The first set of EQ-5D-5L values over health-related preferences in Latin America

July 13, 2017  


Francisco E. Estevez-Carrizo,  MD 

Vice-president, HTA Committee, National Board of Armed Forces Health, Montevideo, Uruguay


The “health-related utilities” are defined as the grade of preference for a given health state, normally expressed by a value between 0 and 1 (death has a ‘utility value’ corresponding to 0; and this same type of value, but corresponding to a very healthy life, is 1). This is very common tool utilized for the economic evaluation of health technologies. The measurement of utilities, or preferences related with health status, could be a complex task if the option is for direct techniques, such as Time Trade Off (TTO), the Standard Gamble, or the Visual Analogic Scale (VAS). However, there is a simpler option to be adopted in these measurements, which consists of using indirect methods by means of the application of generic questionnaires for valuation of the ‘Health Related Quality of Life’ (HRQOL), such as the EuroQol of five dimensions (EQ-5D), the Short Form of six dimensions (SF-6D), and the Health Utility Index (HUI).

The “National Institute for Health and Care Excellence” (NICE), of England and Wales, has established that results in the cost-effectiveness analyzes should be informed using Quality Adjusted Life Years (or QALYs), and the utilities should be derived from the preferences measured in the populations of interest, but using for such measurement the EQ-5D.

The EQ-5D is a generic instrument used to measure the quality of life communicated by patients or healthy contributors. This scale assigns a preference value (utility) to the time lived in a given health state. Many countries, including some in Latin America, have derived demographic values by means of the EQ-5D-3L. However, the EuroQOL has recently introduced a new version of this instrument, transforming it in the EQ-5D-5L, which increases the accuracy of measurements. Although not all investigators agree with the QALYs as a suitable measure for quality of life, this tool is the most utilized one by the majority of authors.

An article about health-related utilities study has been published, by Dr. Federico Augustovski et al. (2016)1 for the first time it estimated preferences from the Uruguayan population. This article describes the survey with which, they carried out the health-related preferences using this new version of EQ-5D.  It’s remarkable that this such type of study has been done in Latin America for the first time.

Instrument characteristics

The classical version of EQ-5D comprises 5 (five) dimensions with 3 (three) levels of severity each, and a visual analogic scale (or VAS, for the English acronym). The dimensions are: “mobility”, “self-validity”, “daily activities”, “pain/discomfort”, and “anxiety/depression”. As referred above, the EuroQol group has developed a new version of this instrument, having the same dimensions but with 5 levels of severity (5L), instead of 3 (3L). This new version was named EQ-5D-5L. The severity states vary between “without problems” up to “extreme inability/problems”, with three intermediate levels (“mild”, “moderate”, and “severe” problems).

This new version describes 3,125 possible health states and it was designed to improve the performance of EQ-5D-3L, mainly by means of the decrease in the “ceiling” effect and increasing the discrimination power of the instrument. The EuroQol group has also developed an evaluation protocol of the EQ-5D-5L, in order to harmonize this instrument across the world. Intending to manage such protocol, the EuroQol group has built a specific software: the EuroQol Valuation Technology (EQ-VT).

Information about Uruguay

Uruguay is a South American country, with a surface of 176,215 square kilometers of mainland and another portion of territorial sea over the Southern Atlantic; it is located between Brazil to the Northeast and Argentina to the Southwest. According to 2011 Census, the Uruguayan population was of 3,286,324 inhabitants. By the end of 2007 the government started a reform in the health system, which included the provision of a basic package of healthcare services, funded by National Health Fund (Fondo Nacional de Salud – FONASA), and managed by the National Board of Health (Junta Nacional de Salud – JUNASA) according to targets already established. Due to such reform, a National Integrated Health System (Sistema Nacional Integrado de Salud – SNIS) was introduced in the country. Since the 80’s a parastatal organization, the National Fund of Resources (Fondo Nacional de Recursos – FNR), was created and financed in large part, by means of FONASA. The FNR has assumed the financial coverage of new medical technologies and, for the last 15 years, has also included the coverage of  high cost medicines for the SNIS.

It’s key for those who are working on economic evaluation of health technologies, to have a reference framework in order to deduce the health-related utilities in the Uruguayan population. The EQ-5D is the ideal tool to evaluate the quality of life changes determined by different interventions; for instance, those induced by medical treatments or diseases. It is, in addition, the cornerstone for outcomes research studies, which results could be comparable with those from other regional or extra-regional countries.

The authors wanted to obtain health-related preferences from Uruguayan general population, using the EQ-5D-5L questionnaire and to derive the utility values related with health. This country was one the first in Latin America to approach an evaluation study of its general population using the instrument EQ-5D-5L


The country was stratified according to its geography; carrying out the study in the capital city and in 2 other regions (departments). The proportions per geographic place, age, gender, and socioeconomic situation of the interviewed reflected the demographic structure of Uruguayan population. The initial target sample was of 1,000 interviewed perople. The potency calculation was based on accuracy requirements for the estimation of TTO average.

The authors have used the standardized interview developed by the EuroQol group. Traditionally, the evaluation methods are based on TTO to assign the health-states utilities. However, considering that the conventional TTO presents some issues, particularly with those health states that are considered worse than death, the EuroQol group defined that a TTO variant which overcomes these difficulties for the EQ-5D-5L, so called “composite time trade-off” (C-TTO).

“Discrete Choice” (DC) method, which implies paired comparisons of EQ-5D-5L health states, was used; and social, demographic and sanitary terminology knowledge was gathered by a survey called “Short Assessment of Health-Literacy” (SAHL-S). The knowledge in sanitary terminology is a concept that reflects the capacity to obtain, process, and understand the information to make appropriate health decision.


The study was started by 21 interviewers who administered the survey using the EQ-VT software. However, shortly after initiating a quality control was carried out, deciding to maintain 11 interviewers with good performance. This resulted in the elimination of 220 surveys conducted by 10 poor performance interviewers. The study sample had 805 respondents aged between 20 and 83 years. 11 interviewees met exclusion criteria, for which the final sample was 794 interviewees.

48% (Forty-eight) of the sample did not experience any problem in whatever EQ-5D-5L dimensions. The older interviewed participants communicated more problems in all dimensions. The self-administered VAS scale has also decreased inasmuch as the age increased, and it was lower in women.

In the study, three models are presented to adjust the data of C-TTO: 1) the OLS model (Ordinary Least Squares) with comparative purposes, 2) the robust model of main effect, and 3) the robust model including the better interaction termination. The authors have compared the OLS, which was traditionally used with approaches of regression more robust. The selected estimates are based on the variability between responders.

This regression method applies a different ponderation to every observation based on how far it is from the median of the population sample (it ponders the extreme values with lower force). This way, the impact of the “outliers” is reduced and there is an attempt to improve the problem of the heteroscedasticity. Both robust models tested in the study gave similar results: the goodness of fit of these two models was similar, being the regression adjustment with “interaction termination”, only 0.4% of the relative improvement, if compared to the model of “main effect”; so, because of that, it was applied the principle of parsimony (having equal performance for two models, the simplest one should be selected) to choose the most adequate. In this way, the authors arrived to the conclusion that the main effect model is the most convenient for the analysis.

When a comparison is done between the utility values of the Uruguayan population with those of the Argentinean population, when transformation of the Argentinean EQ-5D-3L into EQ-5D-5L (called “crosswalk” method) was done, the Uruguayan values are slightly higher along the entire spectrum of the value set.

Discussion and Comments

Uruguay is a country with advantages to carry out this type of health states preference evaluation. Many characteristics determine this particularity: 1) The small territory without connection problems. 2) Small and relatively homogeneous population. 3) Availability of SNIS which allows universal coverage and referral physicians programs. 4) The FNR manages and funds high cost treatments for the entire SNIS. 5) Despite of being some marginalized groups the socioeconomic differences represents a lowe percentage of the entire population and 6) Population minimal education level is relatively homogeneous in the entire trerritory.

The authors raise that one of the study problems was that the sample was not taken randomly, but using “quotas”. Although the sample was, in general, representative of the sociodemographic characteristics of Uruguayan population, the young individuals and those having higher education were somewhat overrepresented. However, the authors clarify that the difference in age was small and, in addition, despite the mildly higher proportion of individuals with higher education in the sample as compared with the general population data, the utility values did not differ significantly as a function of the education levels.

The comparison of utility values in the Uruguayan population with the same values in the Argentinean population has shown that the Uruguayan values are slightly higher along the entire spectrum of severity. This is explained by the changes in the descriptive system of versions 3L and 5L of EQ-5D. As the less severe levels (“misery index”=6) as the more severe ones (“misery index=21) were less severe for 5L than for 3L. So that, just as the extreme values have improved, the entire scale has moved upward for the values derived from EQ-5D-5L.

The average index derived from EQ-5D-5L for the general population in Uruguay is: 0.954. The older age groups had worse quality of life related to health and the men had values slightly higher than women did, even though this difference has not achieved statistical significance. The population norms derived for Uruguay were consistent and similar to the international population norms published previously.

The authors comment that, in their view, this is the first study that provides a set of population values derived from the EQ-5D-5L in Latin America. Just a few countries, all over the world, have conducted evaluation studies with this new tool. It is, therefore, extremely important the development of this study in our country. Having the set of population preferences will allow carrying out the economic evaluations, applying the quality of life related to health with higher accuracy and confidence. On the other hand, this new tool will allow the conduction of comparisons with other regional or extra-regional countries inasmuch as the EQ-5D-5L establishes itself as a generic tool to derive preferences related to health.


Augustovski F, Rey-Ares L, Irazola V, et al. An EQ-5D-5L value set based on Uruguayan populations preferences. Qual Life Res (2016) Feb;25(2):323-33