Investigating the evolution of model-based health economic evaluations and changes in health-related quality of life in osteoarthritis

Osteoarthritis (OA) is one of the most common chronic joint diseases, and is characterized by joint pain, stiffness, swelling, loss of function and disability, which in turn, negatively impact individuals‚ÄövÑv¥ health-related quality of life (HRQoL) and incur a substantial socio-economic burden. Currently, there is no cure for OA, but many treatments and approaches are available to help relieve disease symptoms and improve patients‚ÄövÑv¥ HRQoL. Health economic evaluation models incorporate clinical, health economic and epidemiological data to compare alternative options in terms of both economic costs and clinical effectiveness to identify the interventions that are best value for money. Despite significant progress made in health economic modelling of OA since 1994 (when the first model-based health economic evaluation was performed), existing models are likely to suffer from limitations related to the choice of model input parameters, discount rates, and model health states/events. There is a need to perform a comprehensive review of health-economic evaluation models of OA to identify their key strengths and weaknesses to provide directions for improvement in the current modelling practice. Health state utility values (HSUVs) measure the strength of a preference for a particular health state, represented as a number between 0 (death) and 1 (optimal health). They are an important input parameter for health economic evaluation models. However, no comprehensive database of OA-related HSUVs for patients with different affected joint sites undergoing different treatments is available to date, which makes the selection of HSUVs to be used in the modelling practice a challenging task. Therefore, it is imperative to generate a HSUVs database for various sub-groups of OA patients to guide HSUVs choices in future health economic modelling of OA interventions. Furthermore, OA and the associated comorbidities (numbers as well as patterns) are linked to worse health outcomes over time. However, the quantification of the long-term impacts of OA and associated comorbidities on HRQoL has not been achieved to date and is an important research area that needs attention. This PhD research thesis (comprising of 5 key chapters) aims to fill these evidence gaps by 1) synthesizing the strengths and weaknesses of existing OA health economic evaluation models; 2)generating a HSUVs database for OA-related conditions; 3) investigating the long-termchanges in OA people‚ÄövÑv¥s HRQoL; and 4) evaluating the impact of numbers and patterns of comorbidities on HRQoL and identifying the most prevalent and influential comorbidity patterns that impact HRQoL in people with OA over a ten year period. The following is a brief overview of each of the five chapters included in this research thesis. Chapter 1 presents a general introduction to OA and health economics. Chapter 2 presents a systematic review of all OA health economic evaluation models and evolution of modelling in the field of OA. This is the first study comprehensively reviewing the evolution of health-economic evaluation models of all OA interventions including prevention, core treatments, adjunct non-pharmacological interventions, pharmacological and surgical treatments to identify the key strengths and limitations facing existing OA health-economic evaluation models and provide directions for improvement for current modelling practice. OA health economic evaluation models have evolved and improved substantially over time, with the focus shifting from short-to-medium-term pharmacological decision-tree models to surgical-focused lifetime Markov models. Indirect costs of OA are frequently not considered, despite using a societal perspective. There was a lack of reporting the sensitivity of model outcomes to input parameters including discount rates, OA definition, and population parameters. Whilst the coverage of OA-related adverse events has improved over time, they are still not comprehensively captured in most health economics models of OA. Chapter 3 presents a systematic review and meta-analysis of HSUVs of people with OA-related conditions. This is the first study comprehensively reviewing OA-related HSUVs and statistically meta-analyzing the HSUVs for different affected joint sites before and after various treatments. The systematic review identified important areas where the current evidence is lacking, namely under-represented multi-attribute utility instruments, geographical locations/ethnicities, affected OA joint sites and treatment options. The meta-analyses generated a HSUVs database for OA patients with different affected joint sites undergoing different treatments that may be applied in future health economic modelling of OA interventions. Chapter 4 describes the impact of OA on HRQoL in the forms of HSUVs and health-dimension scores and investigates the longitudinal changes in HRQoL of people with OA compared to those without OA using an Australian population-based longitudinal cohort. Compared to participants without OA, HSUs for those with OA were 0.07 (95% confidence interval: -0.09, -0.05) units lower on average over ten years. HSUs for participants with knee and/or hip OAwere similar to those with other types of OA at 2.5 years follow-up and then diverged, with HSUs of the former being as much as 0.09 units lower than the latter. Those with OA had lower scores for psychological wellness, independent living and social relationships compared to those without OA. Independent living and social relationships were mainly impacted by knee and/or hip OA with the effect on the former increasing over time. In summary, OA negatively impacts multiple facets of HRQoL, but with different intensity and timing. Interventions to improve HRQoL should be tailored to specific OA types, health dimensions, and times. Support to maintain psychological wellness should be provided irrespective of OA type and duration. However, support to maintain independent living could be more relevant to knee and/or hip OA patients living with the disease for longer. People with OA are more likely to have comorbidities than people without OA. Comorbidities are associated with worse health outcomes and increased economic burden. Chapter 5 presents the results of investigating the impact of numbers and patterns of comorbidities on HRQoL and identifying the most prevalent and influential comorbidity patterns that impact HRQoL in people with OA over ten years. Having more comorbidities negatively impacted OA patients‚ÄövÑv¥ long-term HRQoL. Compared with comorbidity-free OA participants, the HSUV of those with 2 or ‚Äöv¢‚Ä¢3 comorbidities were -0.07 and -0.13 units lower respectively over ten years, largely driven by reduced scores for independent living, social relationships, and psychological wellness. The types and combinations of comorbidities vary in effect sizes and health dimensions influenced. Comorbidity patterns including ‚ÄövÑv=cardiovascular and non-OA musculoskeletal‚ÄövÑv¥ were most influential and were associated with up to 0.13 units lower HSUV, mostly through negative impacts on independent living (up to -0.12), psychological wellness (up to -0.08) and social relationship (up to -0.06). The optimal management and prevention of cardiovascular and non-OA musculoskeletal conditions may yield improvements in OA patients‚ÄövÑv¥ HRQoL. The findings are also helpful to guide the adjustment of HSUVs input for comorbidity numbers and patterns in the future OA health economic models of Australians and similar populations with alternative comorbidity profiles. In summary, this thesis synthesizes the strengths and weakness of existing OA health economic evaluation models and provides a HSUV database for OA-related conditions, which will be helpful in the development of an improved model and guiding the choice of HSUVs in the future. This thesis also bridged important research gaps by investigating the long-term changes in OA people‚ÄövÑv¥s HRQoL and evaluating the impact of numbers and patterns of comorbidities on HRQoL and identifying the most prevalent and influential comorbidity patterns on HRQoL in people with OA over ten years, which will help to improve the future management of OA and generate the HSUVs inputs for modelling practice.