On 18 June, the RSS Primary Healthcare special interest group hosted an afternoon of presentations and discussion on the topic of Patient Reported Outcome Measures (PROMs) in Health Economics.
The event, held at the Royal Statistical Society’s headquarters at Errol Street, started with presentations from experts in this field.
First to speak was Professor Nancy J Devlin of the Office of Health Statistics, who provided an introduction to the area, covering the role of PROMs in economic evaluation and the specific issues of EQ-5D-5L, status preference tests, time trade off, utilities and the weightings used to calculate scores.
Next was Professor David Parkin from the Department of Primary Care & Public Health Sciences, King’s College London, who spoke about Using PROMs to Monitor Healthcare Performance and how specifically the EQ-5D-5L and its scores and states are used within the NHS PROMs Initiative.
Lastly, Oliver Rivero-Arias, associate professor at the National Perinatal Epidemiology Unit, Nuffield Department of Population Health, University of Oxford, discussed mapping non-preference onto preference-based PROMs. Whether via direct or indirect methods, the purpose is to create algorithms to link other measures (case studies included HAQ, PDQ-39 and OHS), socio-demographics and other relevant data to the EQ-5D-3L.
A number of questions were then put forward for discussion following the presentations:
- Are utilities used for estimating QALYs (from EQ-5D profile data) appropriate for summarising EQ5D?
- What are the PROs & CONs of NHS PROMs data?
- Can we make better use of PROMs data than simply looking at summary indices?
- What can be done to improve the predictions of poor health states of mapping algorithms?
In response to the question of whether utilities for estimating QALYs are appropriate for summarising EQ-5D, one discussion group provided the blunt answer of ‘no’, adding that too much important information is lost by summating weighted scores. There were questions over whether a (5x5) matrix of the individual components of each EQ-5D item could be used for predicting improvement or deterioration. There was also a worry that the NHS PROMs Initiative, whilst useful, was limited by the fact that it only considered four procedures (hip and knee replacements, and hernia and varicose vein repairs).
Much debate was on the topic of the calculation of the weights created from stated preference studies (via discrete choice experiments, time trade off and regression analysis). This evolved from beliefs that, whilst these weights are based on population averages, there must surely be many and various differences in the participants surveyed. This work is done separately by country, which may take some of this variability into account. For instance, it is known that the Japanese provide a much different set of weights than here, due to the fact that (in general) there is an expectation for enduring pain in their culture. It was also noted that older members of Indian subcontinent would generally tend to score their health much worse than younger members. Adaptation to pain, or other dimensions of the EQ-5D-5L may cause potential under-estimating of scores, and there may be an impact of varying different co-morbidities, and also health literacy.
The excellent presentations and discussions that followed largely went to show that this is a field that has clearly made huge strides but that there are still issues and concerns, even at the heart of the matter when finding the best techniques to create a score. However, there is a definite need to embed patient perspectives into health economics and not just rely on clinical endpoints.
Presentations from the session are available on the PRIMSTAT website.