Cost-benefit model system of chronic diseases in Australia to assess and rank prevention and treatment options

In most high and middle income countries chronic diseases, such as cancer, heart disease, diabetes and arthritis are strongly associated with morbidity, disability and premature mortality. Their rapid growth and by now epidemic proportions are considered to be major challenges for health policy, as they account for a considerable proportion of public and private health expenditures – World Health Organisation (WHO) 2005a; Begg et al. (2007); Lopez et al. (2006).

Although currently Australians’ health is among the best in the world – with one of the highest life expectancies at birth and at age 65 years – there are concerns about the future due to the country’s obesity prevalence being among the worst in the OECD (AIHW 2006a). Obesity is a major risk factor for chronic diseases (Flegal et al., 2005; AIHW 2004e), such as cardiovascular disease, Type2 diabetes, high blood pressure, certain cancers, sleep apnoea, osteoarthritis, psychological disorders and social problems (WHO 2000). A growing proportion of obese persons and the ageing of the population are likely to increase Australia’s already high chronic diseases and comorbidy prevalences.

In Australia, eight major chronic diseases are considered high priority for policy attention: the National Health Priority Areas (NHPAs) of asthma; cancer; cardiovascular disease; diabetes; injuries; mental health; arthritis and musculoskeletal conditions; and dementia. In 2003, NHPAs accounted for 73% of total burden of diseases resulting, among 65 to 69 year olds, in a 14.7% loss of healthy years lived (Begg et al., 2007). In 2005 chronic diseases affected 80% of older Australians; accounted for 50% of all deaths; and were responsible for 70% of total health expenditures (AIHW 2006c). Because lifestyle patterns – eg unhealthy diets, lack of physical activity, excess alcohol and tobacco consumption – are major risk factors (Yach et al., 2004), chronic disease prevention and treatment are not only of medical concern, but also of social, family-level and personal interest (Seymour 2007; Griffith 2007; Eckersley 2004).

Models able to assess the benefits and costs of potential chronic disease interventions are now used worldwide, partly because of increasing awareness of their usefulness in priority setting, and partly because the alternatives to modelling – such as large-scale long-term clinical trials – are considerably more costly and time consuming. However, there is little in the literature that considers all chronic diseases simultaneously and places these into broad medical, behavioural, social and economic contexts. Earlier research tended to focus on the medical issues associated with a single disease, with a few recent articles considering one comorbid condition as well. Rarely have all major chronic diseases been studied simultaneously, although many share common lifestyle risk factors. Accounting for all chronic diseases is important because quality of life has been shown to decline and health expenditures to increase with comorbidities (Walker 2007a; Shwartz et al., 1996).

At the policy level, there has been recent recognition of the need to target the prevention of chronic diseases, and to limit their negative economic, workforce participation, productivity and quality of life impacts. For example improvement and management of prevention and care are among the stated aims of the November 2005 Australian National Chronic Disease Strategy (Dowrick 2006). Also, in April 2007 the Council of Australian Governments allocated additional funds for chronic disease strategies.

The project aims to:

• model at the national level the links between health risk factors and chronic diseases, taking account of comorbidities;

• establish how these links vary by demographic and socioeconomic factors;

• build models of expenditures relative to benefits; and

• obtain rankings for prevention and/or treatment interventions for chronic diseases, comorbidities and/or risk factors.

The key aim of building such a model system is to improve current decision making by providing a more complete view of chronic disease costs and benefits under different prevention and/or treatment scenarios. The purpose of this paper is to document the data and modelling methods proposed for the project. Details are in Walker et al. (2008).

Building the model system requires: (a) the development of a set of person-level demographic, socioeconomic, lifestyle and health variables at a particular point in time; and (b) the projection into the future of both disease-specific incidences and prevalences, and the progression at the level of the individual of chronic diseases and comorbidities. Part (a) involves the bringing together, in a coherent manner, individual-level cross-sectional data from several sources, while Part (b) requires use of disease-specific longitudinal data to estimate the incidence and progression of chronic diseases. We propose modelling (a) and (b) separately, and then link the two parts, so that the ‘big picture’ as well as the ‘detail’ associated with the tracking of individuals’ health can be analysed simultaneously. Policy interventions can then be assessed in terms of broad population-wide variables (eg worse or improved obesity patterns; population screening options); and of disease specific medical treatment (eg surgery or pharmaceuticals).

The proposed model-system, HealthAgeingMod, has two parts: an ‘Umbrella’ static microsimulation model and ‘Chronic disease’ econometric or epidemiology-type sub-models (Figure 1). It is planned that initially the model-system will account for only two chronic diseases: cardiovascular disease (CVD) and diabetes (Type 2). Key reasons for choosing CVD and diabetes are that: (a) they are major contributors to Australia’s total burden of disease (Begg et al., 2007); (b) for people with diabetes CVD is a common complication accounting for 60% of deaths; and (c) CVD and diabetes share common risk factors such as physical inactivity and obesity. Once the benefits of HealthAgeingMod are proven with these two disease-specific sub-models, then further diseases may be added (eg cancers, arthritis, mental health). However, an ability to simultaneously study CVD and diabetes will already be a significant improvement on the traditional disease-by-disease analyses.

Figure 1

Download asset Open asset

In Figure 1, the Umbrella model’s base-year dataset is drawn from the Australian Bureau of Statistics’ (ABS) 2004–05 National Health Survey (NHS05) – (ABS 2006a,b,c,d). Because NHS05 only covers private dwellings (ie mainly households), we propose to complement its unit record dataset with the institutionalised population in ABS’s 2003 Survey of Disability Ageing and Carers (SDAC03) – (ABS 2005a; 2004a,b,c). The list of NHS05 demographic, socioeconomic, risk factor, disease, disability and quality of life variables selected for use in the Umbrella model (in the ‘2005’ Box) is quite comprehensive. However, some NHS05 variables will need to be modified and variables not in NHS05 to be imputed.

We propose that the Umbrella model be able to project 20 years ahead, at 5-year intervals – ie that, from 2005, it project for 2010, 2015, 2020 and 2025. The 5-year intervals will allow simulation of diabetes and CVD screening interventions that are recommended to be repeated every five years. The disease specific sub-models will project ahead at yearly intervals, so disease specific projections that are likely to be needed yearly (eg onset of diseases, deaths) can be stored during the simulation phase. The population projections will be estimated so that the model’s weighted base population lines up, to the extent possible, with published ABS population projections (ABS 2005b). Then the Umbrella Model will consult each chronic disease model to estimate, probabilistically, how each person’s risk factor profiles and chronic disease(s) would develop over the 5-year period being considered, and whether the person would have any new chronic diseases. An important challenge will be to find appropriate ways to project the evolution of individual-level risk factors and comorbidities, and then ensure that in aggregate these individual- level time profiles line up with published aggregate statistics.

Running the Umbrella Model in ‘default’ mode (ie no policy change) will provide the baseline simulation (Figure 1, ‘Base-Scenario Outcomes’ Box). To allow for the simulation of policy interventions, a ‘Specify Scenario parameters’ box will need to be added to the Umbrella model, and filled-in prior to Scenario simulations. The baseline and scenario results will then be compared in terms of differences in health outcomes – eg Disability Adjusted Life Years (DALYs) avoided – and in health expenditures. The Outcomes module will also present year-by-year net benefit and cost streams, and compute cost effectiveness indicators (eg cost/DALY avoided).

For both the Umbrella and the sub-models, most of the methods proposed are ‘current best practice’. However, there will be challenges, such as: developing new methods to model comorbidities; modifying traditional methods to accommodate much improved up-coming datasets; and linking the elements of HealthAgeingMod so that both the disease and comorbidity details can be validated against published aggregate statistics. The methods eventually used will be documented once HealthAgeingMod is built. Those proposed in this paper are described below.

The Umbrella model will be based on microdata (that is person-level information), and will make use of microsimulation techniques. Because models based on microdata deal with the individual, use of such models allows the impact of policy changes to be examined in far greater detail than is possible with more traditional approaches. Since the SAS programming language has proved to be particularly suitable for microsimulation projects, we propose to use SAS for building the Umbrella model, as well as its linkages with the disease-specific models.

The Umbrella model will represent the full Australian population through use of the ‘weights’ embedded in NHS05 (ABS 2006b). The proportion of the total population with a particular chronic disease is generally small (eg 3.5% for diabetes; 2% for both angina and other ischaemic heart diseases – ABS 2006a). Because of this we propose that each survey individual with one or more of the chronic diseases be transformed into a number of unit records, summing to the ‘weight’ initially attached to that survey individual. Enumeration of all persons with the chronic disease will allow a much finer specification of the health variables imputed onto the Umbrella model’s base dataset (eg measured glucose levels) than what would be possible without such enumeration. The sub- models will be disease-specific incidence and/or progression models, predicting changes over time in the health states and risk factors of each individual included in the Umbrella model’s population. Disease-specific models can be of varied types: eg econometric, transition probability, parametric, proportional hazard or Weibull models. The model-system will also make use of standard cost-benefit and cost effectiveness methods (Jena and Philipson 2007).

As noted above, the Umbrella model will be of the microsimulation type. Although microsimulation was first proposed in the 1950s (Orcutt 1957; Orcutt et al., 1961), its general use had to wait until computer technology improved and large microdatasets became routinely available. While many full population-based microsimulation models had been, or are being, developed worldwide, most fall within the tax and social security fields (see review in Zaidi and Rake 2001). The few that account for health tend to be either of the socioeconomic type, which use a broad indicator of health as a covariate within the larger picture, or of the disease-specific type, designed to study disease specific treatment options so as to assess their cost effectiveness (see reviews in Lymer 2009 and Walker 2009).

Major model building requirements include ability to: (a) account for comorbidities. For this the Umbrella model will need to track individuals over time – from the onset of their first chronic disease; the contracting of their second, third etc chronic diseases; until death.; (b) represent all Australians, so that broad public health policy initiatives can be studied; and (c) simultaneously consider a wide range of social, economic, health and person-level variables. While many of these are now routinely available in ABS surveys, indicators of the important linkages between physical health, mental health and general wellbeing are in general significantly under- represented (Appleby 2006; Wilkinson 2005; WHO 2005b). These requirements suggest a dynamic microsimulation model based on nationally representative longitudinal unit record data. However, in Australia such datasets tend to be cross-sectional in nature. Also, although such datasets do contain a wide range of variables, the aggregate nature of many of the disease-level variables means that they do not fully meet our requirements. Hence we propose a static microsimulation Umbrella model, with its base-year population projected 20 years ahead at 5-year intervals.

5.4 Projecting 20 years ahead

The Umbrella model’s projection module will estimate: (1) the ‘ageing’ of the model’s base population at 5-year intervals (preferably by age, sex and broad health status/comorbidity); (2) the health state transitions for each person in that population in each time interval; and (3) the value of the comorbidity index for the period. Figure 2 presents a flowchart of the projection module.

Figure 2

Download asset Open asset

For step (1) we propose the ‘reweighting’ method, using the ABS’s GREGWT optimising software. Briefly, the optimising ‘reweighting’ method involves changing the original sample survey weights, so that application of the new weights (from GREGWT) reproduces the population distribution forecast by ABS (2005b) for the 5-year time period being considered. Step (2) will up-date the health states of each individual in the Umbrella model’s base population by querying the relevant disease specific sub-models. Step (3) will compute health benefits (in DALYs) and track each individual’s comorbidity pattern.

This latter task will be particularly challenging and time consuming as it will require development of new methods to adequately account for comorbidities. Extensive discussions with peers in Australia and overseas are planned, particularly regarding the strengths and weaknesses of possible ‘comorbidity’ options. Another challenge for the projections module will be to ensure that both its disease and its comorbidity projections can be validated against published official health statistics. It may be that the best way to do this would be to align the overall model-generated disease progression and comorbidity patterns with external targets. Should this be required, there will probably be a need to develop an appropriate automated alignment method, as methods for aligning microsimulation models are still at an early stage of development (Walker 2009).

A number of models have been reported in the literature which simulate the incidence and/or progression of CVD and diabetes (eg Mui 1999; Clarke et al., 2004; Colagiuri and Walker 2008). In this project we aim to make use of existing disease-specific models, since the challenge is to build a complex and validated model-system and not the development of disease-specific models. Our chronic disease models will need to be able to carry out four functions: (1) up-date the ‘risk factors’ of the person being processed; (2) predict who will become ‘newly diagnosed’ (with diabetes and/or CVD) within the period being considered; (3) for those already diagnosed, predict the progression of their disease(s); and (4) keep records of the risk factors, health status and comorbidity status of the person being processed – with the records being accessible by the Umbrella model.

The steps proposed for the Umbrella to Diabetes and CVD model linkages are as follows.

Every time the Projection module processes an individual in the Umbrella model’s population, the module first queries whether the person has diagnosed diabetes. If yes, then it obtains risk factor and disease progression estimates from the UKPDS model and updates these in the person’s Umbrella model records. If no, then the Projection module obtains up-dated risk factor estimates for people without diabetes, and uses these when querying the Diabetes incidence module.

If the person is found to have ‘new’ diabetes, then his/her records are up-dated in the Umbrella model. From then onwards, CVD for this person with diabetes will be assessed and progressed via the UKPDS model.

If however the person is found not to have ‘new’ diabetes, then the Projection module queries the CVD incidence module to determine whether the person has newly diagnosed fatal or non-fatal CHD or stroke. If yes, then the person’s records in the Umbrella model are updated. If no, then the Projection module moves on to process the next person in the Umbrella model’s population.

Acceptance and use of HealthAgeingMod for policy relevant applications crucially depends on it being convincing validated against publicly available benchmark statistics. We propose that validations be carried out both at cross-sections and over time. Zaidi and Rake (2001) note that when models are constructed from several sub-modules, there will be multiple sources of error and thus there will be many levels at which validation could occur. They suggest that in such cases consideration should be given to use of multiple-module validation techniques. Although multiple-module validation is rarely used, we propose to consider it in this project as it could offer additional insights into the workings of the Umbrella model.

As initial steps, we propose validation in four phases once the prototype is built and had been aligned. This would involve:

• comparing the estimates obtained from use of overseas material (ie the UKPDS model and the Anderson CVD equations) with the few available and/or forthcoming Australian benchmarks;

• checking that, once data transformations and imputations had been completed, the aggregate statistics generated by HealthAgeingMod for its base year closely match the related external benchmarks (eg by the ABS and AIHW);

• for diabetes and CVD, comparing model system projections with the trends indicated by published cross sectional benchmarks (such as the NHSs). Checks against external statistics are proposed for disease incidence, prevalence, deaths, health outcomes (DALYs) and health expenditures;

• HealthAgeingMod outputs from ‘test’ scenario simulations are proposed to be compared with published results produced by other models that simulated similar scenarios.

One limitation of our model system – as of other models and most statistical collections – is their inability to fully replicate the real world. Nevertheless, models can handle considerable complexity and, if they contain key variables and their inter-relationship, then model simulations can prove to be very valuable to decision makers. We aim to ensure our model system’s policy usefulness by including all the variables identified in the literature as key drivers of disease incidence, prevalence and health expenditures.

Another limitation arises from use of the Monte Carlo method, since this method introduces randomness into the model’s outputs. That is, different runs of the model – with identical parameters but using different random number seeds – will produce different outputs. To assess the importance of the related stochastic variation it will be necessary to execute several runs until the results ‘converge’ within set bounds. Earlier researchers found that, with microsimulation models, the number of repetitions required to achieve ‘convergence’ was relatively small – four runs in Pudney and Sutherland (1993; 1994) and six runs in Walker et al. (2006a,b). A further limitation may arise if attribution of burden of disease to a particular risk factor (eg obesity) is required. In such cases we will endeavour to choose the attribution method that is least likely to create limitations. Such a choice may however be restricted due to unavailability of appropriate data (Hartge 2006; Steeland and Armstrong 2006).

Possible applications include: (1) simulating the impact of various lifestyle interventions (eg obesity/overweight, smoking, alcohol consumption) on health outcomes and health care costs associated with individual chronic diseases and with comorbidities (diabetes and CVD initially); (2) comparing such analyses across chronic diseases individually, and the diseases combined (eg across groups of 2, 3, 4, or 5+ illnesses). The aim would be to identify key comorbidity patterns and the intervention points most likely to be effective; (3) simulating the impacts of various combined lifestyle and disease-specific treatment options, and carrying out cost-benefit analyses so that these highly complex options can be assessed and ranked.

The proposals in this paper show that the tasks set for the project are not only complex, but also require considerable creativity and innovation. Key novel elements are that the proposals cover both the broad socioeconomic and the detailed disease specific aspects; that they account for several chronic diseases, modelling the onset and progression of each; and that the onset or progression of these diseases depend on their often common risk factors. The proposals thus allow simulation of complex chronic disease policy reforms that can combine medical treatment options (eg a new drug being used for diabetes and/or new hospital procedures for stroke management) with lifestyle changing options (eg diet and/or exercise) and with socioeconomic reform options (eg improving the health of poorer population groups).

Another important novel element is HealthAgeingMod’s ability to assess the full benefits of interventions that target risk factors common to several chronic diseases. For example, in traditional models (Eastman et al. (1997a, b); Clarke et al. (2004); Colagiuri and Walker (2008)), the estimated benefits from single-disease-prevention interventions are limited to that disease itself, even when the intervention affects a risk factor – such as obesity – that is common to several chronic diseases. HealthAgeingMod can simultaneously analyse risk factor interventions that impact on multiple chronic diseases, so its benefit estimates are more comprehensive than those of traditional models. At the policy level, availability of such an improved model is likely to encourage consideration of more complex interventions that simultaneously target multiple-chronic diseases.

Another novel element, which is difficult to handle analytically and to place in a nation-wide context, is the estimation of the number of chronic diseases that individuals are likely to accumulate as they age. The difficulty arises in part from national health data collections tending to focus on single diseases, and in part from the number of diseases a person has being rarely available in data sources and being hard to model.

During the model building phase further creativity and innovation will be needed in: (1) data collection, selection and linkage; (2) approaches and methods chosen for the building of HealthAgeingMod; and (3) broadening of the boundaries of the health sectors analysed – the coverage comprising individuals’ lifestyles, individuals as patients, medical treatment options, and government policy initiatives in health care as well as in prevention. However, our proposals show that while there are constraints on what is achievable in the short term, already planned data collections and methodological progress could in future result in significant improvements in models such as HealthAgeingMod.

In view of the above we expect that, once built, the proposed model-system and its applications will demonstrate the benefits of a system-wide approach to chronic disease and comorbidity prevention and treatment. Given the considerable quality of life benefits from prevention, as well as lesser demand for doctors, pharmaceuticals and hospital services, an important goal for HealthAgeingMod applications will be to clearly indicate the relative merits of prevention options versus treatment options.

1. 1
   National Health Survey 2004–05, Summary of Results

   1. ABS (Australian Bureau of Statistics)

   (2006)

   National Health Survey 2004–05, Summary of Results, Cat No 4364.0, Canberra.

   • Google Scholar
2. 2
   National Health Survey 2004–05

   1. ABS (Australian Bureau of Statistics)

   (2006)

   National Health Survey 2004–05, Confidentialised Unit Record Files.

   • Google Scholar
3. 3
   National Health Survey 2004–05: CURF Information Paper

   1. ABS (Australian Bureau of Statistics)

   (2006)

   National Health Survey 2004–05: CURF Information Paper, Cat No 4324.0 and National Health Survey 2004–05 Questionnaire, Cat No 4363.0.55.002.

   • Google Scholar
4. 4
   National Health Survey 2004–05, User Guide

   1. ABS (Australian Bureau of Statistics)

   (2006)

   National Health Survey 2004–05, User Guide, Cat No 4363.0.55.001.

   • Google Scholar
5. 5
   Disability Ageing and Carers, 2003, Confidentialised Unit Record Files Information Paper (Reissue)

   1. ABS (Australian Bureau of Statistics)

   (2005)

   Disability Ageing and Carers, 2003, Confidentialised Unit Record Files Information Paper (Reissue), Cat No 4430.0.00.001.

   • Google Scholar
6. 6
   Projections of the Populations of Australia: 2004–2101

   1. ABS (Australian Bureau of Statistics)

   (2005)

   Projections of the Populations of Australia: 2004–2101, Cat no 3222.0.

   • Google Scholar
7. 7
   Disability Ageing and Carers, 2003

   1. ABS (Australian Bureau of Statistics)

   (2004)

   Disability Ageing and Carers, 2003, Confidentialised Unit Record Files.

   • Google Scholar
8. 8
   Disability Ageing and Carer, 2003

   1. ABS (Australian Bureau of Statistics)

   (2004)

   Disability Ageing and Carer, 2003, Summary of Findings, Cat no. 4430.0.

   • Google Scholar
9. 9
   Disability Ageing and Carer, 2003

   1. ABS (Australian Bureau of Statistics)

   (2004)

   Disability Ageing and Carer, 2003, User Guide, Cat no. 4431.0.55.001.

   • Google Scholar
10. 10
    Diabetes in Australia: A Snapshot

    1. ABS (Australian Bureau of Statistics)

    (2004)

    Diabetes in Australia: A Snapshot, Cat no 4820.0.55.001.

    • Google Scholar
11. 11
    Census of Population and Housing: Socioeconomic Indexes for Areas 2001. Information Paper

    1. ABS (Australian Bureau of Statistics)

    (2003)

    Census of Population and Housing: Socioeconomic Indexes for Areas 2001. Information Paper, Cat no 2039.0.

    • Google Scholar
12. 12
    National Health Survey 2001

    1. ABS (Australian Bureau of Statistics)

    (2003)

    National Health Survey 2001, User Guide, Cat No 4363.0.55.001.

    • Google Scholar
13. 13
    Health risk factors, Australia 2001

    1. ABS (Australian Bureau of Statistics)

    (2003)

    Health risk factors, Australia 2001, Cat No 4812.0.

    • Google Scholar
14. 14
    National Health Survey 2001, Summary of Results, Cat No 4364.0

    1. ABS (Australian Bureau of Statistics)

    (2002)

    National Health Survey 2001, Summary of Results, Cat No 4364.0.

    • Google Scholar
15. 15
    Australia’s Health 2006

    1. AIHW (Australian Institute of Health and Welfare)

    (2006)

    Australia’s Health 2006, AIHW cat No. AUS 73, Canberra.

    • Google Scholar
16. 16
    Socioeconomic inequalities in cardiovascular disease in Australia: Current picture and trends since 1992

    1. AIHW (Australian Institute of Health and Welfare)

    (2006)

    Socioeconomic inequalities in cardiovascular disease in Australia: Current picture and trends since 1992, AIHW Bulletin 3, Canberra.

    • Google Scholar
17. 17
    Chronic diseases and associated risk factors in Australia 2006

    1. AIHW (Australian Institute of Health and Welfare)

    (2006)

    Chronic diseases and associated risk factors in Australia 2006, AIHW Cat No PHE 81, Canberra.

    • Google Scholar
18. 18
    Health expenditure Australia 2004–05

    1. AIHW (Australian Institute of Health and Welfare)

    (2006)

    Health expenditure Australia 2004–05, Canberra.

    • Google Scholar
19. 19
    Obesity and workplace absenteeism among older Australians

    1. AIHW (Australian Institute of Health and Welfare)

    (2005)

    Obesity and workplace absenteeism among older Australians, Canberra.

    • Google Scholar
20. 20
    Health expenditure Australia 2003–04

    1. AIHW (Australian Institute of Health and Welfare)

    (2005)

    Health expenditure Australia 2003–04, Canberra.

    • Google Scholar
21. 21
    Health system expenditure on disease and injury in Australia 2000–01

    1. AIHW (Australian Institute of Health and Welfare)

    (2005)

    Health system expenditure on disease and injury in Australia 2000–01, Canberra.

    • Google Scholar
22. 22
    Australia’s Health 2004

    1. AIHW (Australian Institute of Health and Welfare)

    (2004)

    Australia’s Health 2004, AIHW Cat. No. AUS 44, Canberra.

    • Google Scholar
23. 23
    Disability and its relationship to health conditions and other factors

    1. AIHW (Australian Institute of Health and Welfare)

    (2004)

    Disability and its relationship to health conditions and other factors, Canberra.

    • Google Scholar
24. 24
    The impact of dementia on the health and aged care systems

    1. AIHW (Australian Institute of Health and Welfare)

    (2004)

    The impact of dementia on the health and aged care systems, Canberra.

    • Google Scholar
25. 25
    Cancer Australia 2001

    1. AIHW (Australian Institute of Health and Welfare)

    (2004)

    Cancer Australia 2001, Canberra.

    • Google Scholar
26. 26
    The relationship between overweight, obesity and cardiovascular disease

    1. AIHW (Australian Institute of Health and Welfare)

    (2004)

    The relationship between overweight, obesity and cardiovascular disease, Cat. No. CVD 29, Canberra.

    • Google Scholar
27. 27
    Heart, Stroke and Vascular Disease-Australian facts 2004

    1. AIHW (Australian Institute of Health and Welfare)

    (2004)

    Heart, Stroke and Vascular Disease-Australian facts 2004, AIHW Cat. No. CVD 27, Canberra.

    • Google Scholar
28. 28
    Older Australians at a Glance

    1. AIHW (Australian Institute of Health and Welfare)

    (2002)

    Older Australians at a Glance, Third Edition, AIHW cat. no. AGE25, AIHW and DoHA, Canberra.

    • Google Scholar
29. 29
    Diabetes: Australian Facts 2002

    1. AIHW (Australian Institute of Health and Welfare)

    (2002)

    Diabetes: Australian Facts 2002, Diabetes Series No. 3 AIHW Cat. No. CVD 20, Canberra.

    • Google Scholar
30. 30
    Diabetes: Australian Facts 2002

    1. AIHW (Australian Institute of Health and Welfare)

    (2002)

    Diabetes: Australian Facts 2002, Diabetes Series No. 3 AIHW Cat. No. CVD 20, Canberra.

    • Google Scholar
31. 31
    Chronic diseases and associated risk factors in Australia 2001

    1. AIHW (Australian Institute of Health and Welfare)

    (2002d)

    Canberra: AIHW.

    • Google Scholar
32. 32
    Health system costs of diseases and injury in Australia 1993–9

    1. AIHW (Australian Institute of Health and Welfare)

    (1998)

    Health system costs of diseases and injury in Australia 1993–9, AIHW Cat. No. HWE 5, Canberra.

    • Google Scholar
33. 33
    Cardiovascular disease risk profiles

    1. K Anderson
    2. P Odell
    3. P Wilson
    4. W Kannel

    (1991)

    Am Heart J 121:293–98.

    • Google Scholar
34. 34
    Data briefing: health, wealth and happiness

    1. J Appleby

    (2006)

    Health Services Journal 25:23.

    • Google Scholar
35. 35
    An economic model of the long-term health care burden of type ii diabetes

    1. A Bagust
    2. PK Hopkinson
    3. W Maier
    4. C Currie

    (2001)

    Diabetologia 44:2140–55.

    • Google Scholar
36. 36
    Being big or growing fast: systematic review of size and growth in infancy and later obesity

    1. J Baird
    2. D Fisher
    3. P Lucas
    4. J Kleijnen
    5. H Roberts
    6. C Law

    (2005)

    British Medical Journal 331:929.

    • Google Scholar
37. 37
    Risk of Cardiovascular and All-Cause Mortality in Individuals With Diabetes Mellitus

    1. E Barr
    2. Ρ Zimmet
    3. Τ Welborn
    4. D Jolley
    5. D Magliano
    6. D Dunstan
    7. A Cameron
    8. T Dwyer
    9. H Taylor
    10. A Tonkin
    11. T Wong
    12. J McNeil
    13. J Shaw

    (2007)

    Impaired Fasting Glucose, and Impaired Glucose Tolerance The Australian Diabetes, Obesity, and Lifestyle Study (AusDiab), Circulation 116:151–157.

    • Google Scholar
38. 38
    The burden of disease and injury in Australia 2003, Australian Institute of Health and Welfare

    1. S Begg
    2. Τ Vos
    3. Β Barker
    4. C Stevenson
    5. L Stanley
    6. A Lopez

    (2007)

    AIHW cat. no. PHE 82.

    • Google Scholar
39. 39
    Psychiatrie and somatic comorbidities are negatively associated with quality of life in physically ill patients

    1. Η Baumeister
    2. Κ Balke
    3. M Härter

    (2005)

    Journal of Clinical Epidemiology 58:1090–1100.

    • Google Scholar
40. 40
    Measuring quality of life in chronic illness: the functional assessment of chronic illness therapy measurement system

    1. D Cella
    2. C Nowinski

    (2002)

    Arch Phys Med Rehabil 83:S10–S17.

    • Google Scholar
41. 41
    FRACP (Department of Endocrinology, Prince of Wales Hospital, NSW, Australia). Estimating the cost of complications of diabetes in Australia using administrative health-care-data

    1. P Clarke
    2. J Leal
    3. C Kelman
    4. M Smith
    5. S Colagiuri

    (2007)

    Value in Health, July, August.

    • Google Scholar
42. 42
    A model to estimate the lifetime health outcomes of patients with Type 2 diabetes: the United Kingdom Prospective Diabetes Study (UKPDS) Outcomes Model (UKPDS no. 68)

    1. P Clarke
    2. A Gray
    3. A Briggs
    4. A Farmer
    5. P Fenn
    6. R Stevens
    7. D Matthews
    8. M Stratton
    9. R Holman

    (2004)

    Diabetologia 47:1747–59.

    • Google Scholar
43. 43
    Using an Economic Model of Diabetes to Evaluate Prevention and Care Strategies in Australia

    1. S Colagiuri
    2. A Walker

    (2008)

    Health Affairs 27:256–268.

    • Google Scholar
44. 44
    Insights into the utilisation of health services in Australia based on linked administrative data. Occasional Papers, New series No 9

    1. Department of Health and Aged Care

    (2000)

    Insights into the utilisation of health services in Australia based on linked administrative data. Occasional Papers, New series No 9, Canberra.

    • Google Scholar
45. 45
    The Chronic Disease Strategy for Australia

    1. C Dowrick

    (2006)

    MJA 185:61–62.

    • Google Scholar
46. 46
    Model of Complications of NIDDM – Model Construction and Assumptions

    1. R Eastman
    2. J Javitt
    3. W Herman
    4. E Dasbach

    (1997a)

    Diabetes Care 20:725–734.

    • Google Scholar
47. 47
    Model of Complications of NIDDM: Analysis of the Health Benefits and Cost-effectiveness of Treating NIDDM with the Goal of Normoglycemia

    1. R Eastman
    2. J Javitt
    3. W Herman
    4. E Dasbach

    (1997b)

    Diabetes Care 20:735–744.

    • Google Scholar
48. 48
    Being better off but feeling worse: what’s happening to people in Australia?

    1. R Eckersley

    (2004)

    New Community Quarterly 2:3–7.

    • Google Scholar
49. 49
    Excess deaths associated with underweight, overweight and obesity

    1. KM Flegal
    2. BI Graubard
    3. DF Williamson
    4. MH Gail

    (2005)

    JAMA 293:1861–67.

    • Google Scholar
50. 50
    Social inequalities and the common mental disorders: a systematic review of the evidence

    1. T Friers
    2. D Melzer
    3. R Jenkins

    (2003)

    Social Psychiatry and Psychiatric Epidemiology 38:229–37.

    • Google Scholar
51. 51
    How effective are expert patient (lay led) education programmes for chronic disease?

    1. C Griffiths
    2. G Foster
    3. J Ramsay
    4. S Eldridge
    5. S Taylor

    (2007)

    BMJ 334:1254–1256.

    • Google Scholar
52. 52
    Modelling Our Future: Population Ageing, Health and Aged Care, 2007, International Symposia in Economic Theory and Econometrics, North Holland, Amsterdam

    1. A Gupta
    2. A Harding

    Modelling Our Future: Population Ageing, Health and Aged Care, 2007, International Symposia in Economic Theory and Econometrics, North Holland, Amsterdam.

    • Google Scholar
53. 53
    Estimating the burden of disease and the benefits of prevention – commentary

    1. P Hartge

    (2006)

    Epidemiology 17:498–9.

    • Google Scholar
54. 54
    Obesity as an independent risk factor for cardiovascular disease: a 26- year follow-up of participants in the Framingham Heart Study

    1. H Hubert
    2. M Feinleib
    3. Ρ McNamara
    4. W Castelli

    (1983)

    Circulation 67:968–77.

    • Google Scholar
55. 55
    Self-rated health and mortality: a review of twenty-seven community studies

    1. E Idler
    2. Y Benyamini

    (1997)

    Journal of Health and Social Behaviour 38:21–37.

    • Google Scholar
56. 56
    ACCU-SIM 2 — An Overview

    1. Institute for Medical Informatics and Biostatistics

    (1997)

    ACCU-SIM 2 — An Overview, Switzerland.

    • Google Scholar
57. 57
    ACCU-SIM 2 — Help File Contents

    1. Institute for Medical Informatics and Biostatistics

    (1997)

    ACCU-SIM 2 — Help File Contents, Switzerland.

    • Google Scholar
58. 58
    The Australian Diabetes, Obesity and Lifestyle Study

    1. International Diabetes Institute

    (2006)

    AusDiab 2005: Tracking the Accelerating Epidemic: Its Causes and Outcomes, The Australian Diabetes, Obesity and Lifestyle Study, Melbourne.

    • Google Scholar
59. 59
    Cost-effectiveness as a price control

    1. A Jena
    2. T Philipson

    (2007)

    Health Affairs 26:696–703.

    • Google Scholar
60. 60
    Obesity and the risk of heart failure

    1. S Kenchaiah
    2. J Evans
    3. D Levy
    4. P Wilson
    5. E Benjamin
    6. M Larson
    7. W Kannel
    8. R Vasan

    (2002)

    N Engl J Med 347:305–13.

    • Google Scholar
61. 61
    Global Burden of Disease and Risk Factors, World Bank, Washington

    1. A Lopez
    2. C Mathers
    3. M Ezzati
    4. D Jamison
    5. C Murray

    (2006)

    Global Burden of Disease and Risk Factors, World Bank, Washington.

    • Google Scholar
62. 62
    APPSIM – Modelling Health: population ageing, health status and health outlays. NATSEM Working Paper No 13, University of Canberra

    1. S Lymer

    (2009)

    APPSIM – Modelling Health: population ageing, health status and health outlays. NATSEM Working Paper No 13, University of Canberra, Australia.

    • Google Scholar
63. 63
    Glucose indices, health behaviours and incidence of diabetes in Australia: the AusDiab study

    1. D Magliano
    2. E Barr
    3. P Zimmet
    4. A Cameron
    5. D Dunstan
    6. S Colagiuri
    7. D Jolley
    8. N Owen
    9. P Phillips
    10. R Tapp
    11. TA Welborn
    12. J Shaw

    (2008)

    Diabetes Care 31:267–72.

    • Google Scholar
64. 64
    Self-rated health: a predictor of mortality among the elderly

    1. J Mossey
    2. E Shapiro

    (1982)

    Am J Public Health 72:800–8.

    • Google Scholar
65. 65
    Projecting coronary heart disease incidence and cost in Australia: Results from the Incidence module of the Cardiovascular Disease Policy Model

    1. S-L. Mui

    (1999)

    Australian and New Zealand Journal of Public Health 23:11–19.

    • Google Scholar
66. 66
    Measuring Quality of Life in Health, Churchill Livingstone, Elsevier, London

    1. R O’Connor

    (2004)

    Measuring Quality of Life in Health, Churchill Livingstone, Elsevier, London.

    • Google Scholar
67. 67
    Policy Implications of Ageing Populations: Introduction and Overview. OECD Working Paper no. 33

    1. OECD (Organisation for Economic Co-operation and Development)

    (1996)

    Policy Implications of Ageing Populations: Introduction and Overview. OECD Working Paper no. 33, Paris.

    • Google Scholar
68. 68
    A new type of socio-economic system

    1. G Orcutt

    (1957)

    Review of Economics and Statistics, 58, 2.

    • Google Scholar
69. 69
    Microanalysis of Socioeconomic Systems: A Simulation Study, Harper and Row, New York

    1. G Orcutt
    2. M Greenberger
    3. J Korbel
    4. A Rivlin

    (1961)

    Microanalysis of Socioeconomic Systems: A Simulation Study, Harper and Row, New York.

    • Google Scholar
70. 70
    Microsimulation Policy Modelling Unit paper No MU 9402

    1. S Pudney
    2. H Sutherland

    (1993)

    Statistical Reliability and Microsimulation: the Role of Sampling, Simulation and Estimation Errors, Microsimulation Policy Modelling Unit paper No MU 9402, University of Cambridge, UK.

    • Google Scholar
71. 71
    How reliable are microsimulation estimates? An investigation of the role of sampling error in a UK tax-benefit model

    1. S Pudney
    2. H Sutherland

    (1994)

    Journal of Public Economics.

    • Google Scholar
72. 72
    wealth and the pursuit of happiness

    1. Seymour L. Health

    (2007)

    Journal of the Royal Society for the Promotion of Health 127:2.

    • Google Scholar
73. 73
    The importance of comorbidities in explaining differences in patient costs

    1. M Shwartz
    2. L Iezzoni
    3. M Moskowitz
    4. A Ash
    5. E Sawitz

    (1996)

    Medical Care 34:767–82.

    • Google Scholar
74. 74
    An overview of methods for calculating the burden of disease due to specific risk factors

    1. K Steenland
    2. B Armstrong

    (2006)

    Epidemiology 17:512–9.

    • Google Scholar
75. 75
    Low Income and Multiple Disadvantage: Analysis of the British Household Panel Survey

    1. M Taylor
    2. R Berthoud
    3. S Jenkins

    (2004)

    London: SEU.

    • Google Scholar
76. 76
    Overweight and obesity from childhood to adulthood: a follow-up of participants in the 1985 Australian Schools Health and Fitness Survey

    1. A Venn
    2. R Thomson
    3. M Schmidt
    4. V Cleland
    5. B Curry
    6. H Gennat
    7. T Dwyer

    (2007)

    Medical Journal of Australia 186:458–460.

    • Google Scholar
77. 77
    Modelling the socioeconomic status to health link in Australia: A dynamic microsimulation approach

    1. A Walker

    (2009)

    LAP Lambert Academic Publishing, Koln, Germany.

    • Google Scholar
78. 78
    Multiple chronic diseases and quality of life: patterns emerging from a large national sample

    1. A Walker

    (2007a)

    Australia, Chronic Illness 2007 3:202–218.

    • Google Scholar
79. 79
    How do health, socio-economic status, education and family connections affect labour force status?

    1. A Walker

    (2007b)

    Australian Journal of Labour Economics.

    • Google Scholar
80. 80
    Cost-Benefit Model System of Chronic Diseases in Australia: to Assess and Rank Prevention and Treatment Options – Proposed Approach. Australian Centre for Economic Research on Health Research Report No 3, Australian National University

    1. A Walker
    2. J Butler
    3. S Colagiuri

    (2008)

    Cost-Benefit Model System of Chronic Diseases in Australia: to Assess and Rank Prevention and Treatment Options – Proposed Approach. Australian Centre for Economic Research on Health Research Report No 3, Australian National University, www.acerh.edu.au/publications/ACERH_RR3.pdf.

    • Google Scholar
81. 81
    Hospital Admissions by Socioeconomic Status: Does the ‘Inverse Care Law’ Apply to Older Australians?’

    1. A Walker
    2. J Pearse
    3. HA Thurecht

    (2006a)

    Australian and New Zealand Journal of Public Health 30:467–73.

    • Google Scholar
82. 82
    Changes in hospitalisation rates and costs – New South Wales, 1996–97 and 2000–01

    1. A Walker
    2. L Thurecht
    3. A Harding

    (2006b)

    Australian Economic Review 39:391–408.

    • Google Scholar
83. 83
    Health Inequalities Across Socioeconomic Groups: Comparing Geographic-Area-Based and Individual-Based Indicators

    1. A Walker
    2. N Becker

    (2005)

    Public Health 119:1057–1150.

    • Google Scholar
84. 84
    Cost–Benefit Model of Diabetes Prevention and Care: Model Construction. Assumptions and Validation, Technical Paper No 28

    1. A Walker
    2. S Colagiuri
    3. M McLennan

    (2003)

    University of Canberra: National Centre for Social and Economic Modelling.

    • Google Scholar
85. 85
    The impact of inequality: how to make sick societies healthier

    1. R Wilkinson

    (2005)

    London: Routledge.

    • Google Scholar
86. 86
    Prevalence, Expenditures, and Complications of Multiple Chronic Conditions in the Elderly

    1. J Wolff
    2. B Starfield
    3. G Anderson

    (2002)

    Arch Intern Med 162:2269–2276.

    • Google Scholar
87. 87
    Preventing chronic diseases: a vital investment

    1. World Health Organisation(WHO)

    (2005)

    Preventing chronic diseases: a vital investment, Geneva.

    • Google Scholar
88. 88
    Mental Health Atlas

    1. World Health Organisation(WHO)

    (2005)

    Mental Health Atlas, Geneva.

    • Google Scholar
89. 89
    Global strategy for the prevention and control of non-communicable diseases

    1. World Health Organisation(WHO)

    (2000)

    Global strategy for the prevention and control of non-communicable diseases, Geneva.

    • Google Scholar
90. 90
    The Global Burden of Chronic Diseases: Overcoming Impediments to Prevention and Control

    1. D Yach
    2. C Hawkes
    3. CL Gould

    (2004)

    JAMA 291:2616–2622.

    • Google Scholar
91. 91
    Dynamic Microsimulation Models: a Review and Some Lessons for SAGE. Discussion paper No 2

    1. A Zaidi
    2. K Rake

    (2001)

    London: ESRC-Sage Research Group.

    • Google Scholar

Author details

1. Agnes E Walker

   Australian National University, Australia

   For correspondence

   Agnes.Walker@anu.edu.au

2. Stephen Colagiuri

   University of Sydney, Australia

   For correspondence

   Stephen.Colagiuri@sydney.edu.au