Midlife body weight status has been found to affect late life dementia outcomes. A cohort projections model was developed to assess the impact of midlife body mass index (BMI) profile on dementia in older Australians.
A baseline projection using age-sex specific dementia prevalence rates was constructed and the results of scenarios that took account of midlife BMI were compared with those from population ageing only.
This modelling predicts that if the rising trend in midlife obesity and declining trend in midlife normal weight in Australia are to be taken into account in projecting future numbers of Australians with dementia then the number of people aged 65 or more years with dementia, by 2050, would be 14% higher than that expected from demographic ageing only. If midlife obesity prevalence was decreased to 20% and normal weight increased to 40% over the period of 2015–2025, then dementia cases among persons aged 65–69 years would be lower by about 10% in 2050 compared with the “doing nothing to stop current trends in obesity” projection.
Citation: Nepal B, Brown LJ, Anstey KJ (2014) Rising Midlife Obesity Will Worsen Future Prevalence of Dementia. PLoS ONE 9(9): e99305. https://doi.org/10.1371/journal.pone.0099305
Editor: Sonia Brucki, University Of São Paulo, Brazil
Received: April 18, 2013; Accepted: May 13, 2014; Published: September 3, 2014
Copyright: © 2014 Nepal et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work has been funded through the Australian Government Department of Health and Ageing Dementia Collaborative Research Centre initiative. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Dementia is highly correlated with age. Its prevalence begins to rise from about 1–2 per cent in individuals in their 60s and nearly doubles every five years thereafter . Projections show that, with the ageing of the Australian population, the number of older persons with dementia is expected to increase considerably in the coming decades , . These projections are, however, based on the assumptions that age-sex specific prevalence rates remain constant into the future. A question arises as to how the increase in obesity and overweight in Australian adults is likely to influence the future numbers of older persons with dementia. Modelling shows that changes in the risk factor profile of the older population can have substantial upward and downward influences on the number of people with dementia depending on the future trend in the risk factor considered . In the case of obesity, it is midlife rather than late life obesity status that has been shown to matter more with regard to the risk of developing dementia . Many industrialised societies including Australia are experiencing an upward trend in obesity prevalence . This trend indicates that the rising tide of obesity is likely to make the dementia ‘epidemic’ larger than that expected on the basis of demographic ageing only. This study models and examines changes in future projections of Australians aged 65 years and over living with dementia based on past and hypothetical trends in the midlife BMI profile of older persons.
Method and Data
The modelling began with estimating dementia prevalence by age, sex and midlife BMI status using information on age-sex specific dementia prevalence and prevalence ratios associated with four BMI groups: underweight, normal weight, overweight and obesity. Then the dementia prevalences were applied to the projected population to obtain the number of people with dementia. Population projections were obtained from the Australian Bureau of Statistics . We generated ‘What if’ scenarios by modifying the midlife BMI profile. The age-sex BMI prevalences were obtained by modelling historical time-series data . For the purposes of this study midlife was taken to be 50 years of age.
The modelling steps are described below.
Step 1: Disaggregating dementia prevalence rates by midlife BMI status
Expressing the prevalence ratio (PR) of dementia in people in category j compared to those in the reference category (suffix 1) as(1)And rearranging, we obtain (2)Now, total prevalent cases (P) are sum of prevalent cases across j categories of a risk factor,By substitution,Rearranging, we obtain(3)Once prevalence rate for the reference category (p1) was estimated, prevalence rates for other categories were estimated using equation (2).
Step 2: Projecting prevalent cases of dementia
People were classified into four categories of body mass index (BMI) using standard definitions: obesity (BMI≥30), overweight (BMI 25 to <30), normal range (BMI 18.5 to <25), and underweight (BMI<18.5). Age-sex-BMI specific prevalence rates were assumed to remain constant into the future. These rates were applied to population projections disaggregated by age, sex and BMI status.
Step 3: Comparing ‘ageing only’ and ‘BMI factored’ projections
An ‘ageing-only’ baseline scenario was derived by applying age-sex specific dementia prevalence to the population projections and results of the ‘BMI-factored’ scenario generated by applying age-sex-BMI specific dementia prevalence on the same population projections compared with the baseline scenario. Absolute and relative differences in projected number of people living with dementia based on the ‘ageing-only’ and midlife BMI factored scenarios were calculated. The BMI-factored baseline scenario (Appendix S1) was based on the projection that obesity prevalence at age 50 years would peak at about 46% for males and 39% for females, and normal weight would decline to 13% . This level of obesity is substantially higher than that in in 2010 when about 33% of Australian males and 30% of Australian females aged 50 years were estimated to be obese.
Step 4: Generating intervention scenarios
Two intervention (“what if”) scenarios were generated by modifying the baseline projection of midlife obesity and normal weight prevalence rates. The results are compared against the ‘ageing only’ and ‘BMI-factored baseline’ scenarios. Thus, four scenarios are examined:
Scenario 1: ageing only: applies age-sex specific dementia prevalence.
Scenario 2: BMI-factored baseline: age-sex-BMI specific dementia prevalence based on baseline midlife trend.
Scenario 3: BMI stabilised: obesity and normal BMI stabilised at 2015 level, an optimistic scenario. The obesity levels were 36% for men and 32.6% for women, and normal weight levels were 22.2% for men and 32.3% for women .
Scenario 4: BMI improves: midlife obesity prevalence reduced to 10% and normal BMI increased to 50% over the period of 2015–2025 and maintained at that level thereafter. It was assumed that the BMI profile of the population changes gradually over the intervention period. This scenario illustrates the outcome of reversing the obesity trend back to levels observed in the 1980s .
Data on dementia prevalence
There is no consensus regarding the use of dementia prevalence estimates that best suit the Australian population. A number of different sets of prevalence estimates have been used in the previous studies of dementia in Australia , , . To capture the high  and low  ranges of the available prevalence estimates, two sets of age-sex specific dementia prevalence rates have been used in this modelling (Table 1). The age-sex specific prevalence rates were split for the four midlife BMI categories, namely, obese, overweight, normal weight and underweight using the following prevalence ratios: 1.64 for midlife obesity, 1.26 for midlife overweight , and 1.2 for midlife underweight  versus midlife normal weight.
Table 2 presents results for the four scenarios based on the low and high prevalence regimes for the total population aged 65 years and over. Column 1 shows the numbers of older persons who would be living with dementia according to the ageing-only scenario and as expected these numbers are vastly different depending on the prevalence rates chosen. Columns 2 to 7 compare the outcomes of three BMI-factored scenarios against the ageing-only scenario. If the rising trend in midlife obesity and declining trend in midlife normal weight were to be taken into account, by 2050, the number of people with dementia would be 14% higher than that expected purely from demographic ageing (column 3). Modifying the midlife BMI profile from 2015 would begin to show some impact from 2035. However the impact can barely be noticed when the entire older population aged 65 years and over is considered because a large fraction of older people had already crossed their midlife when the intervention began (columns 4–7).
Taking the 65–69 year age group as an example, the impact of modifying midlife BMI profile becomes apparent in individuals of this age after 2035 (Figure 1). Compared with Scenario 1 (ageing-only), Scenario 2 (BMI factored baseline) would produce 8–9% more prevalent cases of dementia, and Scenario 3 (BMI stabilised at 2015 level) would produce 6% more cases in people in their late 60s. Scenario 4 (BMI-improved) would reduce dementia prevalence cases close to the level expected from the ageing-only scenario. Compared with Scenario 2 (doing nothing to the prevailing obesity trend), Scenario 4 intervention (reducing obesity prevalence to 20% and raising normal weight to 40%) would lower dementia cases in the 65–69 year age group by 4–5% in 2035 and 9–10% in 2050.
Discussion and Conclusions
This work illustrates the potential impact of the historical continuing upward trend in midlife obesity on future dementia numbers in Australia. In the current environment characterised by escalating prevalence of obesity and falling prevalence of normal weight, our modelling shows that, in 2050, there could be 14% more people aged 65 years and over living with dementia than that estimated from demographic ageing alone.
Interventions aimed at containing obesity rates in middle-aged adults at the present level are likely to have only a marginal impact on the size of the dementia epidemic expected in the near future. A substantial reduction in the present obesity level in middle-aged Australians will be needed to contain future dementia numbers even to the size expected from demographic ageing alone. Assuming that a decade is needed to implement the BMI interventions and factoring that a decade and a half is needed for these cohorts to age to 65 years, about two decades are elapsed before the impact of the interventions become noticeable. To illustrate this, we looked at the cross-section of people aged 65–69 years. If an intervention is implemented over the period 2015–2025, its impact on dementia becomes noticeable only after 2035. In an illustrative scenario, we estimated that reducing obesity prevalence to 20% and raising normal weight prevalence to 40% by 2025 would reduce dementia cases by 4–5% in 2035 and 9–10% in 2050.
In this modelling, we focused on relative changes rather than absolute number of cases of dementia. By using two sets of prevalence estimates representing a high and a low prevalence regime, we illustrated the uncertainty in the estimates of number of people living with dementia arising from the choice of one or other prevalence rates. It shows that the model is fairly robust to the choice of prevalence rates. The model was developed in Excel 2010 and provides options for replacing the input data such as age-sex specific dementia prevalence rates and prevalence ratios, and for modifying intervention scenarios by altering BMI profile and the intervention period.
This modelling exercise is a modest attempt to incorporate a modifiable lifestyle risk factor into the projection of dementia and therefore advances previous works based on hypothetical assumptions about delaying or eliminating onset of dementia , . This work adds a new perspective to a similar work that examined the impact of modifying life style factors on dementia  by looking at the impact of changes in midlife body weight profile of birth cohorts.
While this modelling demonstrates that modifying midlife obesity helps contain the future dementia epidemic, the limitations of the study need to be taken into account when interpreting the results. The main limitation is that only BMI was factored in the model. Other lifestyle and biomedical risk factors such as smoking, physical activity, diabetes, cholesterol and blood pressure that have been found to influence the incidence of dementia  are not included in the model. Modelling the joint effect of multiple risk factors was not possible due to the lack of historical prevalence data and prevalence ratios for such combinations. Yet even without modelling the joint impact of multiple factors, the findings of this study remain useful as midlife obesity and overweight have been confirmed as influencing the occurrence of dementia in late life independent of a history of some of the other major risk factors such as diabetes and vascular diseases .
In conclusion, interventions aimed at reducing obesity and increasing normal BMI in middle-aged people can mitigate the effect of rising obesity on late life dementia. However, the impacts only begin to be noticed in the long term. The lag time needs to be taken into account at the time of planning of such interventions.
Conceived and designed the experiments: LB BN KA. Performed the experiments: LB BN. Analyzed the data: LB BN. Contributed reagents/materials/analysis tools: LB BN KA. Wrote the paper: LB BN KA.
- 1. Anstey K, Burns R, Birrell C, Steel D, Kiely K, et al. (2010) Estimates of probable dementia prevalence from population-based surveys compared with dementia prevalence estimates based on meta-analyses. BMC Neurology 10: 62.
- 2. Deloite Access Economics (2011) Dementia across Australia: 2011–2050. Canberra: Alzheimer's Australia.
- 3. Australian Institute of Health and Welfare (2007) Dementia in Australia: National Data Analysis and Development. Canberra. Canberra: AIHW. Cat. No. AGE 53.
- 4. Nepal B, Brown L, Ranmuthugala G (2010) Modelling the impact of modifying lifestyle risk factors on dementia prevalence in Australian population aged 45 years and over, 2006–2051. Australasian Journal on Ageing 29: 111–116.
- 5. Anstey KJ, Cherbuin N, Budge M, Young J (2011) Body mass index in midlife and late-life as a risk factor for dementia: a meta-analysis of prospective studies. Obesity Reviews 12: e426–e437.
- 6. Finucane MM, Stevens GA, Cowan MJ, Danaei G, Lin JK, et al. (2011) National, regional, and global trends in body-mass index since 1980: systematic analysis of health examination surveys and epidemiological studies with 960 country-years and 9·1 million participants. Lancet 377: 557–567.
- 7. Australian Bureau of Statistics (2008) Population Projections, Australia, 2006 to 2101. Canberra: Australian Bureau of Statistics. Cat No.3222.0.
- 8. Nepal B, Brown L (2013) Projection of older Australians with a history of midlife obesity and overweight 2010–2050. Obesity 21: 2579–2581.
- 9. Australian Institute of Health and Welfare (2013) AIHW analysis of the 1980, 1983 and 1989 Risk Factor Prevalence Surveys, 1995 National Nutrition Survey and 1999–2000 Australian Diabetes, Obesity and Lifestyle (AusDiab) Study. Canberra: AIHW.
- 10. Jorm AF, Dear KBG, Burgess NM (2005) Projections of future numbers of dementia cases in Australia with and without prevention. Australian and New Zealand Journal of Psychiatry 39: 959–963.
- 11. Lobo A, Launer LJ, Fratiglioni L, Andersen K, Di Carlo A, et al. (2000) Prevalence of dementia and major subtypes in Europe: A collaborative study of population-based cohorts. Neurology 54: S4–9.
- 12. Fitzpatrick AL, Kuller LH, Lopez OL, Diehr P, O'Meara ES, et al. (2009) Midlife and late-life obesity and the risk of dementia: Cardiovascular Health Study. Archives of Neurology 66: 336–342.
- 13. Vickland V, McDonnell G, Werner J, Draper B, Low LF, et al. (2010) A computer model of dementia prevalence in Australia: Foreseeing outcomes of delaying dementia onset, slowing disease progression, and eradicating dementia types. Dementia and Geriatric Cognitive Disorders 29: 123–130.
- 14. Kivipelto M, Ngandu T, Laatikainen T, Winblad B, Soininen H, et al. (2006) Risk score for the prediction of dementia risk in 20 years among middle aged people: a longitudinal, population-based study. Lancet Neurology 5: 735–741.
- 15. Xu WL, Atti AR, Gatz M, Pedersen NL, Johansson B, et al. (2011) Midlife overweight and obesity increase late-life dementia risk: A population-based twin study. Neurology 76: 1568–1574.