5D based preference weights are used, encouraged or required by government agencies in applications for reimbursement or market access for medicines has grown to include many developing countries including Brazil, Chile, Colombia, Egypt, Iran, Malaysia, Thailand and several Eastern European countries.4-11.
Recently, a 5-level instrument (EQ-5D-5L) was developed by the EuroQol Group which includes two intermediate levels (slight and severe) as levels 2 and 4. These were added to make the instrument more sensitive. EQ-5D-5L index values have been estimated for the Trinidad and Tobago population using a crosswalk algorithm on the EQ-5D-3L values.12 A set of EQ-5D-5L population norms has also been published for Trinidad and Tobago but no EQ-5D-3L population norms have been published for Trinidad and Tobago.13
The EQ-5D 3 level and 5 level health classification systems, along with the associated visual analogue scale (EQ VAS) values and EQ-5D index values are now being used to capture health outcomes data in several clinical and clinico-economic studies in Trinidad and Tobago. The disutility associated with certain illnesses and socioeconomic factors has been estimated using the EQ-5D-5L instrument in one large study of the general population.14 The purpose of this paper is to provide a set of EQ-5D-3L population norms for Trinidad and Tobago to be used by researchers who are using (or wish to use) the EQ-5D-3L instrument.
The EQ-5D-3L Self-Reported health instrument was included in two Trinidad and Tobago studies. The first survey of these two took place in 2013 and included 283 respondents.15 The second survey was the Trinidad and Tobago EQ-5D-3L valuation study undertaken in 2015 which included 307 respondents.1 Both of the surveys used the same EQ-5D-3L Self-Reporting instrument that had been linguistically validated for Trinidad and Tobago in 2013. Both of the surveys were also conducted by the same principal investigator using the same survey firm for which interviewers received the same training on the instrument. In both cases, the samples were generally representative of the Trinidad and Tobago population along age, gender, ethnicity and area of residence.
Interviews were conducted face-to-face. For the 2013 survey, streets were randomly selected from each of the 5 geographic regions served by the 5 Regional Health Authorities (RHAs). For the 2015 survey, enumeration districts were selected from Central Statistical Office of Trinidad and Tobago Enumeration District maps to ensure representativeness of the sample. Streets were then randomly selected in these enumeration districts. Alternate homes were visited and respondents were selected on the basis of being at least 18 years old and having the next, or most recent birthday in the household. Call back cards were left whenever the person with the next or most recent birthday was not available. Once 160 interviews had been completed in the 2013 survey, and once 220 had been completed in the 2015 survey, comparisons were made with the national population and the survey company was given demographic guidelines by age, gender, education and ethnicity for the remaining respondents to bring the sample as close to the national population as possible.
Both surveys went on to present respondents with various EQ-5D valuation tasks as part of an investigation into valuation methods which are discussed elsewhere.1, 15, 16 The self-reported health questions were asked at the start of both surveys, so there would have been no difference in exposure to the EQ-5D system to the two groups up to the point at which the self-reported health questions were answered.
There were no exclusions of data from any respondents in producing the EQ-5D-3L norms reported in this study.
Means were calculated for EQ VAS, index values and ceiling effects for each demographic category based on age group, gender, education, ethnicity, and geographic region (based on within which RHA the respondent resided). Mean EQ VAS and index values were also obtained for age-gender sub-groups as were the rates of reporting each of the 3 levels on each of the 5 dimensions.
EQ-5D states were tabulated in order to identify the most commonly observed states and the most frequently observed problems within the 5 dimensions of EQ-5D.
All analyses were conducted using STATA version 14 (Stata Corp, College Station, Texas).
Table 1 shows the demographic characteristics of the sample compared with the population of Trinidad and Tobago. The sample is approximately representative of the population in age group, region and gender. The lower education group is over-represented and afro-ethnicity is under-represented. Table 1 also shows the mean index and EQ VAS values along with ceiling levels for each demographic group. The ceiling is the percentage of respondents in each demographic group that reported being in state 11111. There were no observations of state 33333, so there was no ‘floor effect’.
Table 1. Demographic characteristics of the sample compared to the Trinidad and Tobago population and mean EQ-5D index values, EQ VAS values and ceiling effects by demographic group.
Table 2. Seven states making up 90% of the sample.
Table 3. Mean EQ-5D Index and EQ VAS values for each age group by gender.
Table 4. Rates of reporting each level on the 5 dimensions of EQ-5D by age group and gender.
The mean index value is 0.931 and the mean EQ VAS score is 83.2. Self-reported health generally declines as age advances and males self-reported higher levels of health than females.
The p-values in Table 1 are associated with ANOVA analyses for the demographic groups (except for gender, where the p values are for t-tests since gender has only two categories). For age group (index values), ethnicity (EQ VAS values), education (index and VAS values) and regional health authority (EQ VAS) values, the data did not pass Bartlett’s test of equal variances, so Welch’s ANOVA was used.
Of the 35=243 states possible in the EQ-5D-3L classification system, 30 states were observed in the combined sample. Seven states accounted for 90% of the sample. These are displayed in Table 2.
The dimensions with the most frequently reported problems are pain/discomfort and anxiety/depression.
As a check for possible overlap in the two sub-samples, the respondents from each sub sample were placed into subgroups based on all five demographic variables in Table 2. These subgroups were then compared to identify any cases of identical demographic combinations in both sub samples. For example, a 33 year old university educated female of afro- ethnicity in village X in the 2013 survey could pair with a university educated female in the 35-44 age group of afro- ethnicity in the same village in 2015 and could in theory be the same individual (but not necessarily). Out of the 590 respondents only 3 such pairs were found-confirming that the two sub-samples comprised different individuals; i.e., only three cases allowed for the possibility that the same individual might have been counted twice (disregarding the possibility that people might also have moved during the two-year period between surveys).
Age and gender are known to be important drivers of health status, so reference values are provided in Table 3 which presents mean index and EQ VAS values and confidence intervals for each age group by gender.
Within each age group, men generally reported higher values than women, and the values generally declined as age increased. Where these trends were not observed (65+ and EQ VAS scores for the youngest age group) the confidence intervals overlap.
Table 4 shows the rates of reporting problems at all 3 levels for the 5 dimensions broken out by age group and gender.
The dimensions with the highest levels of reported problems were pain/discomfort, anxiety/depression, and mobility. Women reported higher levels of problems than men on all three of these dimensions.
In Table 1 the mean EQ VAS and index values were 83.1 and 0.931 which are very close to the respective values of 83.6 and 0.950 that were observed in the Trinidad and Tobago EQ-5D-5L Population Norms study.13 These values are relatively high when compared with the ranges of values observed in a review of 27 published EQ-5D-3L population norms studies where the EQ VAS scores ranged from 71.1 (Hungary) to 83.7 (Denmark) and the EQ-5D-3L index values ranged from 0.815 (Finland) to 0.951 (China).17 Comparisons of overall EQ-5D population norms between countries should always be taken with caution since not all population norms studies are based on representative samples. Relatively high self-reported health using EQ-5D-5L has also been reported in other Caribbean countries.18 The relative values that people place on states of health are subject to many influences including factors associated with national culture.19 These higher values found in Trinidad and Tobago and other Caribbean countries highlight the importance of using local population norms as a base-line when working with EQ-5D patient data.
The mean EQ VAS values and the ceiling effects show consistent declines as age increases. The index values show a similar pattern except for when moving from the 18-24 to the 25-34 age group and from the 55-64 to the 65+ age group. However, in Table 3 the difference between these two means is not statistically significant at the 5% level for either gender. Males were observed to have higher index, EQ VAS values and ceiling levels than females although the difference between EQ VAS values for the two genders was not statistically significant at the 5% level.
In Trinidad and Tobago, the average life expectancy at birth of women is 5 years higher than that of men.20 Despite this, the EQ-5D health outcomes for men in this study are all higher than for women. This finding of lower average life expectancy and higher self-reported health is very common. In most countries women live longer but men self-report higher levels of health using EQ-5D. In one review of published EQ VAS data, men reported higher scores than women in all but three countries.13 Several possible drivers of these differences have been suggested and investigated. These include fibromyalgia, muscular and pain disorders among women21,22,23, psychosocial factors and differences in perception between genders24.
Differences between ethnic groups were non-significant at the 5% level as were differences by regional health authority. For educational attainment, ceiling effects, index values and EQ VAS values were highest in the technical/vocational group and lowest in the lowest education group, but the F-statistic was only significant to the 5% level for the index values (not for the EQ VAS values). The general trend of measures in Table 1 rising with education level are similar to the findings of the EQ-5D-5L population norms although direct comparability is complicated by vocational and high school education combined into one educational level in the EQ-5D-5L population norms. Education level is known to affect health status and the lowest levels of education are usually associated with the lowest levels of health status.13,17,18 Several reasons have been put forward for this finding including lifestyle choices and compliance with treatment.25
In the EQ-5D-5L population norms study for Trinidad and Tobago, the top 10 EQ-5D-5L states accounted for 89% of the sample. In this EQ-5D-3L study, Table 2 shows that 6 EQ-5D-3L states accounted for 89.5% of the sample. This difference is attributable to the increased sensitivity of the 5 level instrument.12 In Table 2 the most common states involving problems at any level (i.e. states other than 11111) were those with moderate pain/discomfort (11121), moderate anxiety/depression (11112) and moderate problems on both of these dimensions together (11122). The same pattern was observed in the EQ-5D-5L population norms with slight problems on pain/discomfort, anxiety/depression and both of these together appearing in the same order for the most frequently observed states with problems at any level. In Table 3 as age rises, the declines in mean index and EQ VAS values are faster for females than for males. However, in the oldest age group (65+) females were observed to have higher mean index and EQ VAS values than males, and in the youngest age group (18-24) females had a higher mean EQ VAS value. The differences between males and females in these age groups are not significant at the 5% level, and confidence intervals for both measures show considerable overlap.
The findings of pain/discomfort, anxiety/depression and mobility having the highest rates of reported problems, with women reporting higher levels of problems on all three dimensions are consistent with the findings of the EQ-5D-5L population norms study for Trinidad and Tobago.
This study had some limitations. The sample size of 590 is small when compared to some other studies (for example the 2,036 respondents in the Trinidad and Tobago EQ-5D-5L population norms study) however EQ-5D-3L population norms have been published with samples of 464 (Greece), 534 (Sweden), and 620 (Japan)17. However combining the two Trinidad and Tobago datasets for this study is the only opportunity that currently exists for the creation of EQ-5D-3L population norms. Marital status is also known to affect EQ-5D self-reported health, however this variable was not necessary for the valuation studies that produced the dataset for this study, so this variable is not included.
The purpose of this study was to produce a set of EQ-5D-3L population norms for Trinidad and Tobago to facilitate researchers and clinicians who are currently using, or wish to use, the 3 level instrument (for example for comparison of patients in specific demographic groups that have been based on EQ-5D-3L). The values presented herein represent base-line values against which clinicians and researchers can compare patients or patient cohorts. The key findings for differences in self-reported health within the Trinidad and Tobago public include: higher health status for men than for women, a faster decline in health status for women as age progresses and lower health status among people of lower educational attainment. These are consistent with findings from other countries17 and with the findings of the EQ-5D-5L population norms for Trinidad and Tobago.13
Ethical Approval statement: Permission for data collection was obtained from the Department of Economics, the University of the West Indies.
Conflicts of Interest statement: None. The author is a member of the EuroQol Research Group.
Informed Consent statement: Not Applicable
Funding statement: No funding was sought or received for this work. The EQ-5D-3L Valuation Study which produced the 2015 survey data that was used in this study was funded by the Ministry of Health of Trinidad and Tobago and by the EuroQol Research Group.
- Bailey H, Stolk E, Kind P. Toward Explicit Prioritization for the Caribbean: An EQ-5D Value Set for Trinidad and Tobago. Value in Health Regional Issues. 2016; 11:60-67.
- NICE Guide to the methods of technology appraisal https://www.nice.org.uk/process/pmg9/resources/guide-to-the-methods-of-technology-appraisal-2013-pdf-2007975843781 (Accessed 8/1/2021).
- Szende A, Oppe M, Devlin N. EQ-5D Value Sets: Inventory, Comparative Review and User Guide. Dordrecht Springer; 2007.
- Ministerio de Salud de Chile. Guía Metológica para la evaluación económica de intervenciones en salud en Chile. http://www.orasconhu.org/case/sites/default/files/files/EE_FINAL_web.pdf (Accessed 8/1/2021).
- Colombia: Instituto de Evaluación Tecnológica en Salud. Manual para la elaboración de evaluaciones económicas en salud. Bogotá D.C.: IETS; 2014.
- Ministry of Health (Brazil): [Methodological guideline: economic assessment guideline: 2nd edition].http://bvsms.saude.gov.br/bvs/publicacoes/diretrizes_metodologicas_diretriz_avaliacao_economica.pdf (Accessed 8/1/2021).
- Elsisi G, Kalo Z, Eldessouki R, Elmahdawy M, Saad A, Ragab S. Recommendations for Reporting Pharmacoeconomic Evaluations in Egypt. Value in Health Regional Issues 2013; 2: 319-327
- Cheraghali AM. Newly Defined Role of Pharmacoeconomics in Iran National Medicine Policy. Shiraz E-Medical Journal. 2016 ; 17(1):e35258. doi: 10.17795/semj35258.
- Ministry of Health (Malaysia): Pharmacoeconomic guideline for Malaysia. https://www.pharmacy.gov.my/v2/sites/default/files/document-upload/pharmacoeconomic-guideline-malaysia.pdf (Accessed 8/6/2021).
- Pattanaphesaj J, Thavorncharoensap M, Ramos- Goñi J. et al The EQ-5D-5L Valuation study in Thailand. Expert Review of Pharmacoeconomics & Outcomes Research. 2018; doi:10.1080/14737167.2018.1494574
- Gulácsi, L., Rotar, A.M., Niewada, M., Löblová, O., Rencz, F., Petrova, G., Boncz, I., Klazinga, N.S.: Health technology assessment in Poland, the Czech Republic, Hungary, Romania and Bulgaria. European Journal of Health Economics 2014; 15(Suppl 1), S13–S25.
- van Hout B, Janssen MF, Feng YS, Kohlmann T, Busschbach J, Golicki D et al. Interim scoring for the EQ-5D-5L: mapping the EQ-5D-5L to EQ-5D-3L value sets. Value in Health 2012;15(5):708-15.
- Bailey H, Janssen MF, La Foucade A, Kind P. EQ-5D-5L population norms and health inequalities for Trinidad and Tobago PLOS One 2019; 14(4): e0214283.
- Braithwaite T, Bailey H, Bartholomew D, Saei A, Pesudovs K, Ramsewak SS et al. Impact of Vision Loss on Health-Related Quality of Life in Trinidad and Tobago Ophthalmology 2019; 126(7):1055-1058.
- Bailey H. A Framework for the Prioritization of Health Programmes for Trinidad and Tobago. PhD Thesis 2013. The University of the West Indies, St. Augustine Campus.
- Bailey, H., Kind, P., & La Foucade, A. Results from an exploratory study to test the performance of EQ-5D-3L valuation subsets based on orthogonal designs, and an investigation into some modeling and transformation alternatives for the utility function. Health economics review 2014; 4:29.
- Szende A, Janssen B, Cabases J. Self-Reported Population Health: An International Perspective Based on EQ-5D. Dordrecht Springer; 2014.
- Bailey H, Janssen M, La Foucade A, Boodraj G, Wharton M. EQ-5D self-reported health in Barbados and Jamaica with EQ-5D-5L population norms for the English-speaking Caribbean. Health and Quality of Life Outcomes 2021;19:97
- Bailey H, Kind P. Preliminary Findings of an Investigation into the Relationship between National Culture and EQ-5D Value Sets. Quality of Life Research 2010; 19:1145–1154
- World Bank website: https://data.worldbank.org/indicator/SP.DYN.LE00.FE.IN?locations=TT (Accessed 8/6/2021)
- Grant J. The epidemiology of chronic generalized musculoskeletal pain. Best Practice and Research: Clinical Rheumatology 2003;17:547–61.
- Malmusi D, Artazcoz L, Benach J, Borrell C. Perception or real illness? How chronic conditions contribute to gender inequalities in self-rated health. European Journal of Public Health 2012;22:781–6.
- Hibbard J, Pope C. Another look at sex differences in the use of medical care: illness orientation and the types of morbidities for which services are used. Women and Health 1986;11:21–36.
- Rollman G, Lautenbacher S. Sex differences in musculoskeletal pain. Clinical Journal of Pain 2001;17:20–4.
- Marmot M. The Health Gap. The Challenges of an Unequal World. London/New York: Bloomsbury; 2015.