A literature review of methods to Measure and value benefits in cost-effectiveness analyses of interventions for patients with either a short life expectancy or receiving palliative care

There is a general concern that generic, preference-based outcome measures of health related quality of life (HRQoL) typically used in economic evaluation, such as EQ-5D, may not be suitable for those receiving palliative care, or more generally for people with short life expectancies. We review economic evaluations with adults or children at the end of life to summarise the range of approaches taken to establish the intervention effect and limitations with the evaluation outcomes and methods reported in the studies. We find that the studies retrieved (n=46) can be categorised into two types of people at the end of life, those specifically receiving PC services (n=22) and those that may not (n=24). The EQ-5D is the most widely used preference-based measure, justified by it being recommended for economic evaluation by NICE and other appraisal bodies. The EQ-5D was generally regarded as not covering important domains relevant to end of life only in the 22 studies that evaluated PC services. There were few limitations reported with current methods for the other set of studies. Therefore, we believe an important focus of future research should be the development of approaches that assess the appropriateness of EQ-5D and other preference-based outcome measures of HRQoL in people receiving PC services. If appropriate, this would enable more widespread usage in economic evaluations of palliative care services, where researchers generally report limitations with existing preference-based measures. We propose two potential directions for future research: (1) a systematic review of the psychometric properties of preference-based measures in PC settings to determine the appropriateness of EQ-5D and commonly used alternatives (such as the SF-12) using existing evidence; and (2) exploring the feasibility of mapping to generate SF-6D utility values from commonly used measures of quality of dying/death designed for use in a palliative care setting (provided the preference-based measure used to generate the utilities is psychometrically valid for this population).