The health and economic burden of multimorbidity in Australia

\\(Background:\\) Multimorbidity is a major challenge facing governments and healthcare systems worldwide due to increasing prevalence globally, the high consumption of resources and the implications for health care. Multimorbidity is commonly defined through a simple count of health conditions, using a cut-off of 2 or more or 3 or more conditions. There is also increasing interest in statistical approaches to definition. No consensus of the definition of multimorbidity yet exists, which hinders comparisons between studies and advancement of the field. The vast majority of studies have been conducted in clinical or age-restricted populations, with few representative population studies available. No studies have been performed to examine the association of multimorbidity and health status using different definitions of multimorbidity. Moreover, multimorbidity and its effect on productivity in the Australian working population, the health of which is central to the economic strength of the country, has not been well studied. \\(Aims:\\) The primary aim of this thesis was to assess the associations of multimorbidity with quality of life, health care service use, productivity losses and the related financial burden, particularly in the working population. Three different crosssectional data sets were interrogated and a systematic review was conducted. A secondary aim was to assess to what extent current large national prevalence surveys in Australia are fit for purpose in surveillance of multimorbidity and its correlates. \\(Methods:\\) The Australian National Survey of Mental Health and Wellbeing 2007 were used in Study I (Chapter 3). The health-related quality of life (HRQoL) scores were measured using the Assessment of Quality of Life (AQoL-4D) instrument. The simple count (2+ & 3+ conditions) and hierarchical cluster methods were used to define/identify clusters of multimorbidity. Linear regression was used to assess the associations between HRQoL and multimorbidity as defined by the different methods. The data derived from the Australian National Health Survey (NHS) 2011-12 was used in Study II (Chapter 4) to understand how Australian employees use health service for a single disease when suffering from multimorbidity. The health service use was reported for each health condition in the NHS and thus it was impossible to tabulate service use by the disorder count. However, the NHS 2011-12 was a large population-based Australian data source including the health status, employment status and health service use of over 10 thousand working adults. This data was the latest available in Australia when we conducted this study. The employee selfreported 2013 data derived from the partnering Healthy@Work (pH@W) survey of all state government employees in Tasmania (including 3,228 Australian employees) was used in Study III (Chapter 5) to assess the associations of multimorbidity on health-related productivity loss by sex as these associations influenced by sex were inconsistent. Data were weighted for non-response. Measures of absenteeism, presenteeism and lost productive time (LPT) were obtained from employees' selfreported data over a 28-day period. Analyses were stratified by sex, and negative binomial models were used to estimate the associations between multimorbidity and the lost productivity time. In Chapter 4 and 5, multimorbidity was defined as the cooccurrence of 2+ chronic conditions out of a pre-specified list depending on the different surveys used. Study IV (Chapter 6) was a systematic review of costs-ofillness (COI) studies of multimorbidity registered with Prospero (an international prospective register of systematic reviews). The search strategy combined key words related to multimorbidity, comorbidity and multiple chronic health conditions. The search was restricted to papers written in English and published since 2000 up to October 2016. The inclusion criteria were peer-reviewed cross-sectional, cohorts and modeling COI studies on multimorbidity, whereas the exclusion criterion was studies focusing on an index disease. The review summarized the current state of evidence and evaluated the quality of cost of illness studies of multimorbidity using the British Medical Journal Checklist for authors and peer reviewers of economic submissions. \\(Results:\\) Study I: HRQoL was negatively associated with multimorbidity regardless of the definition of multimorbidity used. Statistically significant clusters were identified through hierarchical cluster analysis and verified by sensitivity analysis. Study II: the prevalence of multimorbidity in the working population was 23.4% using two cut-off count method. Multimorbid employees with arthritis had higher adjusted arthritis-specific GP visit rates compared to employees with arthritis alone. Similarly, multimorbid employees with CVD had higher adjusted CVD-specific specialist visit rates and CVD-specific other health professional visits than employees with CVD alone. Study III: the positive association of multimorbidity and LPT, and the significant differences in LPT between men and women reporting multimorbidity were identified. Both sexes with multimorbidity were more likely to have greater productivity loss due to absenteeism or presenteeism compared to those without, but female employees with multimorbidity were more likely to have lost productivity days due to presenteeism and absenteeism, compared to their male counterparts. The mean number of total days of health-related lost productive time in the past 4 weeks was 1.2 (SD=2.4) and 1.7 (SD=3.5) for male and female employees with multimorbidity, respectively, compared to 0.6 (SD=2.2) and 0.6 (SD=1.8) for males and females without multimorbidity. Both sexes with multimorbidity were more likely to have greater productivity loss due to absenteeism or presenteeism compared to those without, but female employees with multimorbidity had 40% and 30% more lost days due to absenteeism (PR=1.4, 95% CI 1.1-1.8) and presenteeism (PR=1.3, 95% CI 1.0-1.6), respectively, compared to their male counterparts. However, there were no significant differences in days lost productivity between male employees with multimorbidity versus without multimorbidity. Study IV found that within 26 included articles, the definition used in the 14 studies that clearly defined multimorbidity was limited to the two cut-off count method. The methodology used to derive costs differed markedly among the studies. Average annual costs per person with multimorbidity ranged from $US 49-$US 252,313. Using a two cut-off count method, the ratios of multimorbidity versus non-multimorbidity costs ranged from 2- 16 within 17 available studies; while using three cut-off method, the ratios ranged from 2-10 within 12 available studies. Among 10 studies providing a breakdown on costs, the largest proportion for multimorbidity was spent on inpatient or medication costs in non-societal perspective studies, and social care costs from the societal perspective. Costs-of-illness studies of multimorbidity were highly heterogeneous. The economic burden of multimorbidity was heavy for all age groups. \\(Conclusions:\\) These findings confirm multimorbidity as a significant public health issue in the general population, as well as in the workforce. Further, these findings provide three notable contributions. The first major contribution is theoretical, and refers to the definitions used for multimorbidity. Comparison of definitions shows that the count method is still useful due to its ease of calculation, but consistency is needed on whether a 2-disorder or 3-disorder cut-off is most useful. Hierarchical clustering could be used as a supplementary tool to capture the specific common clusters of multimorbidity. A uniform definition of multimorbidity is needed. The second major contribution is practical. This thesis has quantified the impact of multimorbidity on health care resource consumption in the Australian workforce and on productivity in a large Australia occupational cohort. The heavy economic burden of multimorbidity as shown in the systematic review suggests that multimorbidity will be more and more important in the future, especially with social changes related to delayed retirement. Finally, the results from the currently available datasets we used highlight the fact that the currently available data restrict the further exploration of multimorbidity. Standardisation of chronic disease surveillance methodologies in national prevalence surveys would aid in epidemiological investigation of multimorbidity in the general population.