I. McRae1, L. Zheng2,4, S. Bourke3, N. Cherbuin1, K.J. Anstey2,4
1. Centre for Research on Ageing Health and Wellbeing, Research School of Population Health, The Australian National University, Canberra, ACT, Australia; 2. Neuroscience Research Australia, Margarete Ainsworth Building, Barker Street, Randwick, Sydney NSW, Australia; 3..Department of Health Services Research and Policy, Research School of Population Health, The Australian National University, Canberra, ACT, Australia; 4. Ageing Futures Institute, School of Psychology, University of New South Wales, Sydney, NSW, Australia
Corresponding Author: Dr Ian McRae, Centre for Research on Ageing Health and Wellbeing, Research School of Population Health, The Australian National University, Canberra, ACT 2600, Australia, Email: Ian.S.McRae@anu.edu.au, Ph: +61 431 929 750
J Prev Alz Dis 2020;
Published online December 15, 2020, http://dx.doi.org/10.14283/jpad.2020.71
Background: Assessment of cost-effectiveness of interventions to address modifiable risk factors associated with dementia requires estimates of long-term impacts of these interventions which are rarely directly available and must be estimated using a range of assumptions.
OBJECTIVES: To test the cost-effectiveness of dementia prevention measures using a methodology which transparently addresses the many assumptions required to use data from short-term studies, and which readily incorporates sensitivity analyses.
DESIGN: We explore an approach to estimating cost-effective prices which uses aggregate data including estimated lifetime costs of dementia, both financial and quality of life, and incorporates a range of assumptions regarding sustainability of short- term gains and other parameters.
SETTING: The approach is addressed in the context of the theoretical reduction in a range of risk factors, and in the context of a specific small-scale trial of an internet-based intervention augmented with diet and physical activity consultations.
MEASUREMENTS: The principal outcomes were prices per unit of interventions at which interventions were cost-effective or cost-saving.
RESULTS: Taking a societal perspective, a notional intervention reducing a range of dementia risk-factors by 5% was cost-effective at $A460 per person with higher risk groups at $2,148 per person. The on-line program costing $825 per person was cost-effective at $1,850 per person even if program effect diminished by 75% over time.
CONCLUSIONS: Interventions to address risk factors for dementia are likely to be cost-effective if appropriately designed, but confirmation of this conclusion requires longer term follow-up of trials to measure the impact and sustainability of short-term gains.
Key words: Dementia, risk factors, cost-effectiveness, interventions, sustainability.
While many studies have addressed the association of lifestyle and vascular factors with dementia, few have addressed whether interventions designed to reduce risk factors are cost-effective (1). This is in part because dementia risk reduction programs are implemented well before the usual age of dementia onset. This means that economic evaluation using simulation modelling requires parameters relevant to a long-term time frame. As most intervention studies to date have 5 years or less of follow-up (1) (exceptions include the planned trials of multi-domain interventions (2)), cost-effectiveness studies require model parameters to be extrapolated well beyond the data observed. Reviews of model-based economic evaluations of dementia interventions (1, 3) have identified very few methods which assess prevention strategies. The types of non-pharmaceutical interventions identified in these reviews mainly focused on early assessment of dementia, screening or diagnosis rather than reduction in risk factors (3).
Short-term cost-effectiveness studies (4) and methodologies have been published which address cost-effectiveness of transitions from mild cognitive impairment (MCI) to dementia (5). However, assessing cost-effectiveness of programs which reduce or treat risk factors (many of which occur in mid-life) requires modelling the impact of interventions over longer time frames (1, 3, 5-9) and requires assumptions on how trial results are sustained over the longer term. In the absence of robust estimates of many of the parameters needed for full Markov or other simulation models, we suggest an alternate approach to estimating the price at which programs are cost-effective. This approach provides transparency in estimating the sensitivity of these prices with highly uncertain parameter estimates.
A 2019 review of health economic evaluations of primary prevention programs for dementia (1) identified three analyses of prevention strategies (6, 8, 9) which modelled dementia progress and costs over the long-term. Noting the range of uncertainties, the review recommended that “extensive sensitivity analysis to examine the impact of assumptions” be implemented. This included assumptions regarding long-term vs short-term outcomes of interventions, the impact of optimal program targeting, and discounting (1). Two of the analyses were partial evaluations which addressed potential cost savings from reductions in dementia levels, but did not address health benefits (usually measured by Quality Adjusted Life Years (QALYs)), so cost-effectiveness was not testable(6, 8). While including an extensive sensitivity analysis, the study which addressed cost-effectiveness (9) required a range of assumptions to estimate parameters including annual risk rates, mortality rates for those with and without dementia and QALY levels by age for people with dementia (9).
Estimated age/gender specific incidence rates(10) for dementia are available for Australia, but the impact of interventions on incidence of dementia at each age is not known, nor are age-specific costs or QALY estimates. Hence, there is value in exploring non-simulation approaches to estimate the cost-effectiveness of interventions which address dementia risk factors using aggregate data. We use an approach based on average lifetime costs of dementia and losses in quality of life per individual who develops dementia. Until long-term parameters can be obtained with confidence, this approach avoids the need for transition probabilities and cost and QALY measures by age. It also gives a direct means of linking costs and benefits and provides a transparent means of undertaking sensitivity analyses of all factors, including parameters reflecting the sustainability of improvements in risk factors, program targeting and discounting.
To demonstrate the proposed approach we draw on two examples (11, 12): (1) a study that estimated the effects of risk reduction through population attributable risk(PAR) and (2) a recent randomized control trial (RCT) which assessed the impact of an on-line dementia prevention program. The RCT has a relatively short follow-up (15 months), so to estimate the long term cost-effective and cost-saving prices we provide a range of different assumptions including the degree to which gains in risk reduction are sustained and how well the program is targeted to people with high likelihoods of progressing to dementia.
Methods and Data
We used available estimates of the proportion of adults aged 65 and over who are expected to develop dementia and then estimated the reduction in prevalence of dementia for a target population from the two example interventions. Savings in costs and QALYs per person generated by the interventions were estimated using the average per person life-time costs of dementia and loss of QALY due to dementia. This enabled us to estimate the maximum price per person for an intervention to be cost saving or cost effective.
The standard measure of cost-effectiveness (technically cost-utility) is the incremental cost per QALY gained (i.e. the Incremental Cost-Effectiveness Ratio or ICER). For the purposes of this study, an intervention with an ICER below $50,000 is considered cost-effective. While Australia has no formal ICER thresholds this is the level most commonly quoted and is consistent with UK, Australian (13) and American(14) literature.
Apart from sensitivity analysis for uncertain parameters, we examined: (1) the impact of program targeting, as an intervention targeted at the highest risk groups has a greater opportunity to reduce dementia prevalence, (2) the impact of “decay” which reflects reduction in the gains from an intervention over time, and (3) the impact of different levels of discounting. Discounting is a means of “valuing down” (1) future financial and health costs as people may prefer to save money (or gain health benefits) now rather than in the future.
Lifetime Costs of Developing Dementia
Lifetime costs for people with dementia are the product of average annual costs of treatment/care and the duration of care. While estimates of duration of dementia vary widely depending mainly on age at diagnosis, international evidence and reviews suggest that a mean of 5 years is appropriate for the duration of care for dementia (15, 16) (noting this may not be the same as the actual duration of dementia) (17)).
The available estimates of costs of dementia take several perspectives. An American study (18) including direct healthcare costs and costs of informal care estimated $260,000 per person in 2015 (all costs in Australian Dollars); while a 2016 Australian analysis found average annual costs of $35,550 per person including indirect costs such as loss of productivity of both patients and carers (10). With 5 years life with dementia, this becomes $177,750 lifetime cost per person. A later Australian study (19) of people with dementia in residential care with a markedly different methodology estimated higher annual costs of $88,000 per year for residential care compared to $55,000 from the earlier study(10). Given the varying results from these studies, we used a figure of $200,000 as baseline, with a range from $150,000 to $300,000 used for sensitivity analyses.
Loss of Quality Adjusted Life Years by People Developing Dementia
The lifetime loss of QALYs for people with dementia includes the loss due to poorer quality of life and the loss due to premature mortality. A conservative median estimated years of life lost to dementia used here as a baseline is 5 years. This is consistent with previous studies (20) and an Australian systematic review (15). Some estimates as high as 9 years of life have been found (7, 21); we use this as the upper limit for sensitivity analysis purposes.
Few generally applicable estimates of QALY values for people with dementia are available (22). Most studies deriving QALYs in a dementia context relate to specific RCTs with specific populations rather than comparing average people with and without dementia. We draw on estimates of average QALYs for people with dementia and the wider aged population (7, 23). With 5 years of life with dementia, and 5 years loss of life due to dementia there is an average loss 1.5 QALYs while alive and 4.2 QALYS due to premature mortality giving a lifetime loss of 5.7 QALYs from the dementia. A previous estimate (7) based on 6 years with dementia and 9 years loss of life led to an estimated 9.4 QALYs lost which we use as an upper level for sensitivity testing.
Prevalence of dementia
Population prevalence data is not required for Example 1 as the predicted outcomes are explicitly in prevalence terms, although it is required for Example 2. While “Australian data on dementia prevalence are lacking” (AIHW 2018 p138 (24)), we use an estimate of 10% for people aged 65 or over from a study using Australian data (10), which is marginally above estimates combining Australian and international data (10, 24). For Example 2, we assume that any reduction in risk will lead to an equivalent reduction in prevalence when the cohort reaches age 65 or over, and that this reduction will apply to the estimated 10% prevalence of dementia in this age group.
People generally value future costs and effects less than current costs and effects and the value diminishes the further into the future they are expected to occur (25). Hence, economic evaluations adjust the value of costs and benefits for the time at which they occur, using discounting (25). Discounting over long periods has major impacts on results of cost-effectiveness studies (1), particularly when comparing program costs at midlife to medical and other savings in later life (26). A range of discount levels are used by different organisations including: a) the use of 3% for both costs and QALYs (9), b) discount rates of 4% for costs and 1.5% for QALYs( 1), c) the use of 5% for both costs and QALYs in Australia by the Medicare Services Advisory Committee (25), and d) a UK recommendation that 3.5% be applied to both costs and QALYs (25).
In the light of extremely low interest rates in Australia and many other countries at present, and the long durations of discounting in this study, we use baseline discount rates of 3% for both costs and QALYs. For sensitivity analysis we include the Australian standard of 5% for both costs and QALYs, and the 4%/1.5% applied in Holland (1).
Simulation approaches apply discounting each year. However, assuming on average no differences between treated and untreated groups before onset of dementia, the discounting will have no material impact on the differences between treatment groups prior to diagnosis (note that while in principle costs change at onset, they are only measured from diagnosis). We, therefore, discount from the average age of commencement of the intervention to approximately the mid-point of the dementia period. To establish the period of discounting we take an average age of diagnosis as being in the early 80s (27-29). Most studies addressing average age at diagnosis show averages from the high 70s to mid 80s, but most commence with aged populations which may lead to some upward bias. We, therefore, include some alternate discounting periods for sensitivity analysis.
Example 1 – Estimates of Dementia Prevention using Population Attributable Risk
Ashby-Mitchell et al.(2017) (11) explored the aggregate Population Attributable Risk (PAR) from a set of known correlates of dementia (midlife obesity, physical inactivity, smoking, low educational attainment, diabetes mellitus, midlife hypertension, depression). They used PAR values to estimate the impact of uniform reductions in these correlates on dementia prevalence. They concluded that a uniform 5% improvement across all risks would, over 20 years, lead to a reduction in the prevalence of dementia of 3.2% or 17,454 people in Australia.
Any intervention which aimed to reduce the risk factors addressed in Example 1 would need to improve obesity levels and hypertension in mid-life so we assume an intervention targeted at the population aged 45 years and over with an average age of around 65 years. Consistent with the modelling in Example 1 this gives a 20-year period from average age at intervention to average age of dementia diagnosis (early 80s) which we use for discounting (15 years used for sensitivity testing).
Example 2 – BBL-GP Intervention
The Body-Brain-Life in General Practice program (BBL-GP) aims to reduce known dementia risk factors using a mixture of on-line training and face-to-face consultations with dietitians and exercise physiologists (12). Results are assessed using an aggregate measure combining a range of known risk factors (the ANU-ADRI (30)) with program participants compared to an active control group. After 62 weeks the BBL-GP participants showed a decline in ANU-ADRI scores of 4.62 units more than the active controls (12). For a population of Australians aged 60-64 years at baseline, a difference in baseline values of 1 point of ANU-ADRI is associated with a difference of 8% in people developing mild cognitive decline (MCI) or dementia after 12 years (31). This suggests a BBL-GP effect of 37% if the 4.62 units improvement is sustained.
This is an upper limit. Firstly, it is unlikely all the gains in risk factors will be sustained (e.g. maintaining weight loss). Secondly, the evidence of the impact of one point of ANU-ADRI on MCI and dementia may be the same as the long-term impact on dementia, but need not be, as there is likely to be a bias towards reducing MCI in those who are least likely to go forward to dementia. In this case the 8% impact of one ADRI point would be an overstatement. Finally, it is not clear if differences in the index obtained from an intervention have the same effect as differences brought about by lifetime experiences. The size of “decay” for any particular intervention is, therefore, driven by a range of factors including the time period between the intervention and the age at which dementia diagnosis is likely. For sensitivity analysis we test a range of different levels of reduction in impact of the BBL-GP program on actual dementia risk, beginning with a 50% reduction and increasing to a 95% reduction. We term this “decay” to reflect both the difficulty in sustaining the intervention’s short-term gains and the other issues described.
The trial population in Example 2 had an average age of 51 years (12), so for discounting purposes there is approximately 30 years to the average age of dementia diagnosis (20 years used for sensitivity testing). The average cost per participant in the BBL-GP trial relative to an active control was $2,700 including set-up costs. The number of participants in this trial was small, and while there are fixed costs of around $200 per person, other expenditures was almost independent of participant numbers. If more fully implemented the program would be expected to be at least quadrupled in size and costs would become $825 per person. We use this figure to assess cost-effectiveness. With a larger implementation, average costs would be further reduced.
Table 1 shows baseline estimates for Example 1 with a target population of all people aged 45 years or over. This suggests that, ignoring program costs and discounting, over the lifetimes of the people protected from dementia by the lifestyle changes there would be savings of $3.5b and 99,488 QALYs. While these savings are large, with a targeting across the whole population, the savings per targeted person are only $342. After allowing for discounting, the maximum cost per targeted person which could lead to a cost saving program is $189, while a cost less than $460 would achieve a cost-effective incremental cost per QALY gained (the ICER) of less than $50,000.
1. (10)= ((4) + (9)*(5))/(6)
Table 2 provides estimates of maximum costs per person for a program to be cost saving or cost-effective under different assumptions on target size, lifetime costs, QALY losses and discount rates. Tests 1-3 show relatively little sensitivity in cost-effective or cost-saving prices to changes in estimated lifetime costs and lifetime QALY losses to dementia, with greater effects of QALY increases than cost increases on the cost-effective price. Test 4 assumes the intervention targets only half the population aged 45 and over and assumes the targeting is so well focused on those at higher risk that the number of people avoiding dementia is unchanged. This generates a much greater change in the maximum acceptable costs than shown in Tests 1-3. Test 5 assumes an intervention targeted at a population of only 10,000 who are at very high risk of developing dementia (25% prevalence rate), and again with 3.2% of the anticipated cases “saved” from dementia (11). The cost-effective price increases to $2,069 (after discounting), more than 4 times the baseline estimate. With such precise targeting the percentage saved would probably be greater than 3.2%, and any increase in this parameter would increase the cost-effective prices proportionately. Table 2 also shows the impact of different discounting rates, with the 4%/1.5% levels having broadly similar results to baseline, but the 5%/5% showing acceptable prices around 60% of 3%/3% meaning interventions are considerably less likely to be cost-effective. Should the duration of discounting (the period from the intervention to average age of diagnosis) be reduced, for the 3%/3% calculation the maximum cost-effective price would increase by 15% meaning more expensive interventions would be cost-effective.
NOTE: * shows variation from baseline
Table 3 provides baseline estimates for Example 2. For presentation purposes the assumed population is 10,000 but results are independent of this number. The discounted program prices at baseline of $3,052 per person to be cost saving and $7,401 per person for the program to be cost-effective are well above the average price per participant of $825 relative to the active control.
Table 4 provides sensitivity testing which in addition to the factors tested for Example 1 tests levels of “decay”, and shows that the targeting, decay and discounting assumptions have the greatest impact on the overall outcomes. The targeting level of 60% was chosen as the trial participants were mainly people with obesity, with the relative risk of developing dementia of 1.6 (11). With an average price of $825, results discounted at 3% and all other factors at baseline level, a decay of up to 88% would be cost-effective, although not 95% (Test 3). With a 60% loading for targeting and the maximum levels of cost savings from preventing dementia and QALY lost to dementia, the intervention would be cost-effective at 95% decay (Test 7). Test 8 shows that with the 60% loading for targeting and other factors at baseline, even at 93% decay from the short term results the program would be cost-effective.
NOTE: * shows variation from baseline
The patterns in these tables show that results are linear with respect to both targeting and “decay”, and less than linear with respect to estimated lifetime costs and QALYs lost to dementia. As for Example 1, discounting has a major impact on the results, although even with relatively high levels of “decay” (80% with all other factors at baseline) the intervention is likely to remain cost-effective with 5%/5% discounting. Should the duration of discounting be reduced, the maximum cost-effective price would increase by 34% for the 3%/3% discounting, although this does not lift any of these prices above $825 for the examples in Table 4.
Our results suggest that multi-domain programs such as the BBL-GP in Example 2 are likely to be cost-effective (unless program impacts decay almost completely over time), while the more generic approach of Example 1 requires tight targeting to at-risk populations to be cost-effective. These results are consistent with prior studies (1, 32) in showing the importance of targeting and sustainability of observed results beyond the period of study follow-up.
The estimated cost of $825 per person in Example 2 would be reduced with wider implementation. Previous studies have estimated cost of dementia risk reduction programs of $200 to $500 per person (9, 33). If Example 2 could be conducted at these lower costs it is more likely to be cost-effective even at high levels of “decay”. Should the duration from intervention to diagnosis of dementia be less than the assumed levels, the effect of discounting would be reduced, and maximum cost-effective program prices increased.
Recalling that “decay” includes other factors as well as the need for participants to maintain lifestyle changes over many years, high levels of decay are possible. Studies with long follow-up are needed to assess actual program effects. Programs which continue to interact with the participants continuously over time are likely to improve effects but increase costs. We also note that improving dementia risk factors would improve a range of other health outcomes (e.g. cardiovascular health, diabetes, mild cognitive impairment), in addition to dementia related outcomes. If the total benefits of risk reduction programs were included, they would be even more likely to be cost-effective.
The main limitation in this and any other analysis of cost-effectiveness of dementia prevention interventions is the uncertainty in many parameters, which has required extensive sensitivity analysis to assess a reasonable range of outcomes. However, the approach taken here integrates sensitivity analysis and facilitates estimation of outcomes under varied assumptions.
The study assumed binary outcomes of dementia against no dementia and did not address the benefit of delay in onset of dementia, which also reduced the likelihood of finding cost-effective outcomes. Dementia related QALY losses prior to diagnosis were not included in the study, leading to a further conservative bias in estimates.
Like all approaches to cost-effectiveness modelling for dementia prevention interventions this study is limited by having only short-term program outcomes (1). The baseline calculations assume (1) in the case of Example 1, that well-established associations between risk factors and dementia are causative; (2) for both examples, changes in risk factors driven by interventions have the same effect as if the level of the risk factor was achieved ”naturally” (e.g. reversing midlife obesity with an intervention has the same effect as achieving a normal weight at midlife without intervention) and; (3) changes in risk from a short-term intervention are sustained over time(e.g. weight does not revert to previous levels). The approach used here however provides a simple and transparent way to test the impact of these ongoing concerns.
To explore the cost-effectiveness of interventions aimed at dementia risk reduction requires a means of extrapolating outcomes from what, to date, have been relatively short-term trials. We examined lifetime costs (in both dollar and QALY terms) of dementia and applied these to projected changes in risks of dementia from two example studies. The results suggest that the multi-domain approach of BBL-GP is highly likely to be cost-effective.
The approach shows further the importance of targeting programs to “at risk” portions of the population and the sensitivity to the sustainability or otherwise of trial results. While these factors are well-known, the approach provides a means of estimating the orders of magnitude of program impacts and reinforces the need for longer-term studies to measure all relevant factors to enable assessment of cost-effectiveness with greater confidence.
Funding Sources: This research was undertaken as part of the Centre for Research Excellence in Cognitive Health, which was funded by the National Health and Medical Research Council grant #1100579. Anstey is funded by NHMRC Fellowship #1102694, Zheng is part supported by the NHMRC Dementia Centre for Research Collaboration. The funders had no role in the design and conduct of this study; in the analysis and interpretation of the data; in the preparation of the manuscript; or in the review or approval of the manuscript.
Acknowledgements: We acknowledge the ARC Centre of Excellence in Population Ageing Research.
Conflict of Interest: Dr McRae, Dr Zheng, Dr Bourke, and Professor Cherbuin declare that they have no conflict of interest. Professor Anstey reports personal fees from StaySharp, outside the submitted work.
Ethical standards: The authors followed the ethical guidelines of the Journal for this manuscript.
1. Handels R, Wimo A. Challenges and recommendations for the health-economic evaluation of primary prevention programmes for dementia. Aging & mental health. 2019 Jan;23(1):53-9.
2. Kivipelto M, Mangialasche F, Ngandu T. World Wide Fingers will advance dementia prevention. Lancet Neurol. 2018 Jan;17(1):27.
3. Nguyen KH, Comans TA, Green C. Where are we at with model-based economic evaluations of interventions for dementia? a systematic review and quality assessment. International psychogeriatrics. 2018 Nov;30(11):1593-605.
4. Meeuwsen E, Melis R, van der Aa G, Golüke-Willemse G, de Leest B, van Raak F, et al. Cost-effectiveness of one year dementia follow-up care by memory clinics or general practitioners: economic evaluation of a randomised controlled trial. PloS one. 2013;8(11):e79797-e.
5. Green C, Handels R, Gustavsson A, Wimo A, Winblad B, Skoldunger A, et al. Assessing cost-effectiveness of early intervention in Alzheimer’s disease: An open-source modeling framework. Alzheimer’s & dementia : the journal of the Alzheimer’s Association. 2019 Oct;15(10):1309-21.
. Lin PJ, Yang Z, Fillit HM, Cohen JT, Neumann PJ. Unintended benefits: the potential economic impact of addressing risk factors to prevent Alzheimer’s disease. Health affairs (Project Hope). 2014 Apr;33(4):547-54.
7. Tsiachristas A, Smith AD. B-vitamins are potentially a cost-effective population health strategy to tackle dementia: Too good to be true? Alzheimer’s & Dementia: Translational Research & Clinical Interventions. 2016;2(3):156-61.
8. van Baal PH, Hoogendoorn M, Fischer A. Preventing dementia by promoting physical activity and the long-term impact on health and social care expenditures. Preventive medicine. 2016 Apr;85:78-83.
9. Zhang Y, Kivipelto M, Solomon A, Wimo A. Cost-effectiveness of a health intervention program with risk reductions for getting demented: results of a Markov model in a Swedish/Finnish setting. Journal of Alzheimer’s disease : JAD. 2011;26(4):735-44.
10. Brown L, Hansnata E, La HA. Economic Cost of Dementia in Australia 2016-2056, Report Prepared for Alzheimer’s Australia. Canberra: NATSEM at the Institute for Governance and Policy Analysis, University of Canberra2017.
11. Ashby-Mitchell K, Burns R, Shaw J, Anstey KJ. Proportion of dementia in Australia explained by common modifiable risk factors. Alzheimer’s Research & Therapy. [journal article]. 2017 February 17;9(1):11.
12. Anstey KJ, Kim S, Pond CD, Cherbuin N, McMaster M, Lautenschlager N, et al. Internet-based Intervention Augmented with Diet and Physical Activity Consultation to Decrease Risk of Dementia in At-risk Adults in a Primary Care Setting: Pragmatic Randomized Controlled Trial. Journal of Medical Internet Research. 2020;Accepted for Publication.
13. Wang S, Gum D, Merlin T. Comparing the ICERs in Medicine Reimbursement Submissions to NICE and PBAC—Does the Presence of an Explicit Threshold Affect the ICER Proposed? Value in Health. 2018 2018/08/01/;21(8):938-43.
14. Neumann PJ, Cohen JT, Weinstein MC. Updating Cost-Effectiveness — The Curious Resilience of the $50,000-per-QALY Threshold. The New England Journal of Medicine. 2014 2014 Aug 28;371(9):796-7.
15. Brodaty H, Seeher K, Gibson L. Dementia time to death: a systematic literature review on survival time and years of life lost in people with dementia. International psychogeriatrics. 2012 Jul;24(7):1034-45.
16. Sachs GA, Carter R, Holtz LR, Smith F, Stump TE, Tu W, et al. Cognitive impairment: an independent predictor of excess mortality: a cohort study. Annals of internal medicine. 2011 Sep 6;155(5):300-8.
17. Savva GM, Arthur A. Who has undiagnosed dementia? A cross-sectional analysis of participants of the Aging, Demographics and Memory Study. Age and Ageing. 2015;44(4):642-7.
18. Jutkowitz E, Kane RL, Gaugler JE, MacLehose RF, Dowd B, Kuntz KM. Societal and Family Lifetime Cost of Dementia: Implications for Policy. J Am Geriatr Soc. 2017 Oct;65(10):2169-75.
19. Gnanamanickam ES, Dyer SM, Milte R, Harrison SL, Liu E, Easton T, et al. Direct health and residential care costs of people living with dementia in Australian residential aged care. International journal of geriatric psychiatry. 2018;33(7):859-66.
20. Haaksma ML, Eriksdotter M, Rizzuto D, Leoutsakos J-MS, Olde Rikkert MGM, Melis RJF, et al. Survival time tool to guide care planning in people with dementia. Neurology. 2020;94(5):e538-e48.
21. Strand BH, Knapskog AB, Persson K, Edwin TH, Amland R, Mjorud M, et al. Survival and years of life lost in various aetiologies of dementia, mild cognitive impairment (MCI) and subjective cognitive decline (SCD) in Norway. PloS one. 2018;13(9):e0204436.
22. Prince MJ, Wimo A, Guerchet MM, et al. World Alzheimer Report 2015- The Global Impact of Dementia London: Alzheimer’s Disease International2015.
23. Orgeta V, Edwards RT, Hounsome B, Orrell M, Woods B. The use of the EQ-5D as a measure of health-related quality of life in people with dementia and their carers. Qual Life Res. 2015;24(2):315-24.
24. Australian Institute of Health and Welfare. Australia’s health 2018. Canberra: Australian Institute of Health and Welfare2018.
25. Attema AE, Brouwer WBF, Claxton K. Discounting in Economic Evaluations. Pharmacoeconomics. 2018;36(7):745-58.
26. Devlin N, Scuffham P. Health today versus health tomorrow: does Australia really care less about its future health than other countries do? Aust Health Rev. 2020 Jun;44(3):337-9.
27. Plassman BL, Langa KM, McCammon RJ, Fisher GG, Potter GG, Burke JR, et al. Incidence of dementia and cognitive impairment, not dementia in the United States. Ann Neurol. 2011;70(3):418-26.
28. Brinks R, Landwehr S, Waldeyer R. Age of onset in chronic diseases: new method and application to dementia in Germany. Population Health Metrics. 2013 2013/05/02;11(1):6.
29. Wolters FJ, Tinga LM, Dhana K, Koudstaal PJ, Hofman A, Bos D, et al. Life Expectancy With and Without Dementia: A Population-Based Study of Dementia Burden and Preventive Potential. Am J Epidemiol. 2019 Feb 1;188(2):372-81.
30. Anstey KJ, Cherbuin N, Herath PM, Qiu C, Kuller LH, Lopez OL, et al. A self-report risk index to predict occurrence of dementia in three independent cohorts of older adults: the ANU-ADRI. PloS one. 2014;9(1):e86141.
31. Cherbuin N, Shaw ME, Walsh E, Sachdev P, Anstey KJ. Validated Alzheimer’s Disease Risk Index (ANU-ADRI) is associated with smaller volumes in the default mode network in the early 60s. Brain imaging and behavior. 2019 Feb;13(1):65-74.
32. Richard E, Andrieu S, Solomon A, Mangialasche F, Ahtiluoto S, Moll van Charante EP, et al. Methodological challenges in designing dementia prevention trials – the European Dementia Prevention Initiative (EDPI). Journal of the neurological sciences. 2012 Nov 15;322(1-2):64-70.
33. Clare L, Nelis SM, Jones IR, Hindle JV, Thom JM, Nixon JA, et al. The Agewell trial: a pilot randomised controlled trial of a behaviour change intervention to promote healthy ageing and reduce risk of dementia in later life. BMC Psychiatry. 2015 2015/02/19;15(1):25.