Health-Related Quality of Life and Statistical Challenges in Trial-Based Economic Evaluations

Healthcare decision-makers around the world are facing tough decisions regarding the allocation of healthcare resources in the context of restrained supplies and escalated costs. To inform such allocation decisions, researchers are called upon to demonstrate not only the effectiveness of healthcare interventions but also their cost-effectiveness. Trial-based economic evaluations seek to provide this information by relating the difference in costs between healthcare interventions to the difference in effects. In this type of study, health-related quality of life (HRQoL) is often used as a metric of health effects. A prerequisite for using results of trial-based economic evaluations in decision-making is that they are valid and reliable, i.e., “scrupulous”. Using less-than-optimal methods for estimating HRQoL and the statistical analysis of cost-effectiveness data can lead to biased conclusions and thus potentially a waste of healthcare resources. Despite this knowledge, the literature indicates that the methodological quality of published trial-based economic evaluations is generally suboptimal. This PhD project aimed, therefore, to address some of the gaps in the validity of HRQoL measures and statistical challenges in applied economic evaluations. This was done by answering methodological research questions that emerged from the analysis of trial-based economic evaluation data and by providing step-by-step guidance on how to conduct and interpret trial-based economic evaluations based on statistically sound methods available in the literature. The following methodological research questions are addressed in this thesis: Health-related quality of life scoring methods: 1a. Does the use of crosswalks instead of EQ-5D value sets impact trial-based economic evaluation results and decision-making? (Chapters 2 and 3) 1b. Does the use of different country-specific EQ-5D value sets impact trial-based economic evaluation results and decision-making? (Chapter 4) Predicting health-related quality of life from condition-specific patient-reported outcome measures (PROMs): 2. Is it valid to predict EQ-5D utility values from a condition-specific PROM for patients with LBP using mapping approaches for use in economic evaluations? (Chapters 5 and 6) Handling missing data in trial-based economic evaluations: 3. Is multiple imputation necessary when using longitudinal linear mixed-models to estimate cost-utility outcomes and what is the impact on trial-based economic evaluation results and decision-making? (Chapter 7) Addressing these methodological research questions alone will not improve the conduct of trial-based economic evaluations. To achieve this, clear guidelines are needed for researchers on how to optimize the methodological quality of their trial-based economic evaluations. Therefore, Chapters 8 and 9 include tutorial papers that provide step-by-step guidance for fellow researchers on how to conduct, analyse, and interpret trial-based economic evaluations.