DSU Report
The incorporation of health benefits in cost utility analysis using the EQ-5D
(November 2010)

Related publication
S Davis, A Wailoo. A review of the psychometric performance of the EQ-5D in people with urinary incontinence. Health and Quality of Life Outcomes 2013; 11 (20)

Executive summary
Economic evaluation is an important component for decision making, underpinning much of the guidance issued by NICE. Typically these economic evaluations have taken the form of cost-utility analyses where health benefits are expressed in terms of Quality Adjusted Life Years (QALYs). One of the most widely used preference based instruments for the assessment of Health Related Quality of Life (HRQoL) that can be used to generate QALYs is the EQ-5D. This is stated to be the preferred instrument for NICE.

The EQ-5D has been the subject of criticisms levelled at NICE. This report considers these criticisms of EQ-5D. It aims to identify the types of claims that have been made about EQ-5D and identify the empirical evidence to support such claims. This is used to inform a series of case studies in different disease areas, where evidence of the performance of EQ-5D is systematically identified and reviewed.

The report identifies many claims that have been made as part of individual technology appraisals and in evidence submitted to the Kennedy review into the value of innovation. Many of these claims relate to the QALY as a measure of outcome per se or the decision rule of QALY maximisation. There were few examples of claims that EQ-5D is inappropriate as a measurement tool within these general frameworks. Where such claims were identified they were rarely supported by empirical evidence. These claims can be broadly categorised as relating to situations where a specific relevant dimension of health is not directly included in the EQ-5D instrument, such as fatigue or sensory impairment, or where the disease course is characterised by flares of unpredictable symptom severity. Several claims were made regarding inappropriateness in broad disease areas such as cancer and mental health.

We conducted case study reviews in the areas of rheumatoid arthritis, asthma and incontinence and refer to a separately funded report on visual disorders.

In general terms we found that there were several studies that suggest EQ-5D is less responsive or sensitive than disease specific outcome measures. This was the case for both preference and non preference based outcomes. Other generic preference based measures do not seem to systematically perform differently to EQ-5D. Where an alternative generic instrument includes a specific dimension of relevance to the disorder, such as the HUI3 in visual disorders, then it is more sensitive to changes than EQ-5D. There are also instances where EQ-5D may be a more appropriate instrument than some disease specific outcome measures.

These data inform assessments derived from psychometrics. Numerous cautions must be considered when making such assessments. There are no definitive tests in this situation. The data provide circumstantial evidence that must be combined with intuition and judgement in order to reach conclusions about the appropriateness or otherwise of EQ-5D or any other instrument. There is no gold standard and assessments are a question of degree. In particular, the requirement of the Institute to make consistent decisions across a broad range of diseases, patients and technologies must be considered.

The case studies also highlight the requirement to review a wide range of literature in assessing EQ-5D. Studies that contribute evidence usually are not designed with an assessment of EQ-5D in mind. Detailed, critical examination of the studies is required to assess their relevance.

Several developments to the EQ-5D are in development including a 5 level variant and the use of "bolt-ons". The former may help to overcome problems where sample sizes are insufficient to detect changes in the standard 3-level EQ-5D. The relevance of the latter to NICE is dependent on how a number of other considerations are resolved. For example, what is required of health state valuation methods in order to achieve consistency in decision making and what is the appropriate conceptual nature of health.  

November, 2010


Other DSU work on EQ-5D

Mapping of EQ-5D: A comparison of direct and indirect methods for the estimation of health utilities from clinical outcomes

TSD10: The use of mapping methods to estimate health state utility values

Other work on EQ-5D by DSU members

Publication: B Pennington, S Davis. Mapping from the Health Assessment Questionnaire to the EQ-5D: The Impact of Different Algorithms on Cost-Effectiveness Results. Value in Health 17(8):762-771 Dec 2014

Wailoo, A, Hernández, M, Philips, C, Brophy, S, Siebert, S. Modeling health state utility values in ankylosing spondylitis: Comparisons of direct and indirect methods. Value in Health 18(4): 425-431 Jun 2015