A New Online Tool for Creating Personal and Social EQ-5D-5L Value Sets
OHE’s Prof Nancy Devlin has been collaborating on a new approach to creating personal and social EQ-5D-5L value sets, based on a web application. OHE’s Prof Nancy Devlin has been collaborating with researchers at the University of Otago (New Zealand)…
OHE’s Prof Nancy Devlin has been collaborating on a new approach to creating personal and social EQ-5D-5L value sets, based on a web application.
OHE’s Prof Nancy Devlin has been collaborating with researchers at the University of Otago (New Zealand) and 1000minds on a new approach for creating personal and social EQ-5D-5L value sets. Preliminary findings are now available. This blog post describes the method and its potential value.
Producing EQ-5D social value sets can be costly and time-consuming. Nancy Devlin and colleagues have been working on a new approach, based on a web application that involves a person answering some simple questions for about 10 minutes. The app can immediately generate a complete set of values for all 3,125 EQ-5D-5L health states for that individual, as well as contributing to a social value set.
Provisional results from developing and pilot-testing the approach in New Zealand were presented at the recent EuroQol Group Academy Meeting in Budapest. Below is the poster and brief ‘pitch’ presented at the Meeting.
The tool’s user-friendliness and online delivery could massively reduce the cost of producing value sets for the EQ-5D-5L, or indeed any other patient-reported outcome (PRO) instrument used to inform Health Technology Assessment (HTA). The method – known as the PAPRIKA method (an acronym for Potentially All Pairwise RanKings of all possible Alternatives) – has strong theoretical foundations and it meets the requirement of being ‘trade-off’ based. The tool can be deployed ‘in the field’ to quickly and cheaply generate the preference data required to produce value sets at the population (social) level.
The approach can also readily be used to elicit individual patient preferences. In the future, the tool could be available on iPads in doctor waiting rooms for individual patients to create their own personal value sets. The easy availability of personal value sets opens up the possibility of HTAs being implemented at the individual patient level – e.g. by incorporating individual patient preferences into treatment decisions.
If you would like to experience the tool yourself – and generate your own EQ-5D-5L value set – click here. This demonstration version of the tool is intended for a ‘knowledgeable’ audience (as you will see at its conclusion). The tool developed for patients, from research based in New Zealand, has more user-friendly instructions and finishes with questions for collecting demographic information too.
How does the tool work? The tool is powered by 1000minds software, which, as mentioned earlier, implements the PAPRIKA method, a type of adaptive conjoint analysis (or discrete choice experiment). Since 2004, this method and software have been used in a wide range of health decision-making areas, including health technology and patient prioritisation, disease classification and diagnosis, and for developing clinical guidelines.
Central to the PAPRIKA method is the participant answering a series of simple pairwise-comparison questions based on choosing between two hypothetical EQ-5D-5L health states defined on just two attributes at a time and involving a trade-off; see the illustrative screenshot below. Such simple questions are repeated with different pairs of hypothetical health states – always involving trade-offs between different combinations of attributes, two at a time – until enough information about the person’s preferences has been collected to determine their weights on the attributes, thereby generating a complete value set for that person. The task below looks like a regular discrete choice experiment but, rather than using a fixed experimental design which is applied to all participants, the PAPRIKA approach learns from and adapts to each individual respondent’s preferences. This enables the approach to produce – efficiently and quickly – a value set for each individual, as well as for the overall sample.
The demonstration version of the tool asks questions based on just three levels per dimension and then interpolates two additional levels per dimension to generate an EQ-5D-5L value set. The tool can also be configured to pose questions using the full five-level system, at the ‘cost’ of requiring the participant to answer more trade-off questions. Which option to apply may depend on how incentivised participants are to answer the questions.
To check data quality, two or three repeated questions can be included to assess the person’s consistency or reliability. Checks can also be automatically made of how much time the person took to answer the questions and for any other evidence that they answered questions unreliably.
In addition, a binary search algorithm is employed to identify any health states worse than dead – necessary for including the possibility of negative health state values (for states worse than dead). A visual analogue scale approach was also tested.
A representative sample of the New Zealand general population is currently being surveyed. In addition to establishing preference weights for the EQ-5D-5L, the preferences of sub-groups will be explored; e.g. weights will be analysed according to characteristics such as age, gender, ethnicity and health status. A paper reporting results from the study will be presented at the EuroQol Scientific Plenary meeting in Lisbon, September 2018.
Possible next steps include trialling the new approach in other countries and testing it against alternative methods for creating value sets. If you are interested in collaborating or supporting this research, please email the principal investigator, Trudy Sullivan.
The use of this approach in New Zealand has some parallels with the work undertaken in the UK by OHE and the University of Sheffield to develop methods for estimating Personal Utility Functions (PUFs). Both are based on eliciting preference data to construct individual utility functions, which can readily be aggregated to calculate social value sets. Both use the same trade-off based approach to locating the individual’s position of the 0 anchor within the descriptive system.
There are also differences. The 1000minds approach enables the collection of data online, very quickly and inexpensively. The PUF approach, in contrast, emphasises the importance of reflection and deliberation in the construction of individual preferences, necessitating a more time-consuming interview administration.
OHE, in collaboration with the University of Otago, University of Sheffield and the University of Technology, Sydney, are now planning a UK study to further develop, test and compare these methods for eliciting personal utility functions. For enquiries about the proposed study, please contact Koonal Shah.
Strictly Necessary Cookies
Strictly Necessary Cookie should be enabled at all times so that we can save your preferences for cookie settings.
If you disable this cookie, we will not be able to save your preferences. This means that every time you visit this website you will need to enable or disable cookies again.
3rd Party Cookies
This website uses Google Analytics to collect anonymous information such as the number of visitors to the site, and the most popular pages.
Keeping this cookie enabled helps us to improve our website.
Please enable Strictly Necessary Cookies first so that we can save your preferences!