New Methods for Analysing the Distribution of EQ-5D Observations
Our latest OHE Research Paper presents new methods for summarising EQ-5D data from patients. Our latest OHE Research Paper has been published, presenting new methods for analysing the distribution of EQ-5D observations, led by Bernarda Zamora. EQ-5D profile data…
EQ-5D profile data – that is, the levels of problems reported on each of the five dimensions of health – are often under-analysed. Where the data are used to estimate QALYs, the profile data are summarised using value sets, and the profile data provided by patients is not itself the focus of analysis. Yet the profile data can yield important insights into the specific aspects of health problems that patients are experiencing, and how these change, for example, as a result of treatment.
One thing to look out for in self-reported health on the EQ-5D (and other ‘generic’ outcome measures) is the extent to which the health states reported by patients cluster together in a small number of profiles or are dispersed evenly over many profiles. This can have implications for interpreting statistical analysis of the corresponding EQ values (Index) data, and for clinical management of patients.
Our latest OHE Research Paper presents new methods that we have developed to summarise the distribution of health states in patient reported outcome (PRO) data, with application to EQ-5D profile data in the UK. They are the Health State Density Index (HSDI), Health State Density Curve (HSDC), and estimated Power Law functions (PLFs). The properties of these are examined and compared with existing methods from information theory (e.g. Shannon’s Index).
These methods can be used to generate evidence on the relative performance of different instruments. For example, the HSDCs and HSDIs can be used to examine the differences between the three- and five-level versions of the EQ-5D with respect to how much clustering there is in the profile data. Because of the more granular descriptions of health provided by the additional levels of the 5L, the profile data from the 5L are less clustered than that from the 3L. The methods we present in this paper provide a simple way of summarising this graphically (HSDC) and via a single number (HSDI).
These methods can also be used by clinicians to help with treatment planning. For example, if patients’ pre-treatment EQ-5D data show that most patients’ health problems are captured by a small number of profiles, this might suggest their health needs are homogeneous and can be managed with standardised treatment protocols. However, if a wide range of EQ-5D profiles is reported by patients, this may suggest a greater need for individualised patient management.
The study was undertaken by Bernarda Zamora, David Parkin, Yan Feng, Mike Herdman, and Nancy Devlin (PI), in collaboration with Andrew Bateman of Cambridgeshire Community Services NHS Trust and University of Cambridge. Preliminary findings were presented at the HESG Winter 2017, PROMs conference 2017, and the 2017 EuroQol Plenary Meeting. The study was funded by the EuroQol Research Foundation.
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!