| Notes |
ISBN 9289050425
9 7 8 9 2 8 9 0 5 0 4 2 5
Health Policy Series No. 47
www.healthobservatory.eu
Health Policy
Series
47
AN EVIDENCE REVIEW AND PROPOSED FRAMEWORK
47
Assessing the
economic costs of
unhealthy diets and
low physical activity
Christine Joy Candari
Jonathan Cylus
Ellen Nolte
An evidence review and proposed framework
Unhealthy diets and low physical activity contribute to many chronic diseases and
disability; they are responsible for some 2 in 5 deaths worldwide and for about 30% of
the global disease burden. Yet surprisingly little is known about the economic costs that
these risk factors cause, both for health care and society more widely.
This study pulls together the evidence about the economic burden that can be linked to
unhealthy diets and low physical activity and explores
• How definitions vary and why this matters
• The complexity of estimating the economic burden and
• How we can arrive at a better way to estimate the costs of an unhealthy diet and low
physical activity, using diabetes as an example
The review finds that unhealthy diets and low physical activity predict higher health care
expenditure, but estimates vary greatly. Existing studies underestimate the true economic
burden because most only look at the costs to the health system. Indirect costs caused
by lost productivity may be about twice as high as direct health care costs, together
accounting for about 0.5% of national income.
The study also tests the feasibility of using a disease-based approach to estimate the
costs of unhealthy diets and low physical activity in Europe, projecting the total economic
burden associated with these two risk factors as manifested in new type 2 diabetes cases
at €883 million in 2020 for France, Germany, Italy, Spain and the United Kingdom alone.
The ‘true’ costs will be higher, as unhealthy diets and low physical activity are linked to
many more diseases.
The study’s findings are a step towards a better understanding of the economic burden
that can be associated with two key risk factors for ill health and they will help policymakers in setting priorities and to more effectively promoting healthy diets and physical
activity.
The editors
Christine Joy Candari was an independent consultant at the time of writing this report.
She is currently Chief Consultant for Health Research, U Consult Us Inc, Manila, The
Philippines
Jonathan Cylus is Research Fellow, European Observatory on Health Systems and
Policies, London School of Economics and Political Science
Ellen Nolte is Head of London Hubs, European Observatory on Health Systems and Policies
DIETS AND LOW PHYSICAL ACTIVITY
ASSESSING THE ECONOMIC COSTS OF UNHEALTHY
Christine Joy Candari, Jonathan Cylus, Ellen Nolte
Cover_WHO_nr47_Mise en page 1 7/06/17 14:31 Page 1
Assessing the economic costs of
unhealthy diets and low physical
activity
An evidence review and proposed framework
The European Observatory on Health Systems and Policies supports and promotes evidence-based
health policy-making through comprehensive and rigorous analysis of health systems in Europe. It
brings together a wide range of policy-makers, academics and practitioners to analyse trends in health
reform, drawing on experience from across Europe to illuminate policy issues.
The Observatory is a partnership hosted by the WHO Regional Office for Europe, which includes the
governments of Austria, Belgium, Finland, Ireland, Norway, Slovenia, Sweden, Switzerland, the United
Kingdom, and the Veneto Region of Italy; the European Commission; the World Bank; UNCAM
(French National Union of Health Insurance Funds); the London School of Economics and Political
Science; and the London School of Hygiene & Tropical Medicine. The Observatory has a secretariat
in Brussels and it has hubs in London (at LSE and LSHTM) and at the Technical University of Berlin.
Assessing the economic costs of
unhealthy diets and low physical
activity
Christine Joy Candari, Jonathan Cylus, Ellen Nolte
Keywords:
DIET – ECONOMICS
SEDENTARY LIFESTYLE
CHRONIC DISEASE – ECONOMICS
CHRONIC DISEASE – PREVENTION AND CONTROL
HEALTH CARE EVALUATION MECHANISMS
DELIVERY OF HEALTH CARE
© World Health Organization 2017 (acting as the host organization for, and secretariat of, the
European Observatory on Health Systems and Policies)
All rights reserved. The European Observatory on Health Systems and Policies welcomes requests for
permission to reproduce or translate its publications, in part or in full.
The designations employed and the presentation of the material in this publication do not imply the
expression of any opinion whatsoever on the part of the European Observatory on Health Systems
and Policies concerning the legal status of any country, territory, city or area or of its authorities, or
concerning the delimitation of its frontiers or boundaries. Dotted lines on maps represent approximate
border lines for which there may not yet be full agreement.
The mention of specific companies or of certain manufacturers’ products does not imply that they are
endorsed or recommended by the European Observatory on Health Systems and Policies in preference
to others of a similar nature that are not mentioned. Errors and omissions excepted, the names of
proprietary products are distinguished by initial capital letters.
All reasonable precautions have been taken by the European Observatory on Health Systems and
Policies to verify the information contained in this publication. However, the published material is
being distributed without warranty of any kind, either express or implied. The responsibility for the
interpretation and use of the material lies with the reader. In no event shall the European Observatory on
Health Systems and Policies be liable for damages arising from its use. The views expressed by authors,
editors, or expert groups do not necessarily represent the decisions or the stated policy of the European
Observatory on Health Systems and Policies or any of its partners.
ISBN 978 92 890 5042 5
Printed in the United Kingdom
Typeset by Tetragon, London
Cover design by M2M
Address requests about publications to: Publications, WHO Regional Office for Europe, UN City,
Marmorvej 51, DK-2100 Copenhagen Ø, Denmark.
Alternatively, complete an online request form for documentation, health information, or for
permission to quote or translate, on the Regional Office web site (http://www.euro.who.int/
pubrequest).
Table of contents
Acknowledgements vii
Foreword viii
List of tables, boxes and figures ix
Summary xi
Chapter 1 Introduction 1
Chapter 2 The economic costs of unhealthy diets and low physical activity: what
does the published literature tell us? 3
2.1 Characteristics of reviewed studies 3
2.2 What the evidence tells us: the economic costs of unhealthy diets 5
2.3 What the evidence tells us: the economic costs of low physical activity 7
2.4 Review of the evidence: a summary 10
Chapter 3 Estimating the economic costs of unhealthy diets and low physical
activity is complex 13
3.1 Defining the concepts: how can we understand ‘unhealthy diet’ and ‘low
physical activity’? 13
3.2 Costing studies differ in key assumptions, influencing estimates for the
economic burden of unhealthy diets and low physical activity 16
3.3 The nature and range of costs considered is likely to underestimate the
‘true’ economic burden of unhealthy diets and physical activity 18
3.4 Conceptual and methodological challenges of estimating the economic
costs of unhealthy diets and low physical activity: a summary 19
Chapter 4 Taking available approaches to determining the economic costs of
unhealthy diets and low physical activity further: a proof-of-concept approach
applied to five European countries 21
4.1 Diabetes as an outcome of unhealthy diets and low physical activity 23
4.2 The principal approach used in this study to estimate the economic costs
that can be associated with unhealthy diets and low physical activity 25
4.3 The estimated total economic costs of unhealthy diets and low physical
activity related to diabetes and its complications 40
vi Assessing the economic costs of unhealthy diets and low physical activity
Chapter 5 Discussion and conclusions 43
5.1 Limitations of the costing framework 45
5.2 Implications for future studies 47
References 49
Appendices 55
Acknowledgements
This study forms part of a wider programme of work that seeks to explore in
further depth the available evidence on the economic costs that can be associated with unhealthy diet and lack of physical activity in Europe, or high‐income
countries more broadly, and their effects on health. A better understanding of the
economic burden associated with these critical risk factors can provide important
pointers to inform evidence‐based decision‐making within the European region
as to the most appropriate policies to improve the health, well‐being and quality
of life of the population.
We gratefully acknowledge the very helpful and insightful comments provided by
Nick Cavill on an earlier draft of this work. We would also like to acknowledge
the very useful comments and suggestions provided by the European Commission,
which further helped inform the discussions presented here. We are indebted
to Charles Normand and Josep Figueras for their constructive comments and
guidance that helped shape the final version of this book.
We would like to thank Sarah Cook for copy-editing and Jonathan North and
Caroline White for the production of this book.
The authors are fully responsible for any errors.
Foreword
Lifestyle related health problems are now a challenge of global proportions. For
example, levels of obesity have doubled since 1980, and it costs a staggering
2.8% of the world’s GDP. Europe is not an exception: half of EU adults today
are either overweight or obese, and rates for children are even more worrisome:
one in three. Unless we act, we will condemn a whole generation to a lifetime
of poor health.
With governments across Europe struggling to curb these ever-rising rates, it is
more important than ever to make the economic case for investing in strategies
that promote health and prevent diseases. Unhealthy diets and lack of physical
activity are risk factors for developing a range of chronic diseases such as diabetes, cancer and cardiovascular disease. They not only reduce people’s quality of
life and life expectancy, but also place a burden on our health systems and our
economies, and on society as a whole.
There is growing evidence that many prevention interventions are cost-effective.
It is therefore surprising that OECD countries still spend an average of only
3% of their health care budgets on disease prevention programmes. In contrast,
already some 7% of EU health budgets are spent on treating chronic diseases
linked to obesity.
This clearly points to a need to change mindsets. If we wish to keep our population
healthy, active and productive, and our health systems strong and resilient in the
long term, and reduce the increasing pressure on national health care budgets,
we need to reset our thinking. We need to focus more on disease prevention and
health promotion to save future expenditure on treatment and cure. This sounds
self-evident – yet it is a surprisingly difficult message to get across.
Underpinning policies with sound evidence is essential. I therefore welcome
this report, which provides a useful contribution to the debate on the increasing
importance of promotion and prevention, and I invite you to read the assessment
of the evidence contained within these pages of the costs of unhealthy diets and
low physical activity.
Xavier Prats Monné
Director General of the Directorate-General for Health and Food Safety
List of tables,
boxes and figures
Tables
Table 1 Annual economic costs of unhealthy diets as reported in the published
literature 6
Table 2 Annual economic costs of low physical activity reported in the published
literature 8
Table 3 Intake of major food groups associated with the lowest risks for chronic
disease 13
Table 4 Population-attributable fractions used in costing studies: low physical
activity 17
Table 5 The quality of diets according to the alternate healthy eating index (AHEI)
2010 27
Table 6 Assignment of AHEI scores on mean dietary intakes of AHEI food
groups in France, Germany, Italy, Spain and the United Kingdom 28
Table 7 Relative risks for diabetes related to unhealthy diets and low physical
activity estimated by Li et al. (2015) 29
Table 8 Estimated prevalence rates of unhealthy diets and low physical activity
in the populations eventually developing diabetes and relative risks
for incident diabetes in France, Germany, Italy, Spain and the United
Kingdom 30
Table 9 Estimated incident diabetic cases in 2020 attributable to unhealthy diet
and low physical activity patterns in 2015 32
Table 10 Estimated diabetes-related health care costs in 2020 attributable to
unhealthy diets and low physical activity patterns in 2015 in France,
Germany, Italy, Spain and the United Kingdom 33
Table 11 Estimated diabetic complication-related costs in 2020 attributable to
unhealthy diets and low physical activity patterns in 2015 in France,
Germany, Italy, Spain and the United Kingdom 34
Table 12 Estimated number of incident type 2 diabetes cases at working age
(15–64 years) that can be attributed to unhealthy diets and low physical
activity in France, Germany, Italy, Spain and the United Kingdom, 2020 35
Table 13 Estimated total number of productivity days lost and cost due to
absenteeism and presenteeism among incident type 2 diabetes cases
at working age that can be attributed to unhealthy diets and low physical
activity and who are expected to be in the labour force in France,
Germany, Italy, Spain and the United Kingdom, 2020 36
Table 14 Estimated total number of productivity days lost and cost due to
absenteeism and presenteeism among incident type 2 diabetes cases
x Assessing the economic costs of unhealthy diets and low physical activity
at working age that can be attributed to unhealthy diets and low physical
activity and who are expected to be outside the formal labour force in
France, Germany, Italy, Spain and the United Kingdom, 2020 38
Table 15 Estimated cost of working years lost due to work disability, early
retirement and premature death among incident type 2 diabetes cases
at working age that can be attributed to unhealthy diets and low physical
activity and who are expected to be in the formal labour force in France,
Germany, Italy, Spain and the United Kingdom, 2020 39
Table 16 Estimated economic cost that can be associated with unhealthy diets
and low physical activity patterns in 2015 as manifested in incident
diabetes and complication in France, Germany, Italy, Spain and the
United Kingdom in 2020 41
Table 17 Estimated total and per capita economic cost that can be associated
with unhealthy diets and low physical activity patterns in 2015 as
manifested in incident diabetes and complications in France, Germany,
Italy, Spain and the United Kingdom in 2020 41
Boxes
Box 1 Population-attributable fraction 5
Box 2 Examples of moderate-intensity and vigorous-intensity physical activity 15
Box 3 Conceptualizing economic costs 19
Box 4 Health risks associated with unhealthy diets and low physical activity 22
Box 5 Computing the fully-adjusted population-attributable fraction (PAF) 31
Figures
Figure 1 Relationship between unhealthy diets and low physical activity with type
2 diabetes and associated costs 24
Summary
Unhealthy diets and low levels of physical activity are among the main risk factors
for major chronic diseases. While this is well documented, the economic burden
associated with these two risk factors remains uncertain. A better understanding
of the economic burden could help inform priority setting and motivate efforts
to promote more effectively healthy diets and physical activity in Europe and
worldwide. This volume seeks to help advance the debate through
• critically reviewing the literature that has sought to estimate the economic costs associated with unhealthy diets and low physical activity
and presenting the range of estimates of economic burden;
• analysing the measurement, methodological and practical issues in
assessing the economic burden from unhealthy diets and low physical
activity; and
• developing a framework for assessing costs and testing the feasibility
of this approach to provide better estimates of the economic burden.
The published evidence overwhelmingly shows that unhealthy diets and low
physical activity are predictive of higher health care expenditure, but estimates vary greatly.
We used a rapid assessment of the evidence that has been published between
January 2000 and February 2016, including a total of 30 studies for detailed
review. Of these studies, six addressed diet, 21 looked at physical activity, and
three considered both. Over half of the studies were set in North America, with
only six set in Europe.
Most studies retrospectively assessed the economic burden of unhealthy diets and
low physical activity. About half adopted a disease-based approach, often looking
at cardiovascular disease, type 2 diabetes and selected cancers. In most cases this
approach used the population-attributable fraction (PAF), which estimates the
proportion of a disease that can be attributed to a particular risk factor.
We found that, overall, estimates presented in the reviewed studies varied widely.
Not all included studies reported national or per capita costs that can be associated with the two risk factors. Of those that did, the estimated health care costs
for unhealthy diets ranged from €3.5 per capita in China to €63 in Australia and
xii Assessing the economic costs of unhealthy diets and low physical activity
€156 in the United Kingdom. For low physical activity, the estimated annual
per capita health care costs ranged from €3 in the Czech Republic to €48 in
Canada. In the United States of America alone, these ranged from less than €1
in two studies up to €185. Only four studies also considered indirect costs in
their estimates for low physical activity, and these varied from €3.7 per capita
in China to between €127 and €224 in Canada.
It is plausible that differences in socioeconomic conditions and health care and
labour costs might lead to some of the differences between cost estimates. But
much of the observed variation results from differences in approaches to measurement, such as what is meant by ‘unhealthy diets’ or ‘low physical activity’;
large differences in the methodological approaches chosen (for example, the use
of a retrospective, disease-based approach or of a prospective approach); differences in the populations studied and underlying data that are being used; and
the range and types of costs considered.
Defining what constitutes an ‘unhealthy diet’ and ‘low physical activity’
remains challenging and different conceptualisations make a comparative
assessment of available evidence difficult.
Definitions of unhealthy diets often refer to those high in specific nutrients
such as saturated fats, salts or sugars. Yet, growing evidence finds that intakes of
specific foods rather than the actual nutrients are most relevant for the development of chronic disease, and studies increasingly look at recommended intakes
of selected food groups, such as fruit, vegetables, nuts and seeds, whole grains,
seafood and unprocessed red meats. This is based on studies that found intake
levels for these food groups to be associated with coronary heart disease, stroke,
type 2 diabetes and certain cancers. But even with this approach there remains
the problem of how to score different intake levels.
These conceptual challenges are reflected in the reviewed studies of unhealthy
diets, which often focused on single food items, such as intakes of fruit and
vegetables, while others considered a range of nutrients and foodstuffs including
saturated- and trans-fat, fruit, vegetables and whole grains. Again others examined
diets that scored low on a measure of dietary diversity, or they analysed dietary
patterns according to national dietary recommendations. A small number of
studies did not define what was meant by unhealthy diet.
There is more consensus about what constitutes low physical activity. For measurement purposes, physical activities are classified into categories of intensity, from
light to moderate to vigorous, depending on the amount of effort required to
perform the activity. Guidelines by the WHO combine intensity with duration,
for example recommending that adults should engage in at least 150 minutes of
moderate-intensity aerobic physical activity throughout the week, in bouts of at
least 10 minutes’ duration, to reduce the risk of chronic disease.
Summary xiii
Yet, as with unhealthy diet, reviewed studies that have assessed the costs of
low physical activity used different definitions, ranging from those focusing on
different thresholds of intensity of activities to those looking just at the duration. Others only considered specific types of physical activity (or lack thereof),
such as leisure-time physical activity, walking, participation in an exercise
programme or sedentary behaviour. Two studies did not define low physical
activity.
These differences in definitions and conceptualisations undermine the comparative
assessment of available cost estimates that can be associated with either risk factor.
Costing studies differ in analytical approaches and in the nature and scope of
data used, influencing estimates for the economic burden of unhealthy diets
and low physical activity.
Just under half of reviewed studies used population-attributable fractions (PAFs)
to assess the contribution of unhealthy diets or low physical activity to a range
of diseases (or death) and this then formed the basis to estimate the economic
burden.
Studies analysed different numbers and combinations of diseases, most commonly
coronary heart disease, stroke, type 2 diabetes and colorectal and breast cancer.
They also varied in the population-attributable fractions used. They commonly
applied the PAFs to contemporary prevalence of unhealthy diets or low physical
activity and then estimated the related disease costs incurred in the same year.
Such an approach overlooks the time lag between exposure to the risk factor
and the development of disease.
There was equally great variation in the range of costs being considered by
individual studies. The majority only looked at the direct health care costs,
and this is likely to greatly underestimate the true economic burden that can
be associated with unhealthy diets or low physical activity. For example, three
Canadian studies that had analysed the costs associated with low physical activity found indirect costs caused by lost productivity to be about twice as high
as direct health care costs, together accounting for between 0.4% and 0.6% of
gross domestic product (GDP).
We developed a framework for estimating the economic costs of unhealthy
diets and low physical activity using a disease-based approach, and applied
it to estimate the economic burden of type 2 diabetes associated with these
two risk factors.
Building on the insights from the critical appraisal of the literature and the review
of measurement and methodological challenges, we developed a framework for
assessing the costs of unhealthy diets and low physical activity. We adopted a
disease-based approach, which incorporates a time perspective to account for the
xiv Assessing the economic costs of unhealthy diets and low physical activity
natural progression of disease. We further estimated incidence rather than prevalence, included costs related to the disease and its complications, and considered
indirect costs of productivity losses as a consequence of absenteeism, presenteeism, work disability, early retirement and premature mortality. The framework
was tested by assessing the costs of type 2 diabetes. Diabetes has been associated
with a high individual, social and economic burden and related expenditure
was estimated to account for some 9% of total health care expenditure in the
European region in 2015. There is the specific advantage of using diabetes as
the disease for testing the framework in that there is a well-established causal
pathway from these risk factors and disease.
We projected the total economic costs that can be associated with unhealthy
diets and low physical activity in 2015 as manifested in incident type 2
diabetes cases in 2020 in five large European countries to be €883 million.
Using this approach, we projected the total economic cost of diabetes that can
be associated with unhealthy diets and low physical activity. The approach takes
the patterns of diet and activity levels in 2015 and projects incident diabetes
cases for the year 2020. The estimated direct and indirect costs associated with
these cases ranged from €82.4 million in Spain to €266.7 million in Germany.
This equates to a per capita cost of €1.77 in Spain to €3.29 in Germany. Relating
costs more specifically to the population projected to develop diabetes in 2020 as
a consequence of unhealthy diets and low physical activity in 2015, the United
Kingdom showed the highest amounts, at €18 953, closely followed by Germany
and France, while Italy had the lowest cost, at just over €10 720.
The total cost in the five high-income countries studied (France, Germany,
Italy, Spain and the United Kingdom) was projected to amount to about €883
million in 2020. The populations in the five countries studied account for
almost two thirds of the total population in the European Union (EU-28). This
would imply a total EU cost of around €1.3 billion, but care must be taken in
any extrapolation given differences in population characteristics, costs of care
and value of lost productivity. While these estimates of the economic costs are
substantial, they represent only a small proportion of health care expenditure
and a very small proportion of GDP. Even on the higher estimates in the sensitivity analysis it is likely that the burden of disease associated with unhealthy
diets and low physical activity as measured by poor health and shortened life
will be at least as important as the financial costs of additional health care and
lost productivity.
It is difficult to compare the findings of the analyses presented here with estimates
published elsewhere since only diabetes costs are estimated. The principal analytical steps used in our analysis are similar to those in the recent Lancet Physical
Activity 2016 Series for low physical activity. Where our model differs is that we
Summary xv
only considered the costs of new cases, which can be causally linked to the risk
factor, and we take account of the expected time lag between exposure to the risk
factor (unhealthy diets, low physical activity) and the development of the disease,
and of complications that arise from diabetes. We also considered a wider range
of indirect costs linked to lost productivity because of work absence, disability,
early retirement and premature death among incident diabetes cases that can
be attributed to unhealthy diets and low physical activity. While our estimates
are restricted to diabetes, they provide a fuller picture of the likely future costs
that can be attributed to contemporary dietary and physical activity patterns.
Where do we go from here?
This study has tested the feasibility of estimating the costs of unhealthy diets and
low physical activity using a disease-based approach. While there are limitations,
it has shown that it is broadly feasible to populate the model with data from a
range of sources, and the results show a reasonable consistency across countries.
While the disease burden from diabetes is not currently as large as that for, for
example, ischaemic heart disease, it is a good exemplar because of the strong relationship between these lifestyle factors and the risk of diabetes. In other chronic
diseases there will be additional challenges in identifying the contribution of
these lifestyle factors and disease risk. Given the very wide range of estimates of
costs from the studies reviewed, this may be a more promising approach.
Chapter 1
Introduction
Unhealthy diets and low physical activity are among the key risk factors for major
chronic, non-communicable diseases such as cardiovascular diseases, cancers and
diabetes. In 2015, diets that are low in fruit and vegetables or high in sugar,
processed foods or sodium were estimated to directly account for 37% of all
deaths and just over a quarter of the total disease burden (disability-adjusted
life years, DALYs) (GBD 2015 Risk Factors Collaborators, 2016). Low levels of
physical activity accounted for another 5% of all deaths and 3.4% of DALYs.
Taken together, these two risk factors were thus responsible for some two in
five deaths worldwide and about 30% of the global disease burden. It is against
this background that several strategies have been launched at European and
global levels since the early 2000s to promote healthy diets and physical activity
and so reduce the related burden of ill-health (Commission of the European
Communities, 2007; Council of the European Union, 2014; WHO Regional
Office for Europe, 2001; WHO Regional Office for Europe, 2005; WHO
Regional Office for Europe, 2015; WHO Regional Office for Europe, 2016;
World Health Organization, 2004; World Health Assembly, 2013).
There is an expectation that effectively promoting healthy diets and physical
activity can help reduce the economic burden associated with chronic diseases,
which was estimated to account for 70–80% of health care budgets, or €700 billion annually across the European Union alone (European Commission, 2014).
However, while there is good evidence about the positive impacts of, for example,
a healthy diet on outcomes such as major cardiovascular events (Estruch et al.,
2013; Stefler et al., 2015; Tong et al., 2016; Liyanage et al., 2016) or about
the association between physical activity and mortality (Samitz et al., 2011;
Woodcock et al., 2011), only a small number of studies have provided robust
estimates for the economic impacts on health care and the wider society that are
directly related to either factor. This is in part because the relationships between
unhealthy diets or low physical activity and health care costs is complex, and it
is the consequences of these behaviours, for example obesity or specific lifestylerelated diseases, that lead to health care costs. However, there may be more direct
effects of unhealthy diets and low physical activity on lost productivity.
Available studies provide widely varying estimates, reflecting the range of assumptions and estimates that inform underlying models (Cecchini & Bull, 2015). For
2 Assessing the economic costs of unhealthy diets and low physical activity
example, Scarborough et al. (2011) estimated the economic burden of ill health
that could be attributed to unhealthy diet to the National Health Service in the
United Kingdom in 2006–07 to be £5.8 billion (€8.5 billion) and £0.9 billion
(€1.3 billion) for low physical activity. Maresova (2014) calculated the financial
cost of low physical activity to public health insurance in the Czech Republic to
be 0.4% of total health care cost in 2008. These figures do not take account of
the wider societal costs that can be attributed to these risk factors. More recently,
Ding et al. (2016) estimated that in 2013 health care costs associated with low
physical activity accounted for an average of 0.6% of total health expenditure
across EU Member States.
This volume seeks to help provide a clearer picture of the economic costs of
unhealthy diets and low physical activity in Europe, to understand the methodological and practical difficulties in assessing costs, and to provide a test case
to show how costs might better be assessed.
Chapter 2 provides a targeted review of the literature that has sought to estimate
the economic costs associated with unhealthy diets and low physical activity. This
provides estimates of the costs as reported in the different studies, and shows
how the estimated costs vary with the different assumptions and methods used.
Chapter 3 reviews the methodological and practical challenges in estimating the
economic costs of unhealthy diets and low physical activity. It looks in detail at
issues of measurement of diet and physical activity, and at how there are strong
interactions between diet and physical activity in terms of risks of disease (and
indeed in different elements of diet), and shows how these challenges affect the
estimates of costs.
Chapter 4 develops a bottom-up framework for assessing the costs of unhealthy
diets and low physical activity using a disease-based approach. This is then tested
out for a disease (type 2 diabetes) for which there is strong evidence that the
disease is related to both diet and the level of physical activity. This provides a
test of concept, and shows how evidence from a range of sources can potentially
be combined to improve our understanding of the economic cost.
Chapter 5 briefly summarises the findings of the work and proposes avenues
for further research.
Chapter 2
The economic costs of unhealthy
diets and low physical activity: what
does the published literature tell us?
This chapter reports the findings of a targeted review of the literature that has
sought to estimate the economic costs associated with unhealthy diets and/or
low physical activity. No previous review has covered costs associated with both
of these, and existing reviews of costs of low physical activity (Kruk, 2014;
Oldridge, 2008) did not provide information on the methodological approach
or a critical evaluation of reviewed studies. The aim of this review is to present
the current best estimates of the economic costs, and the strengths and weaknesses of the available studies.
Drawing on the principles of a rapid evidence assessment (Khangura et al.,
2012), we carried out a targeted search of PubMed, the National Library of
Medicine’s Medline and pre-Medline database (NCBI, 2016). We identified
studies using medical subject headings (MeSH) as follows (‘/’ indicating ‘or’):
‘sedentary lifestyle/leisure activities/motor activity’ or ‘diet/food’ in combination
with ‘health care costs[statistics and numerical data]/costs and cost analysis/
public health[economics]/cost of illness’. We limited our search to studies that
were published between January 2000 and February 2016 and that were in the
English language.
We included original studies estimating the costs that can be associated with
unhealthy diets or low physical activity. We did not consider studies that focused
on populations with established disease (e.g. people with osteoarthritis), intervention studies or those that compared different populations with different levels
of physical activity or differing dietary behaviours, except where these were
quantified further. Given the overall scarcity of studies, we adopted an inclusive
approach and we did not formally assess the quality of included studies. We
excluded editorials, commentaries or letters.
2.1 Characteristics of reviewed studies
This section briefly summarises the key characteristics of reviewed studies; a
detailed overview is presented in Appendix 1.
4 Assessing the economic costs of unhealthy diets and low physical activity
The PubMed searches identified a total of 3661 records (diet: 2347; physical
activity: 1314) and, following screening of abstracts and titles, we considered
38 studies for full-text review. Of these, 30 studies were considered eligible for
inclusion in the review. Six addressed diet, 21 examined physical activity, and
three considered both.
Eleven of the included studies were set in the United States of America (Ackermann
et al., 2003; Anderson et al., 2005; Bachmann et al., 2015; Bland et al., 2009;
Carlson et al., 2015; Chevan & Roberts, 2014; Daviglus et al., 2005; Garrett
et al., 2004; Martinson et al., 2003; Wang et al., 2005; Wang et al., 2004), five
in Canada (Alter et al., 2012; Janssen, 2012; Katzmarzyk, 2011; Katzmarzyk et
al., 2000; Krueger et al., 2015), three each in the United Kingdom (Allender
et al., 2007; Rayner & Scarborough, 2005; Scarborough et al., 2011) and in
Australia (Collins et al., 2011; Doidge et al., 2012; Peeters et al., 2014), two in
China (Popkin et al., 2006; Zhang & Chaaban, 2012) and one each in Brazil
(Codogno et al., 2015), Germany (Idler et al., 2015), Czech Republic (Maresova,
2014), Japan (Kuriyama et al., 2004) and Taiwan, China (Lo et al., 2013). One
study assessed the economic costs of disease-related malnutrition in health care
settings in Ireland (Rice & Normand, 2012).
The majority of studies provided a retrospective assessment of the economic
burden that can be associated with either unhealthy diets or low physical activity
or both, while nine studies adopted a prospective approach by following a cohort
of people over a defined period of time (Alter et al., 2012; Bachmann et al., 2015;
Bland et al., 2009; Chevan & Roberts, 2014; Collins et al., 2011; Kuriyama et
al., 2004; Lo et al., 2013; Martinson et al., 2003; Peeters et al., 2014).
About half of the reviewed studies adopted a disease-based approach to estimate
the economic burden that can be associated with unhealthy diets or low physical
activity, most frequently cardiovascular diseases (coronary heart disease, stroke,
hypertension), type 2 diabetes, and colon and female breast cancer (Allender et
al., 2007; Daviglus et al., 2005; Doidge et al., 2012; Garrett et al., 2004; Janssen,
2012; Katzmarzyk, 2011; Katzmarzyk et al., 2000; Krueger et al., 2015; Maresova,
2014; Popkin et al., 2006; Rayner & Scarborough, 2005; Scarborough et al.,
2011; Wang et al., 2004; Zhang & Chaaban, 2012). The disease-based approach
typically, although not always (Daviglus et al., 2005; Wang et al., 2004), uses
the population-attributable fraction (PAF) to quantify the contribution of the
individual risk factor (unhealthy diet, low physical activity) to the burden of a
given disease or death (Box 1).
The remaining studies used a generic, non-disease-based approach, where
unhealthy diets or low physical activity data of each individual were linked
to health care cost data, regardless of the type of disease or diagnosis. Such
an approach is typically followed by regression techniques to identify possible
The economic costs of unhealthy diets and low physical activity 5
associations between the presence or absence of the risk factor and the magnitude of costs (Ackermann et al., 2003; Alter et al., 2012; Anderson et al.,
2005; Bachmann et al., 2015; Bland et al., 2009; Carlson et al., 2015; Chevan
& Roberts, 2014; Codogno et al., 2015; Collins et al., 2011; Idler et al., 2015;
Kuriyama et al., 2004; Lo et al., 2013; Martinson et al., 2003; Peeters et al.,
2014; Rice & Normand, 2012; Wang et al., 2005).
Studies using a disease-based approach based on population-attributable fractions
reported costs that can be associated with unhealthy diets and/or low physical
activity in aggregate terms, for example national costs. Conversely, studies that
adopted a generic approach tended to report the costs as ‘additional costs’, that
is, additional to the costs a non-exposed individual would otherwise incur, in
per capita terms. Two studies reported risk estimates, in this case, odds ratio,
illustrating the strength of the association between unhealthy diet or low physical
activity and costs (Chevan & Roberts, 2014; Codogno et al., 2015).
2.2 What the evidence tells us: the economic costs of
unhealthy diets
Drawing on those studies that have reported aggregate costs, the annual economic
costs of unhealthy diets ranged from €1.4 billion in Australia (AU$ 2 billion)
(Doidge et al., 2012) to €4.5 billion in China (US$ 4.2 billion) (Popkin et al.,
2006) and €8.5–9.5 billion in the United Kingdom (£5.8–6 billion) (Rayner &
Scarborough, 2005; Scarborough et al., 2011) (see Appendix 2 for conversion
Box 1 Population-attributable fraction
The population-attributable fraction (PAF) generally refers to the proportion of cases for a given
outcome of interest that can be attributed to a given risk factor among the entire population.
Specifically, the PAF is a function of the proportion of individuals in the population who are exposed
to the factor of interest (Pexp), for example, unhealthy diet, and the relative risk (RR) of a particular
outcome given that exposure, for example, the development of type 2 diabetes. If the exposure
variable is dichotomous (i.e. the risk factor is present or absent), the mathematical formula reads:
For example, if the relative risk for the effect of a given exposure on a disease outcome was
approximately 5, and we can infer from a population survey that about 20% of the population
was exposed to this risk factor, the proportion of all disease cases in the population that can be
attributed to the risk factor is calculated as: PAF = 0.05 x (20-1) / (0.05 x (20-1) + 1) = 0.95 /
1.95 = 49%.
PAF (%) =
Pexp (RR−1)
[Pexp (RR−1)] + 1
6 Assessing the economic costs of unhealthy diets and low physical activity
rates of currencies applied) (Table 1). Taking account of the population size, the
per capita annual economic costs that can be associated with unhealthy diets is
estimated to range from €143 to €156 for the United Kingdom, €63 for Australia
and €3.5 for China. All of these costs were health care costs only.
Table 1 Annual economic costs of unhealthy diets as reported in the published
literature
Country
Estimated annual
economic costs
of unhealthy diets
(per capita*)
Definition of
unhealthy diets
Perspective of
cost estimation
Population
base Source
Australia €1.4 billion
(€63)
Low levels of dairy
consumption
Direct health care
costs, not specified
General
population
Doidge et al.
(2012)
China €4.5 billion
(€3.5)
Diet high in saturated
and trans-fat, low in
fruit, vegetables and
whole grains plus
heavy alcohol drinking
Direct health care
costs, not specified
General
population
Popkin et al.
(2006)
United
Kingdom
€8.5 billion
(€143) Not defined Direct health care
costs, not specified
General
population
Rayner &
Scarborough
(2005)
€9.5 billion
(€156) Not defined Direct health care
costs, not specified
General
population
Scarborough et
al. (2011)
Note: * per capita costs calculated using United Nations population data (United Nations, 2015).
Reviewed studies varied widely in their definition of unhealthy diets, often
focusing on single food items, such as dairy consumption (Doidge et al., 2012)
or intakes of fruit or vegetables (Bland et al., 2009; Daviglus et al., 2005), while
Popkin et al. (2006) considered a range of nutrients and foodstuffs including
high consumption of saturated- and trans-fat, and low consumption of fruit,
vegetables and whole grains, as well as heavy alcohol use. Lo et al. (2013) examined diets that scored low on a measure of dietary diversity, while Collins et al.
(2011) analysed dietary patterns according to Australian national dietary recommendations. Two studies did not define unhealthy diets specifically (Rayner &
Scarborough, 2005; Scarborough et al., 2011), and Rice & Normand (2012)
analysed the cost of malnutrition in health care settings.
Similarly, studies also varied in relation to the data that were used to assess diet
and costs. A small number of studies used national surveys of self-reported food
intake (Doidge et al., 2012; Popkin et al., 2006), while others did not collect data
on dietary patterns specifically but instead used readily calculated populationattributable fractions published elsewhere. For example, Rayner & Scarborough
(2005) and Scarborough et al. (2011) drew on PAFs produced as part of the
Global Burden of Disease studies (Ezzati et al., 2004; Murray & Lopez, 1997).
The economic costs of unhealthy diets and low physical activity 7
All reviewed studies estimated disease-based direct health care expenditures and
only Rice & Normand (2012) also included social care costs. All estimated costs
applied to the general population; the only exception was the study by Bland
et al. (2009), which estimated the costs for a sample of members of one health
insurance plan in the United States (n = 7983 individuals).
2.3 What the evidence tells us: the economic costs of low
physical activity
Turning to the economic costs that can be associated with low physical activity,
fourteen studies reported aggregate costs. These ranged from €29 million per annum
in the Czech Republic (Kč 700 million) (Maresova, 2014) to €1.32–1.68 billion
in the United Kingdom (£0.9–1.06 billion) (Scarborough et al., 2011; Allender et
al., 2007), €1.3–7.9 billion in Canada (C$ 2.1–10.8 billion) (Katzmarzyk et al.,
2000; Krueger et al., 2015; Janssen, 2012; Katzmarzyk, 2011), €1.8–4.9 billion
in China (US$ 1.7–6.8 billion) (Popkin et al., 2006; Zhang & Chaaban, 2012)
and €90.5 million–€57.7 billion in the United States (US$ 83.6 million–79 billion) (Garrett et al., 2004; Anderson et al., 2005; Carlson et al., 2015) (Table 2).
Taking account of population, the estimated annual per capita health care costs
ranged from €3 in the Czech Republic to €48 in Canada. In the United States
alone, per capita health care cost estimates ranged from less than €1 in two studies
up to €185. Only four studies also considered indirect costs in their estimates, and
these varied from €3.7 per capita in China to between €127 and €224 in Canada.
As illustrated in Table 2, and similar to studies analysing unhealthy diets, low
physical activity or physical inactivity was defined differently across studies, which
makes it difficult to compare estimates. Four studies defined physical inactivity
as not meeting recommendations for moderate- and vigorous-intensity types of
physical activity (Idler et al., 2015; Janssen, 2012; Maresova, 2014; Zhang &
Chaaban, 2012), while Carlson et al. (2015) considered only moderate-intensity
activity and Garrett et al. (2004) only vigorous-intensity activity. Four studies
defined physical inactivity as not meeting the recommended duration (which varied
across studies), regardless of intensity (Alter et al., 2012; Anderson et al., 2005;
Bland et al., 2009; Martinson et al., 2003). Katzmarzyk et al. (2000), Garrett et
al. (2004), Katzmarzyk (2011) and Krueger et al. (2015) only considered leisuretime physical inactivity, Popkin et al. (2006) only considered sedentary behaviour,
Kuriyama et al. (2004) focused on walking and Ackerman et al. (2003) defined
physical activity as participation in an exercise programme. Bachmann et al. (2015)
and Peeters et al. (2014) conceptualised activity in terms of metabolic equivalents
achieved while performing a physical activity, and Wang et al. (2004; 2005) used
small increases in heart rate or heavy breathing induced by physical activity as a
measure of activity. Allender et al. (2007) and Scarborough et al. (2011) did not
define physical inactivity specifically.
8 Assessing the economic costs of unhealthy diets and low physical activity
Country
Annual economic
costs of low
physical activity
(per capita*)
Definition of low
physical activity Perspective of cost estimation Population
base Source
Czech
Republic
€29 million
(€2.8)
Less than 150 minutes of
moderate activity or less
than 75 minutes of vigorous
activity a week or less than
180 minutes of walking
weekly or any combination
resulting in less than 600-
MET minutes a week on at
least three days per week
or no physical activity at all.
Direct health care costs, not
specified
General
population
Maresova
(2014)
United
Kingdom
€1.32 billion
(€22) Not defined Direct health care costs, not
specified
General
population
Scarborough
et al. (2011)
€1.68 billion
(€28) Not defined
Direct health care costs: inpatient
and outpatient, primary care,
pharmaceutical and net community
care service expenditures
General
population
Allender et
al. (2007)
Canada
€1.3 billion
(€43)
Leisure-time energy
expenditure of less than
12.6 kilojoules per kg
bodyweight per day
Direct health care costs, not
specified
General
population
Katzmarzyk
et al. (2000)
€4.3 billion
(€127)
Less than 150 minutes of
moderate-vigorous physical
activity per week
Direct health care costs: hospital,
drugs, physician care, care in other
institutions, ‘additional’ direct
health care expenditures
Indirect productivity costs: income
lost due to sickness absence,
injury-related work disability
and premature deaths before
retirement
General
population
Janssen
(2012)
€5.4 billion
(€160)
Leisure-time energy
expenditure of <1.5 kcal
per kg bodyweight per day
Direct health care costs: hospital,
drugs, physician care, care in other
institutions, other professional fees,
public health, health research and
pre-payment administration costs
Indirect productivity costs: income
lost due to premature death and
short-term and long-term disability
General
population
Katzmarzyk
(2011)
€7.9 billion
(€224)
Average leisure-time energy
expenditure of less than 1.5
kcal per kg per day over the
past three months
Direct health care costs: hospital,
physician services, other health
care professional services
(excluding dental services), drugs,
health research and ‘other’ health
care expenditures
Indirect productivity costs: income
lost due to short-term disability,
long-term disability and premature
mortality (before retirement)
General
population
Krueger et al.
(2015)
Table 2 Annual economic costs of low physical activity reported in the published
literature
The economic costs of unhealthy diets and low physical activity 9
Data on physical activity were typically self-reported (assessed through surveys).
Janssen (2012) measured physical activity using accelerometers, and Bachmann
et al. (2015) assessed cardiorespiratory fitness as a measure of habitual physical
activity using a treadmill test. Four studies estimated indirect costs associated with
low physical activity, in addition to health care costs (Janssen, 2012; Katzmarzyk,
2011; Krueger et al., 2015; Zheng et al., 2012). Analyses mostly applied to the
general population, although six studies estimated costs for specific populations,
such as members of a health plan in the United States (Ackermann et al., 2003;
Anderson et al., 2005; Bachmann et al., 2015; Garrett et al., 2004; Idler et al.,
2015; Wang et al., 2005). Idler et al. (2015) analysed the relationship between
physical activity and health care and (parental) productivity costs among children
aged 9 to 12 years.
Country
Annual economic
costs of low
physical activity
(per capita*)
Definition of low
physical activity Perspective of cost estimation Population
base Source
China
€1.8 billion
(€1.4) Sedentary behaviour Direct health care costs, not
specified
General
population
Popkin et al.
(2006)
€4.9 billion
(€3.7)
Less than 30 minutes of
moderate-intensity activity
for five days per week or
less than 20 minutes of
vigorous-intensity activity
for three days per week
Direct health care costs: hospital,
drugs, physician care, other
professional services, public
health, health research, prepayment administration costs
Indirect productivity costs: income
lost due to sickness absence,
injury-related work disability
and premature deaths before
retirement
General
population
Zhang &
Chaaban
(2012)
The
United
States
€90.5 million
(€0.3)
Less than 20 minutes
of vigorous leisure-time
physical activity for at
least three days a week or
no leisure-time physical
activity at all
Direct health care costs:
inpatient and outpatient facility,
professional, x-ray, laboratory and
pharmaceuticals, including out-ofpocket payments
Sample
population
(n =
1.5 million)
Garrett
et al. (2004)
€212.4 million
(€0.8)
Less than four days of
physical activity lasting
30 minutes or more or no
physical activity at all
Direct health care costs: hospital
and professional fees, excluding
pharmaceuticals
Sample
population
(n =
200 000)
Anderson
et al. (2005)
€26.5 billion
(€98)
Less than 30 minutes of
activity increasing the heart
rate for ≥5 times per week
or less than 20 minutes
of activity substantially
increasing the heart rate for
≥3 times per week
Direct health care costs: inpatient
stays and outpatient visits,
medications and home care,
including out-of-pocket payments
General
population
Wang et
al. (2004)
€57.7 billion
(€185)
Less than 150 minutes per
week of moderate physical
activity or none at all
Direct health care costs: inpatient,
and outpatient, emergency room,
office-based, dental, vision and
home health care, including
prescription drugs
General
population
Carlson et al.
(2015)
Note: * per capita costs calculated using United Nations population data (United Nations, 2015).
10 Assessing the economic costs of unhealthy diets and low physical activity
Two studies reported the costs of low physical activity in combination with
other risk factors. Alter et al. (2012) estimated the incremental health care costs
that can be associated with obesity and additional risk factors, including low
physical activity (described as sedentary) (data not shown in Table 2 as authors
reported per capita costs only). They found that the cumulative additional costs
attributable to overweight and obesity alone were small when compared with
matched normal-weight adults. However, costs increased significantly with
other risk factors. Thus, for obese individuals who were also physically inactive,
health care costs exceeded those of normal-weight, healthy individuals over an
11.5-year period by around €3700 (C$ 4080; p = 0.003). Kuriyama et al. (2004)
estimated that low physical activity increased monthly per capita health care
costs among adults in Japan by 8%, from €172 for adults without lifestyle risk
to €185, with further increases where low physical activity was combined with
obesity (by 16.4%) (1995–2001). Idler et al. (2015) focused on children aged
9 to 12 years and found that low physical activity increased health care costs by
€6 per child annually, but decreased the productivity costs (i.e. earnings lost due
to parental absence from work) by €11 per physically inactive child. However,
the relationship between physical activity and costs in this age group was not
statistically significant. Similarly, Ackermann et al. (2003) found no significant
difference in costs between physically active and inactive members of a health
plan in the United States.
Three studies analysed the comparative impact of unhealthy diets and low physical
activity on cost. Popkin et al. (2006) and Scarborough et al. (2011) estimated
that the cost associated with unhealthy diets exceeded that associated with low
physical activity by a factor of 1.5 to 5. Bland et al. (2009) found that low physical
activity but not unhealthy diet (as measured by low fruit and vegetable consumption) was significantly associated with higher short-term medical costs. The latter
study collected primary data on diet and physical activity among members of a
health plan in the United States to estimate medical costs, whereas both Popkin
et al. (2006) and Scarborough et al. (2011) based their analyses on published
data on population-attributable fractions and studies are not easily comparable.
2.4 Review of the evidence: a summary
In summary, of the 30 studies reviewed, 27 found a significant association
between diet and/or physical activity and costs, with unhealthy diets and low
physical activity predictive of higher health care expenditure. The only exception was the study by Collins et al. (2011), which reported a healthy diet to
be predictive of higher health care costs; the authors noted, however, that the
findings of their study of women were likely confounded by charges incurred for
routine screening services (e.g. cervical and breast cancer screening), with those
The economic costs of unhealthy diets and low physical activity 11
with higher dietary index scores more likely to use these services than those with
poorer scores. Three studies did not find a significant association between diet
or physical activity and costs.
Studies that did report costs associated with the two risk factors found the annual
cost of unhealthy diets to range from €3 to €148 per capita and for low physical
activity from €3 to €181 per capita. The highest health care cost estimates are
equivalent to between 2% and 6% of health spending in the countries. The review
shows that there is a very wide range of estimates, and these are very sensitive to
the measures of diet and activity and the ways in which the studies were carried
out. The next section reviews these methodological and measurement challenges
in assessing the costs of unhealthy diets and low physical activity.
Chapter 3
Estimating the economic costs
of unhealthy diets and low
physical activity is complex
We have shown in the review of published studies that the estimated economic burden associated with unhealthy diets or low physical activity varies
widely. Reasons for this variation include differences in the definition of
what constitutes unhealthy diets or low physical activity; the methodological
approach chosen (such as the method to calculate population-attributable
fractions); and the range and types of costs considered. We discuss each aspect
in turn.
3.1 Defining the concepts: how can we understand
‘unhealthy diet’ and ‘low physical activity’?
Our evidence review illustrates that ‘unhealthy diets’ and ‘low physical activity’
have been conceptualised and interpreted differently, making a comparative
assessment of available studies difficult. For example, definitions of unhealthy
diets often refer to those high in specific nutrients such as saturated fats, salts or
sugars, but growing evidence suggests that intakes of specific foods rather than
macro- or micro-nutrients are most relevant for the development of chronic
disease (Morgan, 2012; Mozaffarian et al., 2011).
Table 3 Intake of major food groups associated with the lowest risks for chronic
disease
Food group Optimal intake levels (mean ± standard deviation)
Fruits 300 ± 30 grams/day
Vegetables 400 ± 40 grams/day
Nuts/seeds 113.4 ± 11.3 grams/week
Whole grains 100 ± 12.5 grams/day
Seafood 350 ± 35 grams/week
Unprocessed red meats 100 ± 10 grams/day
Processed meats 0
Source: Micha et al., 2015.
14 Assessing the economic costs of unhealthy diets and low physical activity
It is against this background that researchers have moved towards identifying and
defining healthy diets based on recommended intakes of selected food groups. For
example, Micha et al. (2015) described optimal consumption levels of selected
food groups, based on probable or convincing evidence about the association
of intake levels and the risk for coronary heart disease, stroke, type 2 diabetes
and certain cancers. An unhealthy diet can be defined as one that does not meet
the recommended intake levels of selected food groups shown in Table 3. Even
with this approach there remains the problem of how to score different levels of
deviation from optimal intake.
Intakes of beneficial dietary factors tend to be positively correlated with each
other and inversely correlated with those considered unhealthy. This correlation could lead to overestimates of the relative risk of each dietary factor and
the total effect of dietary risks at the population level (GBD 2015 Risk Factors
Collaborators, 2016). Instead of individually assessing the risks associated with
selected food groups, an alternative approach is to examine dietary patterns.
Such an approach considers the balance among all food groups, including those
that are recommended for frequent consumption and those that are not. It also
accommodates different eating patterns, so allowing for variation depending on
cultural, ethnic or personal preferences, or the costs and availability of certain
foods. Examples include the Healthy Eating Index (HEI), which measures
adherence to the 2005 dietary guidelines in place in the United States, and the
Alternate Healthy Eating Index (AHEI), which is based on foods and nutrients
predictive of chronic disease risk (Chiuve et al., 2012). Diets which score highly
on either the HEI or the AHEI were shown to be associated with a significant
reduction in the risk of all-cause mortality, cardiovascular disease, cancer and
type 2 diabetes by around 20%, highlighting their relevance for population
health (Schwingshackl & Hoffmann, 2015). The AHEI is discussed further in
the context of our proposed costing framework below.
In contrast to diet, there is more consensus about what constitutes low physical
activity. The World Health Organization (2010) has defined physical inactivity
as the “absence of physical activity or exercise” (p. 53), and it recommends that
adults meet the guidelines of at least 150 minutes of moderate-intensity or at least
75 minutes of vigorous-intensity aerobic physical activity throughout the week,
in bouts of at least 10 minutes’ duration, in order to improve cardiorespiratory
and muscular fitness and bone health, and to reduce the risk of chronic disease
and depression (World Health Organization, 2010). Physical activity includes
activities undertaken while working, playing, carrying out household chores,
travelling, and leisure-time activities but is distinct from exercise.
For measurement purposes, physical activities are classified into different categories, ranging from light to moderate to vigorous intensity, depending on the
Estimating the economic costs of unhealthy diets and low physical activity is complex 15
amount of effort (i.e. kilocalories of energy) required to perform the activity.
This effort is measured in terms of the Metabolic Equivalents (METs), which
is the ratio of a person’s metabolic rate while doing the activity relative to their
resting metabolic rate, and the intensity is then assessed by multiples of METs
spent on a given activity (Box 2) (World Health Organization, 2004). However,
as we have seen in the preceding section, studies that have assessed the costs of
low physical activity vary in their use of intensity thresholds, or intensity is not
taken account of altogether, which makes it difficult to interpret the evidence
as highlighted.
Interpretation of the evidence is further complicated where low physical activity is conceptualised as sedentary behaviour, as for example in the study
by Popkin et al. (2006). Individuals that are not meeting physical activity
guidelines may be wrongly classified as ‘sedentary’. Defined as “any waking
behaviour with low energy expenditure (≤1.5 metabolic equivalents) while
in a sitting or reclining posture” (p. 540) (Sedentary Behaviour Research
Network, 2012), growing evidence suggests that prolonged sedentary time is
independently associated with deleterious health outcomes independent of
the level of physical activity (Biswas et al., 2015). Disentangling these relationships will be important, with for example Ekelund et al. (2016) showing
that moderate levels of physical activity reduced the increased mortality risks
associated with high sitting time.
Box 2 Examples of moderate-intensity and vigorous-intensity physical activity
One MET is equivalent to an energy consumption of 1 kilocalorie per kilogram of bodyweight per
hour (i.e. energy cost of sitting quietly).
Examples of moderate-intensity physical
activity
Equivalent to approximately 3–6 METs;
requires a moderate amount of effort and
noticeably accelerates the heart rate
Examples of vigorous-intensity physical
activity
Equivalent to approximately >6 METs;
requires a large amount of effort and
causes rapid breathing and a substantial
increase in heart rate
• Brisk walking
• Dancing
• Gardening
• Housework and domestic chores
• Traditional hunting and gathering
• Active involvement in games and sports
with children/walking domestic animals
• General building tasks (e.g. roofing, thatching,
painting)
• Carrying/moving moderate loads (<20kg)
• Running
• Walking/climbing briskly up a hill
• Fast cycling
• Aerobics
• Fast swimming
• Competitive sports and games (e.g.
traditional games, football, volleyball,
hockey, basketball)
• Heavy shovelling or digging ditches
• Carrying/moving heavy loads (>20kg)
Source: World Health Organization, 2004.
16 Assessing the economic costs of unhealthy diets and low physical activity
3.2 Costing studies differ in key assumptions, influencing
estimates for the economic burden of unhealthy diets
and low physical activity
As noted earlier, reviewed studies considered different numbers of diseases to
derive estimates of the economic burden that can be associated with unhealthy
diets or low physical activity. The most common conditions considered include
coronary heart disease, stroke, type 2 diabetes and colorectal and breast cancer,
with Garrett et al. (2004) also adding mood and anxiety disorders as directly
related to individual physical activity patterns in adults. The range and combination of disease groups varied among studies, as did the main approaches to
estimating costs, including the use of population-attributable fractions; consideration of lag times between exposure to a given risk factor, the development
of disease and ensuing cost; and the conceptualisation of economic cost itself,
which we briefly discuss here.
3.2.1 Use of population-attributable fractions
Fourteen reviewed studies used population-attributable fractions (PAFs) to assess
the contribution of unhealthy diets or low physical activity to a range of diseases
(or death) as a basis to estimate the economic burden that can be attributed to
either risk factor.
As noted earlier, the PAF generally refers to the proportion of cases for a given
outcome of interest that can be attributed to a given risk factor among the entire
population. Calculation of PAFs commonly uses a relative risk that has been
adjusted for potential confounders of the association between risk factors and
outcomes, such as age or sex, and the prevalence of exposure in the population
under investigation (the partially-adjusted method). However, where confounding or effect modification affects the relative risk, the estimate of the attributable fraction is potentially biased even if the relative risk has been adjusted for
confounding (Benichou, 2001). This is likely to be the case for the unhealthy
diet and low physical activity–disease relationships, with multiple factors, in
addition to age and sex, such as family history or physiological risk factors such
as weight found to confound the association (Laaksonen et al., 2009; Li et al.,
2015; Montonen et al., 2005).
Baliunas (2011) compared the partially-adjusted method for estimating population-attributable fractions with the fully-adjusted approach, which stratifies
the relative risk according to confounder or effect modifier. The comparison was
applied to mortality from lung cancer, ischaemic heart disease, chronic obstructive pulmonary disease and cerebrovascular disease related to smoking. It found
that the partially-adjusted method overestimated the attributable fractions by
Estimating the economic costs of unhealthy diets and low physical activity is complex 17
10%. The majority of studies reviewed in this volume have used the partiallyadjusted method, and cost estimates are therefore likely to be biased, although
the direction of the bias is not clear. The use of a fully-adjusted method would
reduce the risk of over- or under-estimating the ‘true’ association between risk
factors and costs.
In addition, population-attributable fractions used in the reviewed studies varied
widely. This is illustrated further in Table 4 for PAFs used for costing studies of
low physical activity.
This variation reflects, to great extent, the source of PAFs and whether they were
adjusted for the population under investigation. For example, Allender et al.
(2007) and Scarborough et al. (2011), in their analyses of the economic burden
of ill health that can be attributed to low physical activity in the United Kingdom
in the mid-2000s, used PAFs that were produced in the context of the first Global
Burden of Disease study for the western European population (Murray & Lopez,
1997). There is uncertainty about the degree to which these PAFs are applicable
to the country they were used for (in this case the United Kingdom) as PAFs
should take account of the underlying prevalence of a risk factor in the population
under study. Although a sensitivity analysis by Scarborough et al. (2011) revealed
little impact of the choice of PAFs on the cost estimates for low physical activity,
population-specific PAFs remain preferable to increase the accuracy of estimates.
3.2.2 Consideration of time lags between risk factor exposure,
disease development and associated costs
Reviewed studies tended to apply population-attributable fractions using the
prevalence of a given risk factor (unhealthy diets, low physical activity) in a given
Table 4 Population-attributable fractions used in costing studies: low physical
activity
Ischaemic heart
disease (%)
Stroke
(%)
Diabetes
type 2 (%)
Breast
cancer (%)
Colon cancer
(%)
Katzmarzyk et al. (2000) 36 20 20 11 20
Garett et al. (2004) 31 31 18 19 31
Allender et al. (2007) 23 – 15 11 16
Katzmarzyk (2011) 18 23 20 13 17
Scarborough et al. (2011) 23 12 15 11 –
Janssen (2012) 26 (m)*
27 (f)
25 (m)*
26(f)
38 (m)*
29 (f) 15 (f)* 24 (m, f)*
Zhang & Chaaban (2012) 12 16 14 – –
Maresova (2014) 7 3 4 3 4
Note: * m – males; f – females
18 Assessing the economic costs of unhealthy diets and low physical activity
year and estimating the related disease costs incurred in the same year (Allender et
al., 2007; Garrett et al., 2004; Janssen, 2012; Katzmarzyk, 2011; Katzmarzyk et
al., 2000; Maresova, 2014; Popkin et al., 2006; Scarborough et al., 2011; Zhang
& Chaaban, 2012). Such an approach ignores the time lag between exposure to
the given risk factor and the development of disease. For example, Weyer et al.
(1999) found that it takes approximately five years between the initial normal
glucose-tolerant stages and the development of clinically verified type 2 diabetes.
Where the development of diabetes can be linked to unhealthy diet or low physical
activity at the outset, associated economic costs would thus be expected to emerge
only after five years at the earliest. Therefore, accurately estimating the cost that
can be attributed to unhealthy diets or low physical activity would need to take
account of the time it takes for the natural progression of exposure to disease.
3.3 The nature and range of costs considered is likely to
underestimate the ‘true’ economic burden of unhealthy
diets and physical activity
Reviewed studies tended only to consider costs that can be associated with
primary disease outcomes, such as type 2 diabetes or coronary heart disease.
Yet, a full costing would also need to take account of complications associated
with the primary disease outcome. Considering, for example, type 2 diabetes:
data from the UK National Diabetes Audit found that within a one-year followup period between 2011–12 and 2012–13, people with type 2 diabetes were
significantly more likely than those without diabetes to be admitted to hospital
for complications such as angina, at 135.1%, heart failure (121.1%), heart
attack (87.6%) or stroke (59.1%) (Health and Social Care Information Centre,
2015). Disregarding the costs associated with the development of complications
will inevitably underestimate the true economic burden that can be associated
with a given risk factor, but this also assumes that it is possible to quantify the
contribution of the risk factor under consideration to the observed complication.
Furthermore, the majority of reviewed studies considered only the direct health
care costs, further underestimating the true economic burden that can be associated with unhealthy diets or low physical activity (see also Box 3). For example,
three Canadian studies that had analysed the costs associated with low physical
activity found indirect costs caused by lost productivity to be about twice as
high as direct health care costs, and together these accounted for between 0.4%
and 0.6% of gross domestic product (GDP) (Janssen, 2012; Katzmarzyk, 2011;
Krueger et al., 2015). It is difficult to generalise from these studies to other
settings and it will be important to broaden existing costing estimates to also
capture indirect costs in order to better understand the size of the burden that
can be associated with the two risk factors.
Estimating the economic costs of unhealthy diets and low physical activity is complex 19
3.4 Conceptual and methodological challenges of
estimating the economic costs of unhealthy diets and low
physical activity: a summary
This section reviewed the measurement and methodological issues in assessing
the economic burden of unhealthy diet and low levels of activity. We note that
the measurement of what constitutes an ‘unhealthy’ diet is made more difficult
by there being positive effects of some foods, negative effects of others and
interactions between the effects of different foods. Calibrating the extent of
deviation from optimal consumption and the effects of this deviation is difficult.
It is also clear that the context should be taken into account in terms of other
population characteristics.
Box 3 Conceptualizing economic costs
Costing methodology generally distinguishes direct, indirect and intangible costs, although these
have been conceptualized in different ways (Johnston et al., 1999). Direct costs typically refer
to costs of health care services as they relate to the prevention, diagnosis and treatment of a
given condition, such as inpatient or outpatient care, rehabilitation, community health services
and pharmaceuticals; direct costs may also include social care costs where relevant (Suhrcke et
al., 2008). Costs considered typically include those associated with service utilization, that is the
use of a particular service over time (for example, physician visits, emergency room or accident
and emergency department visits, hospital (re-)admissions, length of hospital stay, number of
hospital days), and the actual cost of providing a particular service (health, nursing, social care),
including the costs of procedures, therapies and medications, or expenditure, that is, the amount
of money paid for the services, and from fees (the amount charged), regardless of cost. There are
also direct costs borne directly by people using the services; these include transportation costs,
out-of-pocket payments for medications and devices, special diets and home help.
Indirect costs typically refer to productivity losses to society because of ill health or its treatment
(Koopmanschap et al., 1995). Commonly considered dimensions include presenteeism,
absenteeism, early retirement and premature mortality. Presenteeism costs refer to the value
of productivity losses accrued by employees who are present at work but are unable to work at
full capacity because of illness (Johns, 2010), measured as the value of reduced work output,
errors on the job and failure to meet the company’s production standards (Schultz & Edington,
2007). Absenteeism costs refer to those costs incurred because of absence from work because
of ill health. Costs related to early retirement refer to potential earnings forgone by not working
to the formal age of retirement, while premature mortality refers to the loss of economic output
calculated as the income that individuals who die before or at a given age will lose over the
period of remaining labour market participation (under or to age 65).
Finally, intangible costs generally describe the psychological burden placed on patients and their
carers, including pain, bereavement, anxiety and suffering (Suhrcke et al., 2008).
20 Assessing the economic costs of unhealthy diets and low physical activity
While there is more consensus about the measurement of physical activity, similar
issues arise in terms of the independent effects of moderate and vigorous activity
and sedentary behaviour, but also the interactions between these. Studies take
broader and narrower perspectives in terms of what costs are included, with some
limited to formal health care costs, and others aiming to take a more societal
view. While current evidence makes it difficult to make accurate comparisons,
it is likely that much of the economic burden comes from non-health care costs,
especially from effects on productivity, absenteeism, presenteeism and other
indirect costs.
Chapter 4
Taking available approaches
to determining the economic
costs of unhealthy diets and low
physical activity further: a proofof-concept approach applied
to five European countries
This chapter develops a bottom-up framework for assessing the costs of unhealthy
diets and low physical activity using a disease-based approach. Building on the
insights from the evidence review and critical appraisal of measurement and
methodological issues in the preceding sections, this section builds a framework
for assessing costs of unhealthy diets and low physical activity. It then applies this
framework as a proof-of-concept to demonstrate the feasibility of undertaking a
comprehensive, bottom-up cost assessment that addresses some of the identified
limitations of existing costing studies.
The evidence review demonstrated that there are essentially two approaches
to estimating the economic burden associated with unhealthy diets and low
physical activity. One involves the use of a disease-based approach, based on
the observation that the two risk factors have been identified to be causally
related to major chronic diseases such as coronary heart disease, stroke, type
2 diabetes and selected cancers (Lee et al., 2012; Micha et al., 2015), which
have been associated with considerable costs to health care systems and wider
society through productivity losses (European Commission, 2014). About half
of the reviewed studies adopted the disease-based approach, typically using the
population-attributable fraction to quantify the contribution of the individual
risk factor (unhealthy diet, low physical activity) to the burden of a given disease
or death and the associated economic costs. An alternative method involves a
generic, non-disease-based approach that uses individual data on unhealthy diets
or low physical activity data and links these to (health care) cost data, regardless
of the type of disease or diagnosis. This second approach requires availability
of and access to individual-level data on dietary and physical activity patterns
along with data on health care use and productivity. Such data are difficult to
access even within individual country settings, let alone in multiple countries.
22 Assessing the economic costs of unhealthy diets and low physical activity
We use a disease-based approach to estimate the costs that can be associated with unhealthy diets and low physical activity. Diseases associated with
these lifestyle factors include coronary heart disease, stroke, type 2 diabetes
and selected cancers (Box 4) (Lee et al., 2012; Micha et al., 2015). As a test
of concept we apply the framework using type 2 diabetes. This was chosen
because of the strong association of type 2 diabetes with either risk factor, as
shown in the literature. Diabetes is already an important public health issue
but perhaps more importantly, globally the number of people with diabetes
has doubled during the past 20 years, making it a growing challenge for health
systems (Zimmet et al., 2014). Estimates for the early 2000s place the costs
for type 2 diabetes in the EU at some 7.4% of total health care expenditure, compared to 12% for cardiovascular diseases (Muka et al., 2015). More
recently, expenditure on diabetes was estimated to account for some 9% of
total health care expenditure in the European region in 2015 (International
Diabetes Federation, 2015).
Taking a disease-based approach, we used population-attributable fractions
to provide population-level quantitative estimates for the economic costs that
Box 4 Health risks associated with unhealthy diets and low physical activity
A wide range of studies have presented convincing or probable evidence for an association
between the intake of selected foods and food groups and major chronic conditions (Micha et
al., 2015). For example, coronary heart disease has been linked to intake of fruit and vegetables,
nuts and legumes, whole grains, fish, and red and processed meat (Afshin et al., 2014; Aune
et al., 2016; Boeing et al., 2012; Micha et al., 2010; Zheng et al., 2012); stroke to fruit and
vegetables, nuts and legumes, fish, and red and processed meat (Afshin et al., 2014; Boeing
et al., 2012; Chowdhury et al., 2012, Micha et al., 2010; Zheng et al., 2012), type 2 diabetes
to nuts and legumes, whole grains, red and processed meats (Afshin et al., 2014; Aune et al.,
2016; Micha et al., 2010; Zheng et al., 2012) and selected cancers to fruit and vegetables, nuts
and legumes, whole grains, fish, and red and processed meats (Aune et al., 2016; Boeing et
al., 2012; Bouvard et al., 2015; World Cancer Research Fund and American Institute for Cancer
Research, 2007). Based on the available evidence, Micha et al. (2015) concluded that “even
modest dietary changes are associated with meaningful reductions in [cardiovascular disease]
morbidity and mortality, type 2 diabetes [and] specific cancer sites” (p. 2), along with major risk
factors, such as hypercholesterolaemia, hypertension and obesity.
These conditions have also been linked to physical activity, with Warburton et al. (2010)
demonstrating that low physical activity was associated with an increased risk for cardiovascular
disease, stroke, hypertension, colon and breast cancer and type 2 diabetes. The authors further
demonstrated that higher levels of physical activity reduced the risk for premature all-cause
mortality.
A proof-of-concept approach 23
can be associated with unhealthy diets and low physical activity. In the light of
the reviewed evidence, we adapted the PAF methodology by (i) using a fulladjustment formula that takes account of confounding between unhealthy diets
and low physical activity and associated diseases; (ii) incorporating a time perspective that takes account of the natural progression from risk factor exposure
to the development of disease; (iii) applying the PAFs to costs based on data
on incidence rather than prevalence; (iv) estimating the direct costs that can be
associated with the primary outcome and those associated with complications
using the annual incidence rates for complications; and (v) considering indirect
costs of productivity losses as a consequence of absenteeism, presenteeism, work
disability, early retirement and premature mortality.
The following sections provide a sample computation of the economic costs of
unhealthy diets and low physical activity in five European countries: France,
Germany, Italy, Spain and the United Kingdom, using type 2 diabetes and its
complications as the primary outcome.
4.1 Diabetes as an outcome of unhealthy diets and low
physical activity
As noted above, we chose type 2 diabetes as the primary outcome because of
the convincingly strong evidence of its association with unhealthy diets and low
physical activity (Afshin et al., 2014; Micha et al., 2010; Warburton et al., 2010;
Zheng et al., 2012). Diabetes mellitus, commonly referred to as diabetes, is a
group of metabolic diseases characterized by high levels of sugar in the blood
for a prolonged period of time.
The most common form of diabetes is type 2, which typically occurs in adults,
although it is increasingly seen in young people, including children (Zimmet
et al., 2014). It results primarily from insulin resistance, and at later stages the
pancreas may also fail to produce sufficient levels of insulin. Insulin resistance
in type 2 diabetes is associated mainly with high bodyweight and low physical
activity. The role of genetic factors in the development of type 2 diabetes tends
to be small compared to lifestyle and clinical factors such as increased bodyweight, elevated liver enzymes, current smoking status and reduced measures
of insulin secretion and action (Lyssenko et al., 2008). Factors that increase the
likelihood of developing diabetes include high consumption of sugar-sweetened
drinks (Malik et al., 2010a; Malik et al., 2010b), of saturated and trans-fatty
acids (Risérus et al., 2009) and of refined grains such as white rice (de Bakker
et al., 2012), which, when metabolised to glucose, increase blood sugar levels.
Low physical activity has also been linked to type 2 diabetes through causing
insulin resistance (Warburton et al., 2010) (Figure 1).
Figure 1 Relationship between unhealthy diets and low physical activity with
type 2 diabetes and associated costs
Source: authors.
Decreased cellular
sensitivity to insulin,
decreasing uptake of
glucose from the blood,
insufficiently compensated
for by the production of
insulin (increased insulin
production at the early
stages)
Genetic factors Lifestyle
High in saturated fats,
trans-fatty acids and
total fat intake
Fat
metabolised
to high levels
of glucose
Alters fatty acid
composition of
cell membranes
Affects gene
expression by
binding to
cell receptors
Low physical activity Unhealthy diets
Impaired glucose tolerance Decreased cellular sensitivity to insulin,
decreasing uptake of glucose from the blood
Decreased
capillary
proliferation
in muscles,
insulinsensitive type
of muscle
fibres and
muscle mass
Ischaemic
heart disease
(IHD)
Retinopathy
Renal failure
Amputation
Type 2 diabetes mellitus
Fatal and non-fatal
myocardial
infarction (MI)
Diabetes-related mortality
Congestive
heart failure
Stroke
Direct health care costs
(Inpatient care, outpatient care, medications,
rehabilitation, long-term care, other direct
expenses)
Indirect (productivity) costs
(Absenteeism, presenteeism, reduced
productivity among the unemployed
population, work disability, early retirement
and premature mortality)
A proof-of-concept approach 25
4.2 The principal approach used in this study to
estimate the economic costs that can be associated with
unhealthy diets and low physical activity
The identification of the economic costs that can be associated with unhealthy
diets and low physical activity as conceptualised in this study involves five
principal steps:
1. determining the prevalence of unhealthy diets and low physical activity among the general population and estimating the prevalence in
populations eventually developing the disease using adjustment factors;
2. determining the population-attributable fraction, or the proportion
of cases attributable to unhealthy diets and low physical activity;
3. estimating the proportion of incident diabetes cases in future year X
that can be attributed to present unhealthy diets and low physical
activity patterns;
4. estimating the average annual per patient health care costs that can
be associated with diabetes and with diabetic complications to yield
diabetes-related direct costs; and
5. estimating the indirect costs that can be associated with diabetes
attributable to unhealthy diets and low physical activity.
We discuss these steps in turn.
4.2.1 Determining the prevalence of unhealthy diets and low physical
activity
Our proposed approach requires knowledge of the prevalence of the risk factor
(here: unhealthy diets, low physical activity) among populations that eventually develop the outcome (here: type 2 diabetes), rather than among the general
population, in order to enhance the accuracy of the estimation of costs that can
be associated with the given risk factor. For low physical activity, we drew on
work by Lee et al. (2012) who calculated adjustment factors for different outcomes (coronary heart disease, type 2 diabetes, breast and colon cancer, and those
who died) to identify the added extent to which low physical activity occurred
in people who eventually developed the outcome in question compared to the
general population. Lee et al. (2012) illustrate this with an example from the
Shanghai Women’s Health Study, where the prevalence of low physical activity
in all women at baseline was 45.4% compared to 51.6% among women who
eventually died, yielding an adjustment factor of 1.14 (51.6/45.4 = 1.14). They
then calculated such an adjustment factor for a large number of original studies,
and, for type 2 diabetes, derived a factor of 1.23 after averaging estimates across
studies (for comparison, adjustment factors for coronary heart disease or colon
26 Assessing the economic costs of unhealthy diets and low physical activity
cancer were 1.20 and 1.22, respectively, and for breast cancer, it was 1.05). We
used the adjustment factor of 1.23 to estimate the prevalence of low physical
activity among people who will eventually develop diabetes in France, Germany,
Italy, Spain and the United Kingdom.
Data on prevalence for low physical activity in the five countries were derived from
the WHO Global Health Observatory database (World Health Organization,
2016). According to this source, the proportion of the general population who
were not active, or who did not meet the recommendations of at least 150
minutes per week of moderate-intensity physical activity or at least 75 minutes
per week of vigorous-intensity physical activity in 2010 were: 23.8% in France,
21.1% in Germany, 33.2% in Italy, 30.5% in Spain and 37.3% in the United
Kingdom. Applying the adjustment factor of 1.23 to these rates yielded estimated prevalence rates of 29.3% in France, 26% in Germany, 40.8% in Italy,
37.5% in Spain and 45.9% in the United Kingdom among populations who
will eventually develop diabetes.
In contrast to low physical activity, determining the prevalence rate for unhealthy
diets is more complex given the limited availability of appropriate data. We drew
on the alternate healthy eating index (AHEI) 2010 developed by Chiuve et al.
(2012). The AHEI assesses components of the diet by assigning scores ranging
from 0 (worst) to 10 (best) (Table 5); scores between 67 and 110 are rated as
adherence to what has been described as a healthy diet that emphasises high
intakes of whole grains, polyunsaturated fatty acids, nuts and fish, and reductions in red and processed meats, refined grains and sugar-sweetened beverages.
As noted above, studies found that diets that scored highly on the AHEI were
shown to be associated with a significant reduction in the risk of all-cause
mortality, cardiovascular disease, cancer and type 2 diabetes by around 20%
(Schwingshackl & Hoffmann, 2015).
We used the European Food Safety Authority (EFSA) Database to estimate
the prevalence of unhealthy diets in the five countries under study. The EFSA
contains information on the mean consumption of over 1500 food items in
European countries, based on data from national dietary surveys (European Food
and Safety Authority, 2015). Based on these data, we derived the mean intake of
AHEI food groups for each country (Appendix 3) and allocated AHEI scores as
shown in Table 6. Assuming that the figures shown in Appendix 3 and Table 6
are representative of the five countries as a whole, we estimate the proportions
of those following an unhealthy diet to be 44% of the general population in
France, 25% in Germany, 33.9% in Italy, 34.6% in Spain and 26.5% in the
United Kingdom, as assessed by scores of 43% (France), 34% (Germany), 58%
(Italy), 48% (Spain) and 37% (the United Kingdom) according to the 2010
alternate healthy eating index.
A proof-of-concept approach 27
Unlike for low physical activity, we were unable to identify studies that provide
adjustment factors that would allow assessment of the degree to which unhealthy
diet is present in cases of the outcome compared to the overall population. Jacobs
et al. (2015) estimated approximate adjustment factors based on a cohort of
white Americans (which formed part of a larger multi-ethnic cohort in Hawaii)
with poor dietary patterns assessed using the 2010 alternate healthy eating
index (31 864 individuals), of whom 7.1% eventually developed diabetes (2274
subjects). The prevalence of unhealthy diets in the study population was 61.3%
Table 5 The quality of diets according to the alternate healthy eating index
(AHEI) 2010
Component
Criteria for
minimum score
(0)
Criteria for
maximum score
(10)
Serving size
for food
components
Vegetables1
(servings per day) 0 ≥5 0.5 cup or 1 cup of green leafy
vegetables
Fruit2
(servings per day) 0 ≥4 1 medium piece or 0.5 cup of
berries
Whole grains (grams per day)
Women 0 75
Men 0 90
Sugar-sweetened beverages
and fruit juice (servings per
day)
≥1 0 8oz.
Nuts and legumes (servings
per day)
0 ≥1 1oz. of nuts or 1 tablespoon (15
ml) of peanut butter
Red and processed meat
(servings per day)
≥1.5 0 4oz. of (red) unprocessed meat or
1.5oz. of processed meat
Trans-fats (% of energy) ≥4 ≥0.5
Omega-3 fatty acids (EPA +
DHA) (mg per day)
0 250 Equivalent to two 4oz servings
per week
Polyunsaturated fatty acids (%
of energy)
≤2 ≥10
Sodium (mg per day)
Women ≥3 337 <1 112
Men ≥5 271 ≤1 612
Alcohol (drinks per day)
Women ≥2.5 0.5–1.5 One drink equivalent to 4oz. wine,
Men 12oz. beer or 1.5oz. liquor ≥3.5 0.5–2.0
TOTAL 0 110
Source: adapted from Chiuve et al., 2012.
Notes: 1 Excludes potatoes (including french fries) because they are not associated with lower risk of chronic
disease in epidemiologic studies; 2
whole fruits only.
28 Assessing the economic costs of unhealthy diets and low physical activity Dietary component AHEI recommended intake per day for maximum score (10 points) MEAN INTAKE PER DAY AHEI SCORE France (2007) Germany (2007) Italy (2005– 6) Spain (2009) United Kingdom (2008) France (2007) Germany (2007) Italy (2005–6) Spain (2009) United Kingdom (2008) Vegetables ≥5 cups green leafy vegetables 0.09 cups 0.12 cups 0.20 cups 0.18 cups 0.03 cups 0 0 0 0 0 Fruit ≥453.44g berries 10.97g 13.66g 2.89g 6.35g 10.81g 0 0 0 0 0 Whole grains ≥82.5g * 12.2g 0.49g 35.29g 6.84g 2.80g 1 0 4 0 0 Sugar-sweetened beverages and fruit juice ≤0.79oz. 4.19oz. 12.05oz. 1.98oz. 4.61oz. 7.91oz. 5 0 8 5 1 Nuts ≥1oz. 0.04oz. 0.10oz. 0.04oz. 0.07oz. 0.04oz. 0 1 0 0 0 Processed meat ≤0.224oz. 1.33oz. 1.76oz. 1.05oz. 1.73oz. 1.06oz. 5 3 6 3 6 Trans-fat ≤0.5% of energy intake (equivalent to ≤1.39g for a 2 500kcal diet per day) 63.5g 68.57g 30.23g 48.31g 33.48g 0 0 0 0 0 Long-chain (n-3) fats (EPA + DHA) ≥0.25g 21.39g 13.76g 31.05g 57.31g 21.1g 10 10 10 10 10 PUFA ≥10% of energy intake (equivalent to ≥27.78g for a 2 500kcal diet per day) 10.92g 2.93g 36.63g 32.6g 1.46g 3 0 10 10 0
Sodium ≤1.36g * 1.5g 0.01g 0.01g 0.17g 0.08g 9 10 10 10 10
Alcohol
≤7oz. wine, ≤21oz. beer,
≤2.6oz. liquor * 2.75oz. 6.33oz. 2.50oz. 3.12oz. 6.43oz. 10 10 10 10 10
TOTAL 43 34 58 48 37
Table 6 Assignment of AHEI scores on mean dietary intakes of AHEI food
groups in France, Germany, Italy, Spain and the United Kingdom
Note: * Median of recommended intakes for men and women
A proof-of-concept approach 29
and among those eventually developing diabetes, the proportion of those with
unhealthy diets was 67.1%; this equates to an adjustment factor of 1.09, which
we used in our study.
4.2.2 Determining the population-attributable fraction, or the proportion
of cases attributable to unhealthy diets and low physical activity
Determining the population-attributable fraction first requires assessment of the
relative risks for developing diabetes that can be associated with unhealthy diets
and low physical activity. We drew on work by Li et al. (2015) who prospectively
assessed the joint association of birth weight and five behavioural risk factors
(smoking status, daily alcohol consumption, body mass index, dietary patterns
and physical activity) in adulthood with incident type 2 diabetes based on a
cohort of almost 150 000 male and female health professionals (median age 45
years), who were followed up for a period of 20–30 years (median follow-up:
24 years) (see also Appendix 4 for a justification of using the study by Li et al.).
During the follow-up period, 11 709 new cases of type 2 diabetes were reported,
equating to an incidence rate of 7.8%. Table 7 shows the adjusted relative risks
for different levels of exposure to unhealthy diets or low physical activity as
estimated by Li et al. (2015).
Table 7 Relative risks for diabetes related to unhealthy diets and low physical
activity estimated by Li et al. (2015)
Risk factor Risk factor category
AHEI score (≥67% classified as healthy diet)
Level of exposure 0–22% 23–44% 45–66% ≥67%
Relative risk 1.15 1.06 1.02 1.00 (ref)
Low physical activity (moderate-to-vigorous intensity physical activity in hours/week)
Level of exposure 0 0.01–1.0 1.0–3.5 ≥3.5
Relative risk 1.28 1.19 1.03 1.00 (ref)
Source: Li et al., 2015.
Table 8 provides a summary overview of the various estimates for the prevalence of unhealthy diets and low physical activity in the populations eventually
developing diabetes in France, Germany, Italy, Spain and the United Kingdom
described above and the associated relative risks for incident diabetes that can
be associated with these risk factors as derived from Li et al. (2015). As the data
on low physical activity among the general population derived from the WHO
Global Health Observatory Database described above do not quantify the average
weekly hours of physical activity performed by individuals, we used data from
the 2013 Eurobarometer survey (TNS Opinion & Social, 2014). It assessed the
30 Assessing the economic costs of unhealthy diets and low physical activity
frequency and duration of sport and physical activity among populations in the
EU and quantified the proportions of the survey population who performed
moderate- and vigorous-intensity physical activity. This showed that the majority
of inactive individuals in the five countries studied performed physical activity
for 0.01 to 1 hour per week, which gives a relative risk for the development of
diabetes that can be associated with low physical activity of 1.19 (Table 7).
Table 8 Estimated prevalence rates of unhealthy diets and low physical activity
in the populations eventually developing diabetes and relative risks for incident
diabetes in France, Germany, Italy, Spain and the United Kingdom
Prevalence in the population
eventually developing diabetes
Relative risks for incident diabetes associated
with unhealthy diets and low physical activity
Unhealthy diets
(%)
Low physical
activity (%) Unhealthy diets Low physical activity
France 48.0 29.3 1.06 1.19
Germany 27.3 26.0 1.06 1.19
Italy 37.0 40.8 1.02 1.19
Spain 37.7 37.5 1.02 1.19
United Kingdom 28.9 45.9 1.06 1.19
Estimates shown in Table 8 allow for the computation of the fully-adjusted
population-attributable fraction for each setting as illustrated in Box 5.
4.2.3 Estimating the proportion of incident diabetes cases in future
year X that can be attributed to present unhealthy diet and low
physical activity patterns
We have noted that there is typically a time lag between exposure to a risk factor
and the development of subsequent disease that can be (part-)attributed to the
risk factor. In the case of type 2 diabetes, Weyer et al. (1999) analysed the major
metabolic abnormalities occurring in the development of diabetes, finding that
insulin action and secretion significantly decrease early in the development
of diabetes, during the transition from normal glucose tolerance to impaired
glucose tolerance, and that these early changes are preceded by increases in
bodyweight. As impaired glucose tolerance progresses to established diabetes,
insulin action and secretion further deteriorate, accompanied by an increase in
the endogenous glucose output, and these changes are associated with further
increases in bodyweight. Weyer et al. observed an average interval of 1.8 (± 0.8)
years between normal glucose tolerance and impaired glucose tolerance, while
the interval between impaired glucose tolerance and established diabetes was
3.3 (± 1.4) years. On average, then, it would require some 5.1 (± 1.4) years for
normal glucose tolerance to develop into type 2 diabetes.
A proof-of-concept approach 31
Weyer et al. (1999) further noted that the early deteriorations in insulin secretion and action are distinct from and additive to abnormalities that overweight
and obese individuals may experience while normal glucose-tolerant. With the
progression from impaired glucose tolerance to diabetes, the defects worsen in
parallel with an increase in endogenous glucose output, implying that being
overweight or obese prior to the observed stages does not alter the interval
between normal glucose tolerance to impaired glucose tolerance and diabetes
and that an interval of about five years is therefore consistent.
Based on these observations, in this study we assume a time lag of five years
between exposure to the risk factors (unhealthy diets, low physical activity) and
the eventual development of type 2 diabetes, starting from a 2015 population
Box 5 Computing the fully-adjusted population-attributable fraction (PAF)
The computation of the fully-adjusted population-attributable fraction (PAF) follows the following
principal formula:
in which:
Pd: prevalence of the risk factor in the population eventually developing the disease
RR: The relative risk for a certain disease associated with the risk factor compared to absence
of risk factor, adjusted for confounding variables
Applying this formula to the prevalence and relative risk estimates for each factor in each country
yields the country- and risk-factor-specific PAFs as follows:
Unhealthy diet Low physical activity
France = 3% 0.48 × (1.06−1)
1.06
PAF (%) = = 5% 0.293 × (1.19−1)
1.19
PAF (%) =
Germany = 2% 0.273 × (1.06−1)
1.06
PAF (%) = = 4% 0.26 × (1.19−1)
1.19
PAF (%) =
Italy = 1% 0.37 × (1.02−1)
1.02
PAF (%) = = 7% 0.408 × (1.19−1)
1.19
PAF (%) =
Spain = 1% 0.377 × (1.02−1)
1.02
PAF (%) = = 6% 0.375 × (1.19−1)
1.19
PAF (%) =
United
Kingdom
= 2% 0.289 × (1.06−1)
1.06
PAF (%) = = 7% 0.459 × (1.19−1)
1.19
PAF (%) =
Pd (RRadj−1)
RRadj
PAF (%) =
32 Assessing the economic costs of unhealthy diets and low physical activity
that is normal glucose tolerant. We drew on estimates of cumulative incidence
of diabetes in the 53 countries of the WHO European region from 2010 to
2020 by Webber et al. (2014) to estimate the annual incidence of diabetes in
2020 in the five countries under study. We then applied the country-specific
PAFs for unhealthy diets and low physical activity (Box 5) to the estimated
2020 incidence of diabetes, which gives the total number of incident cases in
2020 that can be attributed to 2015 levels of unhealthy diet and low physical
activity (Table 9).
Table 9 Estimated incident diabetic cases in 2020 attributable to unhealthy diet
and low physical activity patterns in 2015
Country Risk factor
Populationattributable
fraction (%)
Projected
incidence in 2020
Number of incident diabetes
cases attributable to risk factor
France
Unhealthy diets 3
151 590
4 548
Low physical activity 5 7 580
TOTAL 12 128
Germany
Unhealthy diets 2
242 571
4 851
Low physical activity 4 9 703
TOTAL 14 554
Italy
Unhealthy diets 1
151 306
1 513
Low physical activity 7 10 591
TOTAL 12 104
Spain
Unhealthy diets 1
106 929
1 069
Low physical activity 6 6 416
TOTAL 7 485
United
Kingdom
Unhealthy diets 2
118 393
2 368
Low physical activity 7 8 288
TOTAL 10 656
4.2.4 Estimating the average annual per patient health care costs
that can be associated with diabetes and with diabetic complications
to yield diabetes-related direct costs
To estimate the direct health care costs associated with the incident diabetes
cases attributable to unhealthy diets and low physical activity, we used data
from the 2015 Diabetes Atlas, which provides estimates for the average annual
per patient diabetes cost in the studied countries in 2015 (International
Diabetes Federation, 2015). We adjusted these figures by the average annual
growth rate in per capita health spending during 2005 to 2013 in each of
the five countries (OECD, 2015). This yielded estimated per patient costs
A proof-of-concept approach 33
in 2020 to range from €3314 in Spain to €6810 in France. Based on these
figures, the direct diabetes-related health care costs that can be associated
with unhealthy diets and low physical activity in the five countries in 2020
are estimated to range from €24 805 290 in Spain to €96 638 560 in
Germany (Table 10).
Table 10 Estimated diabetes-related health care costs in 2020 attributable to
unhealthy diets and low physical activity patterns in 2015 in France, Germany,
Italy, Spain and the United Kingdom
Country Risk factor
Number of incident
cases in 2020 attributable
to the risk factor in 2015
Estimated per
patient diabetes
cost in 2020 (€) *
Estimated total
health care
cost in 2020 (€)
France
Unhealthy diets 4 548 6 810 30 971 880
Low physical activity 7 580 6 810 51 619 800
TOTAL 12 128 82 591 680
Germany
Unhealthy diets 4 851 6 640 32 210 640
Low physical activity 9 703 6 640 64 427 920
TOTAL 14 554 96 638 560
Italy
Unhealthy diets 1 513 3 935 5 953 655
Low physical activity 10 591 3 935 41 675 585
TOTAL 12 104 47 629 240
Spain
Unhealthy diets 1 069 3 314 3 542 666
Low physical activity 6 416 3 314 21 262 624
TOTAL 7 485 24 805 290
United
Kingdom
Unhealthy diets 2 368 5 292 12 531 456
Low physical activity 8 288 5 292 43 860 096
TOTAL 10 656 56 391 552
Source: International Diabetes Federation, 2015, adjusted for average annual growth rate in per capita health
spending during 2005 to 2013 at 1.2% in France, 2.4% in Germany, 0.55% in Italy, 0.95% in Spain and
1.75% in the United Kingdom (OECD, 2015).
We further estimated the direct health care costs that can be attributed to diabetic complications arising from these attributable cases. We drew on Hayes et
al. (2013) who reported annual diabetes complications incidence rates based
on the UK Prevention Diabetes Study Outcomes Model 2 (UKPDS-OM2) to
estimate the number of incident cases of complications in 2020. Specifically,
Hayes et al. derived models to predict the annual risk and incidence of a range of
outcomes of diabetes, including myocardial infarction, stroke, congestive heart
failure, ischaemic heart disease, amputation, blindness, renal failure and ulcer
(Appendix 5). We first applied these incidence rates to the cases attributable to
unhealthy diets and low physical activity in each country to estimate the number
34 Assessing the economic costs of unhealthy diets and low physical activity
of incident cases of complications arising in 2020. We then applied the average
annual per patient cost for each complication as derived from the published
evidence to estimate the direct health care costs that can be associated with these
complications (Table 11). Appendix 6 provides a detailed breakdown by country
of the estimated number of incident cases of diabetes-related complications and
related costs in 2020.
Table 11 Estimated diabetic complication-related costs in 2020 attributable
to unhealthy diets and low physical activity patterns in 2015 in France, Germany,
Italy, Spain and the United Kingdom
Estimated number of incident cases of
diabetes-related complications in 2020
attributable to the risk factor
Estimated complication-related
total health care cost in 2020 (€)
Unhealthy
diets
Low physical
activity
Unhealthy
diets
Low physical
activity Total cost
France 181 302 2 646 685 4 411 142 7 057 827
Germany 193 386 3 910 441 7 821 688 11 732 129
Italy 60 422 492 429 3 447 004 3 939 433
Spain 43 255 516 414 3 099 453 3 615 867
United Kingdom 94 330 1 874 299 6 560 047 8 434 347
Accordingly, we estimate that the total number of incident cases of complications arising from diabetes cases that can be attributed to unhealthy diets and
low physical activity in 2020 would be 483 in France, 579 in Germany, 482 in
Italy, 298 in Spain and 424 in the United Kingdom. The associated direct health
care costs ranged from some €3.6 million in Spain to €11.7 million in Germany.
Taken together, the total direct health care costs linked to incident diabetes and
its complications in 2020 that can be attributed to unhealthy diets and low physical activity in 2015 are estimated to be €89.6 million in France, €108.4 million
in Germany, €51.6 million in Italy, €28.4 million in Spain and €64.8 million
in the United Kingdom (see also below for a detailed breakdown of numbers).
4.2.5 Estimating the indirect costs that can be associated with
diabetes attributable to unhealthy diets and low physical activity
We considered two principal categories of indirect costs that can be associated
with diabetes attributable to unhealthy diets and low physical activity: first,
productivity loss in the workplace due to absenteeism and presenteeism, including among the unemployed population; second, productivity forgone because
of work disability, early retirement or premature death. We first identified the
proportion of diabetes cases in 2020 attributable to unhealthy diets and low
A proof-of-concept approach 35
physical activity who are of working age, using the average sex-specific proportions of incident type 2 diabetes cases in the United Kingdom for the period
1991–2010 as provided by Holden et al. (2013) (Appendix 7). We then computed
the percentage of annual incident type 2 diabetes cases by five-year age group
and sex in the United Kingdom for the period 1991–2010 (Appendix 8) and
applied these to the total number of diabetes cases attributable to unhealthy diets
and low physical activity in each country, again by sex and five-year age group
(Appendix 9). This allowed us to compute the total number of incident type 2
diabetes cases at working age (15–64 years) that can be attributed to unhealthy
diets and low physical activity in each country by 2020 (Table 12).
Table 12 Estimated number of incident type 2 diabetes cases at working age
(15–64 years) that can be attributed to unhealthy diets and low physical activity in
France, Germany, Italy, Spain and the United Kingdom, 2020
Unhealthy diets Low physical activity Total
France 1 517 2 528 4 045
Germany 1 618 3 236 4 854
Italy 505 3 532 4 037
Spain 357 2 140 2 497
United Kingdom 790 2 764 3 554
We then computed the proportion of those of working age who are expected
to be participating in the labour force. In the United Kingdom, labour force
data for 2013 and 2014 suggest that the proportion of working age people with
diabetes was 71.3% compared with 72.3% (2013) and 73.5% (2014) in the
general population (Department for Work and Pensions, 2015). Based on these
observations and in the absence of labour force participation data specific to
people living with diabetes in the other countries studied, we assumed labour
participation rates in the diabetic population to be similar to the general population, using annual national labour force participation rates for each of the five
countries for the period 2000 to 2015 by five-year age group (OECD, 2016).
We acknowledge that this is a conservative estimate that likely overestimates
the ‘true’ proportion of people with diabetes who are in gainful employment
across the five countries. Using this assumption, we computed the number of
incident cases in 2020 attributable to unhealthy diets and low physical activity
at working age who are expected to be in or out of the labour force by five-year
age group (Appendix 10).
Based on these data, we first estimated the indirect costs that can be linked to
incident cases of diabetes that can be attributed to unhealthy diets and low
physical activity. We drew on a systematic review by Breton et al. (2015), which
36 Assessing the economic costs of unhealthy diets and low physical activity
reported that people with diabetes lose an average of 11.9 days of productivity
per year due to the disease. This included absenteeism (work time lost) and
presenteeism (work time impaired). Using this figure, we estimated the total
number of productivity days lost due to absenteeism and presenteeism in the five
countries. We then applied this estimate to data on average daily salary in 2020,
using hourly labour cost data collected by Eurostat (Eurostat, 2015), adjusted
to 2020 figures (based on the average rate of hourly salary increase every five
years in the five countries from 2000 to 2015). Assuming an eight-hour working day, we estimate the total cost of productivity lost due to absenteeism and
presenteeism that can be attributed to diabetes because of unhealthy diets and
low physical activity in 2020 to range from €2.7 million in Spain to €9.7 million in Germany (Table 13).
Table 13 Estimated total number of productivity days lost and cost due to
absenteeism and presenteeism among incident type 2 diabetes cases at working
age that can be attributed to unhealthy diets and low physical activity and who are
expected to be in the labour force in France, Germany, Italy, Spain and the United
Kingdom, 2020
Country Risk factor
Estimated number of
incident type 2 diabetes
cases at working
age attributable
to the risk factor
(in the labour force)
Estimated total
number of days
of productivity
lost in 2020
Average
daily
wage in
2020 (€)
Estimated
cost of
productivity
lost in 2020
(€)
France
Unhealthy diets 893 10 621 224.64 2 385 959
Low physical activity 1 488 17 702 224.64 3 976 599
TOTAL 6 362 558
Germany
Unhealthy diets 1 105 13 144 246.96 3 246 097
Low physical activity 2 209 26 291 246.96 6 492 864
TOTAL 9 738 961
Italy
Unhealthy diets 265 3 149 152.00 478 643
Low physical activity 1 852 22 043 152.00 3 350 501
TOTAL 3 829 144
Spain
Unhealthy diets 216 2 574 150.16 386 576
Low physical activity 1 298 15 451 150.16 2 320 178
TOTAL 2 706 754
United
Kingdom
Unhealthy diets 545 6 488 220.16 1 428 341
Low physical activity 1 907 22 699 220.16 4 997 459
TOTAL 6 425 800
A proof-of-concept approach 37
We further estimated the costs of productivity lost that can be linked to individuals with diabetes who are of working age but are outside the formal labour force.
Productivity losses among this population relate to lost unpaid contributions to
national productivity such as time spent providing child care, household activities
and other activities such as volunteering in the community (American Diabetes
Association, 2013). To determine the value of such losses, we again applied Breton
et al.’s estimate of 11.9 productivity days lost per year per diabetic person to the
number of cases attributable to unhealthy diets and low physical activity who are
of working age but are outside the labour force. This gives the total number of
productivity days lost per year among the diabetic population outside the formal
labour force, to which we then applied the average minimum daily wage in 2020
as derived from Eurostat (2015) for the period 2000 to 2015, adjusted to 2020
figures. Again assuming an eight-hour working day, we estimate the total cost
of productivity lost due to absenteeism and presenteeism that can be attributed
to diabetes because of unhealthy diets and low physical activity among those
outside the formal labour force in 2020 to range from €0.4 million in Spain to
€1.8 million in Italy (Table 14).
Secondly, we computed the indirect costs that can be attributed to work disability, early retirement and premature deaths. We drew on work by Herquelot
et al. (2011), which prospectively assessed the impact of diabetes on the risks
of work disability, early retirement and premature death in a cohort of 20 625
employees of a national gas and electricity company in France, among whom
2.4% (506 individuals) had developed diabetes. They estimated the number
of working years lost among diabetic persons aged 35–60 years to be 1.1
years per person. Of these, 0.09 years were attributed to work disability, 0.7
years to early retirement and 0.28 years to premature death. We applied these
estimates to the number of diabetes cases attributable to unhealthy diets and
low physical activity for those aged 35–60 who are expected to participate
in the labour force. This yielded an estimate of the total number of working
years lost due to work disability, early retirement and premature death because
of unhealthy diet- and physical inactivity-related diabetes (Appendix 11). We
then combined the total number of working years lost with the average annual
salary in 2020, which we estimated using the average 2015 annual salary data
collected by Eurostat (Eurostat, 2015), adjusted to 2020 values (based on the
average rate of salary increase every five years from 2000 to 2015) (Table 15).
38 Assessing the economic costs of unhealthy diets and low physical activity
Table 14 Estimated total number of productivity days lost and cost due to
absenteeism and presenteeism among incident type 2 diabetes cases at working
age that can be attributed to unhealthy diets and low physical activity and who are
expected to be outside the formal labour force in France, Germany, Italy, Spain
and the United Kingdom, 2020
Country Risk factor
Estimated number
of incident type
2 diabetes cases
at working age
attributable to
the risk factor
(outside the formal
labour force)
Estimated total
number of days
of productivity
lost in 2020
Minimum
daily
wage in
2020 (€)
Estimated cost
of productivity
lost in 2020 (€)
France
Unhealthy diets 624 7 429 75.68 562 252
Low physical activity 1 041 12 382 75.68 937 086
TOTAL 1 499 338
Germany
Unhealthy diets 513 6 109 70.80 432 513
Low physical activity 1 027 12 219 70.80 865 116
TOTAL 1 297 629
Italy
Unhealthy diets 240 2 856 77.20 220 482
Low physical activity 1 680 19 992 77.20 1 543 377
TOTAL 1 763 859
Spain
Unhealthy diets 140 1 668 33.92 56 590
Low physical activity 841 10 013 33.92 339 646
TOTAL 396 236
United
Kingdom
Unhealthy diets 245 2 911 92.88 270 339
Low physical activity 856 10 182 92.88 945 746
TOTAL 1 216 085
A proof-of-concept approach 39
Table 15 Estimated cost of working years lost due to work disability, early
retirement and premature death among incident type 2 diabetes cases at working
age that can be attributed to unhealthy diets and low physical activity and who are
expected to be in the formal labour force in France, Germany, Italy, Spain and the
United Kingdom, 2020
Country
Risk factor to
which cases are
attributable
Average
annual salary
in 2020 (in €)*
Estimated
costs of
disability in
2020 (in €)
Estimated
costs of early
retirement in
2020 (€)
Estimated costs
of premature
death in 2020
(€)
France
Unhealthy diets 41 131.99 3 304 085 25 698 440 10 279 376
Low physical activity 41 131.99 5 506 808 42 830 733 17 132 293
TOTAL 8 810 894 68 529 172 27 411 669
Germany
Unhealthy diets 41 523.63 4 127 868 32 105 639 12 842 256
Low physical activity 41 523.63 8 256 587 64 217 897 25 687 159
TOTAL 12 384 455 96 323 537 38 529 415
Italy
Unhealthy diets 32 021.69 762 619 5 931 485 2 372 594
Low physical activity 32 021.69 5 338 336 41 520 393 16 608 157
TOTAL 6 100 956 47 451 878 18 980 751
Spain
Unhealthy diets 31 377.75 610 939 4 751 747 1 900 699
Low physical activity 31 377.75 3 666 776 28 519 369 11 407 748
TOTAL 4 277 715 33 271 116 13 308 446
United
Kingdom
Unhealthy diets 49 359.15 2 421 903 18 837 024 7 534 810
Low physical activity 49 359.15 8 473 719 65 906 700 26 362 680
TOTAL 10 895 622 84 743 725 33 897 490
Note: * estimated from 2015 average annual salary adjusted using the average percentage of increase in salary
every five years from 2000 to 2015 as derived from OECD (2016).
40 Assessing the economic costs of unhealthy diets and low physical activity
4.3 The estimated total economic costs of unhealthy
diets and low physical activity related to diabetes and its
complications
Table 16 summarises the total economic cost of diabetes in the five countries
in 2020 that can be associated with unhealthy diets and low physical activity
patterns in 2015. Costs associated with low physical activity tended to be higher
than those associated with unhealthy diets in all countries, although differences
varied, ranging from some 67% higher in France to a factor of six in Spain and
seven in Italy. Indirect costs that can be associated with either risk factor were
higher than direct health care costs in all countries, by between 25% in France
and up to 100% in the United Kingdom.
The differences in total cost across countries reflect, to a great extent, differences
in population sizes, health care costs and labour costs. When related to the
population projected to develop diabetes in 2020 as a consequence of unhealthy
diets and low physical activity in 2015, the United Kingdom showed the highest
cost, at €18 953, closely followed by Germany and France, while Italy had the
lowest cost, at just over €10 720 (Table 17).
4.3.1 Sensitivity analysis
We carried out a limited set of sensitivity analyses in order to better understand the
likely range of cost estimates provided here. We repeated the above analyses using
the lower and upper values (i.e. 95% confidence intervals) for those parameters
where they were available. This was the case for the prevalence of low physical
activity in the general population and the adjustment factor for low physical
activity. Applying the lower value of these two parameters decreased the total
estimated costs by between 27% and 63%. The highest impact was on estimates
for Italy, reducing the total costs by 63% (from €146 million to €54.3 million),
followed by Spain at 59% (€94.1 million to €39.0 million), Germany at 58%
(€266.7 million to €112.0 million), France at 29% (€151.7 million to €107.4 million) and the United Kingdom at 27% (€202.0 million to €147.8 million).
Using the higher values of the same parameters increased the total costs by
between 26% and 160%. The greatest effect was seen for Germany, where the
total costs more than doubled (from €266.7 million to €694.1 million). A
doubling of costs was also observed for Spain (€94.1 million to €212.3 million)
and Italy (€146 million to €307.9 million). In France estimated costs rose by
29% (€151.7 million to €196.2 million) and in the United Kingdom by 26%
(€202 million to €254.3 million).
A proof-of-concept approach 41
Table 16 Estimated economic cost that can be associated with unhealthy diets
and low physical activity patterns in 2015 as manifested in incident diabetes and
complication in France, Germany, Italy, Spain and the United Kingdom in 2020
Country Risk factor Direct health care cost (€) Indirect cost (€) Total cost (€)
France
Unhealthy diets 33 618 565 42 230 112 75 848 677
Low physical activity 56 030 942 70 385 985 126 416 927
TOTAL 202 265 604
Germany
Unhealthy diets 36 121 081 52 774 976 88 896 057
Low physical activity 72 249 608 105 530 564 177 780 172
TOTAL 266 676 229
Italy
Unhealthy diets 6 446 084 9 765 824 16 211 908
Low physical activity 45 122 589 68 415 902 113 538 491
TOTAL 129 750 398
Spain
Unhealthy diets 4 059 080 7 706 874 11 765 955
Low physical activity 24 362 077 46 253 942 70 616 019
TOTAL 82 381 974
United
Kingdom
Unhealthy diets 14 405 755 30 455 391 44 861 146
Low physical activity 50 420 143 106 686 304 157 106 447
TOTAL 201 967 593
Table 17 Estimated total and per capita economic cost that can be associated
with unhealthy diets and low physical activity patterns in 2015 as manifested in
incident diabetes and complications in France, Germany, Italy, Spain and the
United Kingdom in 2020
Country Total cost
Cost per head of
total estimated
population in
2020 (€)
Cost per
incident case of
type 2 diabetes
in 2020 (€)
Cost per incident case of
type 2 diabetes attributable
to unhealthy diets and low
physical activity in 2020 (€)
France 202 265 604 2.97 1 334 16 678
Germany 266 676 229 3.29 1 099 18 323
Italy 129 750 398 2.14 858 10 720
Spain 82 381 974 1.77 770 11 006
United Kingdom 201 967 593 3.01 1 706 18 953
Chapter 5
Discussion and conclusions
This study sought to contribute to a better understanding of the economic
burden that can be associated with unhealthy diets and low levels of physical
activity in order to help inform priority setting and motivate efforts to promote
more effectively healthy diets and physical activity in Europe and worldwide.
We did so through critically reviewing the available evidence on the economic
costs associated with unhealthy diets and low physical activity; discussing the
measurement, methodological and practical issues for estimating the economic
burden from unhealthy diets and low physical activity; and developing a framework for assessing costs and testing the feasibility of this approach to provide
better estimates of the economic burden.
We showed that the majority of reviewed studies found a significant association between diet and/or physical activity and costs, with unhealthy diets and
low physical activity predictive of higher health care expenditure. Studies that
did report costs that can be associated with the two risk factors estimated the
annual cost of unhealthy diets to range from €3 to €148 per capita and for
low physical activity from €3 to €181 per capita. The highest health care cost
estimates were equivalent to between 2% and 6% of health spending in the
countries. We noted that there is a very wide range of estimates, and these are
very sensitive to the measures of diet and activity and the ways in which the
studies were carried out.
Costing studies differ widely in their analytical approaches and in the nature and
scope of data used, influencing estimates for the economic burden of unhealthy
diets and low physical activity. Particular challenges arise from measuring
unhealthy diets given the different effects of foods and the interactions between
these effects. Calibrating the extent of deviation from optimal consumption and
the effects of this deviation is difficult. It is also clear that the context should be
taken into account in terms of other population characteristics. While there is
more consensus about the measurement of physical activity, similar issues arise in
terms of the independent effects of moderate and vigorous activity and sedentary
behaviour, but also the interactions between these. Further, studies take broader
and narrower perspectives in terms of what costs are included, with some limited to formal health care costs, and others aiming to take a more societal view.
While current evidence makes it difficult to make accurate comparisons, much
44 Assessing the economic costs of unhealthy diets and low physical activity
of the economic burden is likely to come from non-health care costs, especially
from effects on productivity, absenteeism, presenteeism and other indirect costs.
Based on a critical appraisal of existing approaches, we developed a framework
for estimating the economic costs of unhealthy diets and low physical activity
using a disease-based approach, with type 2 diabetes mellitus chosen as a disease
for which both are risk factors. The aim was to demonstrate the feasibility of
undertaking a comprehensive, disease-based, bottom-up cost assessment drawing on the best available data as identified from a rapid review of the published
evidence that addresses some of the limitations of existing costing studies. Our
choice of type 2 diabetes as the exemplar outcome was motivated by its consistently strong association with either risk factor as shown in the literature.
Using this approach, we projected the total economic costs that can be associated with unhealthy diets and low physical activity patterns in 2015 as manifested in incident diabetes cases in 2020 to range from €82.4 million in Spain
to €266.7 million in Germany. This equates to a per capita cost of €1.77 in
Spain to €3.29 in Germany. Relating costs more specifically to the population
projected to develop diabetes in 2020 as a consequence of unhealthy diets and
low physical activity in 2015, the United Kingdom showed the highest cost, at
€18 953, closely followed by Germany and France, while Italy had the lowest
cost, at just over €10 720.
The total cost in the five high-income countries studied (France, Germany, Italy,
Spain and the United Kingdom) was estimated to amount to about €883 million. The populations in the five countries studied account for almost two thirds
of the total population in the European Union (EU-28), which would imply a
total EU cost of around €1.3 billion, but care must be taken in any extrapolation given differences in population characteristics, costs of care and value of
lost productivity. While these estimates of the economic costs are substantial,
they represent only a small proportion of health care expenditure and a very
small proportion of GDP. Even on the higher estimates in the sensitivity analysis
it is likely that the burden of disease associated with unhealthy diets and low
physical activity as measured by poor health and shortened life will be at least as
important as the financial costs of additional health care and lost productivity.
It is difficult to compare the findings of the analyses presented here with estimates
published elsewhere since only diabetes costs are estimated. Scarborough et al.
(2011) calculated the cost of unhealthy diets and low physical activity in the
United Kingdom in 2006–07 to be €9.8 billion (€8.5 billion and €1.3 billion
for unhealthy diets and low physical activity, respectively). Ding et al. (2016),
in their recent assessment of the global economic costs that can be associated
with low physical activity, provided estimates of direct and indirect costs ranging
from $(Int) 1.4 billion in Italy to $(Int) 2.6 billion in Germany. Again, data
Discussion and conclusions 45
are difficult to compare as analyses by Ding et al. considered a wider range of
disease outcomes (coronary heart disease, stroke, type 2 diabetes, breast cancer,
colorectal cancer), and the cost estimates are not easily transferable.
The principal analytical steps employed in the present analysis are similar to
those used by Ding et al. (2016) for low physical activity in that we calculated
country-specific adjusted population-attributable fractions based on available
prevalence data and relative disease risks causally linked to either risk factor in
order to estimate the total number of cases for the outcome (here: diabetes)
in each country. Ding et al. used the same definition of low physical activity
that we used, and drew on the same data sources for prevalence of low physical
activity and estimated the same relative risk-adjusted population-attributable
fractions that we calculated for our analysis. We also estimated health care
costs as well as indirect costs that can be associated with disease developed as a
consequence of the risk factor and similar to Ding et al. we drew on the latest
estimates of diabetes-related health care costs provided by the International
Diabetes Federation (International Diabetes Federation, 2015). Where our
model differs is that we only considered the costs of incident cases, that is, new
cases, which can be causally linked to the risk factor, whereas Ding et al. calculated costs on the basis of prevalence data. Also, our approach takes account of
the expected time lag between exposure to the risk factor (unhealthy diets, low
physical activity) and development of the disease and complications. Further, we
considered a wider range of indirect costs linked to lost productivity because of
work absence, disability, early retirement and premature death among incident
diabetes cases that can be attributed to unhealthy diets and low physical activity.
Conversely, Ding et al. only considered lost productivity that can be associated
with premature death. We therefore believe that our estimates provide a fuller
picture of the likely future costs that can be attributed to contemporary dietary
and physical activity patterns.
5.1 Limitations of the costing framework
As noted, the costing model as proposed here presents a ‘proof of concept’
approach, drawing on the best available data as identified from a rapid review of
the published evidence and providing point estimates only (although we present
data from a limited sensitivity analysis). Clearly, there is uncertainty associated
with each input parameter, namely prevalence rates of unhealthy diets and low
physical activity in the general population, adjustment factors, relative risks and
average per patient disease and productivity costs, which will all impact on the
estimated effect size of predicted incident diabetes cases and cost estimates. A
fully costed model would consider ranges as input parameters as a reflection of
variation at baseline, using for example probabilistic modelling such as Monte
46 Assessing the economic costs of unhealthy diets and low physical activity
Carlo simulation, and also employ sensitivity analyses in order to better understand the influence of the various input parameters on cost estimates.
A major challenge presents the availability of suitable data on the prevalence of
unhealthy diets and low physical activity that are comparable across countries
and over time. For example, the prevalence data for low physical activity used
for the five countries were obtained from the WHO Global Health Observatory
Database, with 2010 prevalence rates as the latest available data. The level of
exposure among physically inactive individuals was not specified so data from
a more recent Eurobarometer survey conducted in 2013 was used to assess the
level of exposure. In order to arrive at comparable estimates there is therefore
a need for more detailed national prevalence data on physical activity in each
country, specifying the type, duration, frequency and intensity to identify the
extent of low physical activity and levels of exposure to risk.
We defined a given dietary pattern as unhealthy, based on a score of <67% on
the 2010 alternate healthy eating index (AHEI) and we used dietary data compiled by the European Food Safety Authority, which draws on national surveys
that are not directly comparable in relation to assessment methods and data
collection instruments. The EFSA database also only provides aggregate-level
data of mean consumption of certain food groups (European Food and Safety
Authority, 2015). In the absence of more detailed data sets, we assumed that
mean intakes as presented in the EFSA database are nationally representative of
a relatively homogenous dietary pattern in each of the five countries. But this
is not necessarily the case and in order to arrive at more precise estimates more
recent, individual-level data on mean intake of the different AHEI food-groups
in different countries in Europe would be needed.
We considered a window period of five years from the occurrence of unhealthy
diets and low physical activity to the development of diabetes and associated
costs incurred. The five-year lag reflects the average latency period and it will
vary according to individual risk profile, with those with other high-risk characteristics, such as genetic predisposition, likely to develop the condition more
quickly. If the costing model was applied more widely this would require a
systematic assessment of the evidence of the range of the latency period, and its
incorporation in the form of a sensitivity analysis.
Likewise, we used average annual per patient costs for diabetes and its complications in previous years, which we adjusted to reflect more closely 2020 prices.
However, average costs are not sensitive to variations in individual patterns of
health care utilisation and a fully costed model would ideally derive a unique set
of average per patient costs for each country that take account of patient characteristics such as age, sex, ethnicity, socioeconomic status and disease severity,
which may influence patterns of use. A particular challenge relates to estimating
Discussion and conclusions 47
the indirect costs that can be associated with diabetes as a manifestation of
unhealthy diets and low physical activity, and within the scope of this study we
applied very crude assumptions, which would need to be revisited for a fully
costed model. Any future modelling exercise would also need to take account
of the long-term care costs associated with diabetes which are estimated to be
substantial, but which we have been unable to address in this work.
Finally, the costing framework as presented here uses only one outcome, namely
type 2 diabetes, as a manifestation of exposure to unhealthy diets and low physical
activity. Yet, unhealthy diets and low physical activity are associated with a range
of other conditions of ill health as highlighted in earlier parts of this volume and
a fully costed model would incorporate these also, along with their sequelae,
guided by the strength of evidence of the association. Ding et al. (2016), in
their analysis of the global economic cost of low physical activity, disaggregated
cost figures by disease category, and diabetes cases that can be attributed to low
physical activity accounted for a large majority of the estimated total health care
costs. Specifically, according to Ding et al., for the five countries considered in the
present study, diabetes cases accounted for approximately half (Italy) up to 85%
(Spain) of the estimated total health care costs. This would mean that estimates
for health care costs provided in the present study are likely to underestimate the
‘true’ health care costs that can be associated with low physical activity-related
diabetes cases by at least one third. However, since Ding et al. (2016) calculated
costs for prevalent cases it is very difficult to generalise from their estimates.
5.2 Implications for future studies
This study has tested the feasibility of estimating the costs of unhealthy diets and
low physical activity using a disease-based approach. While there are limitations,
it has shown that it is broadly feasible to populate the model with data from a
range of sources, and the results show a reasonable consistency across countries.
While the disease burden from diabetes is not currently as large as that for, for
example, ischaemic heart disease, it is a good exemplar because of the strong relationship between these lifestyle factors and the risk of diabetes. In other chronic
diseases there will be additional challenges in identifying the contribution of
these lifestyle factors and disease risk. Given the very wide range of estimates of
costs from the studies reviewed, this may be a more promising approach.
References
Ackermann R, Cheadle A & Sandhu N (2003). Community exercise program use and changes
in healthcare costs for older adults. Am J Prev Med, 25:232–7.
Afshin A, Micha R, Khatibzadeh S & Mozaffarian D (2014). Consumption of nuts and legumes
and risk of incident ischemic heart disease, stroke, and diabetes: a systematic review and
meta-analysis. Am J Clin Nutr, 100:278–88.
Al Tunaji H, Davis J, Dawn M & Khan K (2014). Population-attributable fraction of type 2
diabetes due to physical inactivity in adults: a systematic review. BMC Public Health, 14:469.
Allender S, Foster C, Scarborough P & Rayner M (2007). The burden of physical activity-related
ill health in the UK. J Epidemiol Community Health, 61:344–48.
Alter D, Wijeysundera H & Franklin B (2012). Obesity, lifestyle risk-factors and health service
outcomes among healthy middle-aged adults in Canada. BMC Health Serv Res, 12:238.
American Diabetes Association (2013). Economic costs of diabetes in the U.S. in 2012. Diabetes
Care, 36:1033–46.
Anderson L, Martinson B, Crain A, Pronk N, Whitebird R, O’Connor P & Fine L (2005). Health
care charges associated with physical inactivity, overweight and obesity. Prev Chronic Dis, 2:A09.
Aune D, Keum N, Giovannucci E, Fadnes L, Boffetta P, Greenwood D, Tonstad S, Vatten L, Riboli
E & Norat T (2016). Whole grain consumption and risk of cardiovascular disease, cancer,
and all cause and cause specific mortality: systematic review and dose-response meta-analysis
of prospective studies. BMJ, 353:i2716.
Bachmann J, Defina L, Franzini L, Gao A, Leonard D, Cooper K, Berry J & Willis B (2015).
Cardiorespiratory fitness in middle age and health care costs in later life. J Am Coll Cardiol,
66:1876–85.
Baliunas D (2011). A comparison of two methods of adjusted attributable fraction estimation
as applied to the four major smoking related causes of death in Canada in 2005. Ph.D.
University of Toronto. URL: https://tspace.library.utoronto.ca/bitstream/1807/32061/1/
Baliunas_Dalia_O_201111_PhD_thesis.pdf (accessed 23 October 2016).
Benichou J (2001). A review of adjusted estimators of attributable risk. Stat Methods Med Res,
10:195–216.
Biswas A, Oh P, Faulkner G, Bajaj R, Silver M, Mitchell M, et al. (2015). Sedentary time and its
association with risk for disease incidence, mortality, and hospitalization in adults: a systematic
review and meta-analysis. Ann Intern Med, 162:123–32.
Bland P, An L, Foldes S, Garrett N & Alesci N (2009). Modifiable health behaviours and shortterm medical costs among health plan members. Am J Health Promot, 23:265–73.
Boeing H, Bechthold A, Bub A, Ellinger S, Haller D, Kroke A, Leschik-Bonnet E, Müller M,
Oberritter H, Schulze M, Stehle P & Watzl B (2012). Critical review: vegetables and fruit in
the prevention of chronic diseases. Eur J Nutr, 51:637–63.
Bouvard V, Loomis D, Guyton K, Grosse Y, Ghissassi F, Benbrahim-Tallaa L, Guha N, Mattock H,
Straif K & International Agency For Research On Cancer Monograph Working Group (2015).
Carcinogenicity of consumption of red and processed meat. Lancet Oncol, 16:1599–600.
Breton M, Guénette L, Amiche M, Kayibanda J, Grégoire J & Moisan J (2015). Burden of diabetes
on the ability to work: A systematic review. Diabetes Care, 36:740–9.
50 Assessing the economic costs of unhealthy diets and low physical activity
Carlson S, Fulton J, Pratt M, Yang Z & Adams E (2015). Inadequate physical activity and health
care expenditures in the United States. Prog Cardiovasc Dis, 57:315–23.
Cecchini M & Bull F (2015). Promoting physical activity. In: McDaid D, Sassi F & Merkur S (eds).
Promoting health, preventing disease. The economic case. Maidenhead: Open University Press.
Chevan J & Roberts D (2014). No short-term savings in health care expenditures for physically
active adults. Prev Med, 63:1–5.
Chiuve S, Fung T, Rimm E, Hu F, Mccullough M, Wang M, Stampfer M & Willett W (2012).
Alternative dietary indices both strongly predict risk of chronic disease. J Nutr, 142:1009–18.
Chowdhury R, Stevens S, Gorman D, Pan A, Warnakula S, Chowdhury S, Ward H, Johnson
L, Crowe F, Hu F & Franco O (2012). Association between fish consumption, long chain
omega 3 fatty acids, and risk of cerebrovascular disease: systematic review and meta-analysis.
BMJ, 345:e6698.
Codogno J, Turi B, Kemper H, Fernandes R, Christofaro D & Monteiro H (2015). Physical
inactivity of adults and 1-year health care expenditures in Brazil. Int J Public Health, 60:309–16.
Collins C, Patterson A & Fitzgerald D (2011). Higher diet quality does not predict lower Medicare
costs but does predict number of claims in mid-aged Australian women. Nutrients, 3:40–8.
Commission of the European Communities (2007). White Paper on A Strategy for Europe on
Nutrition, Overweight and Obesity related health issues. Brussels: European Commission.
Council of the European Union (2014). Council conclusions on nutrition and physical activity.
URL: http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:52014XG0708(0
1)&from=EN (accessed 23 October 2016).
Daviglus M, Liu K, Pirzada A, Yan L, Garside D, Wang R, Van Horn L, Manning W, Manheim L,
Dyer A, Greenland P & Stamler J (2005). Relationship of fruit and vegetable consumption in
middle-aged men to medicare expenditures in older age: the Chicago Western Electric Study.
J Am Diet Assoc., 105:1735–44.
De Bakker D, Struijs J, Baan C, Raams J, De Wildt J, Vrijhoef H & Schut F (2012). Early results
from adoption of bundled payment for diabetes care in the Netherlands show improvement
in care coordination. Health Aff (Millwood), 31:426–33.
Department for Work and Pensions (2015). Labour Force Survey analysis of disabled people by
region and main health problem. URL: https://www.gov.uk/government/uploads/system/
uploads/attachment_data/file/406369/labour-force-survey-disabled-people.pdf (accessed 15
June 2016).
Ding D, Lawson K, Kolbe-Alexander T, Finkelstein E, Katzmarzyk P, Van Mechelen W, Pratt M
& Lancet Physical Activity Series 2 Executive Committee (2016). The economic burden of
physical inactivity: a global analysis of major non-communicable diseases. Lancet, 388:1311–24.
Doidge J, Segal L & Gospodarevskaya E (2012). Attributable risk analysis reveals potential
healthcare savings from increased consumption of dairy products. J Nutr, 142:1772–80.
Ekelund U, Steene-Johannessen J, Brown W, Fagerland M, Owen N, Powell K, Bauman A, Lee
I, Lancet Physical Activity Series 2 Executive Committee & Lancet Sedentary Behaviour
Working Group (2016). Does physical activity attenuate, or even eliminate, the detrimental
association of sitting time with mortality? A harmonised meta-analysis of data from more
than 1 million men and women. Lancet, 388:1302–10.
Estruch R, Ros E, Salas-Salvadó J, Covas M, Corella D, Arós F, Gómez-Gracia E, et al. (2013).
Primary prevention of cardiovascular disease with a Mediterranean diet. N Engl J Med,
368:1279–90.
European Commission (2014). The 2014 EU Summit on Chronic Diseases. Brussels, 3 and 4 April
2014, Conference Conclusions. URL: http://ec.europa.eu/health/major_chronic_diseases/
docs/ev_20140403_mi_en.pdf (accessed 23 October 2016).
European Food and Safety Authority (2015). The EFSA Comprehensive European Food
Consumption Database. URL: https://www.efsa.europa.eu/en/food-consumption/
comprehensive-database (accessed 15 June 2016).
References 51
Eurostat (2015). Hourly labour costs. URL: http://ec.europa.eu/eurostat/statistics-explained/
index.php/Hourly_labour_costs (accessed 16 May 2016).
Ezzati M, Lopez A, Rodgers A, et al. (2004). Comparative quantification of health risks. Geneva:
World Health Organization.
Garrett N, Brasure M, Schmitz K, Schultz M & Huber M (2004). Physical inactivity: direct cost
to a health plan. Am J Prev Med, 27:304–9.
GBD 2015 Risk Factors Collaborators (2016). Global, regional, and national comparative risk
assessment of 79 behavioural, environmental and occupational, and metabolic risks or clusters
of risks, 1990–2015: a systematic analysis for the Global Burden of Disease Study 2015.
Lancet, 388:1659–1724.
Hayes A, Leal J, Gray A, Holman R & Clarke P (2013). UKPDS outcomes model 2: a new
version of a model to simulate lifetime health outcomes of patients with type 2 diabetes
mellitus using data from the 30 year United Kingdom Prospective Diabetes Study: UKPDS
82. Diabetologia, 56:1925–33.
Health and Social Care Information Centre (2015). National Diabetes Audit 2012–2013. Report
2: Complications and mortality. URL: http://content.digital.nhs.uk/catalogue/PUB16496/
nati-diab-audi-12-13-rep2.pdf (accessed 25 October 2016).
Herquelot E, Gueguen A, Bonenfant S & Dray-Spira R (2011). Impact of diabetes on work
cessation: data from the GAZEL cohort study. Diabetes Care, 34:1344–9.
Holden S, Barnett A, Peters J, Jenkins-Jones S, Poole C, Morgan C & Currie C (2013). The
incidence of type 2 diabetes in the United Kingdom from 1991 to 2010. Diabetes Obes
Metab, 15:844–52.
Idler N, Teuner C, Hunger M, Holle R, Ortlieb S, Schulz H, et al. (2015). The association between
physical activity and health care costs in children – results from the GINIplus and LISAplus
cohort studies. BMC Public Health, 15:437.
International Diabetes Federation (2015). IDF Diabetes Atlas, Seventh edition. Brussels:
International Diabetes Federation.
Jacobs S, Harmon B, Boushey C, Morimoto Y, Wilkens L, Le Marchand L, Kröger J, Schulze M,
Kolonel L & Maskarinec G (2015). A priori-defined diet quality indexes and risk of type 2
diabetes: the Multiethnic Cohort. Diabetologia, 58:98–112.
Janssen I (2012). Health care costs of physical inactivity in Canadian adults. Appl Physiol Nutr
Metab, 37:803–6.
Johns G (2010). Presenteeism in the workplace: a review and research agenda. J Org Behav,
32:519–42.
Johnston K, Buxton M, Jones D & Fitzpatrick R (1999). Assessing the costs of healthcare
technologies in clinical trials. Health Technol Assess, 3:1–76.
Katzmarzyk P (2011). The Economic Costs Associated with Physical Inactivity and Obesity in
Ontario. Health and Fitness Journal of Canada, 4:31–40.
Katzmarzyk P, Gledhill N & Shephard R (2000). The economic burden of physical inactivity in
Canada. CMAJ, 163:1435–40.
Khangura S, Konnyu K, Cushman R, Grimshaw J & Moher D (2012). Evidence summaries: the
evolution of a rapid review approach. Syst Rev, 1:10.
Koopmanschap M, Rutten F, Van Ineveld B & Van Roijen L (1995). The friction cost method
for measuring indirect costs of disease. J Health Econ, 14:171–89.
Krueger H, Krueger J & Koot J (2015). Variation across Canada in the economic burden attributable
to excess weight, tobacco smoking and physical inactivity. Can J Public Health, 106:e171–7.
Kruk J (2014). Health and economic costs of physical inactivity. Asian Pac J Cancer Prev, 15:7499–
503.
Kuriyama S, Hozawa A & Ohmori K (2004). Joint impact of health risks on health care charges:
7-year follow-up of National Health Insurance beneficiaries in Japan (the Ohsaki Study).
Prev Med, 39:1194–9.
52 Assessing the economic costs of unhealthy diets and low physical activity
Laaksonen M, Knekt P & Rissanen H (2009). The relative importance of modifiable potential
risk factors of type 2 diabetes: a meta-analysis of two cohorts. Eur J Epidemiol, 25:115–24.
Lee I, Shiroma E, Lobelo F, Puska P, Blair S & Katzmarzyk P (2012). Effect of physical inactivity
on major non-communicable diseases worldwide: an analysis of burden of disease and life
expectancy. Lancet, 280:219–29.
Li Y, Ley S, Tobias D, Chiuve S, Vanderweele T, Rich-Edwards J, Curhan G, Willett W, Manson J,
Hu F & Qi L (2015). Birth weight and later life adherence to unhealthy lifestyles in predicting
type 2 diabetes: prospective cohort study. BMJ, 351:h3672.
Liyanage T, Ninomiya T, Wang A, Neal B, Jun M, Wong M, Jardine M, Hillis G & Perkovic V
(2016). Effects of the Mediterranean diet on cardiovascular outcomes – A systematic review
and meta-analysis. PLoS One, 11:e0159252.
Lo Y, Wahlqvist M, Chang Y, Kao S & Lee M (2013). Dietary diversity predicts type of medical
expenditure in elders. Am J Manag Care, 19:e415–23.
Lyssenko V, Jonsson A, Almgren P, Pulizzi N, Isomaa B, Tuomi T, Berglund G, Altshuler D,
Nilsson P & Groop L (2008). Clinical risk factors, DNA variants, and the development of
type 2 diabetes. N Engl J Med, 359:2220–32.
Malik V, Popkin B, Bray G, Després J & Hu F (2010a). Sugar sweetened beverages, obesity, type
2 diabetes and cardiovascular disease risk. Circulation, 121:1356–64.
Malik V, Popkin B, Bray G, Després J, Willett W & Hu F (2010b). Sugar-sweetened beverages and
risk of metabolic syndrome and type 2 diabetes: A meta-analysis. Diabetes Care, 33:2477–83.
Maresova K (2014). The costs of physical inactivity in the Czech Republic in 2008. J Phys Act
Health, 11:489–94.
Martinson B, Crain A & Pronk N (2003). Changes in physical activity and short-term changes
in health care charges: a prospective cohort study of older adults. Prev Med, 37:319–26.
Micha R, Wallace S & Mozaffarian D (2010). Red and processed meat consumption and risk of
incident coronary heart disease, stroke, and diabetes mellitus: a systematic review and metaanalysis. Circulation, 121:2271–83.
Micha R, Khatibzadeh S, Shi P, Andrews K, Engell R, Mozaffarian D & Global Burden of Diseases
Nutrition and Chronic Diseases Expert Group (Nutricode) (2015). Global, regional and
national consumption of major food groups in 1990 and 2010: a systematic analysis including
266 country-specific nutrition surveys worldwide. BMJ Open, 5:e008705.
Montonen J, Knekt P, Härkänen T, Järvinen R, Heliövaara M, Aromaa A & Reunanen A (2005).
Dietary patterns and the incidence of type 2 diabetes. Am J Epidemiol, 161:219–27.
Morgan P (2012). Back to the future: the changing frontiers of nutrition research and its relationship
to policy. Proc Nutr Soc, 71:190–7.
Mozaffarian D, Appel L & Van Horn L (2011). Components of a cardioprotective diet: new
insights. Circulation, 12:2870–91.
Muka T, Imo D, Jaspers L, Colpani V, Chaker L, Van Der Lee S, Mendis S, Chowdhury R, Bramer
W, Falla A, Pazoki R & Franco O (2015). The global impact of non-communicable diseases
on healthcare spending and national income: a systematic review. Eur J Epidemiol, 30:251–77.
Murray C & Lopez A (1997). Global mortality, disability and the contribution of risk factors:
global burden of disease study. Lancet, 349:1436–42.
NCBI (2016). Pubmed. URL: http://www.ncbi.nlm.nih.gov/pubmed.
OECD (2015). Focus on Health Spending: OECD Health Statistics 2015. URL: http://www.
oecd.org/health/health-systems/Focus-Health-Spending-2015.pdf (accessed 16 May 2016).
OECD (2016). Labour force participation rate. URL: https://data.oecd.org/emp/labour-forceparticipation-rate.htm (accessed 15 June 2016).
Oldridge N (2008). Economic burden of physical inactivity: healthcare costs associated with
cardiovascular disease. Eur J Cardiovasc Prev Rehabil, 15:130–9.
Peeters G, Mishra G, Dobson A & Brown W (2014). Health care costs associated with prolonged
sitting and inactivity. Am J Prev Med, 46:265–72.
References 53
Popkin B, Kim S, Rusev E, Du S & Zizza C (2006). Measuring the full-economic costs of diet,
physical activity and obesity-related chronic diseases. Obes Rev, 7:271–93.
Rayner M & Scarborough P (2005). The burden of food-related ill health in the UK. J Epidemiol
Community Health, 59:1054–7.
Rice N & Normand C (2012). The cost associated with disease-related malnutrition in Ireland.
Public Health Nutr, 15:1966–72.
Risérus U, Willett W & Hu F (2009). Dietary fats and prevention of type 2 diabetes. Prog Lipid
Res, 48:44–51.
Samitz G, Egger M & Zwahlen M (2011). Domains of physical activity and all-cause mortality:
systematic review and dose-response meta-analysis of cohort studies. Int J Epidemiol, 40:1382–
400.
Scarborough P, Bhatnagar P, Wickramasinghe KK, Allender S, Foster C & Rayner M (2011). The
economic burden of ill health due to diet, physical inactivity, smoking, alcohol and obesity in
the UK: an update to 2006–07 NHS costs. J Public Health (Oxf), 33:527–35.
Schultz A & Edington D (2007). Employee health and presenteeism: a systematic review. J Occup
Rehabil, 17:547–79.
Schwingshackl L & Hoffmann G (2015). Diet quality as assessed by the Healthy Eating Index,
the Alternate Healthy Eating Index, the Dietary Approaches to Stop Hypertension score,
and health outcomes: a systematic review and meta-analysis of cohort studies. J Acad Nutr
Diet, 115:780–800.
Sedentary Behaviour Research Network (2012). Letter to the editor: standardized use of the terms
“sedentary” and “sedentary behaviours”. Appl Physiol Nutr Metab, 37:540–2.
Stefler D, Malyutina S, Kubinova R, Pajak A, Peasey A, Pikhart H, Brunner E & Bobak M
(2015). Mediterranean diet score and total and cardiovascular mortality in Eastern Europe:
the HAPIEE study. Eur J Nutr, [Epub ahead of print].
Suhrcke M, Fahey D & Mckee M (2008). Economic aspects of chronic disease and chronic disease
management. In: Nolte E & McKee M (eds) Caring for people with chronic conditions: A health
system perspective. Maidenhead: Open University Press.
TNS Opinion & Social (2014). Special Eurobarometer 412. Sport and physical activity. Brussels:
European Commission.
Tong T, Wareham N, Khaw K, Imamura F & Forouhi N (2016). Prospective association of the
Mediterranean diet with cardiovascular disease incidence and mortality and its population
impact in a non-Mediterranean population: the EPIC-Norfolk study. BMC Med, 14:135–46.
United Nations (2015). World population prospects, the 2015 revision. URL: https://esa.un.org/
unpd/wpp/Download/Standard/Population/ (accessed 30 November 2016).
Wang F, Mcdonald T, Reffitt B & Edington D (2005). BMI, physical activity, and health care
utilization/costs among Medicare retirees. Obes Res, 13:1450–7.
Wang G, Pratt M & Macera C (2004). Physical activity, cardiovascular disease, and medical
expenditures in U.S. adults. Ann Behav Med, 28:88–94.
Warburton D, Charlesworth S, Ivey A, Nettlefold L & Bredin S (2010). A systematic review of the
evidence for Canada’s Physical Activity Guidelines for Adults. Int J Behav Nutr Phys Act, 7:39.
Webber L, Divajeva D, Marsh T, Mcpherson K, Brown M, Galea G & Breda J (2014). The future
burden of obesity-related diseases in the 53 WHO European-Region countries and the impact
of effective interventions: a modelling study. BMJ Open, 4:e004787.
Weyer C, Bogardus C, Mott D & Pratley R (1999). The natural history of insulin secretory
dysfunction and insulin resistance in the pathogenesis of type 2 diabetes mellitus. J Clin
Invest, 104:787–94.
WHO Regional Office for Europe (2001). The First Action Plan for Food and Nutrition Policy
WHO European Region 2000–2005. Copenhagen: World Health Organization.
WHO Regional Office for Europe (2005). Steps to health. A European framework to promote
physical activity for health. Copenhagen: World Health Organization.
54 Assessing the economic costs of unhealthy diets and low physical activity
WHO Regional Office for Europe (2015). European Food and Nutrition Action Plan 2015–2020.
Copenhagen: World Health Organization.
WHO Regional Office for Europe (2016). Physical activity strategy for the WHO European
Region 2016–2025. Copenhagen: World Health Organization.
Woodcock J, Franco O, Orsini N & Roberts I (2011). Non-vigorous physical activity and all-cause
mortality: systematic review and meta-analysis of cohort studies. Int J Epidemiol, 40:121–38.
World Cancer Research Fund & American Institute for Cancer Research (2007). Food, nutrition,
physical activity, and the prevention of cancer: a global perspective. Washington DC: American
Institute for Cancer Research.
World Health Assembly (2013). Follow-up to the Political Declaration of the High-level Meeting
of the General Assembly on the Prevention and Control of Non-communicable Diseases.
WHA66.10. URL: http://apps.who.int/gb/ebwha/pdf_files/WHA66/A66_R10-en.pdf
(accessed 23 October 2016).
World Health Organization (2004). Global Strategy on Diet, Physical Activity and Health. Geneva:
World Health Organization.
World Health Organization (2010). Global recommendations on physical activity for health.
Geneva: World Health Organization.
World Health Organization (2016). Global Health Observatory (GHO) data. URL: http://www.
who.int/gho/en/ (accessed 15 June 2016).
Zhang J & Chaaban J (2012). The economic cost of physical inactivity in China. Prev Med, 56:75–8.
Zheng J, Huang T, Yu Y, Hu X, Yang B & Li D (2012). Fish consumption and CHD mortality:
an updated meta-analysis of seventeen cohort studies. Public Health Nutr, 15:725–37.
Zimmet P, Magliano D, Herman W & Shaw J (2014). Diabetes: a 21st century challenge. Lancet
Diabetes Endocrinol, 2:56–64.
Appendices
Appendix 1
Summary overview of key
characteristics of included studies
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
DIET
Daviglus et al.
(2005)
To examine
whether fruit
and vegetable
intake in middle
age is related
to health care
costs (total
and diseasespecific) in old
age.
The
United
States
Based on
diseasespecific costs
(cardiovascular
disease and
cancer) and
overall national
health insurance
costs.
Intake of fruit and vegetables were determined
from previous dietary histories and classified
into one of three groups: <14 cups per month,
14–42 cups per month and >42 cups per
month. Combined fruit and vegetable intake
was classified as either low, medium or
high. Using Medicare claims data, individual
mean annual health care costs in old age and
ten-year cumulative costs prior to death were
estimated. The costs were linked to previous
fruit and vegetable intake and examined across
three intake groups to determine possible
associations.
Fruit and vegetable intake:
Dietary interview (using
Burke’s comprehensive
dietary history method)
Health care costs:
estimated from national
health insurance claims
Direct
health
care costs
(USD)
1984–2000 Although most relationships were not
statistically significant, higher fruit
and vegetable intakes were generally
associated with lower mean annual
health care costs. Mean annual costs
for cardiovascular disease for high
vs low intake were US$ 3 128 vs
US$ 4 223; for cancer: US$ 1 352
vs US$ 1 640 and for total costs:
US$ 10 024 vs US$ 12 211. Trends
were generally similar for fruit or
vegetable intake alone. This trend
also applied to fruit and vegetable
intake and ten-year cumulative
costs before death (US$ 92 757 total
cumulative costs for high intake vs
US$ 132 713 for low intake).
Rayner &
Scarborough
(2005)
To quantify the
disease and
financial burden
of ill health
related to food.
United
Kingdom
Disease-based:
cardiovascular
disease, type 2
diabetes, other
hormonal and
immune system
diseases,
cancer, dental
caries and
digestive
system
diseases.
PAFs for diseases due to risk factors published
by the WHO (2004) were applied to disease
DALYs to estimate the proportion of disease
burdens attributable to unhealthy diets. The
same PAFs were applied to NHS disease costs.
PAFs: Global Burden of
Disease (GBD) study
(1997), WHO Comparative
Quantification of Health
Risks (2004)
Disease DALYS: WHO
Report on Diet, Nutrition
and the Prevention of
Chronic Diseases (2003)
NHS cost data: extrapolated
from estimates from an
earlier study for 1992/93
Direct
health
care costs
(GBP)
2002 Food-related disease accounted for
about 10% of morbidity and mortality
in the United Kingdom (as measured
by DALYs).
Food-related disease was associated
with an annual health care cost of £6
billion, or about one third of the total
NHS costs.
Appendix 1 57
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
DIET
Daviglus et al.
(2005)
To examine
whether fruit
and vegetable
intake in middle
age is related
to health care
costs (total
and diseasespecific) in old
age.
The
United
States
Based on
diseasespecific costs
(cardiovascular
disease and
cancer) and
overall national
health insurance
costs.
Intake of fruit and vegetables were determined
from previous dietary histories and classified
into one of three groups: <14 cups per month,
14–42 cups per month and >42 cups per
month. Combined fruit and vegetable intake
was classified as either low, medium or
high. Using Medicare claims data, individual
mean annual health care costs in old age and
ten-year cumulative costs prior to death were
estimated. The costs were linked to previous
fruit and vegetable intake and examined across
three intake groups to determine possible
associations.
Fruit and vegetable intake:
Dietary interview (using
Burke’s comprehensive
dietary history method)
Health care costs:
estimated from national
health insurance claims
Direct
health
care costs
(USD)
1984–2000 Although most relationships were not
statistically significant, higher fruit
and vegetable intakes were generally
associated with lower mean annual
health care costs. Mean annual costs
for cardiovascular disease for high
vs low intake were US$ 3 128 vs
US$ 4 223; for cancer: US$ 1 352
vs US$ 1 640 and for total costs:
US$ 10 024 vs US$ 12 211. Trends
were generally similar for fruit or
vegetable intake alone. This trend
also applied to fruit and vegetable
intake and ten-year cumulative
costs before death (US$ 92 757 total
cumulative costs for high intake vs
US$ 132 713 for low intake).
Rayner &
Scarborough
(2005)
To quantify the
disease and
financial burden
of ill health
related to food.
United
Kingdom
Disease-based:
cardiovascular
disease, type 2
diabetes, other
hormonal and
immune system
diseases,
cancer, dental
caries and
digestive
system
diseases.
PAFs for diseases due to risk factors published
by the WHO (2004) were applied to disease
DALYs to estimate the proportion of disease
burdens attributable to unhealthy diets. The
same PAFs were applied to NHS disease costs.
PAFs: Global Burden of
Disease (GBD) study
(1997), WHO Comparative
Quantification of Health
Risks (2004)
Disease DALYS: WHO
Report on Diet, Nutrition
and the Prevention of
Chronic Diseases (2003)
NHS cost data: extrapolated
from estimates from an
earlier study for 1992/93
Direct
health
care costs
(GBP)
2002 Food-related disease accounted for
about 10% of morbidity and mortality
in the United Kingdom (as measured
by DALYs).
Food-related disease was associated
with an annual health care cost of £6
billion, or about one third of the total
NHS costs.
58 Assessing the economic costs of unhealthy diets and low physical activity Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
DIET
Rice &
Normand
(2012)
To establish the
annual public
expenditures for
patients with
disease-related
malnutrition
(DRM).
Republic
of Ireland
Based on overall
public health
and social care
costs.
Prevalence of DRM in multiple health care
settings were estimated from age-standardised
comparisons between available Irish data and
large-scale United Kingdom surveys. Frequency
of health care utilisation among adults with
DRM was estimated, to which relevant unit
costs were applied to obtain DRM-related
costs.
Prevalence of DRM: earlier
studies which used the
Malnutrition Universal
Screening Tool, British
Association for Parenteral
and Enteral Nutrition
Screening Week Surveys
(2007, 2008, 2010)
Patient activity levels
and costs: Department
of Health and Children,
Health Service Executive,
Economic and Social
Research Institute
Direct
health
care costs
(EUR)
2007 The annual public health and social
care costs associated with diseaserelated malnutrition among adults
was over €1.4 billion or 10% of the
health care budget. The majority of
the costs were from acute hospital
or residential care settings (70%).
The additional cost of DRM per adult
patient was €5357.
Collins et al.
(2011)
To examine
whether higher
diet quality was
associated with
lower health
care claims and
costs among
middle-aged
adults.
Australia Based on
national health
insurance costs.
Individual dietary patterns were assessed
using a survey consistent with national dietary
recommendations. Based on survey results, the
Australian Recommended Food Score (ARFS)
was calculated and grouped into quintiles (1 =
highest score; 5 = lowest score). Median fiveyear cumulative Medicare costs and six-year
cumulative number of Medicare claims were
linked to ARFS and examined across quintiles
to determine possible associations between
diet quality, number of health care claims and
costs.
Dietary intake: Survey
(using the Dietary
Questionnaire for
Epidemiological Studies)
Number of health care
claims and costs: Medicare
Australia
Direct
health
care costs
(AUD)
Costs:
2002–2006
Number
of claims:
2002–2007
There was a statistically significant
association between five-year
cumulative health care costs and the
ARFS, with individuals with dietary
scores in the highest quintile having
higher health care costs compared
to those in the lowest quintile (cost
difference of AU$ 110). The trend
was not consistent across quintiles.
Individuals in the highest quintile had
a lower number of claims compared
to those in the lowest quintile
(significant difference of 10 claims).
This trend was not consistent in the
medium term.
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
DIET
Doidge et al.
(2012)
To quantify the
potential effects
of increasing
dairy product
consumption to
recommended
levels in terms
of population
health impact
and direct
health care
costs.
Australia Disease-based:
type 2 diabetes,
ischaemic heart
disease, stroke,
osteoporosis,
obesity,
hypertension.
PAFs for diseases related to low dairy
consumption were identified based on fine
levels of dairy consumption (i.e. increments of
0.1 standard serving per day), and distribution
among the population and corresponding RRs.
Two variant-formulas were used to calculate
the PAFs, depending on data availability.
Resulting PAFs were applied to disease DALYs
and health care costs to determine proportions
potentially avoided by increasing dairy
consumption.
Dairy consumption levels
and population distribution:
Australian National Nutrition
Survey (1995)
RRs: Earlier studies
representing the “highest
level of evidence” (as per
criteria by Phillips et al.
(2009))
Disease DALYs and costs:
Australian Institute of
Health and Welfare, earlier
studies, government
reports, primary analysis of
publicly available databases
and government reports
Direct
health
care costs
(AUD)
2010–2011 Increasing dairy consumption to
recommended levels can potentially
prevent 18.4% of the incident
cases of obesity, 10.2% of type 2
diabetes, 5% of ischaemic heart
disease, 16.2% of stroke and 8.3%
of hypertension.
Increasing dairy consumption can
potentially save AU$ 2.0 billion in
direct health care costs, while also
saving an additional 75 012 DALYs.
The amount comprises 1.7% of total
direct health care expenditures and
is comparable with total spending
on public health (AU$ 2.0 billion in
2009–2010).
Lo et al.
(2013)
To assess the
relationship
between dietary
quality and
medical care
utilisation and
expenditures
among
populations
aged ≥65 years.
China
(Taiwan)
Based on
national health
insurance costs.
Individual dietary intakes were assessed
through a 24-hour dietary recall. Dietary quality
was scored from 0 points (lowest) to 6 points
(highest) through the Dietary Diversity Score
(DDS) method. National health insurance claims
in the succeeding eight years were linked to
DDS scores and examined across quartiles
(DDS scores of ≤3, 4, 5, 6) to identify possible
associations between diet quality and level of
medical utilisation and costs.
Dietary intake: National
Elderly Nutrition and Health
Survey (1999–2000)
Medical care utilisation
and costs: National health
insurance (covering >99%
of the population)
Direct
health
care costs
(TWD)
1999–2006 Participants with better diet quality
(as indicated by higher DDS) had
lower utilisation of and costs for
emergency and hospitalisation
services. Average annual
emergency costs were NT$ 2 330
vs NT$ 1 560 for DDS of ≤3 vs DDS
of 6. Hospitalisation costs were
NT$ 47 600 vs NT$ 35 100. For
preventive and dental services,
however, higher DDS predicted
greater utilisation (0.25 and 0.5
times) and costs (+NT$ 270 and
+NT$ 420) compared to the lowest
DDS. Overall expenditures were
still lower for those with higher
DDS at NT$ 64 200 (DDS of 6) vs
NT$ 68 300 (DDS ≤3).
Appendix 1 59
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
DIET
Doidge et al.
(2012)
To quantify the
potential effects
of increasing
dairy product
consumption to
recommended
levels in terms
of population
health impact
and direct
health care
costs.
Australia Disease-based:
type 2 diabetes,
ischaemic heart
disease, stroke,
osteoporosis,
obesity,
hypertension.
PAFs for diseases related to low dairy
consumption were identified based on fine
levels of dairy consumption (i.e. increments of
0.1 standard serving per day), and distribution
among the population and corresponding RRs.
Two variant-formulas were used to calculate
the PAFs, depending on data availability.
Resulting PAFs were applied to disease DALYs
and health care costs to determine proportions
potentially avoided by increasing dairy
consumption.
Dairy consumption levels
and population distribution:
Australian National Nutrition
Survey (1995)
RRs: Earlier studies
representing the “highest
level of evidence” (as per
criteria by Phillips et al.
(2009))
Disease DALYs and costs:
Australian Institute of
Health and Welfare, earlier
studies, government
reports, primary analysis of
publicly available databases
and government reports
Direct
health
care costs
(AUD)
2010–2011 Increasing dairy consumption to
recommended levels can potentially
prevent 18.4% of the incident
cases of obesity, 10.2% of type 2
diabetes, 5% of ischaemic heart
disease, 16.2% of stroke and 8.3%
of hypertension.
Increasing dairy consumption can
potentially save AU$ 2.0 billion in
direct health care costs, while also
saving an additional 75 012 DALYs.
The amount comprises 1.7% of total
direct health care expenditures and
is comparable with total spending
on public health (AU$ 2.0 billion in
2009–2010).
Lo et al.
(2013)
To assess the
relationship
between dietary
quality and
medical care
utilisation and
expenditures
among
populations
aged ≥65 years.
China
(Taiwan)
Based on
national health
insurance costs.
Individual dietary intakes were assessed
through a 24-hour dietary recall. Dietary quality
was scored from 0 points (lowest) to 6 points
(highest) through the Dietary Diversity Score
(DDS) method. National health insurance claims
in the succeeding eight years were linked to
DDS scores and examined across quartiles
(DDS scores of ≤3, 4, 5, 6) to identify possible
associations between diet quality and level of
medical utilisation and costs.
Dietary intake: National
Elderly Nutrition and Health
Survey (1999–2000)
Medical care utilisation
and costs: National health
insurance (covering >99%
of the population)
Direct
health
care costs
(TWD)
1999–2006 Participants with better diet quality
(as indicated by higher DDS) had
lower utilisation of and costs for
emergency and hospitalisation
services. Average annual
emergency costs were NT$ 2 330
vs NT$ 1 560 for DDS of ≤3 vs DDS
of 6. Hospitalisation costs were
NT$ 47 600 vs NT$ 35 100. For
preventive and dental services,
however, higher DDS predicted
greater utilisation (0.25 and 0.5
times) and costs (+NT$ 270 and
+NT$ 420) compared to the lowest
DDS. Overall expenditures were
still lower for those with higher
DDS at NT$ 64 200 (DDS of 6) vs
NT$ 68 300 (DDS ≤3).
60 Assessing the economic costs of unhealthy diets and low physical activity Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Katzmarzyk et
al. (2000)
To estimate the
mortality and
economic costs
of physical
inactivity and
the effect of a
10% reduction
in inactivity
levels on costs.
Canada Disease-based:
coronary artery
disease, stroke,
hypertension,
colon cancer,
breast cancer,
type 2 diabetes
mellitus,
osteoporosis.
PAFs were calculated based on RRs from
earlier meta-analyses and large prospective,
epidemiological studies. The PAFs were
applied to total numbers of premature deaths
per disease and disease costs. A hypothetical
scenario was assumed where inactivity
prevalence was reduced by 10%, under which
new PAFs were calculated and applied to
disease costs.
Physical activity prevalence:
Physical Activity Monitor
Survey, Canadian Fitness
and Lifestyle Research
Institute (1999)
RRs: Earlier meta-analyses
and prospective studies
Premature mortality:
Statistics Canada (1995)
Cost data: Canadian Health
Expenditures Database,
Economic Burden of Illness
Canada (1993), American
Heart Association (1999),
earlier studies
Direct
health
care costs
(CAD)
1999 Physical inactivity was responsible
for 10.3% (21 000) of deaths from all
causes in 1995, through the seven
diseases.
Physical inactivity accounted for
C$ 2.1 billion or 2.5% of total direct
health care costs. The amount
represented 25.5% of the total costs
of treating the seven diseases.
Reducing the prevalence of physical
inactivity by 10% would reduce
direct health care expenditures by
C$ 150 million per year.
Ackermann et
al. (2003)
To determine
if changes in
health care
utilisation
and costs for
Medicareeligible
enrolees of a
large health
maintenance
organization
(HMO) are
related to
their choice to
participate in
a community
exercise
programme
offered as a
health benefit.
The
United
States
Based on
private health
insurance costs.
A retrospective, matched cohort study was
conducted to determine if changes in health
care costs for Medicare-eligible enrolees
(aged ≥65 years) choosing to participate in the
exercise programme were different from those
of similar individuals who did not participate.
Three enrolees who never attended the
exercise programme were randomly selected
as controls for each participant matching
on age and gender. Changes in health care
utilisation and costs were compared between
the intervention and control groups.
Exercise programme
participation and cost data:
Group Health Cooperative
of Puget Sound, Decision
Support System (integrates
clinical information, units of
service and actual costs)
Direct
health
care costs
(USD)
1997–2000 Overall, no significant differences
were found in the total health care
costs between exercise programme
participants and non-participants.
However, compared to those who
did not participate, programme
participants had 4.9% lower
hospitalisations and displayed
a stronger trend towards lower
inpatient costs (i.e. US$ 708
lower). Among those who had high
programme usage (i.e. ≥1 visit per
week), annual total health care
costs were US$ 1 057 lower and
the hospitalisation risk was 7.9%
lower compared to controls. These
significant effects among higher
users were observed at a relatively
low attendance rate (average of 1.74
visits per week) and under normal
daily operating conditions of the
programme.
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Martinson et
al. (2003)
To examine
the impact of
changes in
physical activity
over the course
of a single year
on short-term
changes in
health care
charges among
members of
a health plan
aged 50 and
above.
The
United
States
Based on
commercial
health insurance
costs.
Health plan members were surveyed twice for
physical activity levels (in August 1995 and
September 1996). Individuals were classified
as active or inactive at each of the two time
points. To account for alternative definitions of
physical activity, five definitions of inactivity
and activity were used:
A: Inactive = 0 days, active = 1+ days
B: Inactive = 0 days, active = 2+ days
C: Inactive = 0–1 days, active = 2+ days
D: Inactive = 0–1 days, active = 3+ days
E: Inactive = 0–1 days, active = 4+ days
A specific definition was used in categorising
an individual’s physical activity changes
between August 1995 and September 1996
into one of five mutually exclusive groups: (1)
inactive to inactive, (2) active to active, (3)
active to inactive, (4) inactive to active and (5)
unclassified (falling within the “gap”” between
the inactive and active categories under B,
D and E). Changes in health care charges
between September 1994 to August 1995
and September 1996 to August 1997 were
examined across 25 physical activity change
groups to determine possible associations.
Physical activity levels:
HealthPartners Minnesota
health plan members’
survey
Health care charges:
commercial health
insurance
Direct
health
care costs
(USD)
1994–1995
and
1996–1997
For physical inactivity/activity
definitions A and B, physically
active groups had larger decreases
in health care charges relative to
individuals who were physically
inactive in both surveys. These
declines ranged from US$ 1 200 to
US$ 1 900. For definitions C, D and
E, declines in total charges were
observed primarily among individuals
who moved from inactivity to activity.
In particular for definition D, the
decline in charges was significantly
larger (in the inactive to active group)
at US$ 2 200. Similar trends are seen
in models using definitions C and E
but the results are not significant.
Garrett et al.
(2004)
To estimate
the costs
of physical
inactivity
among
members of
the Blue Cross
Blue Shield
health plan in
Minnesota.
The
United
States
Diseasebased: heart
disease, stroke,
hypertension,
type 2 diabetes,
colon cancer,
breast cancer,
osteoporosis,
depression,
anxiety.
PAFs were calculated based on RRs from
earlier meta-analyses of published studies,
stratified by level of exposure: inactive and
irregularly active. The PAFs were applied to
disease costs.
Physical activity
prevalence: Behavioural
Risk Factors Surveillance
System (2000)
RRs: Earlier meta-analyses
of published studies
Cost data: Blue Cross and
Blue Shield health plan
Direct
health
care costs
(USD)
2000 Physical inactivity was estimated
to account for US$ 83.6 million of
total medical expenditures among a
health plan population of 1.5 million
members. The cost per member was
US$ 56.
Appendix 1 61
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Martinson et
al. (2003)
To examine
the impact of
changes in
physical activity
over the course
of a single year
on short-term
changes in
health care
charges among
members of
a health plan
aged 50 and
above.
The
United
States
Based on
commercial
health insurance
costs.
Health plan members were surveyed twice for
physical activity levels (in August 1995 and
September 1996). Individuals were classified
as active or inactive at each of the two time
points. To account for alternative definitions of
physical activity, five definitions of inactivity
and activity were used:
A: Inactive = 0 days, active = 1+ days
B: Inactive = 0 days, active = 2+ days
C: Inactive = 0–1 days, active = 2+ days
D: Inactive = 0–1 days, active = 3+ days
E: Inactive = 0–1 days, active = 4+ days
A specific definition was used in categorising
an individual’s physical activity changes
between August 1995 and September 1996
into one of five mutually exclusive groups: (1)
inactive to inactive, (2) active to active, (3)
active to inactive, (4) inactive to active and (5)
unclassified (falling within the “gap”” between
the inactive and active categories under B,
D and E). Changes in health care charges
between September 1994 to August 1995
and September 1996 to August 1997 were
examined across 25 physical activity change
groups to determine possible associations.
Physical activity levels:
HealthPartners Minnesota
health plan members’
survey
Health care charges:
commercial health
insurance
Direct
health
care costs
(USD)
1994–1995
and
1996–1997
For physical inactivity/activity
definitions A and B, physically
active groups had larger decreases
in health care charges relative to
individuals who were physically
inactive in both surveys. These
declines ranged from US$ 1 200 to
US$ 1 900. For definitions C, D and
E, declines in total charges were
observed primarily among individuals
who moved from inactivity to activity.
In particular for definition D, the
decline in charges was significantly
larger (in the inactive to active group)
at US$ 2 200. Similar trends are seen
in models using definitions C and E
but the results are not significant.
Garrett et al.
(2004)
To estimate
the costs
of physical
inactivity
among
members of
the Blue Cross
Blue Shield
health plan in
Minnesota.
The
United
States
Diseasebased: heart
disease, stroke,
hypertension,
type 2 diabetes,
colon cancer,
breast cancer,
osteoporosis,
depression,
anxiety.
PAFs were calculated based on RRs from
earlier meta-analyses of published studies,
stratified by level of exposure: inactive and
irregularly active. The PAFs were applied to
disease costs.
Physical activity
prevalence: Behavioural
Risk Factors Surveillance
System (2000)
RRs: Earlier meta-analyses
of published studies
Cost data: Blue Cross and
Blue Shield health plan
Direct
health
care costs
(USD)
2000 Physical inactivity was estimated
to account for US$ 83.6 million of
total medical expenditures among a
health plan population of 1.5 million
members. The cost per member was
US$ 56.
62 Assessing the economic costs of unhealthy diets and low physical activity Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Kuriyama et
al. (2004)
To examine the
joint impact
of physical
inactivity,
including
smoking and
obesity, on
direct health
care charges.
Japan Based on
national health
insurance costs.
Risk factor status of participants in terms of
BMI, physical activity and smoking status were
assessed. Participants’ health insurance claims
and costs were then followed-up prospectively
for seven years. They were classified into eight
risk groups (i.e. different combinations of the
three risk factors) and examined in terms of
monthly per capita health care charges to
determine the cost impact of different risk
factor combinations.
Physical activity, BMI,
smoking data: study
interview
Cost data: national health
insurance (covers 35% of
the Japanese population)
Direct
health
care costs
(USD)
1995–2001 Participants without risk (i.e. never
smoking, with normal BMI and
physically active) had mean monthly
per capita health care charges
of US$ 171.6. Compared to this
group, the presence of physical
inactivity alone increased per capita
costs by 8% (US$ 185.3), smoking
and physical inactivity by 31.4%
(US$ 225.4) and obesity and physical
inactivity by 16.4% (US$ 199.8).
Presence of all three risk factors
increased per capita costs by 42.6%
(US$ 244.7).
Wang et al.
(2004)
To estimate
the costs of
cardiovascular
disease (CVD)
associated
with physical
inactivity.
The
United
States
Disease-based:
CVD (including
coronary
heart disease,
hypertension,
stroke and
rheumatic heart
disease).
Individual medical expenditure data were linked
to physical activity status in the previous year.
Physical activity was categorised as active
and inactive. Medical expenditures on CVD
associated with inactivity were derived by
comparing mean medical costs between
population groups stratified by CVD status
and physical activity status and obtaining the
difference.
CVD status and physical
activity: National Health
Interview Survey (1995)
Cost data: Medical
Expenditure Panel Survey
(1996)
Direct
health
care costs
(USD)
1996 Among 7.3 million cases of CVD,
1.1 million or 15.3% were associated
with physical inactivity. The total
medical expenditure of persons
with CVD was US$ 41.3 billion, of
which US$ 5.4 billion (13.1%) was
associated with physical inactivity.
Applying the percentages to the
national health and economic
burden, 9.2 million CVD cases in the
United States were associated with
physical inactivity in 2001, costing
US$ 23.7 billion in direct medical
expenditures.
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Anderson et
al. (2005)
To estimate
the proportion
of total health
care charges
associated
with physical
inactivity,
overweight and
obesity among
populations
aged 40 and
older.
The
United
States
Based on
commercial
health insurance
costs.
A predictive model of health care charges
was developed through preliminary analysis
of data from 8 000 health plan members. Five
variables were assessed in association with
health care charges: physical activity, BMI,
chronic disease status (i.e. with diabetes and/
or hypertension or none) and smoking status.
Counterfactual health care charges were also
estimated by reclassifying all individuals as
normal weight and physically active, leaving
other characteristics unchanged. Model cells
for hypothetical 200 000 health plan members
were then developed and defined by status of
BMI, physical activity and other co-variates.
Total health care charges were estimated by
multiplying the percentage of the health plan
population in each cell with the predicted
charges per cell and summing all cells.
Counterfactual estimates were computed by
using counterfactual charges. The difference
between current risk profile total charges and
counterfactual total charges were computed
as charges associated with physical inactivity,
overweight and obesity. The same calculation
was performed but using national population
percentages to identify national cost estimates.
BMI, physical activity
and other variable data:
HealthPartners Minnesota
health plan survey (1995)
Health care charges:
commercial health
insurance
National population
percentages: National
Health Interview Survey
(2001)
Direct
health
care costs
(USD)
1996–1999 For a health plan with 200 000 white
members aged 40 and above, total
annualised health care charges of
US$ 1.12 billion was estimated,
23% of which (US$ 236 million) was
associated with physical inactivity
and overweight or obesity. The three
sub-populations with the largest
charges associated with physical
inactivity, overweight and obesity
were:
1) men aged 50–64 without chronic
disease (US$ 44.7 million)
2) men aged 65 and older with
chronic disease (US$ 43.7 million)
3) men aged 40–49 with no chronic
disease (US$ 41.7 million).
At the national level, the percentage
of national health care charges
associated with physical inactivity,
overweight and obesity was
estimated to be 27%.
Appendix 1 63
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Anderson et
al. (2005)
To estimate
the proportion
of total health
care charges
associated
with physical
inactivity,
overweight and
obesity among
populations
aged 40 and
older.
The
United
States
Based on
commercial
health insurance
costs.
A predictive model of health care charges
was developed through preliminary analysis
of data from 8 000 health plan members. Five
variables were assessed in association with
health care charges: physical activity, BMI,
chronic disease status (i.e. with diabetes and/
or hypertension or none) and smoking status.
Counterfactual health care charges were also
estimated by reclassifying all individuals as
normal weight and physically active, leaving
other characteristics unchanged. Model cells
for hypothetical 200 000 health plan members
were then developed and defined by status of
BMI, physical activity and other co-variates.
Total health care charges were estimated by
multiplying the percentage of the health plan
population in each cell with the predicted
charges per cell and summing all cells.
Counterfactual estimates were computed by
using counterfactual charges. The difference
between current risk profile total charges and
counterfactual total charges were computed
as charges associated with physical inactivity,
overweight and obesity. The same calculation
was performed but using national population
percentages to identify national cost estimates.
BMI, physical activity
and other variable data:
HealthPartners Minnesota
health plan survey (1995)
Health care charges:
commercial health
insurance
National population
percentages: National
Health Interview Survey
(2001)
Direct
health
care costs
(USD)
1996–1999 For a health plan with 200 000 white
members aged 40 and above, total
annualised health care charges of
US$ 1.12 billion was estimated,
23% of which (US$ 236 million) was
associated with physical inactivity
and overweight or obesity. The three
sub-populations with the largest
charges associated with physical
inactivity, overweight and obesity
were:
1) men aged 50–64 without chronic
disease (US$ 44.7 million)
2) men aged 65 and older with
chronic disease (US$ 43.7 million)
3) men aged 40–49 with no chronic
disease (US$ 41.7 million).
At the national level, the percentage
of national health care charges
associated with physical inactivity,
overweight and obesity was
estimated to be 27%.
64 Assessing the economic costs of unhealthy diets and low physical activity Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Wang et al.
(2005)
To explore the
relationship of
physical activity
with shortterm health
care costs
across BMI
groups among
individuals
aged 65 and
above.
The
United
States
Based on
national health
insurance and
indemnity/
PPO insurance
costs.
Physical activity of participants was assessed
by the number of times per week they engage
in physical activity “hard enough to induce
heavy breathing and fast heart beat” for at
least 20 minutes. Participants were classified
into three groups: sedentary, moderately active
and very active. They were also divided by BMI
group (normal weight, overweight, obese). In all
three BMI groups, health care costs were linked
to physical activity levels to determine possible
associations. Analysis was also done in specific
age groups (65–69, 70–74 and 75+ years).
Physical activity prevalence:
study interview (using
the modified Health Risk
Appraisal questionnaire)
Cost data: national health
insurance (Medicare),
indemnity/PPO plans
Direct
health
care costs
(USD)
2001–2002 Higher levels of physical activity
predicted lower short-term health
care costs for older individuals
across the three BMI groups.
Moderately active retirees had
US$ 1 456, US$ 1 731 and
US$ 1 177 lower total health
care costs than their sedentary
counterparts in the normal weight,
overweight and obese groups,
respectively. The very active
retirees had US$ 1 823, US$ 581
and US$ 1 379 lower costs than
moderately active retirees in the
same BMI groupings, respectively.
The same association between
physical activity and short-term
health care costs applies when
analysis is done in specific age
groups (65–69, 70–74, 75+ years).
Allender et al.
(2007)
To estimate
the health
and economic
burden of
physical
inactivity.
United
Kingdom
Disease-based:
ischaemic
heart disease,
ischaemic
stroke, breast
cancer, colon/
rectum cancer,
type 2 diabetes.
PAFs published by the WHO (2004) stratified
by sex were applied to disease DALYs in the
WHO EUR-A region to determine health burdens
attributable to physical inactivity. The same
PAFs were applied to NHS disease costs.
Disease DALYs: WHO (2003)
Cost data: extrapolated
from estimates from earlier
studies for 1992/93
Direct
health
care costs
(GBP)
2002 Physical inactivity was estimated to
be associated with 3% of DALYs lost.
Physical inactivity was estimated to
be associated with £1.06 billion in
direct health care costs.
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Katzmarzyk
(2011)
To estimate
the economic
burden of
physical
inactivity.
Canada Disease-based:
coronary heart
disease, stroke,
hypertension,
colon cancer,
breast cancer,
type 2 diabetes,
osteoporosis.
PAFs were calculated based on summary
RRs derived in a previous meta-analysis of
prospective longitudinal studies. The PAFs were
applied to disease health care and productivity
costs.
Physical activity prevalence:
Canadian Community
Health Survey (2009)
RRs: earlier meta-analysis
of studies
Cost data: Economic
Burden of Illness in Canada
(1993, 1998), Health
Canada (2002), National
Cancer Institute of Canada
(2002), Canadian Institute
for Health Information
(2010), American Heart
Association (2002), US
Centers for Disease
Prevention and Control,
earlier studies
Direct
health
care and
indirect
productivity
costs (CAD)
2009 Physical inactivity in Ontario
was associated with a cost of
C$ 3.4 billion (C$ 1.02 billion in
direct costs and C$ 2.34 billion in
indirect costs). After extrapolation
to the national situation, estimated
costs reach C$ 8.6 billion
(C$ 2.6 billion direct, C$ 5.95 billion
indirect).
Appendix 1 65
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Katzmarzyk
(2011)
To estimate
the economic
burden of
physical
inactivity.
Canada Disease-based:
coronary heart
disease, stroke,
hypertension,
colon cancer,
breast cancer,
type 2 diabetes,
osteoporosis.
PAFs were calculated based on summary
RRs derived in a previous meta-analysis of
prospective longitudinal studies. The PAFs were
applied to disease health care and productivity
costs.
Physical activity prevalence:
Canadian Community
Health Survey (2009)
RRs: earlier meta-analysis
of studies
Cost data: Economic
Burden of Illness in Canada
(1993, 1998), Health
Canada (2002), National
Cancer Institute of Canada
(2002), Canadian Institute
for Health Information
(2010), American Heart
Association (2002), US
Centers for Disease
Prevention and Control,
earlier studies
Direct
health
care and
indirect
productivity
costs (CAD)
2009 Physical inactivity in Ontario
was associated with a cost of
C$ 3.4 billion (C$ 1.02 billion in
direct costs and C$ 2.34 billion in
indirect costs). After extrapolation
to the national situation, estimated
costs reach C$ 8.6 billion
(C$ 2.6 billion direct, C$ 5.95 billion
indirect).
66 Assessing the economic costs of unhealthy diets and low physical activity Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Alter et al.
(2012)
To estimate
cumulative
outcomes
and costs
associated
with obesity
alone and in
combination
with physical
inactivity,
including
smoking and
psychosocial
distress among
middle-aged
adults.
Canada Based on public
health insurance
costs.
Risk factor status in terms of BMI, physical
activity, psychosocial status and smoking
status were assessed through a telephone
survey. Individual risk factor data were
linked with health care utilisation and costs
and followed longitudinally for 11.5 years.
Propensity-matching was done where each
exposed individual was matched in 1:1 fashion
to a healthy, normal weight person, based
on age, gender, socioeconomic status, comorbidity and non-BMI risk factors. Differences
in costs between the exposed and unexposed
matched pairs were tallied and averaged
throughout the follow-period and attributed to
the corresponding risk factor.
Risk factor data: National
Population Health Survey
(1994–96)
Cost data: public health
insurance, Canadian
Institutes of Health
Information, Ontario Case
Costing Initiative, Ontario
Drug Benefits formulary
Incidence of diabetes and
hypertension: Ontario
diabetes and hypertension
databases
Mortality: registered
persons database
Direct
health
care costs
(CAD)
1994/95/96
–
2005/06/07
Cumulative costs associated with
obesity alone were C$ 8 294.67
per person, not significantly
higher compared to costs among
propensity-matched normal weight
controls (C$ 7 323.59 per person).
Obesity in combination with other
lifestyle factors was associated
with significantly higher cumulative
expenditures as compared with
normal-weight healthy matched
controls. Excess costs were
estimated to be:
Overweight + physically inactive:
C$ 1 095.30
Obese + physically inactive:
C$ 4 079.47
Overweight + physically inactive +
smoking: C$ 2 026.10
Obese + physically inactive +
smoking: C$ 2 632.91
Overweight + physically inactive +
distressed: C$ 1 868.57
Obese + physically inactive +
distressed: C$ 7 156.94
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Janssen
(2012)
To estimate
the economic
burden of
physical
inactivity.
Canada Disease-based:
coronary artery
disease, stroke,
hypertension,
colon cancer,
female breast
cancer, type
2 diabetes,
osteoporosis.
PAFs were calculated based on summary
RRs derived in earlier meta-analyses of RRs
in prospective cohort studies, stratified by
sex. PAFs were applied to disease direct and
indirect costs.
Physical activity prevalence:
Canadian Health Measures
Survey (2007–2009)
RRs: earlier meta-analyses
of studies
Cost data: Economic
Burden of Illness in Canada
(2000) extrapolated to
2009 values
Direct
health
care and
indirect
productivity
costs (CAD)
2009 Physical inactivity costs C$ 6.8 billion
(C$ 2.4 billion in direct costs and
C$ 4.3 billion in indirect costs). Direct
costs represent 3.8% of the overall
health care costs.
Zhang and
Chaaban
(2012)
To estimate
the economic
burden of
physical
inactivity.
China Disease-based:
coronary heart
disease, stroke,
hypertension,
cancer, type 2
diabetes.
Impacts of physical inactivity were assessed
through direct mechanisms (inactivity to
NCDs) and indirect mechanisms (inactivity
to overweight/obesity to NCDs). PAFs were
calculated for each disease related to physical
inactivity, overweight and obesity, based on
RRs from earlier meta-analyses of prospective
studies and single cohort studies. PAFs related
to overweight and obesity were multiplied by
12% (i.e. the proportion of overweight and
obesity attributable to physical inactivity). All
three sets of PAFs were summed and the totals
applied to disease direct and indirect costs.
Physical activity,
overweight, obesity
prevalence: Chinese
Behavioural Risk Factors
Surveillance Survey (2007)
RRs: Earlier meta-analyses
and single cohort studies
Cost data: National Health
Service Survey (2003)
extrapolated to 2007
Direct
health
care and
indirect
productivity
costs (USD)
2007 Economic costs associated
with physical inactivity were
US$ 6.7 billion (US$ 3.5 billion for
direct costs and US$ 3.3 billion for
indirect costs). This was equivalent
to 15.2% of the total costs of the
five diseases, 5.3% of total NCD
costs and 3.8% of total costs for all
diseases.
Appendix 1 67
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Janssen
(2012)
To estimate
the economic
burden of
physical
inactivity.
Canada Disease-based:
coronary artery
disease, stroke,
hypertension,
colon cancer,
female breast
cancer, type
2 diabetes,
osteoporosis.
PAFs were calculated based on summary
RRs derived in earlier meta-analyses of RRs
in prospective cohort studies, stratified by
sex. PAFs were applied to disease direct and
indirect costs.
Physical activity prevalence:
Canadian Health Measures
Survey (2007–2009)
RRs: earlier meta-analyses
of studies
Cost data: Economic
Burden of Illness in Canada
(2000) extrapolated to
2009 values
Direct
health
care and
indirect
productivity
costs (CAD)
2009 Physical inactivity costs C$ 6.8 billion
(C$ 2.4 billion in direct costs and
C$ 4.3 billion in indirect costs). Direct
costs represent 3.8% of the overall
health care costs.
Zhang and
Chaaban
(2012)
To estimate
the economic
burden of
physical
inactivity.
China Disease-based:
coronary heart
disease, stroke,
hypertension,
cancer, type 2
diabetes.
Impacts of physical inactivity were assessed
through direct mechanisms (inactivity to
NCDs) and indirect mechanisms (inactivity
to overweight/obesity to NCDs). PAFs were
calculated for each disease related to physical
inactivity, overweight and obesity, based on
RRs from earlier meta-analyses of prospective
studies and single cohort studies. PAFs related
to overweight and obesity were multiplied by
12% (i.e. the proportion of overweight and
obesity attributable to physical inactivity). All
three sets of PAFs were summed and the totals
applied to disease direct and indirect costs.
Physical activity,
overweight, obesity
prevalence: Chinese
Behavioural Risk Factors
Surveillance Survey (2007)
RRs: Earlier meta-analyses
and single cohort studies
Cost data: National Health
Service Survey (2003)
extrapolated to 2007
Direct
health
care and
indirect
productivity
costs (USD)
2007 Economic costs associated
with physical inactivity were
US$ 6.7 billion (US$ 3.5 billion for
direct costs and US$ 3.3 billion for
indirect costs). This was equivalent
to 15.2% of the total costs of the
five diseases, 5.3% of total NCD
costs and 3.8% of total costs for all
diseases.
68 Assessing the economic costs of unhealthy diets and low physical activity Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Chevan &
Roberts
(2014)
To examine
the association
between
leisure-time
physical activity
levels and
short-term
health care
expenditures.
The
United
States
Based on shortterm public and/
or private health
care costs.
Physical activity was assessed by the number
of minutes of vigorous- and moderate-intensity
activity completed per week. By equating
2 minutes of moderate activity to 1 minute
of vigorous activity, physical activity was
categorised in two ways. The first categorised
participants as to: (1) whether they met the
physical activity guidelines for strength and
aerobic activity, (2) met guidelines only for
strength, (3) met guidelines only for aerobic
activity and (4) not meeting guidelines for both.
The second focused on aerobic activity and
split up participants as completing 0, <75,
75–149, 150–299 or >300 minutes per week.
The two categorisations characterised weekly
leisure-time physical activity levels which were
linked to individual health care expenditures to
determine possible associations.
Physical activity prevalence:
National Health Interview
Survey (2006, 2007)
Cost data: Medical
Expenditures Panel
Survey (2007–2008 and
2008–2009)
Direct
health
care costs
(USD)
2007–2009 Whether in terms of meeting the
physical activity guidelines (i.e. for
strength and aerobic activity) or the
number of minutes spent on aerobic
activity per week, not a single level
of leisure-time physical activity
had a significant association with
all types of short-term health care
expenditures.
Maresova
(2014)
To estimate
the economic
burden of
physical
inactivity.
Czech
Republic
Disease-based:
coronary
heart disease,
ischaemic
stroke, diabetes
type 2, female
breast cancer,
colon cancer.
PAFs calculated from disease-specific RRs
derived from WHO (2004) stratified by age
group (15–69, 70–79, 80+) and physical
activity level (level 1: inactive; level 2:
insufficiently active). The PAFs were applied to
disease costs.
Mortality data: Czech
Statistical Office (CZSO)
Morbidity (DALYs): WHO
GBD study (2004)
Physical activity
prevalence: Czech Republic
European Health Interview
Survey (EHIS)
Cost data: health insurance
(covering approx. 75% of
health care expenditure in
the Czech Republic); earlier
studies
Direct
health
care costs
(CZK)
2008 Physical inactivity accounted for
2 442 of all deaths (2.3%) and
18 065 DALYs (1.24%).
Physical inactivity accounted for
almost CZK 700 million (~€29
million) of public health insurance
expenditure (0.4% of total
expenditure).
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Peeters et al.
(2014)
To estimate
the costs
associated
with prolonged
sitting and
physical
inactivity
among middleaged women.
Australia Based on
national health
insurance costs.
Participants were surveyed in terms of time
spent sitting, walking and in moderate and
vigorous leisure-time activities. Sitting time
was categorised as low, moderate and high.
Physical activity was categorised as inactive,
low, moderate and high. Combined sitting
time and physical activity or “activity patterns”
were categorised into active + low sitting time,
active + high sitting time, inactive + low sitting
time and inactive + high sitting time. Medicare
costs averaged over the survey year ±1 year
were used to calculate the annual costs.
Annual median costs were linked to sitting
time, physical activity and activity patterns
and examined across groups to determine
possible associations. Analysis was also done
by BMI strata to examine for potential effect
modifications of BMI.
Sitting time and physical
activity: Australian
Longitudinal Study on
Women’s Health (2001,
2005, 2007, 2010)
Cost data: National health
insurance
Direct
health
care costs
(AUD)
2010 The annual median costs for
highly active vs inactive individuals
were AU$ 689 vs AU$ 741, with a
significant difference of AU$ 94. In
terms of sitting time, annual median
costs for people with low sitting time
vs high sitting time were AU$ 671
vs AU$ 709, with a difference of
AU$ 16. No statistically significant
associations were found between
sitting time and costs. A high sitting
time did not add to inactivityassociated costs (AU$ 110 higher
for inactive people with high sitting
time vs active people with low sitting
time). Although costs are higher
for overweight and obese groups
compared to normal weight, the
effects of physical activity on costs
were similar across BMI ranges,
suggesting no BMI interaction
effects.
Appendix 1 69
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Peeters et al.
(2014)
To estimate
the costs
associated
with prolonged
sitting and
physical
inactivity
among middleaged women.
Australia Based on
national health
insurance costs.
Participants were surveyed in terms of time
spent sitting, walking and in moderate and
vigorous leisure-time activities. Sitting time
was categorised as low, moderate and high.
Physical activity was categorised as inactive,
low, moderate and high. Combined sitting
time and physical activity or “activity patterns”
were categorised into active + low sitting time,
active + high sitting time, inactive + low sitting
time and inactive + high sitting time. Medicare
costs averaged over the survey year ±1 year
were used to calculate the annual costs.
Annual median costs were linked to sitting
time, physical activity and activity patterns
and examined across groups to determine
possible associations. Analysis was also done
by BMI strata to examine for potential effect
modifications of BMI.
Sitting time and physical
activity: Australian
Longitudinal Study on
Women’s Health (2001,
2005, 2007, 2010)
Cost data: National health
insurance
Direct
health
care costs
(AUD)
2010 The annual median costs for
highly active vs inactive individuals
were AU$ 689 vs AU$ 741, with a
significant difference of AU$ 94. In
terms of sitting time, annual median
costs for people with low sitting time
vs high sitting time were AU$ 671
vs AU$ 709, with a difference of
AU$ 16. No statistically significant
associations were found between
sitting time and costs. A high sitting
time did not add to inactivityassociated costs (AU$ 110 higher
for inactive people with high sitting
time vs active people with low sitting
time). Although costs are higher
for overweight and obese groups
compared to normal weight, the
effects of physical activity on costs
were similar across BMI ranges,
suggesting no BMI interaction
effects.
70 Assessing the economic costs of unhealthy diets and low physical activity Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Bachmann et
al. (2015)
To evaluate
the association
of health care
costs in later
life with cardiorespiratory
fitness (an
objective
measure
of habitual
physical
activity) in
mid-life after
adjustment for
cardiovascular
risk factors.
The
United
States
Based on public
and/or private
health care
costs.
Participants’ cardiorespiratory fitness at
mid-life (mean age: 49 years) was assessed
from a treadmill test and measured in terms
of maximal metabolic equivalents (METs)
achieved, categorised into age- and sexspecific quintiles of fitness. The quintiles
were combined into three fitness groupings,
into which participants were assigned: low
fit (quintile 1), moderate fit (quintiles 2 and
3) and high fit (quintiles 4 and 5). Health care
costs were followed up later in old age for an
average period of 6.5 years, from the date of
initiating Medicare coverage until death or at
the end of follow-up. Average annual health
care costs were examined across fitness
groupings to determine possible associations.
Treadmill test data: Cooper
Center Longitudinal Study
Cost data: national health
insurance (Medicare),
Carrier, Durable Medical
Equipment, Home Health
Agency, Hospice, Center
for Medicare and Medicaid
Services
Direct
health
care costs
(USD)
1999–2009 Health care costs among those
aged 65 years and above were
significantly associated with
cardiorespiratory fitness in midlife. Compared to men with low
cardiorespiratory fitness, men with
high cardiorespiratory fitness had
significantly lower health care costs
(i.e. US$ 12 811 vs US$ 7 569).
The same trend applied to women
(US$ 10 029 vs US$ 6 056).
Average annual health care costs
were incrementally lower per MET
achieved (i.e. 6.8% and 6.7%
decrease in costs per MET achieved
in mid-life in men and women,
respectively). The associations
persisted when analysis was done
between cohorts who died during the
follow-up and those who survived.
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Carlson et al.
(2015)
To examine
the association
between
leisure-time
aerobic
physical activity
and health care
expenditures
with and
without
adjusting for
obesity status.
The
United
States
Based on public
and/or private
health care
costs.
Physical activity was assessed through the
frequency and, if applicable, the duration
of leisure-time activity participated in for at
least 10 minutes at a time in vigorous- and
light- or moderate-intensity activities. Minutes
of moderate-intensity equivalent activity were
calculated by counting 1 minute of vigorousintensity activity as 2 minutes of light- or
moderate-intensity activity. Participants were
classified into three activity levels: 1) active, 2)
insufficiently active and 3) inactive. Physical
activity levels were linked to individual health
care costs to determine mean and per cent
differences in expenditures. The sum of
the differences in costs was divided by the
total predicted expenditures for all groups to
determine the percentage of aggregate costs
associated with inadequate levels of physical
activity. Two models were used: with and
without adjustment for BMI.
Physical activity prevalence:
National Health Interview
Survey (2004–2010)
Cost data: Medical
Expenditure Panel Survey
(2006–2011)
Direct
health
care costs
(USD)
2006–2011 Higher levels of leisure-time aerobic
physical activity were significantly
associated with lower health care
costs. There was a mean difference
of 29.9% in annual per capita
costs between active and inactive
groups, as inactive individuals paid
an additional cost of US$ 1 437.
Insufficiently active individuals
paid an additional cost of US$ 713
compared to active individuals
(mean difference: 15.4%). After
adjusting for BMI, the means of
annual expenditure and per cent
differences decreased but remained
significant (26.6% or US$ 1 313
between active and inactive groups
and 12.1% or US$ 576 between
insufficiently active and inactive
groups). Before adjustment for
BMI, the percentage of aggregate
health care expenditures associated
with inadequate levels of physical
activity was 12.5% (US$ 131 billion)
and remained significant at 11.1%
(US$ 117 billion) after adjusting for
BMI. When adults with any reported
difficulty walking due to a health
problem were excluded from the
analysis, the percentage was 8.7%
(US$ 79 billion).
Appendix 1 71
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Carlson et al.
(2015)
To examine
the association
between
leisure-time
aerobic
physical activity
and health care
expenditures
with and
without
adjusting for
obesity status.
The
United
States
Based on public
and/or private
health care
costs.
Physical activity was assessed through the
frequency and, if applicable, the duration
of leisure-time activity participated in for at
least 10 minutes at a time in vigorous- and
light- or moderate-intensity activities. Minutes
of moderate-intensity equivalent activity were
calculated by counting 1 minute of vigorousintensity activity as 2 minutes of light- or
moderate-intensity activity. Participants were
classified into three activity levels: 1) active, 2)
insufficiently active and 3) inactive. Physical
activity levels were linked to individual health
care costs to determine mean and per cent
differences in expenditures. The sum of
the differences in costs was divided by the
total predicted expenditures for all groups to
determine the percentage of aggregate costs
associated with inadequate levels of physical
activity. Two models were used: with and
without adjustment for BMI.
Physical activity prevalence:
National Health Interview
Survey (2004–2010)
Cost data: Medical
Expenditure Panel Survey
(2006–2011)
Direct
health
care costs
(USD)
2006–2011 Higher levels of leisure-time aerobic
physical activity were significantly
associated with lower health care
costs. There was a mean difference
of 29.9% in annual per capita
costs between active and inactive
groups, as inactive individuals paid
an additional cost of US$ 1 437.
Insufficiently active individuals
paid an additional cost of US$ 713
compared to active individuals
(mean difference: 15.4%). After
adjusting for BMI, the means of
annual expenditure and per cent
differences decreased but remained
significant (26.6% or US$ 1 313
between active and inactive groups
and 12.1% or US$ 576 between
insufficiently active and inactive
groups). Before adjustment for
BMI, the percentage of aggregate
health care expenditures associated
with inadequate levels of physical
activity was 12.5% (US$ 131 billion)
and remained significant at 11.1%
(US$ 117 billion) after adjusting for
BMI. When adults with any reported
difficulty walking due to a health
problem were excluded from the
analysis, the percentage was 8.7%
(US$ 79 billion).
72 Assessing the economic costs of unhealthy diets and low physical activity Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Codogno et al.
(2015)
To analyse the
association
between
physical
inactivity
in different
domains and
public health
expenditures
and identify
whether
clustering
of physical
inactivity in
these domains
contributes
to increased
health care
costs.
Brazil Based on public
primary health
care costs.
Participants’ primary health care costs
were estimated in the last 12 months prior
to interview on physical activity. Individual
habitual physical activity was assessed in three
domains (i.e work, sports and leisure-time)
using the Baecke Questionnaire and scored.
Scores were categorised into quartiles: the
bottom quartile with scores ≤25, middle
quartiles with scores ≥25 but ≤75 and the high
quartile with scores ≥75. Participants were
analysed according to in how many domains
of physical activity they were categorised in
the bottom quartile (i.e. from 0 to 3 times). This
frequency of being in the bottom quartile was
correlated with annual expenditure data.
Cost data: Basic Healthcare
Unit, Sao Paolo, Brazil
Physical activity: study
interview
Direct
health
care costs
(USD)
2009 Lower physical activity (i.e. scores
in the bottom quartile) in two or all
of the domains was associated with
higher expenditures (odds ratio for
two domains: 1.75 and for three
domains: 2.12). Those in the bottom
quartile for all three domains had
the highest overall expenditures
(OR: 2.28). For specific domains,
lower physical activity at work and
in sports is associated with higher
health care expenditures related to
medicine discharge (OR for work:
1.58, OR for sport: 1.57). Lower
physical activity in leisure-time was
still associated with higher overall
expenditures (OR: 1.53).
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Idler et al.
(2015)
To analyse the
relationship
between
physical activity
and health care
and (parental)
productivity
costs among
children aged 9
to 12 years.
Germany Based on public
and/or private
health care and
parental work
absence costs.
Information on children’s physical activity and
frequency of health care utilisation and costs
were provided by parents in an interview (using
a self-administered questionnaire). Children
were grouped into two categories based on
weekly hours spent on moderate and vigorous
physical activity: <7 hours per week and ≥7
hours per week. Direct health care and indirect
productivity costs (i.e. from parental absence
at work) were examined across the two
physical activity groups to determine possible
associations.
Physical activity, frequency
of health care utilisation:
The German Infant Study
on the Influence of
Nutrition Intervention plus
Air Pollution and Genetics
on Allergy Development
(GINIplus) study and the
Influence of Lifestyle
Factors on Development of
the Immune System and
Allergies in East and West
Germany plus Air Pollution
and Genetics on Allergy
Development (LISAplus)
study
Direct medical unit costs,
number of days of parental
absence from work and
costs per day: Working
Group Methods in Health
Economic Evaluation (AG
MEG) national costing
guide, earlier studies
Direct
health
care and
indirect
productivity
costs (EUR)
2007 The mean annual cost for a child with
a higher level of physical activity (i.e.
≥7 hours per week) vs a child with
lower activity level (i.e. <7 hours per
week) were €392 vs €398 for direct
costs and €138 vs €127 for indirect
costs. However, there were no
statistically significant associations
noted between physical activity and
health care utilisation and costs
among children aged 9 to 12 years.
Different directions of estimates
were noticeable throughout the cost
components.
Appendix 1 73
Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Idler et al.
(2015)
To analyse the
relationship
between
physical activity
and health care
and (parental)
productivity
costs among
children aged 9
to 12 years.
Germany Based on public
and/or private
health care and
parental work
absence costs.
Information on children’s physical activity and
frequency of health care utilisation and costs
were provided by parents in an interview (using
a self-administered questionnaire). Children
were grouped into two categories based on
weekly hours spent on moderate and vigorous
physical activity: <7 hours per week and ≥7
hours per week. Direct health care and indirect
productivity costs (i.e. from parental absence
at work) were examined across the two
physical activity groups to determine possible
associations.
Physical activity, frequency
of health care utilisation:
The German Infant Study
on the Influence of
Nutrition Intervention plus
Air Pollution and Genetics
on Allergy Development
(GINIplus) study and the
Influence of Lifestyle
Factors on Development of
the Immune System and
Allergies in East and West
Germany plus Air Pollution
and Genetics on Allergy
Development (LISAplus)
study
Direct medical unit costs,
number of days of parental
absence from work and
costs per day: Working
Group Methods in Health
Economic Evaluation (AG
MEG) national costing
guide, earlier studies
Direct
health
care and
indirect
productivity
costs (EUR)
2007 The mean annual cost for a child with
a higher level of physical activity (i.e.
≥7 hours per week) vs a child with
lower activity level (i.e. <7 hours per
week) were €392 vs €398 for direct
costs and €138 vs €127 for indirect
costs. However, there were no
statistically significant associations
noted between physical activity and
health care utilisation and costs
among children aged 9 to 12 years.
Different directions of estimates
were noticeable throughout the cost
components.
74 Assessing the economic costs of unhealthy diets and low physical activity Author/s Study objective Country Principal approach Methodology Data Type of costs Base-year Principal findings
PHYSICAL ACTIVITY
Krueger et al.
(2015)
To identify
the economic
burden of
physical
inactivity and
the potential
reduction in
economic costs
if all provinces
in Canada
achieved a
prevalence
rate of physical
inactivity
equivalent
to that of the
province with
the lowest rate.
Canada Disease-based:
malignant
and other
neoplasms,
endocrine,
nutritional
and metabolic
diseases,
cardiovascular,
respiratory
infections,
digestive and
musculoskeletal
diseases.
PAFs were derived based on sex-specific RRs
from an earlier study and applied to disease
costs. For indirect costs of physical inactivity,
the ratio of direct and indirect costs for each
diagnostic category from an earlier study were
applied to PAF-based direct costs to generate
equivalent indirect cost data. A second set of
PAFs was derived using the prevalence rate
of physical inactivity in British Columbia (the
province with the lowest prevalence rate) and
applied to populations living in other provinces
to determine potential reductions in the
economic burden.
Physical activity prevalence:
Canadian Community
Health Survey (2012)
RRs: earlier study for 2001
Cost data: National Health
Expenditure Database,
Economic Burden of Illness
in Canada (EBIC) online tool
(2008)
Ratio of direct to indirect
costs: EBIC (1998)
Direct
health
care and
indirect
productivity
costs (CAD)
2013 Economic costs associated with
physical inactivity in 2013 (using
actual provincial prevalence rates)
were C$ 10.8 billion (males:
C$ 5.2 billion, females: C$ 5.6 billion;
direct costs: C$ 3.3 billion, indirect
costs: C$ 7.5 billion). If all provinces
were to achieve the low prevalence
rate of physical inactivity in British
Columbia, 14% (C$ 1.5 billion) of the
costs could potentially be avoided,
reducing the attributable costs to
C$ 9.3 billion.
DIET AND PHYSICAL ACTIVITY
Popkin et al.
(2006)
To determine
the economic
costs of
unhealthy diets
and physical
inactivity
through its
direct effects on
disease risks as
well as indirect
effects, through
overweight and
obesity.
China Disease-based:
coronary heart
disease, type
2 diabetes,
hypertension,
stroke,
cancers of the
breast, colon,
oesophagus,
endometrium,
lungs, stomach
and bladder.
PAFs related directly to unhealthy diets and
physical inactivity were calculated based on
RRs from earlier meta-analyses of studies
and applied to disease costs. PAFs related to
overweight and obesity were also calculated, of
which the proportion attributable to unhealthy
diets and physical inactivity was also obtained
and applied to disease costs.
Physical activity, overweight
and obesity prevalence:
China Health and Nutrition
Survey
RR: earlier meta-analyses
of studies
Cost data: National Survey
of Health Services for China
(1998)
Direct
health
care costs
(USD)
2000 and
2025
In the year 2000 unhealthy diets
accounted for US$ 4.2 billion
of direct health care costs
(US$ 3.4 billion for direct effects on
disease risks and US$ 0.83 billion for
indirect effects through overweight
and obesity). Physical inactivity
accounted for US$ 1.7 billion
(US$ 1.3 billion for direct effects
and US$ 0.35 billion for indirect
effects through overweight and
obesity). Costs predicted for 2025
did not increase considerably at
US$ 6.1 billion (US$ 3.9 billion for
unhealthy diets and US$ 2.2 billion
for physical inactivity).
Appendix 2
Average annual currency exchange
rates used to identify the EUR
equivalent of costs in studies
Cost values in non-European currencies were converted into Euros, using the
average of yearly currency exchange rates from the OANDA website (https://
www.oanda.com/currency/average) during the base year/s for cost estimation in
each study. The following table shows the average exchange rate used in each of
the studies, according to the base year/s for cost estimation.
Study Base year/s for
cost estimation
Original
currency used
Equivalent to EUR
per one unit
of original currency
Rayner & Scarborough (2005); Allender et al. (2007) 2002 GBP 1.59
Scarborough et al. (2011) 2006–2007 GBP 1.47
Lo et al. (2013) 1999–2006 TWD 0.03
Collins et al. (2011) 2002–2006 AUD 0.59
Peeters et al. (2014) 2010 AUD 0.69
Doidge et al. (2012) 2010–2011 AUD 0.72
Daviglus et al. (2005) 1984–2000 USD 0.96
Kuriyama et al. (2004) 1995–2001 USD 1.00
Anderson et al. (2005) 1996–1999 USD 0.90
Bland et al. (2009) 1999–2000/2001 USD 1.03
Bachmann et al. (2015) 1999–2009 USD 0.87
Garrett et al. (2004); Popkin et al. (2006) 2000 USD 1.08
Wang et al. (2005) 2001–2002 USD 1.09
Carlson et al. (2015) 2006–2011 USD 0.73
Zhang & Chaaban (2012) 2007 USD 0.73
Maresova (2014); Kruk (2014) 2008 CZK 0.04
Alter et al. (2012) 1994–2007 CAD 0.91
Katzmarzyk et al. (2000) 1999 CAD 0.63
Katzmarzyk (2011); Janssen (2012) 2009 CAD 0.63
Krueger et al. (2015) 2013 CAD 0.73
Note: Average annual exchange rates from USD to EUR in 1984 to 2000 only reflect the average of rates
from 1990 to 2000 as earlier figures were not available. Costs in USD reported by Martinson et al. (2003)
and Wang et al. (2004) were not converted into EUR as average exchange rates during the base year/s for cost
estimation were not available from the sources used.
Appendix 3
Mean intake of AHEI food groups
in France, Germany, Italy, Spain
and the United Kingdom
Country Dietary component Mean intake
(grams/day)
Mean intake (cups or
ounces/day)d
France
(n = 23 048
consumers;
44% of surveyed
population)
2007 Individual and
National Study on
Food Consumption
Vegetablesa 20.99 0.09
Fruitb 10.97 0.05
Whole grains 12.2
Sugar-sweetened beverages and
fruit juice 118.84 4.19
Nuts 1.13 0.04
Processed meat 37.59 1.33
Trans-Fat 63.5
Long-chain (n-3) fats (EPA + DHA) 21.39
Polyunsaturated fatty acid 10.92
Sodium 1.5
Alcoholc 78.08 2.75
Germany
(n = 54 710
consumers;
25% of surveyed
population)
2007 National
Nutrition Survey
Vegetablesa 27.31 0.12
Fruitb 13.66 0.06
Whole grains 0.49
Sugar-sweetened beverages and
fruit juice 341.65 12.05
Nuts 2.95 0.10
Processed meat 49.82 1.76
Trans-Fat 68.57
Long-chain (n-3) fats (EPA + DHA) 13.67
Polyunsaturated fatty acid 2.93
Sodium 0.01
Alcoholc 52.02 1.83
Appendix 3 77
Country Dietary component Mean intake
(grams/day)
Mean intake (cups or
ounces/day)d
Italy
(n = 18 035
consumers;
33.9% of surveyed
population)
2005–2006 National
Food Consumption
Survey
Vegetablesa 45.62 0.20
Fruitb 2.89 0.01
Whole grains 35.29
Sugar-sweetened beverages and
fruit juice 56.03 1.98
Nuts 1.06 0.04
Processed meat 29.87 1.05
Trans-Fat 30.23
Long-chain (n-3) fats (EPA + DHA) 31.05
Polyunsaturated fatty acid 36.63
Sodium 0.01
Alcoholc 70.88 2.50
Spain
(n = 6 940
consumers;
34.6% of surveyed
population)
2009 Spanish Agency
for Food Safety Survey
Vegetablesa 40.17 0.18
Fruitb 6.35 0.03
Whole grains 6.84
Sugar-sweetened beverages and
fruit juice 130.65 4.61
Nuts 1.99 0.07
Processed meat 48.97 1.73
Trans-Fat 48.31
Long-chain (n-3) fats (EPA + DHA) 57.31
Polyunsaturated fatty acid
Sodium
Alcoholc 1.01
United Kingdom
(n = 7 046
consumers;
26.5% of surveyed
population)
2008 National Diet and
Nutrition Survey
Vegetablesa 6.96 0.03
Fruitb 10.81 0.05
Whole grains 2.80
Sugar-sweetened beverages and
fruit juice 224.12 7.91
Nuts 1.07 0.04
Processed meat 30.1 1.06
Trans-Fat 33.48
Long-chain (n-3) fats (EPA + DHA) 21.1
Polyunsaturated fatty acid 1.46
Sodium 0.08
Alcoholc 86.7 3.06
Source: European Food and Safety Authority, 2015.
Notes: a Vegetables refers to green leafy vegetables; b fruit refers to berries and small fruits; c alcohol refers to
either beer and beer-like beverages, wines or liquors, whichever has the highest mean intake; d converted into
units used in the AHEI: 240 g = 1 cup for liquids, 226.72 g = 1 cup for solids and 28.35 g = 1 oz for both
solids and liquids.
Appendix 4
Quality of studies that have
quantified the association
between unhealthy diets and low
physical activity and diabetes
Quality criterion as
defined by Al Tunaji
et al. (2014)
Quality assessment
Montonen et al. (2005) Laaksonen et al. (2009) Li et al. (2015)
The exposure (risk
factor) is clearly
defined
An unhealthy diet
was defined as a
‘conservative’ dietary
pattern characterised by
consumption of butter,
potatoes and whole milk
Physical inactivity was
defined as less than 30
minutes of occasional
or regular leisure-time
exercise per day
Poor diet was defined as a
dietary score in the 2010
alternate healthy eating
index belonging to the third
to fifth quintiles (i.e. score
of <67%).
Physical inactivity was
defined as those not
meeting 150 minutes per
week of moderate-intensity
exercise or 75 minutes per
week of vigorous-intensity
exercise
The exposure was
measured objectively
No; the exposure was
measured using a
one-year dietary history
interview
No; the exposure
was self-reported in
a health interview or
a self-administered
questionnaire
No; the exposure was
measured using a food
frequency questionnaire
and interviews on lifestyle
habits and medical history
The outcome
(disease) was clearly
defined
The outcome was defined
as type 2 diabetes; no
diagnostic criteria were
presented
The outcome was defined
as type 2 diabetes
according to the WHO
diagnostic criteria
The outcome was defined
as type 2 diabetes,
according to national
diagnostic criteria
The outcome was
ascertained by
objective measures
or, if self-reported,
confirmed by other
measures
The outcomes were
identified from a
nationwide registry of
patients receiving drug
reimbursement (including
for diabetes). Study
participants were linked
to this register by unique
social security codes
The outcomes were
identified from a
central register of all
patients receiving drug
reimbursement (including
for diabetes). Study
participants were linked
to this register by unique
social security codes
The outcomes were selfreported and confirmed by
a validated supplementary
questionnaire (validated
through hospital records)
Appendix 4 79
Quality criterion as
defined by Al Tunaji
et al. (2014)
Quality assessment
Montonen et al. (2005) Laaksonen et al. (2009) Li et al. (2015)
The analysis was
based on raw data
from a prospective or
cohort study
Yes Yes Yes
The follow-up time
was provided
23 years 10 years 20–30 years
Full adjustments
were made
Adjustments were made
for age, sex, body mass
index, energy intake,
smoking status, family
history of diabetes,
geographic area,
serum cholesterol and
hypertension
Adjustments were
made for age, sex, other
lifestyle risk factors (e.g.
BMI, smoking, alcohol
consumption and serum
vitamin D) or components
of metabolic syndrome
(BMI, blood pressure,
serum triglyceride levels,
serum HDL cholesterol,
fasting glucose)
Adjustments were made for
sex, ethnicity (Caucasian
yes/no), marriage status,
living status (alone yes/no),
family history of diabetes,
menopausal status (pre- or
post-menopausal; never,
past or current menopausal
hormone use), and for
other lifestyle risk factors
assessed in the study (i.e.
BMI, smoking status, daily
alcohol consumption)
Appendix 5
Annual incidence rate of
diabetic complications from the
UKPDS Outcomes Model 2
Complication Annual incidence rate (%)
First myocardial infarction 1.13
Second myocardial infarction 0.19
First stroke 0.56
Second stroke 0.09
Congestive heart failure 0.39
Ischaemic heart disease 0.83
First amputation 0.19
Second amputation 0.06
Retinopathy/Blindness 0.3
Renal failure 0.13
Ulcer 0.11
Source: Hayes et al., 2013.
Appendix 6
Estimated diabetes complicationrelated costs in 2020 attributable
to unhealthy diets and low
physical activity in 2015
Country Complication
Estimated number
of incident cases
of diabetes-related
complications in
2020 attributable to
the risk factor
Estimated
per patient
costs in
2020 (€)
Estimated diabetes
complication-related
costs in 2020 (€)
Unhealthy
diets
Low
physical
activity
Unhealthy
diets
Low
physical
activity
France
First myocardial infarction 51 86 19 097 981 441 1 635 734
Second myocardial infarction 9 14 19 097 165 021 275 035
First stroke 25 42 14 396 366 649 611 081
Second stroke 4 7 14 396 58 926 98 210
Congestive heart failure 18 30 4 838 85 813 143 021
Ischaemic heart disease 38 63 3 200 120 795 201 325
First amputation 9 14 39 191 338 657 564 429
Second amputation 3 5 39 191 106 944 178 241
Retinopathy/Blindness 14 23 468 6 385 10 642
Renal failure 6 10 69 186 409 055 681 759
Ulcer 5 8 1 399 6 999 11 665
Total 181 302 2 646 685 4 411 142
82 Assessing the economic costs of unhealthy diets and low physical activity
Country Complication
Estimated number
of incident cases
of diabetes-related
complications in
2020 attributable to
the risk factor
Estimated
per patient
costs in
2020 (€)
Estimated diabetes
complication-related
costs in 2020 (€)
Unhealthy
diets
Low
physical
activity
Unhealthy
diets
Low
physical
activity
Germany
First myocardial infarction 55 110 22 465 1 231 448 2 463 150
Second myocardial infarction 9 18 22 465 207 058 414 158
First stroke 27 54 29 032 788 672 1 577 506
Second stroke 4 9 29 032 126 751 253 528
Congestive heart failure 19 38 9 030 170 838 341 711
Ischaemic heart disease 40 81 5 002 201 397 402 836
First amputation 9 18 33 068 304 784 609 632
Second amputation 3 6 33 068 96 248 192 515
Retinopathy/Blindness 15 29 15 650 227 754 455 556
Renal failure 6 13 86 975 548 490 1 097 094
Ulcer 5 11 1 312 7 001 14 003
Total 193 386 3 910 441 7 821 688
Italy
First myocardial infarction 17 120 12 580 215 079 1 505 553
Second myocardial infarction 3 20 12 580 36 164 253 146
First stroke 8 59 5 994 50 786 355 502
Second stroke 1 10 5 994 8 162 57 134
Congestive heart failure 6 41 3 363 19 844 138 908
Ischaemic heart disease 13 88 2 091 26 259 183 810
First amputation 3 20 9 266 26 637 186 459
Second amputation 1 6 9 266 8 412 58 882
Retinopathy/Blindness 5 32 5 021 22 790 159 532
Renal failure 2 14 39 220 77 142 539 993
Ulcer 2 12 694 1 155 8 085
Total 60 422 492 429 3 447 004
Appendix 6 83
Country Complication
Estimated number
of incident cases
of diabetes-related
complications in
2020 attributable to
the risk factor
Estimated
per patient
costs in
2020 (€)
Estimated diabetes
complication-related
costs in 2020 (€)
Unhealthy
diets
Low
physical
activity
Unhealthy
diets
Low
physical
activity
Spain
First myocardial infarction 12 73 22 638 273 460 1 641 273
Second myocardial infarction 2 12 22 638 45 980 275 966
First stroke 6 36 5 447 32 608 195 709
Second stroke 1 6 5 447 5 241 31 453
Congestive heart failure 4 25 5 834 24 323 145 981
Ischaemic heart disease 9 53 2 592 22 998 138 031
First amputation 2 12 17 366 35 272 211 698
Second amputation 1 4 17 366 11 139 66 852
Retinopathy/Blindness 3 19 3 669 11 766 70 621
Renal failure 1 8 36 679 50 973 305 932
Ulcer 1 7 2 258 2 655 15 936
Total 43 255 516 414 3 099 453
United
Kingdom
First myocardial infarction 27 94 8 823 236 089 826 313
Second myocardial infarction 4 16 8 823 39 696 138 938
First stroke 13 46 5 131 68 041 238 144
Second stroke 2 7 5 131 10 935 38 273
Congestive heart failure 9 32 48 330 446 337 1 562 180
Ischaemic heart disease 20 69 39 119 768 860 2 691 012
First amputation 4 16 14 114 63 502 222 256
Second amputation 1 5 14 114 20 053 70 186
Retinopathy/Blindness 7 25 1 890 13 427 46 993
Renal failure 3 11 63 949 196 861 689 012
Ulcer 3 9 4 030 10 497 36 741
Total 94 330 1 874 299 6 560 047
Appendix 7
Number of incident type 2
diabetes cases attributable to
unhealthy diets and low physical
activity in each country by sex
Country
Number of type 2 diabetes cases attributable to the risk factor
Unhealthy diets Low physical activity
Males Females Total Males Females Total
France 2 507 2 041 4 548 4 178 3 402 7 580
Germany 2 674 2 177 4 851 5 348 4 355 9 703
Italy 834 679 1 513 5 838 4 753 10 591
Spain 589 480 1 069 3 536 2 880 6 416
United Kingdom 1 305 1 063 2 368 4 568 3 720 8 288
Appendix 8
Average percentage of annual
incident type 2 diabetes cases in
the United Kingdom by sex and
five-year age group, 1991–2010
Age range
Average percentage of annual total
incident type 2 diabetes cases
Males (%) Females (%)
0–4 0.16 0.14
5–9 0.13 0.14
10–14 0.13 0.16
15–19 0.25 0.46
20–24 0.32 0.79
25–29 0.46 0.97
30–34 0.77 1.19
35–39 1.30 1.61
40–44 2.39 2.33
45–49 3.94 3.40
50–54 5.90 5.00
55–59 8.04 7.05
60–64 10.79 9.56
65–69 12.57 11.61
70–74 13.31 12.95
75–79 13.28 13.61
80–84 11.64 12.42
85–89 9.25 10.56
90+ 5.38 6.03
Appendix 9
Estimated number of diabetes cases
attributable to unhealthy diets and
low physical activity by sex and fiveyear age group in France, Germany,
Italy, Spain and the United Kingdom
Country Agegroup
Number of diabetes cases attributable to the risk factor
Unhealthy diets Low physical activity
Males Females Total Males Females Total
France
0–4 4 3 7 7 5 12
5–9 3 3 6 5 5 10
10–14 3 3 7 5 5 10
15–19 6 9 16 10 16 26
20–24 8 16 24 13 27 40
25–29 12 20 31 19 33 52
30–34 19 24 44 32 40 73
35–39 33 33 65 54 55 109
40–44 60 48 107 100 79 179
45–49 99 69 168 165 116 280
50–54 148 102 250 247 170 417
55–59 202 144 345 336 240 576
60–64 270 195 466 451 325 776
65–69 315 237 552 525 395 920
70–74 334 264 598 556 441 997
75–79 333 278 611 555 463 1 018
80–84 292 254 545 486 423 909
85–89 232 216 447 386 359 746
90+ 135 123 258 225 205 430
TOTAL 2 507 2 041 4 548 4 179 3 401 7 580
Appendix 9 87
Country Agegroup
Number of diabetes cases attributable to the risk factor
Unhealthy diets Low physical activity
Males Females Total Males Females Total
Germany
0–4 4 3 7 9 6 15
5–9 3 3 7 7 6 13
10–14 3 3 7 7 7 14
15–19 7 10 17 13 20 33
20–24 9 17 26 17 34 52
25–29 12 21 33 25 42 67
30–34 21 26 46 41 52 93
35–39 35 35 70 70 70 140
40–44 64 51 115 128 101 229
45–49 105 74 179 211 148 359
50–54 158 109 267 316 218 533
55–59 215 153 368 430 307 737
60–64 289 208 497 577 416 993
65–69 336 253 589 672 506 1 178
70–74 356 282 638 712 564 1 276
75–79 355 296 651 710 593 1 303
80–84 311 270 582 623 541 1 163
85–89 247 230 477 495 460 955
90+ 144 131 275 288 263 550
TOTAL 2 674 2 177 4 851 5 349 4 354 9 703
Italy
0–4 1 1 2 9 7 16
5–9 1 1 2 8 7 14
10–14 1 1 2 8 8 15
15–19 2 3 5 15 22 36
20–24 3 5 8 19 38 56
25–29 4 7 10 27 46 73
30–34 6 8 15 45 57 102
35–39 11 11 22 76 77 152
40–44 20 16 36 140 111 250
45–49 33 23 56 230 162 392
50–54 49 34 83 344 238 582
55–59 67 48 115 469 335 804
60–64 90 65 155 630 454 1 084
65–69 105 79 184 734 552 1 286
70–74 111 88 199 777 616 1 393
75–79 111 92 203 775 647 1 422
80–84 97 84 181 680 590 1 270
85–89 77 72 149 540 502 1 042
90+ 45 41 86 314 287 601
TOTAL 834 679 1 513 5 838 4 752 10 591
88 Assessing the economic costs of unhealthy diets and low physical activity
Country Agegroup
Number of diabetes cases attributable to the risk factor
Unhealthy diets Low physical activity
Males Females Total Males Females Total
Spain
0–4 1 1 2 6 4 10
5–9 1 1 1 5 4 9
10–14 1 1 2 5 5 9
15–19 1 2 4 9 13 22
20–24 2 4 6 11 23 34
25–29 3 5 7 16 28 44
30–34 5 6 10 27 34 61
35–39 8 8 15 46 46 92
40–44 14 11 25 85 67 152
45–49 23 16 40 139 98 237
50–54 35 24 59 209 144 353
55–59 47 34 81 284 203 487
60–64 64 46 109 382 275 657
65–69 74 56 130 445 334 779
70–74 78 62 141 471 373 844
75–79 78 65 144 470 392 862
80–84 69 60 128 412 358 769
85–89 55 51 105 327 304 631
90+ 32 29 61 190 174 364
TOTAL 589 480 1 069 3 537 2 879 6 416
United
Kingdom
0–4 2 1 3 7 5 13
5–9 2 1 3 6 5 11
10–14 2 2 4 6 6 12
15–19 3 5 8 11 17 29
20–24 4 8 13 15 29 44
25–29 6 10 16 21 36 57
30–34 10 13 23 35 44 79
35–39 17 17 34 59 60 119
40–44 31 25 56 109 87 196
45–49 51 36 88 180 126 306
50–54 77 53 130 270 186 456
55–59 105 75 180 367 262 630
60–64 141 102 242 493 356 849
65–69 164 123 287 574 432 1 006
70–74 174 138 311 608 482 1 090
75–79 173 145 318 607 506 1 113
80–84 152 132 284 532 462 994
85–89 121 112 233 423 393 815
90+ 70 64 134 246 224 470
TOTAL 1 305 1 063 2 368 4 569 3 719 8 288
Appendix 10
Estimated number of diabetes cases
attributable to unhealthy diets and
low physical activity who are expected
to be in and out of the formal labour
force by five-year age group
Country
Working
age
range
Average
annual
labour force
participation
rate (%)
Number of diabetes cases attributable to the risk factor
Unhealthy diets Low physical activity
Total
In the
labour
force
Out of the
labour
force
Total
In the
labour
force
Out of the
labour
force
France
15–19 14.94 16 2 14 26 4 22
20–24 60.90 24 15 9 40 25 16
25–29 86.52 31 27 4 52 45 7
30–34 87.68 44 38 6 73 64 9
35–39 88.98 65 58 7 109 97 12
40–44 89.58 107 96 11 179 160 19
45–49 88.59 168 149 19 280 248 32
50–54 83.87 250 210 40 417 349 67
55–59 62.14 345 215 131 576 358 218
60–64 17.69 466 82 383 776 137 639
TOTAL 1 517 893 624 2 528 1 488 1 041
Germany
15–19 30.16 17 5 12 33 10 23
20–24 70.17 26 18 8 52 36 15
25–29 81.62 33 27 6 67 55 12
30–34 86.16 46 40 6 93 80 13
35–39 87.88 70 61 8 140 123 17
40–44 89.64 115 103 12 229 206 24
45–49 88.99 179 160 20 359 319 40
50–54 85.01 267 227 40 533 453 80
55–59 74.50 368 275 94 737 549 188
60–64 38.10 497 189 307 993 378 615
TOTAL 1 618 1 105 513 3 236 2 209 1 027
90 Assessing the economic costs of unhealthy diets and low physical activity
Country
Working
age
range
Average
annual
labour force
participation
rate (%)
Number of diabetes cases attributable to the risk factor
Unhealthy diets Low physical activity
Total
In the
labour
force
Out of the
labour
force
Total
In the
labour
force
Out of the
labour
force
Italy
15–19 12.48 5 1 5 36 5 32
20–24 49.95 8 4 4 56 28 28
25–29 70.99 10 7 3 73 52 21
30–34 79.36 15 12 3 102 81 21
35–39 80.59 22 18 4 152 123 30
40–44 79.73 36 29 7 250 200 51
45–49 77.45 56 43 13 392 303 88
50–54 70.71 83 59 24 582 412 170
55–59 50.48 115 58 57 804 406 398
60–64 22.50 155 35 120 1 084 244 840
TOTAL 505 265 240 3 532 1 852 1 680
Spain
15–19 23.57 4 1 3 22 5 17
20–24 63.32 6 4 2 34 22 12
25–29 85.63 7 6 1 44 38 6
30–34 87.39 10 9 1 61 54 8
35–39 85.69 15 13 2 92 79 13
40–44 83.66 25 21 4 152 127 25
45–49 80.22 40 32 8 237 190 47
50–54 73.15 59 43 16 353 258 95
55–59 60.20 81 49 32 487 293 194
60–64 35.39 109 39 71 657 232 424
TOTAL 357 216 140 2 140 1 298 841
United
Kingdom
15–19 52.83 8 4 4 28 15 13
20–24 74.60 13 9 3 44 33 11
25–29 84.25 16 14 3 57 48 9
30–34 84.52 23 19 4 79 67 12
35–39 84.87 34 29 5 119 101 18
40–44 86.02 56 48 8 196 168 27
45–49 85.97 88 75 12 306 263 43
50–54 83.29 130 108 22 456 379 76
55–59 71.93 180 129 50 629 452 177
60–64 44.74 242 108 134 849 380 469
TOTAL 790 545 245 2 763 1 907 856
Appendix 11
Estimated total number of working
years lost due to work disability, early
retirement and premature death among
incident type 2 diabetes cases at working
age that can be attributed to unhealthy
diets and low physical activity and
who are expected to be in the formal
labour force in France, Germany, Italy,
Spain and the United Kingdom, 2020
Country Risk factor
Estimated number of incident
type 2 diabetes cases aged 35–60
attributable to the risk factor (in
the labour force)
Total number of working years lost
due to
work
disability
early
retirement
premature
death
France
Unhealthy diets 893 80 625 250
Low physical activity 1 488 134 1 041 417
Germany
Unhealthy diets 1 105 99 773 309
Low physical activity 2 209 199 1 547 619
Italy
Unhealthy diets 265 24 185 74
Low physical activity 1 852 167 1 297 519
Spain
Unhealthy diets 216 19 151 61
Low physical activity 1 298 117 909 364
United
Kingdom
Unhealthy diets 545 49 382 153
Low physical activity 1 907 172 1 335 534
ISBN 9289050425
9 7 8 9 2 8 9 0 5 0 4 2 5
Health Policy Series No. 47
www.healthobservatory.eu
Health Policy
Series
47
AN EVIDENCE REVIEW AND PROPOSED FRAMEWORK
47
Assessing the
economic costs of
unhealthy diets and
low physical activity
Christine Joy Candari
Jonathan Cylus
Ellen Nolte
An evidence review and proposed framework
Unhealthy diets and low physical activity contribute to many chronic diseases and
disability; they are responsible for some 2 in 5 deaths worldwide and for about 30% of
the global disease burden. Yet surprisingly little is known about the economic costs that
these risk factors cause, both for health care and society more widely.
This study pulls together the evidence about the economic burden that can be linked to
unhealthy diets and low physical activity and explores
• How definitions vary and why this matters
• The complexity of estimating the economic burden and
• How we can arrive at a better way to estimate the costs of an unhealthy diet and low
physical activity, using diabetes as an example
The review finds that unhealthy diets and low physical activity predict higher health care
expenditure, but estimates vary greatly. Existing studies underestimate the true economic
burden because most only look at the costs to the health system. Indirect costs caused
by lost productivity may be about twice as high as direct health care costs, together
accounting for about 0.5% of national income.
The study also tests the feasibility of using a disease-based approach to estimate the
costs of unhealthy diets and low physical activity in Europe, projecting the total economic
burden associated with these two risk factors as manifested in new type 2 diabetes cases
at €883 million in 2020 for France, Germany, Italy, Spain and the United Kingdom alone.
The ‘true’ costs will be higher, as unhealthy diets and low physical activity are linked to
many more diseases.
The study’s findings are a step towards a better understanding of the economic burden
that can be associated with two key risk factors for ill health and they will help policymakers in setting priorities and to more effectively promoting healthy diets and physical
activity.
The authors
Christine Joy Candari was an independent consultant at the time of writing this report.
She is currently Chief Consultant for Health Research, U Consult Us Inc, Manila, The
Philippines
Jonathan Cylus is Research Fellow, European Observatory on Health Systems and
Policies, London School of Economics and Political Science
Ellen Nolte is Head of London Hubs, European Observatory on Health Systems and Policies
DIETS AND LOW PHYSICAL ACTIVITY
ASSESSING THE ECONOMIC COSTS OF UNHEALTHY
Christine Joy Candari, Jonathan Cylus, Ellen Nolte
Cover_WHO_nr47_Mise en page 1 13/06/17 13:21 Page 1 |