«Tackling Obesities: Future Choices – Modelling Future Trends in Obesity & Their Impact on Health 2nd Edition Government Office for Science ...»
Future Choices – Modelling
Future Trends in Obesity &
Their Impact on Health
Government Office for Science
Tackling Obesities: Future Choices –
Modelling Future Trends in Obesity and
the Impact on Health
Klim McPherson, University of Oxford
Tim Marsh, National Heart Forum
Martin Brown, Martin Brown Associates
This report has been produced by the UK Government’s Foresight Programme.
Foresight is run by the Government Ofﬁce for Science under the direction of the Chief Scientiﬁc Adviser to HM Government. Foresight creates challenging visions of the future to ensure effective strategies now.
Details of all the reports and papers produced within this Foresight project can be obtained from the Foresight website (www.foresight.gov.uk). Any queries may also be directed through this website.
This report was commissioned by the Foresight programme of the Government Ofﬁce for Science to support its project on Tackling Obesities: Future Choices.
The views are not the ofﬁcial point of view of any organisation or individual, are independent of Government and do not constitute Government policy.
Contents 1 Introduction 1 2 Background 2 3 Methods and procedures 5
3.1 Cross-sectional analysis 5
3.2 Microsimulation of BMI growth 7
3.3 Longitudinal BMI model 8
3.4 Disease cost model 9
3.5 Software 10
3.6 Statistical error analysis 10
3.7 Using the programs 11 Part One: Using the Obesity 1 program 4 BMI Trends: ﬁndings 13
4.1 Gender 13
4.2 Morbid obesity 15
4.3 Age 15
4.4 Social class 17
4.5 Ethnicity 19
4.6 Regional variations 19
4.7 Interactions 21 Part Two, Using the microsimulation program, Obesity 2 5 Disease attributable to BMI and costs
Authors’s note: Definitions Body Mass Index is a continuum and the population’s BMI distribution is moving inexorably upwards. Risks of disease and incapacity increase with weight gain through the overweight range (BMI 25-29.9) and increase further with obesity (BMI 30+). The quantitative modelling captures increased risk across the spectrum of raised BMI. In the research literature the word ‘obesity’ is taken to refer either generally to a raised BMI or specifically to a level of BMI greater than or equal to 30. In this report ‘obesity’ refers to the latter unless the context unambiguously implies the more general and commonly used meaning.
Tackling Obesities: Future Choices – Modelling future trends in Obesity and the impact on Health 1 Introduction Rates of obesity and overweight have increased sharply in the UK since the mid-1980s and are projected to continue to rise until 20101. The purpose of this modelling exercise, commissioned by the Foresight programme as part of the Tackling Obesities: Future Choices project, is to project the growth, or otherwise, of obesity rates through to 2050 and to predict the consequences for health, health costs and life expectancy.
A cell-based simulation was planned in order to test the effects of changing the main determinants of obesity on obesity rates. Two developments changed this approach. First, although attempts have been made to model the impacts of policy interventions by a team in Australia2, the early Foresight systems mapping work3 clearly demonstrated that the determinants of obesity were too complex for such a modelling process to be reliable. Secondly, close inspection of 12 years of data from the Health Survey for England1 demonstrated extraordinary order and consistency in obesity trends. The case for making reliable projections from these data, entirely independent of identifying possible determinants, was irresistible, and this is the course that has been pursued. This does not exclude the possibility of incorporating epidemiological impact analysis into any future iterations of this model.
This report considers the following questions in turn:
• What will be the likely distribution of overweight and obesity across the population over the next 40 years?
• What will be the likely health and cost consequences of these extrapolated overweight and obesity trends?
• How much might these consequences be altered by effective interventions to reduce body mass index (BMI) across the population or in targeted subgroups?
Part 1 of the report addresses the likely obesity levels that will be seen in 2050 by supposing that the trends observed between 1993 and 2004 continue until 2050.
Part 2 allows wide-ranging changes in the predicted trajectories of BMI rates among any specified subgroups of the population and calculates the consequences in terms of the rates of related diseases, health service costs and life expectancy (other, BMI-unrelated, determinants of these indices remaining constant).
The results reported in Part 2 have been produced using a microsimulation commissioned specifically by the Foresight Tackling Obesities: Future Choices project for this purpose. The microsimulation models the population of England from the mid-1990s to the end of the 21st century. It grows the population from its current age, gender and disease distributions. Its predictions of the future are based on current birth and death rates. Obesity-related disease and death rates are allowed to change, consequent on changing BMI.
Foresight Tackling Obesities: Future Choices Project
The Turner Commission on a new pensions settlement4 for the 21st century noted:
‘Poor lifestyle trends such as increasing obesity among young adults and children may in the long-term reduce the increase in life expectancy, but over the next 30 years they could make the burden on the working population worse, since they may reduce the number of healthy working-age people more than they reduce the number of elderly pensioners.’4 The timescale of the report anticipates pension policy until 2050, when today’s youth will be nearing pension entitlement. its predictions rely on trends in current death rates. it is too early for the current rise in obesity to have had a major impact on these trends and it is certainly too early for rising childhood obesity and its known consequences later in life to have had any impact on them.
Therefore two significant determinants of pension policy are not addressed by the Commission: morbidity in the medium term and life expectancy in the longer term, consequent on currently changing obesity levels. These could change current trends in mortality once predicted obesity trends affect people reaching an age that brings a greater risk of dying. The effect this might have on life expectancy, as well as health service costs is of considerable interest. Rising obesity levels will almost certainly quite dramatically affect rates of disease caused or influenced by obesity.
The most recent Health Survey for England5 shows that:
One in four adults is now obese.
For men, this figure has nearly doubled since 1993 (13%, rising to 24% in 2004).
For women, the increase is slightly lower (16% rising to 24% from 1993 to 2004).
Using Health Survey for England data and applying international Obesity Task Force (iOTF) definitions6, around 5% of 11–15-year-old boys and 11% of 11–15-year-old girls are considered to be obese. The more commonly used definitions in the UK from 1990 UK Growth Charts (85th and 95th percentiles) show one in four 11-15-year-old boys as being obese.
Obesity prevalence for the period 1995–2004 increased from 14% to 24% for boys and from 15% to 26% for girls (UK Growth Chart definitions).
Obesity prevalence in boys aged 2–10 increased from 10% in 1995 to 16% in 2004 and in girls from 10% in 1995 to 11% in 2004 (UK Growth Chart definitions).
Around 10% of 6–10-year-old boys and girls were shown as obese in 2004 (using iOTF definitions6).
Tackling Obesities: Future Choices – Modelling future trends in Obesity and the impact on Health These trends are broadly mirrored throughout western Europe, while in the USA similar rises were observed some 6–10 years earlier. in 1986, 1 in 200 adult Americans had a BMi 40 and this is now 1 in 50. The rate of increase in BMi 40 is twice as rapid as for BMi 30. Nonetheless, currently 1 in 5 Americans is now obese (BMi 30).
Obesity is related pathologically to a number of common morbid conditions (see 10 Appendix 1). Most of these conditions are uncommon while young, but become prevalent in middle and later life. Current trends suggest that around 8% of obese 1–2-year-old children will be obese when they become adults, while 80% of children who are obese at age 10–14 will become obese adults, particularly if one their parents is also obese.7 Adjusting for parental obesity, the odds ratio of an obese 1–2-year-old being obese as an adult is 1:3, i.e. 30% more likely than a non-obese child. While for a child obese at age 15–17 years, the odds ratio is 17 fold. Among very obese children aged 10–14, the unadjusted odds ratio is 44 fold.
Clearly, the increasing prevalence of obesity in childhood8, is very likely to translate into greatly increased levels of obesity among adults, rendering them more susceptible to chronic, life-threatening illness.
in adults, obesity increases the likelihood of type 2 diabetes dramatically – by up to 80 times that of the non-obese. Diabetes is a predisposition for hypertension and coronary heart disease as well other morbidity. Obesity increases the risk of coronary heart disease by 2–3 times and, although BMi may not be a strong independent risk factor, other measures of obesity, such as waist:hip ratio, certainly are. Mortality from cancer among non-smoking obese people is elevated by around 40% compared to non-obese people. Among post-menopausal women, obesity is a significant risk factor for breast cancer. Of course, obesity is associated with many less serious but debilitating conditions such as shortness of breath, back pain, reduced mobility and poor quality of life, as well as an increased psychological and social burden.9 Modelling the current effects by evidence-based extrapolation and incorporating and attributing the epidemiology of related diseases allows straightforward estimates to be made for the time development of incidence and death rates over the next 50 years – at least for those conditions most closely associated with obesity. This is done by making basic assumptions about plausible rates of change in childhood obesity rates and tracking individuals into adulthood, using established likelihoods from current trends. These, in turn, can be used to compare predicted illness and mortality rates, under various assumptions, with those that arise from demographic extrapolation from existing current mortality rates (as yet unaffected by rapid, and unprecedented, changes in childhood obesity) during the coming half-century. These figures could be used to revise estimates of the healthy working population, for example, by removing the dead and accounting for the sick in a manner that incorporates known and current changes in obesity.
Foresight Tackling Obesities: Future Choices Project
A great deal of effective public health policy depends on reliable information on what the future might hold, without policy change and with it. This, in turn, depends on our understanding of what health policies are feasible with what consequences, given an understanding of the contemporary causes of obesity, particularly environmental ones. This report contributes to the scientific understanding of the predictable effects of changing obesity. For public health, reliable long-term predictions are vital.
Tackling Obesities: Future Choices – Modelling future trends in Obesity and the impact on Health 3 Methods and procedures To predict future levels of obesity in the English population to 2050 and beyond, a two-part modelling process was undertaken. Different, but complementary, methods were used for each part, with each method having its own computer program. The ﬁrst program, Obesity 1 (obesity_distribution.exe10), implements a cross-sectional and regression analysis; the second program, Obesity 2 (obesity.
exe11), implements a longitudinal analysis using a microsimulation.
Using standard epidemiological methods, the implications (disease incidence and deaths) of these BMI distributions for the future health of the UK population can be estimated. The microsimulation allows for constraints on future BMi distributions to be applied, simulating the effects of successful obesity policy interventions. The consequent changes in obesity-related diseases are predicted.
By utilising a basic disease-cost model, the implications for NHS expenditure in the long term can be estimated (see Section 5.4).
The microsimulation model can also be used to estimate – in principle, to the end of this century – period life expectancy for any year and cohort and by gender.
This can be done under assumptions of either no change in BMI distribution or predicted changes in the distribution, and with or without specified interventions.
Examples of possible implications for life expectancy are described in Section 5.4.
3.1 Cross-sectional analysis Using the annual datasets of the Health Survey for England 1993–2004, we estimated the distribution of obesity, at all ages, for both genders as well as by ethnicity, social class group and geographical region.
The dataset is large (typically 10,000–20,000 records per year) and, especially for BMI, represents good-quality data. The Obesity 1 program is capable of sorting the dataset and implements non-linear regression analysis methods to derive BMI distributions for the projected English population in future years. The distributions are provided in either graphical format (Figures 1 and 3–8 are examples) or in spreadsheet format and are used as the basis for the longitudinal modelling of the Obesity 2 program.