- Title
- Using population health data to measure healthcare costs of arthritis for Australian women
- Creator
- Lo, Thomas King Tong
- Relation
- University of Newcastle Research Higher Degree Thesis
- Resource Type
- thesis
- Date
- 2015
- Description
- Research Doctorate - Doctor of Philosophy (PhD)
- Description
- Objective: The aims of this thesis were to: 1) to systematically examine the methods for identifying individuals with arthritis; 2) and systematically examine the methods for estimating healthcare cost using individual-level data; 3) using these valid methods to calculate the healthcare cost of arthritis in older Australian women with arthritis; 4) examine the past trends of the healthcare cost; and 5) analyse the explanatory factors of the healthcare cost of arthritis. Methods: The thesis was grouped into two parts. Each part consists of three independent studies using different study methods. In Part One, various case-definitions of arthritis using population-based data were examined in detail. The studies include a systematic review of published prevalence studies and two other studies designed to assess the performance of two different case-definitions of arthritis. Part Two consists of a systematic review of published cost of arthritis studies that identify the most appropriate measurement methods for the cost of arthritis, followed by two other studies that apply the these methods (with the best arthritis case-definition identified from Part One) to estimate the healthcare cost of arthritis and to assess the explanatory factors for cost. The studies were based on data for participants in the Australian Longitudinal Study on Women’s Health. The healthcare utilisation and cost information were obtained from the linked Medicare Australia datasets. Results: In Part One, the systematic review found that self-reported arthritis was the most common case-definition in recent prevalence of arthritis studies. Examination of the agreement between self-reported diagnosed arthritis and musculoskeletal symptoms found that there was adequate agreement between these two measures. Utility of healthcare administrative data for identification of arthritis was also explored by means of case identification algorithms systematically built using elements from Medicare data. The algorithms were found to have no better than fair agreement with self-reported arthritis. Overall, results indicate that self-reported diagnosed arthritis is the best option for a case-definition to study the economic burden of arthritis in the Australian context. In Part Two, the results from the systematic review indicated that the incremental cost method is most appropriate for accounting for the costs of comorbidities in individuals with arthritis. Results also indicated that gamma regression and quantile regression statistical methods should be adopted in the cost of arthritis studies in this thesis. Accordingly, the mean adjusted incremental Medicare cost among older Australian women with arthritis was estimated at (AUD 2012) $502.59 per person per year in 2009. Results also indicate that the cost of comorbidities accounts for a considerable proportion of the Medicare costs. Cost distribution was severely positively skewed (as illustrated by the estimates at several percentiles), where the top 10% of the population incurred 300% greater costs. Longitudinal analysis did not find significant changes in arthritis costs between 2003 and 2009. Results also show that explanatory variables had statistically significantly different effects on healthcare cost at different percentiles of cost. Specifically, quantile regression found that for each increase in the SF-36 physical component score (PCS), Medicare cost decreased by $47, $71, $137, and $195 at the 50th, 75th, 90th, and 95th percentiles respectively. And for each increase in the number of comorbid conditions, Medicare cost increased by $195, $200, $419, and $639 at the 50th, 75th, 90th, and 95th percentiles respectively. Moreover, generalized estimating equations predict thatneed variables (such as the SF-36 PCS score and the number of comorbid conditions) have less influence on cost than do predisposing characteristics (such as area of residence), and enabling factors (such as complementary health insurance coverage). Conclusion: This thesis makes a substantial contributions in three main areas. First, it provides robust evidence to show that self-reported diagnosed arthritis is the best case-definition option for population-based epidemiological studies. Second, this thesis extends our understanding of the healthcare cost of arthritis by employing advanced methods in cost research. It was possibly the first study in Australia that assessed the cost of arthritis (all forms combined) using individual-level data, and the incremental cost method to account for the impact of comorbidity on the cost of arthritis. An assessment of the cost trends over time using longitudinal data also provides insights into the dynamics of the healthcare cost of arthritis. Third, this thesis sets new directions and methods for future research on the determinants of healthcare cost. It provides evidence that individuals with different levels of healthcare utilisation are heterogeneous groups, and that their healthcare costs are influenced by different sets of explanatory factors at different degrees between these sub-groups. It also provides evidence that traditional regression methods, which produce a single rate of change (a slope) as indicated by the regression coefficient, are incapable of accurately describing the relationships between the explanatory variable and costs across the entire cost distribution. Moreover, it demonstrates that quantile regression is a very useful tool, not only in the estimation of the adjusted cost of arthritis at multiple percentiles, but also in the assessment of the explanatory factors of cost in population sub-groups. The findings of this thesis lead to a more accurate understanding of the economic burden of arthritis and provide important insights into the determinants of healthcare costs in Australians with arthritis.
- Subject
- arthritis; claims database; economic burden; healthcare costs; patient-reported outcomes
- Identifier
- http://hdl.handle.net/1959.13/1063100
- Identifier
- uon:17202
- Rights
- Copyright 2015 Thomas King Tong Lo
- Language
- eng
- Hits: 590
- Visitors: 703
- Downloads: 0