Relationship between the intake of total flavonoid and ...

93
Relationship between the intake of total flavonoid and flavan-3-ols, and 5-year cardiovascular disease risk scores in New Zealand adults Kirsten Sim A thesis submitted in fulfillment of the requirements for the degree of Masters in Dietetics At the University of Otago Dunedin, New Zealand June 2013

Transcript of Relationship between the intake of total flavonoid and ...

Relationship between the intake of total flavonoid

and flavan-3-ols, and 5-year cardiovascular disease

risk scores in New Zealand adults

Kirsten Sim

A thesis submitted in fulfillment of the requirements for the degree of

Masters in Dietetics

At the University of Otago

Dunedin, New Zealand

June 2013

! i!

Abstract

Objective: The aim of the study was to expand the New Zealand Food Composition

Database (NZFCD) to include flavonoid data, in order to determine the relationship

between the intake of (a) total flavonoids and (b) flavan-3-ols, and 5-year cardiovascular

disease (CVD) risk score in New Zealand adults.

Design: The present study used data from the New Zealand Adult Nutrition Survey

(NZANS) 08/09 - a national population-based nutrition survey, conducted in a sample of

4271 participants. Flavonoid and Flavan-3-ol intake was assessed using 24-hour food

recalls. Data on blood pressure, lipid status, diabetes status (HbA1c), smoking status,

ethnicity, age and gender were used to determine 5-year CVD risk scores. Each food in the

FOODfiles dataset of the NZFCD was updated using the flavonoid databases: USDA and

Phenol explorer. To determine the relationship between flavonoid intake and CVD risk,

logistic regression was used to investigate relationships between categories of X intake

with categories of CVD risk.

Results: Those in the high total flavonoid intake category were less likely to have a high

CVD risk score than those with a low total flavonoid intake, after adjustment for age, body

mass index (BMI), sex, ethnicity and New Zealand Deprivation Index (NZdep). Similarly,

those with a high intake of flavan-3-ols were less likely to have a high CVD risk score.

Those with a low intake of flavan-3-ols, were more likely to have a high CVD risk score.

Foods that contributed most to the quantitative intake of flavan-3-ols were sugar/sweets,

fruit, alcoholic and non-alcoholic beverages.

Conclusions: New Zealand adults with a higher intake of total flavonoids and in particular

flavan-3-ols are likely to have a lower CVD risk score. Therefore, consuming foods higher

in flavan-3-ols, such as cocoa products, fruit and tea, may help to reduce the risk of

developing CVD, as long as a balanced diet is followed.

! ii!

Preface

The flavonoid database was developed jointly by the candidate and Carly Fry (Master of

Dietetics student). We were jointly responsible for:

• Matching foods from the FOODfiles list of 2710 foods to that from the Flavonoid

database from the United States Department of Agriculture (USDA), which

contained flavonoid values for approximately 500 foods. If foods could not be

matched to this database, they were matched to an alternate database named

Phenol Explorer, which also contained flavonoid values for approximately 500

foods.

• Developing composites for foods in which no match could be made. We used

packet ingredients and website recipes to compile a composite. Refer to the

methods section for more detail on the development of composites.

• Determining which foods were cooked and/or processed, so retention factors could

be applied due to loss of flavonoids from the raw to cooked food.

• Determining foods that needed to be concentrated.

• Checking energy content of matched foods to ensure they were a close enough

match, and making changes where necessary.

The candidate was also responsible for interpreting the statistical analyses carried out by

Dr Jill Haszard, and for writing the thesis. Cecilia Sam was responsible for coding the

flavonoid database, and thus matching flavonoid values to all foods and composites. She

was also responsible for developing the coding for CVD risk score, so flavonoid intake

could be compared to CVD risk.

! iii!

Acknowledgements

I would like to thank the following people for their assistance:

i. Dr Paula Skidmore and Liz Fleming, my two supervisors, for their guidance,

suggestions and support throughout the project.

ii. Dr Jill Haszard for her input in the statistical methods and analysis of my data.

iii. Cecilia Sam, for coding the database and making it possible to generate results.

iv. The participants from the NZANS, and all those from the University of Otago

Department of Human Nutrition who helped conduct this study.

v. Charlie Blakey and Liz Fleming, who were responsible for updating the web based

program Kai-culator with flavonoid values, in order to determine flavonoid intake

of the NZANS participants. Without their input none of this would have been

possible.

! iv!

1 Table of Contents

1 Introduction ....................................................................................................... 1

2 Literature Review .............................................................................................. 3

2.1 Introduction ................................................................................................... 3

2.1.1 CVD Risk Assessment Tools ................................................................. 4

2.1.2 CVD Risk Assessment in New Zealand................................................. 6

2.1.3 CVD and Diet ........................................................................................ 6

2.2 Flavonoids ................................................................................................... 14

2.2.1 Flavonoids and Chronic Disease .......................................................... 15

2.2.2 Flavonoids Mechanisms of action........................................................ 16

2.3 Flavan-3-ol .................................................................................................. 16

2.3.1 Flavan-3-ol and CVD ........................................................................... 17

2.3.2 Flavan-3-ol Mechanisms of Action...................................................... 21

2.4 Conclusions ................................................................................................. 35

3 Objective statement .......................................................................................... 36

3.1 Hypothesis ................................................................................................... 36

4 Participants and Methods ................................................................................ 37

4.1 Study Design ............................................................................................... 37

4.2 New Zealand Adult Nutrition Survey 2008/2009 Methods ........................ 37

4.2.1 Participants .......................................................................................... 37

4.2.2 24-Hour Food Recalls .......................................................................... 37

4.2.3 Matching to Nutrient Data.................................................................... 38

4.2.4 Anthropometry ..................................................................................... 39

4.2.5 Blood Pressure...................................................................................... 39

4.2.6 Blood Samples...................................................................................... 39

! v!

4.2.7 Diabetes ............................................................................................... 40

4.2.8 New Zealand Index of Deprivation ...................................................... 40

4.2.9 Ethnic Groups....................................................................................... 40

4.3 CVD Risk Score ............................................................................................. 41

4.4 Flavonoid Database Development............................................................... 43

4.4.1 Flavonoid Databases............................................................................. 44

4.4.2 Matching FOODfiles to USDA............................................................ 44

4.4.3 Matching FOODfiles to Phenol Explorer............................................. 44

4.4.4 Retention Factors.................................................................................. 44

4.4.5 Flavonoid Classes and Subclasses........................................................ 45

4.4.6 Composite Development ...................................................................... 45

4.4.7 Retention factors for composites ......................................................... 48

4.4.8 Adjustments for Concentrated Foods ................................................... 50

4.4.9 Flavonoid Calculations ........................................................................ 50

4.5 Participant Categories.................................................................................. 50

4.6 Statistical Methods ..................................................................................... 51

5 Results .............................................................................................................. 53

5.1 Response Rate and Data Used for Analyses ............................................... 53

5.2 Sample Characteristics ............................................................................... 53

5.3 Differences in Flavonoid and Flavan-3-ol Intake Between Age

Categories, BMI, Sex, Prioritised Ethnicity and NZdep Quintile ............... 53

5.4 Cardiovascular Disease Risk Categories .................................................... 54

5.4.1 Mild CVD Risk Category .................................................................... 54

5.4.2 Moderate to High CVD Risk Category ............................................... 54

5.4.3 High CVD Risk Category .................................................................... 54

5.5 Main Food Sources of Flavonoids and Flavan-3-ol .................................... 58

! vi!

5.6 Intake of Flavonoids and Flavan-3-ols for each CVD Risk Category ........ 59

5.7 Total Flavonoid Intake and CVD Risk ....................................................... 59

5.8 Flavan-3-ol Intake and CVD Risk .............................................................. 61

6 Discussion and Conclusions............................................................................. 65

6.1 Main Findings.............................................................................................. 65

6.2 Strengths and Limitations of the Present Study .......................................... 68

6.2.1 Strengths............................................................................................... 68

6.2.2 Limitations............................................................................................ 70

6.3 Implications for Future Research ................................................................ 71

6.4 Conclusions ................................................................................................. 72

7 Application to Dietetic Practice ....................................................................... 73

8 References ........................................................................................................ 74

9 Appendix .......................................................................................................... 79

! vii!

List of Tables

Table 2.1 Worldwide cardiovascular risk assessments scores .............................. 8

Table 2.2 Studies investigating flavan-3-ol and CVD............................................ 22

Table 5.1 Demographic data for NZANS participants by categories of CVD risk

and total flavonoid and flavan-3-ol intake ............................................ 56

Table 5.2 Main food groups contributing to total flavonoid intake and flavan-3-ol

intake ..................................................................................................... 58

Table 5.3 Mean total flavonoid and flavan-3-ol intake by CVD risk category in

unadjusted analyses ............................................................................... 59

Table 5.4 Relationships between groups of total flavonoid intake and CVD risk

categories. Results are presented as Odds Ratio (CI) for being in a

particular total flavonoid intake group for each CVD risk category,

compared to the low total flavonoid intake group.................................. 60

Table 5.5 Relationships between groups of total flavonoid intake and CVD risk

categories. Results are presented as Odds Ratio (CI) for being in a

particular total flavonoid intake group for each CVD risk category...... 61

Table 5.6 Relationships between groups of flavan-3-ol intake and CVD risk

categories. Results are presented as Odds Ratio (CI) for being in a

particular flavan-3-ol intake group for each CVD risk category,

compared to the low flavan-3-ol intake group ....................................... 62

Table 5.7 Relationships between groups of flavan-3-ol intake and CVD risk

categories. Results are presented as Odds Ratio (CI) for being in a

particular flavan-3-ol intake group for each CVD risk category............ 64

! viii!

List of Figures

Figure 3.1 New Zealand Cardiovascular Risk Charts.............................................. 42

Figure 3.2 Composite for a Black Bean and Beef Stir-fry Using Foods From

FOODfiles .............................................................................................. 47

Figure 3.3 Retention Factors for Cooked versus Uncooked Ingredients in a

Composite using foods from FOODfiles ............................................... 49

!

! ix!

List of Abbreviations

ANOVA analysis of variance

ASSIGN Scottish Intercollegiate Guidelines Network to Assign Preventative Treatment Score

BMI body mass index

BP blood pressure

CI confidence interval

CHD coronary heart disease

CVD cardiovascular disease

DALYs Disability Adjusted Life Years

DASH Dietary Approaches to Stop Hypertension

DBP diastolic blood pressure

DHA docosahexaenoic acid

ECG electrocardiogram

EDTA ethylene diamine tetra-acetic acid

EPA eicosapentaenoic acid

EPIC Europoean Prospective Investigation into Cancer and Nutrition

eNOS endothelial nitric oxide synthase

FFQ food frequency questionnaire

FMD flow mediated dilatation

GRADE Grading of Recommendations Assessment, Development and Evaluation

HDL high density lipoprotein

HOMA-IR Homeostasis Model of Assessment - Insulin Resistance

IHD ischemic heart disease

ISI insulin sensitivity index

LDL low density lipoprotein

MI myocardial infarction

MOH Ministry of Health

NADH nicotinamide adenine dinucleotide

! x!

NO nitric oxide

NOS nitric oxide synthase

NZ New Zealand

NZANS New Zealand Adult Nutrition Survey 2008/2009

NZFCDB New Zealand Food Composition Database

OR odds ratio

PFR Plant and Food Research

PROCAM Prospective Cardiovascular Münster score

QRISK QRESEARCH Cardiovascular Risk score

QUICKI Quantitative Insulin Sensitivity Check Index

RCT randomised controlled trial

RNO sum of nitrosylated and nitrosated species

SCORE Systematic Coronary Risk Evaluation score

SE standard error

SES socioeconomic status

SBP systolic blood pressure

T2DM type two diabetes mellitus

TC total cholesterol

NZDep New Zealand Deprivation Index

UK United Kindgdom

US United States

USDA United States Department of Agriculture

WHO World Health Organisation

!

! 1!

1 Introduction

Over the past century, CVD has gone from being a relatively uncommon disease, to one

of the leading causes of morbidity and mortality worldwide (1). Improved health care and

pharmacological methods to decrease the rates of CVD mortality over the past decade

have been successful and there has been a decline in the rates observed in New Zealand

(2). However, as the population ages the number of people burdened by CVD has

increased. There are various economic, cultural and social influences that continue to

increase certain risk factors for CVD (1). As this disease bears such a great burden on

society, researchers continue to address this matter, with the ultimate goal of further

reducing the prevalence of CVD. Along with risk factors such as obesity, diabetes

mellitus, tobacco use, and hypertension, it has been demonstrated by various

epidemiological studies that diet plays a significant role in the development of CVD (1-3).

There is convincing evidence that high consumption of linoleic acid, fish and fish oils

(EPA and DHA), vegetables and fruits (including berries), potassium, and a low to

moderate alcohol intake is likely to decrease the risk of developing CVD (2).

Epidemiological evidence as to the health benefits of other non-nutrients in foods is

advancing, in particular the possible health benefits of the phenolic substances found in

plant foods: flavonoids (2,4).

Flavonoids are found in various plant foods such as fruit, vegetables, soy products, nuts

and seeds, spices, cocoa, and beverages such as tea and wine. Previous research has

demonstrated that flavonoids have a wide range of potential health benefits such as

antioxidant, anti-carcinogenic and anti-inflammatory effects (5). Studies have investigated

the effects of total flavonoid intake, as well as the effects of specific flavonoids. There is

growing evidence that suggests there may be an inverse association between the flavonoid

! 2!

subclass flavan-3-ols, and CVD (6). Flavan-3-ols are one of the seven major flavonoid

subclasses, and particularly high levels are found in apples, black tea, wine, blueberries,

cocoa and cocoa products (6). It has been proposed that flavan-3-ol may have specific

effects on vasculature, in particular enhancing the activity of endothelial nitric oxide

synthase (eNOS), which promotes endothelial relaxation, thus improving vascular health

(7). Furthermore, it may have antioxidant and anti-inflammatory properties, which all

help to explain its potential effect on reducing the risk of CVD (8). The health and well-

being of many New Zealanders could potentially be improved by increasing their

consumption of flavonoids, as long as a balanced diet high in fruits, vegetables and whole

grains, and low consumption of foods high in added sugar, sodium and fat (particularly

trans and saturated fats) is followed.

The aim of this thesis is to develop a flavonoid database so the intake of flavonoids in

New Zealand adults can be determined. This will be used to investigate whether a higher

intake of flavonoids, in particular flavan-3-ols, is associated with lower 5-year CVD risk

score in New Zealand adults.

! 3!

2 Literature Review

2.1 Introduction

CVD is defined as any abnormal condition characterised by dysfunction of the heart and

blood vessels and includes coronary heart disease (CHD) or ischemic heart disease (IHD),

cerebrovascular disease or stroke, and peripheral artery disease (2). CVD is an ongoing

burden on society, and is a leading cause of death for adults worldwide (2). In New

Zealand, CVD accounts for 40% of all deaths annually (9). The two leading causes of

death from CVD are IHD and cerebrovascular disease (2). Worldwide, CVD accounts for

approximately 30% of all deaths, resulting in approximately 17.5 million deaths annually

(10). It is predominantly a burden in developed countries, but rates are also increasing in

developing countries (2,6,9). Decreases in rates of CVD can be seen in countries where

active attempts have been made to reduce cardiovascular risk factors, and where resources

and facilities are available to treat those suffering acute cardiovascular events (2). In New

Zealand age-standardized CVD mortality has halved over the past 30 years as a result of

active attempts to reduce CVD risk factors, in part as a result of increased rates of

prescription medications such as statins and lower rates of smoking (2). However, rates

still remain high, and burden certain populations within New Zealand more than others.

Maori, Pacific Islanders and those of low socioeconomic status bear the greatest burden of

CVD in New Zealand (11). Mortality rates from IHD in the Maori population are 2.5

times that of non-Maori, and Maori are 1.5 times more likely to be hospitilised due to

IHD than non-Maori. Further, stroke mortality is 1.7 times higher for Maori than non-

Maori, and stroke hospitilisation rates are twice as high (12).

Many individuals are affected and die prematurely as a result of this largely preventable

! 4!

disease (13). Approximately a quarter of Disability Adjusted Life Years (DALYs) lost by

New Zealanders are as a result of CVD. IHD results in approximately 73,804 DALYs,

with stroke resulting in 30115 DALYs (14). Furthermore, there is a significant economic

burden on the health sector as a result of CVD. A study by Scott et al (1993) estimated

that in New Zealand, direct medical costs attributed to CVD were $179 million, and

indirect costs such as practitioners fees, and the need to supplement wages were between

$14-$24 million (15). This study took place between 1989 and 1992, thus costs are likely

to be higher in today’s economy. There are also various intangible and indirect costs of

CVD due to pain, suffering and anxiety experienced by not only individuals suffering

from CVD, but their families too.

2.1.1 CVD Risk Assessment Tools

There are various well established CVD risk factors including obesity, hypertension,

dyslipidaemia, smoking and diabetes (6,16,17). Due to the synergistic interaction of CVD

risk factors, various multivariate risk prediction tools have been developed worldwide to

yield estimates of absolute CVD risk by combining risk factor information (Table 2.1)

(18,19). There have also been disease-specific formulations developed to predict specific

components of CVD such as stroke or CHD (19). In terms of primary care, individual

health care interventions should be targeted at those with a high total or absolute

cardiovascular risk instead of single risk factor levels above traditional thresholds (20).

Risk assessment tools help identify those with many marginal risk factors which act

synergistically to have potentially harmful overall effects. It also helps to ensure those

with one isolated risk factor are not unnecessarily alarmed, as having a single risk factor

does not necessarily mean they have a high risk of CVD (16,19,21,22).

! 5!

Risk assessment tools are derived from data from cohort studies that assess the

relationship between risk factors and risk of CVD. The first risk assessment tool was

developed from the Framingham Heart study in 1967, which was based on a population of

mostly Caucasian Americans. It has since been modified to determine total CVD risk, as

well as risk for outcomes such as CHD and stroke, as it recognised that different risk

factors contribute to these diseases (23). The Framingham risk score has been validated in

both African-American and Caucasian populations in the United States, and with

calibration, can be used on certain populations in Europe, the Mediterranean and Asia

(19). However, this tool is not necessarily reliable at predicting CVD risk in other

populations, as risk assessment depends on the background risk of the population that is

being assessed (24). Environment and behavioral differences between countries results in

different rates of CVD morbidity and mortality, therefore unique risk scores need to be

used for each population (2,18,25).

As shown in Table 2.1, various risk prediction tools have been produced since the

Framingham risk score including the Reynolds risk score, the Scottish Intercollegiate

Guidelines Network to assign preventative treatment (ASSIGN) score, the Systematic

Coronary Risk Evaluation (SCORE) score, the Prospective Cardiovascular Münster

(PROCAM) score, and the QRESEARCH Cardiovascular Risk (QRISK1 and QRISK2)

algorithms (6,18). These risk scores have been adapted for use in primary care as simple

charts, tables and computer programs or tools, for ease of use by health care professionals

in establishing those at greater risk of CVD (24). These risk scores are different as they

have been developed to detect CVD risk in each specific population. They take into

account different risk factors that are most important for the particular population they are

estimating CVD risk for.

! 6!

2.1.2 CVD Risk Assessment in New Zealand

Identifying those with increased CVD risk is important if New Zealand is to reduce levels

of CVD morbidity and mortality. As discussed previously, using multiple risk factors is

more advantageous than investigation of single factors (26). Therefore, in New Zealand,

risk assessment charts have been developed which assess an individual’s absolute 5-year

CVD risk, so individuals at ‘high risk’ of cardiovascular events can be identified and

receive appropriate lifestyle advice and/or drug treatments (26,27). These risk charts were

developed based on the risk equation used from the Framingham Heart study (26). The

New Zealand risk score charts are believed to predict 5-year cardiovascular events in New

Zealand in men 35-74 years, and women 35-69 years, more accurately than the

Framingham risk score (26). This is because it takes into account population specific

information such as higher CVD risk in Maori, Pacific and Indo-Asian populations within

New Zealand, and it calibrates the formula accordingly.

2.1.3 CVD and Diet

Primary prevention of CVD is predominately attainable by maintaining a healthy lifestyle,

including physical activity and a well balanced diet (4,21). There is convincing evidence

from various intervention and epidemiological studies that diet can be modified to reduce

CVD risk (2,28). For example, the Dietary Approaches to Stop Hypertension (DASH)

trials demonstrated that relatively high intake of fruits, vegetables and low-fat dairy, and

low intake of sodium, resulted in improvements in BP (2,3,9,10). The Oslo trial in 1989,

which consisted of 1232 males with a total cholesterol (TC) between 7.5 – 9.8, found that

with a total and saturated fat reduction, and higher intake of fibre rich fruit, vegetables

and wholegrains, TC was reduced by an average of 13 percent (2,29).

! 7!

Various epidemiological studies demonstrate how dietary patterns can influence the risk

of CVD. For example, there are lower rates of CVD in those who consume a

‘Mediterranean diet’, which is low in total and saturated fat, and high in oleic and alpha-

linoleic acids, fruits, vegetables and wholegrains (2,30). For example, the Lyon Diet Heart

Study, a randomised controlled trial (RCT) consisting of 605 participants, demonstrated

the beneficial effects of the ‘Mediterranean-style diet’(25). The experimental group were

instructed to eat more bread, more vegetables, fruit at least once daily, more fish, less red

meat (replace with poultry), margarine instead of butter and cream, and to use rapeseed

and olive oil. Those in the control group received general dietary advice. After 46 months

of follow up it was noted that there were significant beneficial effects in the experimental

group consuming the ‘Mediterranean diet’, with 50-70% lower risk of recurrent heart

disease than those in the control group (25). The so called ‘Western Diet’ with higher than

recommended intakes of saturated and trans-fats, salt and sugar, and low intakes of

cardioprotective foods such as fruit and vegetables is a contributing factor in the high

rates of CVD observed in New Zealand (2).

It is likely that specific nutrients in foods (such as potassium in fruit in vegetables, or

linoleic acid, DHA and EPA in fish and fish oils) can reduce risk of CVD (2,31). There

are also other non-caloric, non-nutrient constituents in plant foods that may have

cardioprotective benefits, one being flavonoids (6,32). Epidemiological data surrounding

the association between flavonoids and CVD is growing (33).

! 8!

Table 2.1 Worldwide cardiovascular risk assessments scores

Risk Assessment Score

Participants Sample size

Years of follow-up

Factors included in risk score

Risk score purpose, and inclusion criteria

Major changes

Limitations Rationale for inclusion

Reference

Framingham Men and women from the original Framingham cohort aged 30-62 years. Conducted in the United States of America.

N=4856 2187 men 2669 women

12 years

Sex, age, serum TC, systolic blood pressure (SBP), relative weight, haemaglobin, smoking status ECG or Left ventricular hypertrophy.

Probability/risk of developing CHD in 12 years, Including: definite myocardial infarction (MI), coronary insufficiency, angina pectoris, and/or death from coronary heart disease.

This study was the original CVD risk study which was used to develop risk assessment score, named the Framingham risk score.

Truett, Cornfield et al. (1967) (23)

Framingham

Men and women from the original Framingham cohort aged 35-74 years. Conducted in the United State of America.

N=not reported

16 years

Sex, age, TC, SBP, smoking status, glucose tolerance, left ventricular hypertrophy.

Probability/risk of developing CHD in 6 years, including: MI, coronary insufficiency, angina pectoris, sudden death, and/or coronary attack.

Glucose tolerance introduced.

Didn’t use weight or haemoglobin. ECG abnormality was restricted to left ventricular hypertrophy.

It recognised that several independent factors lead to CVD and they must all be assessed to develop a risk score.

Gordon et al. (1971) (34)

! 9!

Risk Assessment Score

Participants Sample size

Years of follow-up

Factors included in risk score

Risk score purpose, and inclusion criteria

Major changes

Limitations Rationale for inclusion

Reference

Framingham

Men and women from the original Framingham cohort aged 35-65 years at baseline. Conducted in the United States if America.

N=5,209 8 years Age, gender, TC, diastolic blood pressure (DBP), smoking, glucose tolerance, left ventricular hypertrophy.

Probability/risk of developing CVD in 8 years: coronary heart disease, brain infarction, intermittent claudication, hypertensive heart failure, total CVD.

Includes glucose intolerance.

Cannot be used once CVD has developed. Only appropriate for those free of CVD.

This enabled assessment of separate CVD outcomes that significantly contribute to mortality, and may require different treatment measures.

Kannel, McGee et al. (1976) (35)

Framingham

Men and women from the original Framingham cohort aged 35-65 years at baseline. Conducted in the United States if America.

N=5,209 16 years

Age, gender, TC, HDL cholesterol, SBP, smoking, glucose tolerance, left ventricular hypertrophy.

Probability/risk of developing Coronary Arterial Disease in 6 years.

HDL cholesterol introduced.

Cannot be used once CVD has developed. Only appropriate for those free of CVD.

This is the first time HDL was considered for CVD risk. It is used in New Zealand’s CVD risk charts today as part of the cholesterol: HDL ratio. This is recognised as a significant CVD risk factor today, so its inclusion is paramount.

Wilson et al. (1987) (36)

! 10!

Risk Assessment Score

Participants Sample size

Years of follow-up

Factors included in risk score

Risk score purpose, and inclusion criteria

Major changes

Limitations Rationale for inclusion

Reference

PROCAM study 1979

Men and Women from PROCAM study in Munster, Germany.

N = 5389 men aged 35-65 years

10 years

Age, LDL cholesterol, smoking status, HDL cholesterol SBP, family history of premature MI, diabetes mellitus, triglycerides.

Global risk of MI. A major coronary event was defined as the occurrence of sudden cardiac death or definite fatal or nonfatal myocardial infarction on the basis of ECG and/or cardiac enzyme changes.

Included information of family history of CHD, triglycerides, and LDL cholesterol, as well as diabetes mellitus.

Didn’t include left ventricular hypertrophy.

This study included only the ‘hard’ end points of definite MI or sudden coronary death. Took into consideration family history and diabetes mellitus, which are two significant risk factors for CVD.

Assmann et el. (2002) (37)

SCORE Pooled dataset of cohort studies from 12 European countries: Finland Russia, Norway, UK (BRHS), UK (Scotland). Denmark, Sweden, Belgium, Germany, Italy, France, and Spain.

N = 88080 women, 117098 men

2.7 million person years

Age, gender, smoking status, TC:HDL ratio or TC against SBP.

10-year fatal cardiovascular risk. Looking at fatal CVD as the only endpoint.

Used TC against SBP.

Only looked at fatal CVD risk, instead of fatal, and non-fatal endpoints such as that in Framingham. Didn’t include other well established risk factors such as diabetes mellitus.

Looked at total CVD risk rather than coronary heart disease risk. Only looking at fatal CVD, as it is difficult to replicate those studies with multiple non-fatal CVD endpoints, which are difficult to ascertain. Also a very large sample size used.

Conroy et at. (2003) (38)

! 11!

Risk Assessment Score

Participants Sample size

Years of follow-up

Factors included in risk score

Risk score purpose, and inclusion criteria

Major changes

Limitations Rationale for inclusion

Reference

ASSIGN Scottish Heart Health Extended Cohort (SHHEC). includes: Women aged 40-59 years across 25 districts of Scotland 1984-87; the Scottish MONICA project recruited in Edinburgh and North Glasgow 1989 and 1995, North Glasgow again in 1989 and 1995, ages 25-64 and 1992, ages 25-74.

N = 6540 men and 6757 women

Range from 10 to 21 years

Age, Scottish Index of Multiple Deprivation, family history, diabetes, smoking status, cigarettes per day (smokers), SBP, TC: HDL ratio.

10 year CVD risk. Endpoints included: deaths from CVD causes, or any hospital discharge diagnosis post-recruitment for coronary heart disease, or cerebrovascular disease or for coronary artery interventions.

Used to mitigate potential unfairness in Framingham and similar risk scores when applied across different social groups in the same population.

This risk score may overestimate the CVD risk as socioeconomic status is not as decisive an endpoint as something like HDL cholesterol, or smoking status. However, it is an important consideration.

This score helps to take into account social deprivation. Social deprivation, or socioeconomic status is a powerful determinant of CVD risk and also decreases chances of receiving medical attention promptly during a coronary event.

Woodward et al. (2007) (39)

! 12!

Risk Assessment Score

Participants Sample size

Years of follow-up

Factors included in risk score

Risk score purpose, and inclusion criteria

Major changes

Limitations Rationale for inclusion

Reference

QRISK Prospective open cohort study using routinely collected data from general practice. UK practices contributing to the QRE-SEARCH database. Aged 35-74 years, registered at 318 practices.

N = 1.8 million patients

17 years

Age, gender, smoking status, SBP, TC: HDL ratio, BMI, family history of coronary heart disease (first degree related aged less than 60), area measure of deprivation, existing treatment with antihypertensive agent.

10 year estimated CVD risk. Endpoints included first recorded diagnosis of cardiovascular disease, MI, CHD, stroke, and transient ischemic attacks.

Includes additional variables, which improve risk estimates for patients with positive family history or those on antihyperten-sive treatment.

Area measure of deprivation may overestimate CVD risk, as it is not a well-defined CVD risk factor.

Likely to provide more appropriate estimates of risk, on the basis of age, gender and social deprivation. It will help to ensure treatments are directed towards those that are most likely to benefit.

Hippisley-Cox et al. (2007) (40)

! 13!

Risk Assessment Score

Participants and development of model

Sample size

Years of follow-up

Factors included in risk score

Risk score purpose, and inclusion criteria

Major changes

Limitations Rationale for inclusion

Reference

New Zealand Risk assessment charts

Individuals Aged 35 to 74 years either enrolled in general electoral rolls or employed by a New Zealand-wide multi-industry corporation, in 1992-1993.

N = 6354 (4638 men and 1716 women)

5 years Age, gender, TC:HDL ratio, SBP, diabetes mellitus, smoking status, ethnicity (Maori, Pacific or Indo-Asian people), family history of premature coronary heart disease or ischemic stroke in a first-degree relative.

5-year cardiovascular risk score. Hospitilisation and mortality from cardiovascular disease.

Only include SBP. Takes into account increased risk in Maori, Pacific and Indo-Asian as having increased risk.

Quite population specific, as accounts for specific ethnic groups within New Zealand.

Including only SBP as the most informative and conventionally measured BP parameter for CVD risk, so used instead of looking at both SBP and DBP. The risk scores are also calibrated so they can be used in different ethnic groups in New Zealand who are at higher risk.

Primary care handbook (2002) (27)

! 14!

2.2 Flavonoids

Flavonoids are phenolic substances that are found in various plant species, mainly fruit,

vegetables, soy products, nuts and seeds, spices, cocoa, and beverages such as tea and wine

(6). It is the glycosylated forms of flavonoids that contribute to the bright colour of many

leaves, flowers and fruits (24,30,41). Flavonoids are formed in plants from the three amino

acids phenylalanine, tyrosine and malonate. The flavan nucleus consists of 15 carbon

atoms arranged in three rings. The different classes have different levels of oxidation, and

substitutions around the rings. The flavonoid subclasses of particular interest in human

health include flavones, flavanones, isoflavones, flavonols, flavanonols, flavan-3-ols, and

anthocyanins, and their oligomeric and polymeric forms, which are found in a variety of

plant foods (6,41). Altogether, there are over 8000 individual flavonoid compounds

(19,20,41).

It has been suggested that flavonoids have antioxidant properties, as they are able to reduce

free radical formation and scavenge free radicals. It is also suggested they have antiviral,

antiallergenic, anti-inflammatory and vasodilating actions (9,30,41,42). It is the oxidative

damage caused by free radicals that plays a role in several diseases such as heart disease

and cancer. There are certain antioxidant systems in the human body that work to prevent

damage from free radicals, which work via endogenous antioxidants that are produced in

the body, and exogenous antioxidants that are obtained from the diet (41). Dietary

antioxidants have been studied intensively for many years, and include vitamins A, C, E,

and carotenoids, which if consumed in high quantities help to reduce the detrimental

impacts of oxidative damage over the life course (6,17,20,41). Over the past decade, it has

been demonstrated that other substances in plant foods may also have antioxidant

properties, such as the plant polyphenols: phenols, phenolic acids, flavonoids, tannins and

! 15!

lignans (9,14,41-44).

Flavonoids have important roles in the physiology of plants, and are also important

components in the human diet. They are considered a 'non-nutrient' as they have no caloric

value, however, they are consumed in much higher quantities than antioxidant vitamins

such as vitamin C (99mg/day for New Zealand men and women) or E (11.5mg/day for

New Zealand males and 9.1mg/day for New Zealand females), with an intake of

approximately 50-800mg/day (12,15,43,45,46). There have been no studies to date

investigating flavonoid consumption in New Zealand adults.

2.2.1 Flavonoids and Chronic Disease

Epidemiological evidence shows that flavonoids play a preventative role in chronic

diseases such as CVD and cancer. Several RCTs have demonstrated that even modest

intakes of flavonoid-rich foods may reduce the risk of certain CVD risk factors

(2,6,8,17,20). Various observational studies have demonstrated the effects of flavonoid

intake on chronic disease. A study by McCullough et al. (2012) - a large prospective cohort

study carried out in the US on a total of 38,180 men and 60,289 women, demonstrated that

higher intakes of flavan-3-ols are associated with a lower risk of CVD (6). There is also

evidence that flavonoids reduce the risk of type 2 diabetes mellitus (T2DM). A study

carried out in the US by Wedick et el. (2012), prospectively evaluated the association

between flavonoid intake and T2DM in 3 large cohorts. The study included a total of

70,359 women in the Nurses' Health study (NHS) and NHS II, and 41,334 men in the

Health Professionals Follow-up Study. It was found that higher intakes of the flavonoid

subclass anthocyanins, was significantly associated with a lower risk of T2DM (45).

! 16!

2.2.2 Flavonoids Mechanisms of Action

The antioxidant mechanism of action for flavonoids works by inhibiting superoxide anion

production, and also the inhibition of the activity of enzymes that are involved in the

generation of reactive oxygen species such as lipoxygenase, glutathione S-transferase and

NADH oxidase (6,8,45). It is these antioxidant properties that reduces the risk of disorders

such as coronary artery disease, stroke, and other vascular diseases, that occur partly as a

result of oxidative stress (7). It has also been demonstrated that flavonoids have anti-

inflammatory actions, increase nitric oxide (NO) production and reduce LDL cholesterol

oxidation (4,6,30). These functions help improve endothelial function leading to

improvements in vasodilation and blood pressure, and thus reduce the risk of CVD. Other

beneficial effects include improvements in insulin secretion and sensitivity, glucose

tolerance and metabolism, and B-cell function, thus decreasing the risk of T2DM

(6,45,47). It is outside the scope of this thesis to review the literature on all flavonoids and

CVD risk factors, therefore the focus of this literature review will be on Flavan-3-ols.

2.3 Flavan-3-ols

Flavan-3-ols are one of the seven major flavonoid subclasses. Monomers of flavan-3-ol

include: (+)-Catechin, (+)-Catechin-3-gallate, (-)-Epicatechin, (-)-Epicatechin-3-gallate, (-)

-Epigallocatechin, (-)-Epigallocatechin-3-gallate, (+)-Gallocatechin, and (+)-

Gallocatechin-3-gallate (48). Common plant sources of include apples, black tea, red wine,

blueberries, cocoa and cocoa products (6,28,30,33). Particularly high levels of flavan-3-ols

are found in cocoa and subsequently chocolate (6,8,28,47). Darker chocolate contains more

cocoa, and thus has a higher content of flavan-3-ols. The protective effect of cocoa on

CVD has largely been ascribed to the high content of flavan-3-ol and its monomers

(epicatechin and catechin) (31,49,50). Subsequently, the majority of research has focused

! 17!

on the effect of flavan-3-ols through intakes of cocoa products, and their effects on health.

2.3.1 Flavan-3-ols and CVD

There has been recent interest in the relationship between different flavonoid subclasses

and the risk of CVD. In particular, there has been evidence from both observational and

experimental research that demonstrates there is an inverse relationship between the intake

of the flavonoid subclass flavan-3-ol, and CVD risk (Table 2.2). Early migration studies

demonstrated the relationship between cocoa intake and CVD. The Kunan Indians from

indigenous islands in Panama had traditionally high cocoa consumption, and very little

incidence of hypertension and CVD. It was observed that when they emigrated to Panama

City, their consumption of cocoa containing beverages decreased by a factor of ~10, and

their mean BP increased SBP by 3.8mmHg, and DBP by 5.6mmHg (50).

As shown in Table 2.2, there have been several observational studies investigating the

relationships between cocoa, chocolate and flavan-3-ol intake; and CVD biomarkers, rates

of CVD events, and rates of CVD mortality. A study by Janszy et al (2008) in Sweden,

demonstrated an inverse dose-dependent relationship between intake of chocolate, and risk

of CVD events in 1169 non diabetic patients hospitalised with a confirmed first acute MI.

Compared to those who never ate chocolate, there was a 27%, 44%, and 66% lower

relative risk of cardiac mortality for those that ate chocolate less than once per month, up

to once per week and twice or more per week respectively (43). This demonstrated even

modest consumption of flavan-3-ol rich chocolate may reduce the risk of CVD events in

those already susceptible to CVD.

A large prospective cohort study by McCullough et al. (2012), which included 38,180 men

! 18!

and 60,289 women, investigated the intake of specific flavonoid subclasses and the risk of

fatal CVD. It was found those in the highest quintile for flavan-3-ol intake had a 17%

lower risk of fatal CVD than those in the in the lowest quintile for flavan-3-ol intake (6). In

comparison, a cohort study from the Iowa Women’s Health Study of 34,489

postmenopausal women who were free from CVD, found there was no relationship

between flavan-3-ol intake and CVD (17). These differences in findings may be due to the

difference in sample size, or potentially a difference in the effects of flavonoids between

genders. In support of these findings from various observational studies, several RCTs

have shown a positive relationship between cocoa and chocolate flavonoids on certain

CVD risk factors such as lipid profile, BP, LDL oxidation, plasma antioxidant activity,

arterial flow mediated dilatation (FMD), and platelet aggregation (28,31,41). Note that

FMD is used as a marker of endothelium function, as it helps to determine the NO-

dependent endothelium-mediated dilatory response in an artery (28,41,51).

There is also evidence that a high intake of flavan-3-ols is likely to have a beneficial effect

on endothelial function in both healthy individuals and those with increased CVD risk.

Heiss et al. (2003), found ingestion of high flavan-3-ol cocoa in individuals with one CVD

risk factor, improved endothelial function via improved FMD 2 hours after ingestion (52).

Similarly, a RCT by Wang-Polagruto et al. (2006), on 32 postmenopausal women with

hypercholesterolemia, demonstrated that there were beneficial vascular effects as a result

of consumption of flavan-3-ol rich chocolate (47). A study by Faridi et al (2008), looking

at endothelial function in 45 obese individuals after consumption of either a cocoa

beverage or a placebo, had similar findings (31). A RCT by Engler et al (2004),

investigated the short term effects of flavonoid-rich dark chocolate intake on endothelial

function in 21 healthy adult subjects. It was found high flavonoid rich chocolate

! 19!

consumption improved endothelium-dependent FMD of the brachial artery, compared to

those consuming low-flavonoid rich chocolate (7). These results are important, as

endothelial function is a key determinant of cardiovascular health. Although sample size in

these studies is small, they are important as they demonstrate a high intake of flavan-3-ols

may have significance in both the prevention of CVD, as well as in the management of

disease in those with established CVD. Further studies with more participants are

necessary to confirm these results.

There have been several studies that have investigated the relationship between intake of

flavan-3-ols and BP, a well-established CVD risk factor. An inverse association between

flavan-3-ols and BP has been demonstrated in several studies. A recent systematic review

of 31 RCTs used a meta-regression model to determine the relationship between the dose

of the flavan-3-ol monomer epicatechin, and BP. It was found that an intake of 25mg of

epicatechin/day (corresponding to 25-30g dark chocolate) significantly reduced both SBP

(-4.2 mmHg) and DBP (-2.1mmHg), which may have clinically significant effects for

CVD (49). Similarly, a systemic review of 42 RCTs and 1297 participants by Hooper et al.

(2012) examined the effect of chocolate, cocoa, and flavan-3-ols on major CVD risk

factors. It was found there were significant reductions in BP after intake of the flavan-3-ol

monomer epicatechin at intakes >50mg/day (4). It has also been demonstrated that a high

intake of flavan-3-ols has beneficial effects on other CVD biomarkers such as lipid profile,

lipid oxidation, insulin sensitivity and resistance (4,26,44). Curtis et al (2012), examined

the relationship of CVD and the combination of flavan-3-ols and isoflavones in

postmenopausal women with T2DM (44). This study showed that combined flavan-3-ols

and isoflavones significantly improved lipoprotein status and markers of insulin sensitivity

(HOMA-IR and QUICKI). It was also observed that the estimated 10 year risk of CHD

! 20!

was reduced in the experimental group (44).

As previously mentioned, there have been several observational studies that have

demonstrated a beneficial effect of increased intake of cocoa containing food and

beverages on the risk of CVD. It is important to note that cocoa products such as chocolate

are often high in sugar and fat, and high intakes of these foods increase the risk of obesity,

and thus CVD. In the German EPIC (European Prospective Investigation into Cancer and

Nutrition) cohort, it was concluded to see the observed effects on CVD risk, one may have

to consume around 50g chocolate/day, which provides ~230kcal. Thus, there is potential

for adverse effects on weight if the diet is not energy balanced (4).

Several RCTs have examined whether cocoa products with and without sugar affected its

relationship to CVD, with findings that suggest both intake of both sugar-sweetened and

sugar-free cocoa products have beneficial effects on endothelial function such as FMD.

However there were close to significant findings that sugar may have attenuated the

beneficial effects (31,53). Therefore, it is important to consider how much chocolate or

cocoa products would be beneficial to recommend in practice, and whether it would be

more beneficial to consider sugar-free cocoa products. The longest RCT was for 18 weeks,

and only 7 trials had an intervention period of more than 6 weeks. It has been suggested

that these outcomes were greatest in acute studies, suggesting there may not be long term

benefits of cocoa consumption (4,28,30,52). GRADE suggests there is low to moderate

quality evidence that there are beneficial effects with increasing doses of flavan-3-ols.

There was no suggestion that increased intakes have any negative effects (4,8). Therefore,

further larger long-term trials are required to give a better understanding of the relationship

between flavan-3-ol and absolute CVD risk.

! 21!

2.3.2 Flavan-3-ols Mechanisms of Action

In vitro studies suggest that flavonoids may have specific vascular effects, and certain

mechanisms of action have been suggested. Endothelium derived NO is a signaling

molecule that ensures vascular homeostasis. Bioactive NO helps reduce platelet adhesion

as well as amplify relaxation of smooth muscle resulting in vasodilatation (49,51). Flavan-

3-ols may have the potential to increase nitric oxide synthase (NOS) activity, which in turn

increases the pool of bioactive NO (4,8,28,41,52). This can explain improvements in BP

and FMD after intake of flavan-3-ol rich cocoa products (8,41,46). It has also been

demonstrated that flavan-3-ols have antioxidant and anti-inflammatory properties, which

help to explain the observed reduction in LDL oxidation and platelet aggregation. It has

also been shown in vitro to improve glucose transport, insulin sensitivity and insulin

resistance (4,8,28,41). These mechanisms of action help to explain why improvements

have been seen in tests such as HOMA-IR and QUICKI, and improved glucose tolerance

in various RCTs.

These mechanisms of action have been supported by evidence in vitro, ex vivo and in

animal studies (28,41,46). However these findings cannot necessarily be translated to in

vivo potentials in humans due to uncertainties surrounding bioavailability (28,41). It has

been demonstrated that a small part of the ingested flavonoid is absorbed, however a large

part is degraded by intestinal microflora to different phenolic acids, and high quantities do

not reach the blood stream (41). Bioavailability may also depend on the kind of food that is

ingested. It has been suggested that proteins may bind to the polyphenols, thereby reducing

their availability (28,41). In comparison, alcohol (such as red wine), or fat (such as dark

chocolate) may increase availability (41). This is important to consider in a clinical setting

when recommending what level of flavan-3-ol intake would be required to show benefits.

! 22!

Table 2.2 Studies investigating flavan-3-ol intake and CVD risk

Reference

Purpose/ Objective

Population, Setting, Sample, Age Range

Study Design Methods/ Measures

Key Findings Conclusions Rationale for inclusion

Limitations and Further Research Needed

Heiss et al. (2003) (52)

To assess the time-course of flavan-3-ol effects on FMD after intake of flavan-3-ol.

N = 26 participants with at least 1 CVD risk factor, with mean age of 45 years, and mean BMI 25. Study conducted in Germany.

Double blind crossover RCT. FMD was measured at 0,2,4,6 hours after ingestion of a 100mL cocoa drink containing 176mg of flavan-3-ol (70mg epicatechin plus catechin, 106mg of procyanidins), or control (100ml cocoa with <10mg flavan-3-ol).

The sum of nitrosylated and nitrosated species (collectively referred to as RNO) were measured. Also measured the FMD of the brachial artery.

Ingestion of 100mL of cocoa with flavan-3-ol increased FMD maximally at 2 hours. Cocoa drink with no flavan-3-ol did not affect FMD. Increase in RNO from 22 to 36 in the group with high flavan-3-ol cocoa (p<0.001), and changes in RNO and FMD were correlated (p = 0.02).

A single dose of 100ml cocoa drink high in flavan-3-ol transiently increases NO bioactivity in plasma, and reverses endothelial dysfunction. There is a correlation between FMD and RNO suggesting the effects on endothelial function are a result of increased NO availability.

This was one of the first studies to demonstrate high intakes of flavan-3-ol are likely to improve vascular function in those with increased risk of CVD. It also depicted the relationship between flavan-3-ol and nitric oxide availability and thus activity.

Very short term study. It therefore doesn’t conclude the long-term clinical effect of high intakes of flavan-3-ol and CVD risk.

! 23!

Reference

Purpose/ Objective

Population, Setting, Sample, Age Range

Study Design Methods/ Measures

Key Findings Conclusions Rationale for inclusion

Limitations and Further Research Needed

Engler et al. (2004) (7)

To determine the effect of flavonoid-rich dark chocolate on endothelial function.

N = 21 healthy adult subjects. Study conducted in California, Unites States of America.

Randomised, double-blind, placebo-controlled trial over 2 week period. Subjects randomly assigned to daily intake of high-flavonoid (213mg procyanidins, 46mg epicatechin), or low-flavonoid dark chocolate. FFQ to determine intake of other antioxidant nutrients.

Triacylglycerol, TC, LDL, HDL, LDL oxidation, epicatechin, total antioxidant capacity. Also endothelium-dependent FMD of brachial artery and BP.

The means change in FMD was significantly different in the high versus low-flavonoid group after 2 weeks of chocolate consumption (p = 0.024). There was a significant increase in plasma epicatechin concentration in the high-flavonoid group (p <0.001), but not the low-flavonoid group. No change in biomarkers of antioxidant and oxidative stress.

Dark chocolate high in flavonoids improved endothelium-dependent vasodilation. It was demonstrated that this was associated with the observed increase in epicatechin concentration after consumption of high-flavonoid chocolate.

First clinical trial to demonstrate improved endothelial function in healthy adults following short-term intake of high flavonoid chocolate.

Small sample size, and short duration of study. This may have been a reason for lack of results of lipid profile and oxidative stress measures.

! 24!

Reference

Purpose/ objective

Population, setting, sample, age range

Study Design Methods/ measures

Key Findings Conclusions Rationale for inclusion

Limitations of Study and further research needed

Grassi et al. (2005) (54)

To compare the effects of different kinds of chocolate bars (either dark or white) on BP, glucose and insulin response, and oral-glucose tolerance tests in healthy subjects.

15 Healthy persons (7 men and 8 women) aged 33.9 +/-7.6years recruited from medical staff. Study conducted in L’Aquila, Italy.

Randomised controlled trial: Participants randomly assigned to 100g dark chocolate bars, which contained ~500mg polyphenols, or 90g white chocolate bars, which contained no polyphenols, for 15 days.

Effects of dark or white chocolate bars on BP and glucose and insulin response to an oral glucose-tolerance test.

HOMA-IR was significantly lower after dark chocolate than after white chocolate ingestion (0.94 +/- 0.42 compared with 1.72 +/- 0.62; p < 0.001). SBP was lower after dark than after white chocolate ingestions (107.5 +/- 8.6 compared with 113.9 +/- 8.4mmHg; p <0.05).

This study demonstrated that dark chocolate – which is high in polyphenols reduces BP, and improves insulin sensitivity in comparison to white chocolate. Indication that dark chocolate may have a protective action on the vascular endothelium.

This demonstrates that chocolate rich in polyphenols are likely beneficial for CVD risk factors, in comparison to a food with a similar nutrient composition but no polyphenols.

Small sample size. Larger scale trial would be needed to confirm these observed effects on BP and insulin sensitivity. Only carried out in healthy subjects, so would be useful to determine impacts on those who have hypertension or are overweight/obese.

! 25!

Reference

Purpose/ Objective

Population, Setting, Sample, Age Range

Study Design Methods/ Measures

Key Findings Conclusions Rationale for inclusion

Limitations and Further Research Needed

Wang-Polagruto et al. (2006) (47)

To determine if chronic dietary intake of flavonoid rich cocoa-based products on endothelial function and cardiovascular health on hy-percholesterolemic post-menopausal women.

N = 32 post-menopausal women with hypercholes-terolemia. Study conducted in California, United States of America

Randomised, double-blind, parallel-arm study. Participants were assigned to either a high-flavanol cocoa beverage (446mg total flavanols), or a low-flavanol cocoa beverage (43mg total flavanols) for 6 weeks.

Endothelial function was determined by brachial artery reactive hyperemia. Plasma was analysed for lipids (TC, HDL, LDL), hormones, total nitrate/nitrite, activation of cellular adhesion markers, and platelet function and reactivity.

There was a significant increase (76%) in brachial artery hyperemic blood flow (p<0.05) after 6 weeks of cocoa ingestion in the group consuming high flavanol cocoa, compared with only a non significant (32%) increase in the low flavanol cocoa group. There was also a 2% increase in FMD. There was a significant decrease in plasma levels of soluble vascular cell adhesion molecule-1.

It was concluded there were beneficial vascular effects of consumption of high flavanol-cocoa products in hypercholes-terolemic women. As the vascular health is a primary regulator of cardiovascular disease, these improvements suggest that high-flavanol products will help reduce the risk of CVD.

This was the first study to look at flavanol consumption in this group of individuals, which is important as this group is at higher risk of CVD. This adds to the body of evidence that a flavanol rich diet provides cardiovascular protection. The study also measured plasma epicatechin concentration so this could be correlated to improved FMD.

Small sample size. Was not clear which specific components of the flavanol-rich chocolate contributed to the observed benefits, so cannot be concluded if it is purely attributable to flavan-3-ol.

! 26!

Reference

Purpose/ Objective

Population, Setting, Sample, Age Range

Study Design Methods/ Measures

Key Findings Conclusions Rationale for inclusion

Limitations and Further Research Needed

Mink et al. (2007) (17)

Using the USDA database, estimate dietary flavonoid intake and determine the relationship between flavonoid intake and CVD mortality.

N = 34489 postmenopausal women from the Iowa Women’s Health Study who were free from CVD. Study conducted in Iowa, United States of America.

Cohort study from the Iowa Women’s Health Study, of postmeno-pausal women who were free from CVD. Completed a FFQ at baseline, and determined intake of 7 subclasses of flavonoids. Intakes were categoriesed into quintiles. Women were followed up after 16 years.

CVD, CHD, stroke and total mortality.

The only flavonoids associated were: anthocyanidins and CHD, CVD and total mortality, and between flavanones and CHD, and between flavones and total mortality. For individual foods, chocolate was associated with CVD (as well as bran, apples or pears or both and red wine, grapefruit and strawberries – either CVD, CHD, stroke, or total mortality).

Certain flavonoids and food groups were associated with reduced risk of CHD, CVD and all causes mortality.

This study showed that flavan-3-ol may not be significantly associated with reduced risk of CVD, but that chocolate (which is high in flavan-3-ol), may reduce risk, therefore it may be other components in chocolate other than flavan-3-ol that acts to reduce CVD risk.

This study is only observational, and they may not have determined all foods high in flavan-3-ol in the FFQ, for example dark chocolate, which will significantly contribute to total flavan-3-ol intake.

! 27!

Reference

Purpose/ Objective

Population, Setting, Sample, Age Range

Study Design Methods/ Measures

Key Findings Conclusions Rationale for inclusion

Limitations and Further Research Needed

Faridi et al. (2008) (31)

To determine the effects of dark chocolate and liquid cocoa intake on BP and endothelial function in obese individuals.

N = 45 adults. Mean age 52 years. Study conducted in Connecticut, United States of America.

Randomized, controlled, single-blind crossover trail. Phase 1 - Participants consumed either a solid dark chocolate bar (22g cocoa powder) or a cocoa-free placebo bar. Phase 2 – participants consumed sugar-free cocoa (22g cocoa powder), sugared cocoa (22g cocoa powder) or a placebo (0g cocoa).

Endothelial function via FMD, BP (SBP and DBP).

Solid dark chocolate and liquid cocoa improved FMD compared with placebo (p <0.001). Dark chocolate and sugar-free cocoa reduced BP compared to placebo (p = 0.01). Sugar-free cocoa improved endothelial function significantly more than regular cocoa (p <0.001).

Both dark chocolate, and liquid cocoa improve endothelial function and BP in overweight individuals. Sugar-free cocoa products may augment these effects.

Various studies have demonstrated benefits of chocolate and cocoa on CVD risk factors due to high flavanol content, but this was the first study that demonstrated that the sugar in these foods might attenuate its effects. This is important, as high sugar intake is associated with obesity, which increases the risk of CVD.

The study was of short duration, so only demonstrated the short term effects of chocolate and cocoa ingestion, and not its long term impacts on CVD risk, or risk factors. Further, it did not specify the amount of flavonoid in the chocolate, or measure blood levels of epicatechin, thus one cannot determine if beneficial effects are a result of these compounds.

! 28!

Reference

Purpose/ Objective

Population, Setting, Sample, Age Range

Study Design Methods/ Measures

Key Findings Conclusions Rationale for Inclusion

Limitations and Further Research Needed

Janszky et al. (2009) (43)

To determine the long-term effects of chocolate consumption in those with established CHD.

N = 1169 non-diabetic patients hospitilised with a confirmed first acute MI. Study conducted in Stockholm, Sweden.

Population-based inception cohort study, in Sweden, as part of the Stockholm Heart Epidemiology Program. Participants self-reported their usual chocolate intake over the past year. 8 year follow up.

Cardiac mortality, and nonfatal outcomes.

There was a significant inverse association between chocolate consumption and cardiac mortality. Hazard ratios for those who consumed chocolate less than once per month, up to once per week and twice or more per week was 0.73, 0.56, and 0.34 respectively.

Chocolate, which is high in flavonoids, reduced the risk of cardiac mortality. This relationship is dose dependent, with higher chocolate intakes associated with lower CVD risk.

This study clearly demonstrated there is a dose-response relationship between high flavonoid chocolate and CVD. This is important as it demonstrates those at higher risk of CVD events may benefit from moderate chocolate consumption.

Causality cannot be determined as the study is subject to various confounding variables. The type of chocolate consumed was not differentiated, so levels of cocoa, and thus flavonoids cannot be determined. May have been more beneficial to determine quantity of dark chocolate consumed.

! 29!

Reference

Purpose/ Objective

Population, Setting, Sample, Age Range

Study Design Methods/ Measures

Key Findings Conclusions Rationale for Inclusion

Limitations and Further Research Needed

Njike et al. (2011) (53)

To determine the effect of intake of sugar-sweetened versus sugar-free cocoa beverages on endothelial function.

N = 44 overweight but otherwise healthy individuals. Study conducted in Connecticut, United States of America.

Randomized, controlled, crossover trial. Participants received a sugar free cocoa (22g/d), a sugar-sweetened cocoa (22g/d cocoa and 91g/d sugar) or placebo (no cocoa, 110g/day sugar. Treatments were administered twice daily for 6 weeks, with a 4 week washout period.

Primary outcome measure was endothelial function via FMD in brachial artery.

Consumption of cocoa-containing beverages (in both sugar-sweetened and sugar-free groups) resulted in improved FMD in comparison to placebo group (p = <0.01). BMI and body weight did not change in the treatment groups versus the placebo.

Both sugar-free and sugar-sweetened cocoa beverages have the potential to improve endothelial function in overweight but healthy participants.

This was the first trial to assess the effects of sugar-free versus sugar-sweetened cocoa on endothelial function. It demonstrated that presence of sugar did not impact the beneficial effects of cocoa on endothelial function.

Doesn’t specify the quantity of flavan-3-ol (or any other flavonoid) in the cocoa beverages. Larger sample size and duration may have seen significant differences between sugar-free and sugar-sweetened beverages on endothelial function and waist circumference, which were close to significance.

! 30!

Reference

Purpose/ Objective

Population, Setting, Sample, Age Range

Study Design Methods/ Measures

Key Findings Conclusions Rationale for Inclusion

Limitations and Further Research Needed

Curtis et al. (2012) (44)

To examine the effect of flavan-3-ol on CVD risk in postmeno-pausal women with type 2 diabetes currently receiving standard care including statin therapy.

N = 93 participants aged 51-74 years receiving standard care for type 2 diabetes. Study conducted in Norfolk, United Kingdom.

Parallel-design, placebo-controlled trail. Subjects consumed 27 g/day (split dose) flavonoid-enriched chocolate (850mg flavan-3-ol [90mg epicatechin] and 100mg isoflavones, or matched placebo for 1 year.

Insulin resistance (HOMA-IR and QUICKI), HbA1c, blood glucose, TC:HDL cholesterol, LDL cholesterol, 10 year coronary heart disease risk (derived from UK Prospective Diabetes Study algorithm) and BP.

Reduction in peripheral insulin resistance (HOMA-IR P = 0.004), insulin sensitivity – (QUICKI p = 0.04) resulting from decreases in insulin levels (p = 0,02). Significant reduction in TC:HDL ratio (p = 0.01) and LDL cholesterol (p = 0.04). 10-year CHD risk was attenuated (p = 0.02). There was no impact on BP, HbA1c or glucose.

This study found that after 1 year intake of flavan-3-ol and isoflavone rich chocolate, CVD risk biomarkers improved suggesting there may be added benefit of flavonoids to the standard CVD risk management in postmeno-pausal type 2 individuals.

This study was the first to look at flavonoid intake and CVD risk in medicated postmeno-pausal women with T2DM, which is particularly important as this group has increased CVD risk. It also looked at 10 – year CHD risk score, instead of exclusively examining CVD risk biomarkers. Was also longer in duration than most other studies.

There were various limitations of this study, as it had a high dropout rate (21%), as it was difficult for them to consume the chocolate for the duration of the study. Levels of flavonoid given may be unachievable in a normal diet, so the extent to which the findings can be portrayed in reality is questionable.

! 31!

Reference

Purpose/ Objective

Population, Setting, Sample, Age Range

Study Design Methods/ Measures

Key Findings Conclusions Rationale for Inclusion

Limitations and Further Research Needed

Ellinger et al. (2012) (49)

To determine the effect of epichatechin ingested via cocoa products on change in BP.

Not defined. Systematic review of 16RCTs of SBP and 15 RCTs on DBP, a nonlinear meta-regression model was used to determine the relationship of epicatechin dose and BP.

SBP and DBP. An intake of 25mg epicatechin/day resulted in a mean reduction of -4.2mmHg in SBP and -2.0mmHg in DBP.

It was found that BP reduction is dependent on the dose of epicatechin. The reductions observed in BP have clinical significance, as only a small amount (25-30g) dark chocolate (providing 25mg epicatechin), is required to significantly reduce SBP and DBP, and thus the risk of CVD.

This study helps to explain why there were no observed effects in certain studies looking at the relationship between epicatechin and BP, and that higher doses are required to have a more beneficial or significant effect.

The studies were mostly done in pre-hypertensive and hyperten-sive individuals, so these results may not be applicable for normotensive individuals.

! 32!

Reference

Purpose/ Objective

Population, Setting, Sample, Age Range

Study Design Methods/ Measures

Key Findings Conclusions Rationale for Inclusion

Limitations and Further Research Needed

Hooper et al. (2012) (4)

To carry out a systematic review of literature concerning the effects of chocolate, cocoa and flavan-3-ols on major CVD risk factors.

N = 1297 Systematic review and meta-analysis of 42 acute or short-short term chronic parallel or crossover RCTs of chocolate, cocoa, or flavan-3-ols.

Outcomes included FMD, lipoprotein concentration, BP, fasting insulin, glucose, HbA1c, insulin resistance (by using HOMA-IR and QUICKI, insulin sensitivity index, mean arterial pressure, and CRP.

Meta-analysis showed there was acute improvement in FMD 2 hours after ingestion of chocolate/cocoa, and after chronic intake. There were also significant reductions in fasting serum insulin concentrations, serum insulin after glucose challenge, HOMA-IR, and ISI after chocolate or cocoa interventions. There were also significant reductions in DBP after chronic intake, as well as marginal significant improvements on LDL and HDL cholesterol. Greater effects…

This systematic review was able to demonstrate that chocolate or cocoa intervention reduces insulin resistance as a result of a reduction in secretion of insulin. They also concluded the strongest relationship between CVD biomarkers and cocoa intake was FMD. There are weaker relationships between DBP, triglycerides and mean arterial pressure. Benefits of HDL were more significant in…

This is an up-to-date review on all studies looking at flavan-3-ol, chocolate and cocoa consumption, and its relationship to cardiovascular disease. This study also demonstrates there may be a dose-dependent effect of flavan-3-ol (epicatechin), and reduction of DBP, so higher levels of intake may be required to observe significant effects.

There were no trials that were at low risk of bias, few were independently funded (i.e. they were funded by industry), which may have effected the conclusions drawn. Therefore results must be interpreted with caution. The studies were all of short duration (<18 weeks). No long term trials have examined the effect of chocolate or cocoa intake and CVD risk biomarkers. As chocolate has the potential…

! 33!

Hooper et al. (2012) continued

seen at doses of epicatechin >50mg/day.

long-term trials. It was concluded the 1.3% increase in FMD is likely to have clinical importance, especially in combination with improved HOMA-IR.

to cause energy imbalance in humans, it would be necessary to determine if long-term intakes did impact CVD risk biomarkers, and also CVD risk.

! 34!

Reference

Purpose/ Objective

Population, Setting, Sample, Age Range

Study Design Methods/ Measures

Key Findings Conclusions Rationale for Inclusion

Limitations and Further Research Needed

McCullough et al. (2012) (6)

To observe the relationship between seven different flavonoid subclasses and CVD.

N = 38,180 men and 60,289 women with mean age of 70 and 69 years respectively. Study conducted in Massachusetts, United States of America.

Prospective cohort study. Cancer Prevention Study II Nutrition Cohort. Completed questionnaires on medical history and lifestyle behaviors, and a validated FFQ of 152 food items. Cox proportional hazards modeling used to determine RR seven subclasses of flavonoids and CVD mortality after a 7-year follow up.

CVD mortality. Mean flavonoid intakes for both men and women were 268mg/d. It was found the subjects (both men and women) with flavonoid intakes in the top (compared to the bottom) quintile had a decreased risk of fatal CVD. Multivariate-adjusted: 17% reduced RR for fatal CVD (men and women) in highest intake of flavan-3-ol, compared to the lowest quintile (p = 0.02)

Supports other observational and experimental findings that flavan-3-ol reduces the risk of CVD. Also concluded even small amounts of flavonoid intake may be beneficial, as many of the observations were nonlinear, with lower risk in those with modest intakes.

The large size of this prospective study enabled the authors to clearly portray the relationship between flavonoid subclasses and the risk of CVD. It helps support previous observational data that higher intakes of flavan-3-ol reduce the risk of fatal CVD.

The data is observational, so causality cannot be concluded. It is possible that flavonoid intake was misclassified, as dark chocolate (a main food containing flavan-3-ol) was not included.

! 35!

2.4 Conclusions

The literature demonstrates there may be an inverse relationship between the intake of a) total

flavonoids, and b) flavan-3-ols, and CVD risk. However, findings are not homogeneous and

there are still gaps in the literature as to dose required, the long-term health benefits of total

flavonoids and flavan-3-ols, and their potential effects on CVD. Further, there is yet to be any

research investigating the relationship between flavonoid intake and CVD in New Zealand.

Therefore, the aforementioned knowledge needs now be applied to the New Zealand

population. Findings from other countries may not translate to New Zealand, and the different

cultural groups who experience different rates of CVD, specifically Maori and Pacific

Islanders. As CVD is one of the leading causes of death in New Zealand, it is of paramount

importance to expand on research that may help reduce its prevalence. No studies have

investigated the relationship between flavan-3-ol intake and 5- year CVD risk score using a

nationally representative sample, or assessed if there is a difference between genders or

ethnicities. The present study aims to achieve these objectives.

! 36!

3 Objective Statement

Literature suggests intake of flavonoids/flavan-3-ols and CVD are related; however no

research has investigated whether this relationship holds true in New Zealand. The

relationships demonstrated from various studies internationally provide insight into what may

be expected in New Zealand, yet results cannot be generalized to this population. Thus the

objectives of this study are as follows:

• To expand the NZFCDB to include flavonoids using the two major flavonoid

databases in use worldwide: The United States Department of Agriculture (USDA)

flavonoid database and Phenol Explorer.

• To estimate the total flavonoid intake and specifically intake of flavan-3-ols in adults

in New Zealand adults using dietary intake data from the NZANS 2008/2009, and to

determine if there are differences in intake between different age groups, BMI

categories, sex, NZdep quintiles and ethnicities.

• To assess the relationship between flavonoid/flavan-3-ol intake and 5-year CVD risk

scores.

Researchers are continuously looking for ways in which diet can influence the risk of chronic

diseases, so findings will be important for both science and society.

3.1 Hypothesis

A higher total intake of total flavonoids and flavan-3-ols is associated with lower 5-year CVD

risk score in New Zealand adults.

! 37!

4 Participants and Methods

4.1 Study Design

The current study is a cross-sectional study design, looking at the relationship between dietary

intake of flavan-3-ol and 5-year cardiovascular disease risk score in participants from the

NZANS.

As this study involves secondary analysis of the NZANS data, no ethical consent was

necessary. The Ministry of Health has given permission to use the NZANS data.

4.2 New Zealand Adult Nutrition Survey 2008/2009 Methods

4.2.1 Participants

The NZANS was a national population-based nutrition survey, conducted on a total of 4721

adult participants. Face-to-face interviews were carried out with each participant to obtain

data on their dietary intake and eating patterns, dietary supplement use, food security, body

size, BP and biochemical measures. Data from 24-hour food recalls is the primary data used

to determine nutrient intake, and ultimately flavonoid intake for the purpose of this study.

This study had a total of 1040 Maori, and 757 Pacific Island participants. A representative

sample was achieved by oversampling such ethnic groups, as well as younger and older age

groups, primarily though ‘screened’ sampling.

4.2.2 24-Hour Food Recalls

During the interviews, 24-hour food recalls were carried out in the participant’s homes.

Interviewers asked each participant to recall all the food and drinks they consumed in the past

! 38!

24 hours, or the previous day. Interviews were aided using the LINZ24 module of the Abbey

software package, a four-step program that helped ensure detailed information was obtained.

The first step was a basic food recall, or a ‘quick list’ of the foods, beverages and supplements

consumed over the chosen 24-hour period. The second step required more detailed

information to be obtained from the participants, including brand or product names, what time

the food was consumed, and where the food was obtained. The third step included questions

about how much food or beverage was consumed. This was done using aids such as food

photographs and volume measures. The fourth and final step consisted of a comprehensive

review of all recorded food and beverages to ensure everything was correct, and give

participants another chance to add anything they may have missed out.

Interviews were conducted in an even spread throughout the week, with a minimum of 10%

during the weekend days to enable data collection representative of food consumption across

the week. Further, 25% of the participants were asked to repeat a 24-hour diet recall within

one month of the first interview to determine any intra-individual variation in food and thus

nutrient intake. This method of dietary assessment was used, as it allowed for an accurate

representation of the food and nutrients consumed by each participant.

4.2.3 Matching to Nutrient Data

Food and beverages that were consumed according to the 24-hour recalls were matched to the

NZFCDB – a database developed and updated by Plant and Food Research Ltd (PFR). Within

this database is an electronic subset of data named FOODfiles, which contains 55 core

components of 2710 foods. This was used as the primary source of food composition data.

There were 11,850 food descriptions reported by the participants, and these food descriptions

! 39!

were firstly matched to the nutrient line of a food in FOODfiles. If no match was found,

overseas food composition data was used. If consumption of the food item was high enough,

and there was no match in any database, the food was analysed. For lower frequency food

items a recipe was developed and the nutrient content calculated by PFR.

4.2.4 Anthropometry

The participants BMI was measured while they were wearing light clothing and no shoes.

Two measurements of height (m), and weight (kg), were taken, and if these differed by >1% a

third was taken. The mean of these two (or three) measurements was used to determine BMI

using the Quetelet formula: BMI = weight (kg) / height (m)2.

4.2.5 Blood Pressure

A total of 4407 participants had their BP measured. In the current study BP measurements

from the NZANS were used to determine the participant’s 5-year CVD risk scores. BP was

measured using an OMRON HRM 907 instrument, which has an automatic cuff inflation, and

records pulse, diastolic and SBP. Blood pressure was measured in triplicate. SBP was the only

BP measurement used to determine 5-year CVD risk score.

4.2.6 Blood Samples

A total of 3349 participants gave a blood sample. These indices were measured using blood

samples collected into three vacutainers from a forearm vein. One 4 ml vacutainer containing

ethylene diamine tetra-acetic acid (EDTA) was transported to Canterbury Health Laboratories

at 4°C to test HbA1c. One 10ml vacutainer with no additive was allowed to separate at the

! 40!

local laboratory and the serum was also transported to Canterbury Health Laboratories at 4°C

and analysed for TC and HDL cholesterol. Participants were required to give informed

consent before any blood samples were collected.

4.2.7. Diabetes

The participant’s diabetic status was determined by the history of diagnosed diabetes and the

HbA1c measurement in blood. Participants that were categorised as having ‘diagnosed

diabetes’, or had an HbA1c of greater than 6.5%, were deemed as diabetic.

4.2.8 New Zealand Index of Deprivation

The NZ Index of Deprivation 2006 is a measure of the level of socioeconomic deprivation in

each defined ‘meshblock’ according to the following Census variables: income, benefit

receipt, transport (access to car), household crowding, home ownership, employment status,

qualifications, support (sole-parent families), and access to a telephone. Individuals are then

placed into quintiles 1-5, with quintile 1 being the least deprived areas, and quintile 5 being

the most deprived areas. Note “meshblocks vary in size from part of a city block to large areas

of rural land. Each meshblock abuts another to cover all of New Zealand” (27,55). There were

a total of 32,173 meshblocks defined in the 2006 Census.

4.2.9 Ethnic Groups

Participants were able to report affiliation with up to nine different ethnic groups. For the

NZANS, total response standard output was used. However, prioritised ethnicity was used for

the purpose of the current study. This method allocates each participant who affiliates with

! 41!

one or more ethnic groups to a single mutually exclusive ethnic group based on the pre-

determined hierarchy system. Ethnicity was classified hierarchically into three ethnic groups:

Maori, Pacific and New Zealand European and Others (NZEO) (55). For example, if the

participants reported being both Maori and Pacific, they were classified as being Maori.

4.3 CVD Risk Score

Based on the self-reported data from the NZANS questionnaire and the blood test results from

the NZANS, an individual’s 5-year absolute CVD risk was calculated using the New Zealand

Cardiovascular Risk Charts (Figure 3.1). The ANS participants were categorized into three

risk levels (mild, moderate to high, and high) according to their sex, diabetic status, smoking

history, the TC:HDL ratio, SBP, and age. As the age category 15-34 years is not included in

the New Zealand Cardiovascular Risk Charts, calculations for someone who is 35 were used.

! 42!

Figure 3-1 New Zealand Cardiovascular Risk Charts

! 43!

4.4 Flavonoid Database Development

As FOODfiles does not contain data on flavonoid values for foods, it was not previously

possible to determine flavonoid intake for participants in the NZANS, and thus in New

Zealand. Therefore, in the present study flavonoid data were estimated for the foods in the

FOODfiles database using information from the two most complete flavonoid databases

worldwide, the flavonoid USDA and the Phenol Explorer databases.

4.4.1 Flavonoid Databases

The ‘USDA Database for the Flavonoid Content of Selected Foods’ was developed in the

United States of America due to the growing evidence and interest in the health effects of

dietary flavonoids. This database contains data on 26 different flavonoid compounds from 5

subclasses (flavonols, flavones, flavanones, flavan-3-ols, and anthocyanidins), found in 500

different foods. The data was obtained through analysis of flavonoid content of food in

various studies, and only included when the particular data was generated by acceptable

analytical procedures. These included methods that lead to good separation of flavonoid

compounds such as column chromatography and high-performance liquid chromatography.

As well as controlling for analytical methods, all food composition data used was evaluated

using a quality index that considered the sampling plan, sample handling, number of samples

and quality control. As this database is known for its accuracy in the estimation of flavonoid

content in food, it was the primary database used to estimate flavonoid values for the foods

the FOODfiles list, in accordance with previous research.

! 44!

4.4.2 Matching FOODfiles to USDA

In order to estimate to flavonoid values for the FOODfiles foods, where possible, direct

matches were made to foods from the USDA database. This was done by making sure the

names of foods being matched were the same, or the food was very similar. In order to

determine if foods were a correct match, energy values (kJ) per 100g of the FOODfiles food,

and the USDA food were compared. If a food was not considered a close enough match then

where appropriate, data from Phenol Explorer database was used, or a composite was

developed.

4.4.3 Matching FOODfiles to Phenol Explorer

There were a number of foods in the FOODfiles list that contained flavonoids, for which there

was no match in the USDA database. In this case, data from the European FCDB Phenol

Explorer was used. This database contains more than 35,000 content values for 500 different

polyphenols in ~500 food and beverages. Flavonoid data was derived from the systematic

collection of more than 60,000 original content values found in more than 1,300 scientific

publications. Each of these publications were critically evaluated before they were included in

the FCDB (52,56).

4.4.4 Retention Factors

It has been reported there is a loss of flavonoids during culinary preparation of raw foods

(5,7). Only a small number of cooked foods have been analysed for their flavonoid value in

each flavonoid FCDB. As many of the foods in FOODfiles are cooked, there was no exact

match in either flavonoid FCDB. If these foods were not included in the updated FOODfiles

database, it would underestimate flavonoid consumption. Therefore, cooked foods in

! 45!

FOODfiles were matched to a raw food in either USDA or Phenol explorer, and adjustments

were made due to loss of flavonoids during the cooking process. There are different retention

factors depending on the cooking method used. Retention factors are; 0.7 when foods are

fried/roasted, 0.35 when microwaved or steamed, and 0.25 when foods are boiled (5,54).

These retention factors were applied to all flavonoid subclasses except isoflavones, as they

have no cooking losses. The same method was used to provide flavonoid values for cooked

foods in the Spanish EPIC cohort by Zamora-Ros et al (2010) (5).

4.4.5 Flavonoid Classes and Subclasses

Names of flavonoid subclasses used in the USDA were matched to those in Phenol explorer.

As names of certain subclasses were different between the two databases, even though they

have the same structure, and thus are the same subclass of flavonoid, chemical structure of

these were checked to ensure the correct match was made. Data on the following flavonoid

Classes were obtained: flavanones, flavones, flavonols, flavan-3-ols, anthocyanidins,

isoflavones, and proanthocyanidins. and subclasses are (-)-Epicatechin, (-)-Epicatechin 3-

gallate, (-)-Epigallocatechin, (-)-Epigallocatechin 3-gallate, (+)-Catechin, (+)-Gallocatechin,

(+)-Gallocatechin 3-gallate, Apigenin, Cyanidin, Eriodictyol, Hesperetin, Isorhamnetin,

Kaempferol, Luteolin, Myricetin, Naringenin, Pelargonidin, Peonidin, Quercetin, Theaflavin,

Theaflavin-3,3'-digallate, Theaflavin-3'-gallate, Theaflavin-3-gallate, Vanillic acid,

Protocatechuic acid, 4-Hydroxybenzoic acid, 2-Hydroxybenzoic acid and 3-Hydroxybenzoic

acid.

4.4.6 Composite Development

There were a number of foods in the FOODfiles list that contained flavonoids, however there

! 46!

was no match in the USDA database or Phenol Explorer. For example, a chocolate-coated

biscuit or a vegetable stir-fry would be rich in flavonoids, yet there was no match for these in

either database. Therefore a number of composites were developed in order to obtain

flavonoid values for these foods. Composites were developed by using either the ingredients

on the packaging of the particular food, or finding a recipe on the internet. Recipes from New

Zealand websites were used when possible, for example, from ‘Women’s Weekly’ or

‘Healthy Food Guide’. The amount or proportion of each ingredient in the composite recipe

was based on that in the chosen recipe. Proportions were often adjusted so energy,

carbohydrate, protein, fat and water were matched as closely as possible to the FOODfiles

food. When recipes were developed using ingredient information from food packaging,

proportions were estimated unless specified on the package. Proportions were again adjusted

to match the nutrient line of the FOODfiles food as closely as possible. When developing

recipes that contained fruit or vegetables, dietary fibre was matched as closely as possible to

give a better indication if proportions were accurate. Recipes from the Internet were

sometimes used in conjunction with packet ingredient information to make estimations more

accurate. Refer to Figure 3.2 for an example of a composite.

! 47!

Food Name Water g/100g

Energy kJ/100g

Protein, total; calculated from total nitrogen g/100g

Fat, total g/100g

CHO, available g/100g

Proportion of recipe/ingredient in a recipe

Nutrient line for food from FOODfiles trying to match: Stirfry, Black Bean Beef, Chinese

80.78 364.12 6.25 5.76 2.8 1.0

Recipe using ingredients from FOODfiles in g/100g Pepper,Red Chilli,flesh,raw 90.4 136.08 1.06 0.4 6.2 0.011

Sauce,soy 72 144.25 5.94 0 2.7 0.06 Ginger,raw,peeled 90.4 109.57 0.88 0.4 4.8 0.011 Garlic,cloves,raw,peeled 64.3 404.09 7.94 0.6 14.93 0.011 Beans,black,seeds,raw 11.02 1208.6 21.63 1.42 47.6 0.012 Onion,flesh,raw 87.92 167.77 1.25 0.12 8.53 0.02 Sherry,dry 81 485.52 0.19 0 1.34 0.01 Beef,schnitzel,lean,raw 69.37 609.34 23.44 5.89 0 0.22 Carrot,raw 89.55 83.42 0.62 0.4 3.49 0.11 Broccoli,fresh,raw 89.4 110.87 3.78 0.5 1.75 0.11 Mushrooms,flesh and stem,raw 93.28 49.53 2.31 0.2 0.21 0.11

Oil,vegetable,blend 0.2 3663 0 99 0 0.04 Courgette,raw 93.13 70.1 1.96 0.22 1.75 0.11 Water,municipal 99.99 0 0 0 0 0.16 Nutrient Line for recipe in g/100g 81.17 353.62 6.86 5.44 1.99 1.0

Figure 3-2 Composite for a black bean and beef stir-fry using foods from FOODfiles

! 48!

4.4.7 Retention factors for composites

Retention factors were also applied to composites. In cases where all foods in the composite

are cooked (for example, in the stir-fry above), then one retention factor was applied to the

whole composite. In some cases, certain ingredients within a composite were not cooked,

such as lettuce and tomatoes in a burger. In this case, retention factors were only applied to

the foods within the composites that were cooked. Refer to Figure 3.3 for an example of a

composite with retention factors applied.

! 49!

Food Name Water g/100g

Energy kJ/100g

Protein, total; calculated from total nitrogen g/100g

Fat, total g/100g

CHO, available g/100g

Proportion of recipe/ ingredient in a recipe

Retention factor

Burger,`Big Mac',McDonald's

46.3 1116.86 13.13 15.6 19.19 1.0

Recipe using ingredients from FOODfiles

Yeast,baker's,dried 5 768.59 39.5 1.5 3.2 0.03

0.7

Flour,wheat,white,standard

13.3 1404.6 9.12 1.15 72.44 0.25 0.7

Flour,soy,full fat 8.25 1572.77 29.52 18.17 24.4 0.004 0.7

Oil,canola 0.2 3663 0 99 0 0.05 0.7

Water,municipal 99.99 0 0 0 0 0.12 N/A*

Salt,iodised,table 0.05 0 0 0 0 0.02 N/A

Tomatoes,assorted variety,flesh,skin and seeds,raw

94.15 72.25 0.75 0.4 2.69 0.04 1.0

Beef,patty, McDonald's

48.3 1398.6 26.88 25.67 0 0.32 N/A

Lettuce,assorted variety,heart,fresh

96.55 61.45 1.13 0.3 1.89 0.04 1.0

Pickle,McDonald's 95.07 12.24 0.29 0.2 0 0.04 1.0

Cheese, Cheddar, Tasty

33.8 1742.2 24.56 36 0 0.04 N/A

Sesame seeds, whole, dried, raw

4.69 2277.14 17.7 49.7 8.54 0.01 0.7

Onion,flesh,raw 87.92 167.77 1.25 0.12 8.53 0.04 1.0

Nutrient Line for recipe in g/100g

47.32 1116.21 13.48 15.55 18.91 1.0

* N/A given as these foods do not contain any flavonoids, so a retention factor is not applicable

Figure 3-3 Retention Factors for Cooked versus Uncooked Ingredients in a Composite using

foods from FOODfiles

! 50!

4.4.8 Adjustments for Concentrated Foods

There were several foods in the FOODfiles list that were a concentrated form of a raw

ingredient, for example tomato sauce, and instant coffee. Such foods will contain higher

levels of flavonoids than the raw food per 100g. Because of this, adjustments were made. The

flavonoid content in the raw food was multiplied by a factor that accounted for how much

more concentrated the food was in comparison to its raw form. For example, instant coffee

powder was 135 times more concentrated than brewed coffee, so the USDA value for coffee,

brewed from grounds was multiplied by 135 for each flavonoid in order to give the value for

coffee powder from the FOODfiles list.

4.4.9 Flavonoid Calculations

Once all FOODfiles foods had been allocated to either a direct match, a composite had been

developed, or it was decided there was no flavonoids in the particular food, flavonoid values

were calculated. This was done using Microsoft Excel where flavonoid values from USDA

and Phenol explorer were inserted into a spreadsheet beside the FOODfiles foods. They were

then adjusted according to retention factors, concentration of foods, and composite foods

proportions as described previously. Once flavonoid values were generated the main food

sources of total flavonoids and flavan3-ols were also identified from the data.

4.5 Participant Categories

For the purposes of this study, data from ANS participants was categorised for various

characteristics. Age was split into five categories: 15-18, 19-30, 31-50, 51-70 and 71+. BMI

was split into four categories: underweight (<18), normal (18-25), overweight (25-30) and

obese (>30). Sex was split into two categories: male and female. Ethnicity was split into three

! 51!

categories: Maori, Pacific and NZEO. Finally, NZdep was split into five categories: quintile 1

(least deprived) - quintile 5 (most deprived).

Participants were split into tertiles for intake of total flavonoid as follows:

• Low < 71.32mg/day

• Moderate intake = 71.32mg – 580.73mg/day

• High intake > 580.73mg/day

Participants were split into tertiles for intake of flavan-3-ols as follows:

• Low intake <1.82 mg/day

• Moderate intake = 1.82mg-49.84 mg/day

• High intake > 49.84 mg/day

4.6 Statistical Methods

Within each category comparisons were made to determine if there were significant

differences for total mean flavonoid intake and mean flavan-3-ol intake; for example it was

determined whether there was a difference in total flavonoid intake between different age

categories. ANOVA was used to determine if there were differences between the categories

and a Bonferonni correction applied for multiple comparisons. Comparisons between

demographic groups were also made within CVD risk category using a chi-squared test; for

example, whether there was a difference between age categories in those in the mild CVD risk

category.

! 52!

In order to determine the relationship between total flavonoid intake and being in a particular

CVD risk category, logistic regression was used to look at the odds ratio of being in each

CVD risk category, using low, medium and high flavonoid intakes. This was then adjusted for

age, BMI, sex, NZdep, and ethnicity. In an additional analysis, those in the low flavonoid

intake group were used as the reference group, and given an OR of 1.0. Those in the moderate

and high intake groups were then compared to those in the low intake group (reference group)

to determine whether flavonoid intake affected the odds of being in a particular CVD risk

category. The same was done for the relationship between intake categories of flavan-3-ols

and CVD risk category. Odds ratios, 95% confidence intervals and p-values were calculated.

All analyses were carried out for those NZANS participants with a complete set of data to

determine their 5-year CVD risk score and who had 24-hour recall data.

! 53!

5 Results

5.1 Response Rate and Data Used for Analyses

The target population for the NZANS was comprised of approximately 3.2 million adults who

were current New Zealand citizen’s aged 15 years and over. The final weighted response rate

was 61%, with a 31% refusal to participate, and 8% uncontactable. A total of 4721

participants were included in the final survey. Of the 4721 participants in the survey, data

from a total of 3126 participants (66%) were used in these analyses. These were the

participants that had completed data for 24-hour diet recalls and all data required for

calculating CVD risk score profile.

5.2 Sample Characteristics

Table 5.1 shows socio-demographic characteristics of the participants included in the

flavonoid dataset, and their mean total flavonoid and flavan-3-ol intake. The sample was

predominantly NZ European (69% of participants), with 19% Maori, and 12% Pacific.

Participants were predominantly female (56%).

5.3 Differences in Flavonoid and Flavan-3-ol Intake Between Age Categories, BMI, Sex,

Prioritised Ethnicity and NZdep Quintile

As shown in Table 5.1, when comparing mean total flavonoid intakes, there were significant

differences in intake between age categories. The highest intake was in those aged 71+ years

(679.8mg/day), and the lowest intake in those aged 15-18 years (117.1mg/day) (p<0.0001).

There were also significantly higher intakes in females compared to males, and in NZ

Europeans compared to other ethnic groups. Similarly, for flavan-3-ol intake, there were

! 54!

significant difference between age categories with the highest intake in those aged 71+ years

(60.7mg/day), and lowest intake in those aged 15-18 years (6.5mg/day). There were also

significantly higher intakes in females compared to males, and in NZ Europeans compared to

other ethnic groups.

5.4 Cardiovascular Disease Risk Categories

5.4.1 Mild CVD Risk Category

As shown in Table 5.1, for those in the mild CVD risk category, there was a significant

difference between age categories with 31.7% in the age category 31-50 years, and the

minority aged 71+ years (10%) (p <0.0001). There were more people in the normal BMI

category than other categories, compared to those who were underweight, overweight or

obese, more females than males and more in quintile 5 (most deprived) than other quintiles.

5.4.2 Moderate to High CVD Risk Category

For those in the a moderate to high CVD risk category, there was a significant difference

between age categories, with more of those aged 71+years (46.3%) in this CVD risk category,

and minority of those aged 15-18 years (0%) (p <0.0001). There were also more people in the

obese BMI category compared to other BMI categories.

5.4.3 High CVD Risk Category

For those in the a high CVD risk category, there was a significant difference between age

categories, with a majority aged 71+ years (67.4%), and a minority aged 15-18, and 19-30

! 55!

(0%) (p <0.0001). There were also more people in the overweight BMI category than other

BMI categories, more males than females, and more NZ Europeans than other ethnicities.

! 56!

Table 5.1 Demographic data for NZANS participants by categories of CVD risk and total flavonoid and flavan-3-ol intake

Total n (%)

n = 3126

Mild CVD risk n (%) a

n = 2010

Moderate to high

CVD risk n (%)

n = 717

High CVD risk n (%)

n = 399

Mean total

flavonoid intake

(mg/day) (SE)

Mean flavan-3-ol

intake (mg/day) (SE)

Age category (y)

15-18 416 (13.3%) 416 (20.7%) 0 (0%) 0 (0%) 117.1 (8.9) 6.5 (0.8)

19-30 376 (12.0%) 366 (18.2%) 10 (1.4%) 0 (0%) 258.8 (34.2) 19.9 (3.2)

31-50 851 (27.2%) 745 (37.1%) 100 (13.9%) 6 (1.5%) 473.4 (27.0) 39.6 (2.5)

51-70 680 (21.8%) 281 (14.0%) 275 (38.4%) 124 (31.1%) 660.8 (32.2) 58.1 (3.1)

71+ 803 (25.7%) 202 (10.0%) 332 (46.3%) 269 (67.4%) 679.8 (24.5) 60.67 (2.3)

P-valueb <0.0001 <0.0001 <0.0001 <0.0001 <0.0001

BMI category (kg/m2)

Underweight (<18) 65 (2.1%) 41 (63.0%) 15 (2.1%) 9 (2.3%) 295.3 (54.8) 23.4 (5.0)

Normal (18-25) 954 (30.5%) 745 (37.1%) 147 (20.5%) 62 (15.5%) 485.5 (30.0) 41.2 (2.8)

Overweight (26-30) 1105 (35.3%) 664 (33.0%) 266 (37.1%) 175 (43.9%) 481.1 (23.1) 41.1 (2.3)

Obese (>30) 1002 (32.1%) 560 (27.9%) 289 (40.3%) 153 (38.3%) 450.4 (28.7) 37.6 (2.7)

P valuec <0.0001 <0.0001 <0.0001 0.0138 0.0106

Sex

Male 1374 (44.0) 732 (36.4%) 336 (46.9%) 306 (76.7%) 430.7 (20.7) 35.7 (2.0)

Female 1752 (56.0%) 1278 (63.6%) 381 (53.1%) 93 (23.3%) 508.0 (21.6) 43.7 (2.1)

P valued <0.0001 0.074 <0.0001 0.0076 0.0038

! 57!

Total n (%)

n = 3126

Mild CVD risk n (%) a

n = 2010

Moderate to high

CVD risk n (%)

n = 717

High CVD risk n (%)

n = 399

Mean total flavonoid

intake (mg/day) (SE)

Mean flavan-3-ol

intake (mg/day) (SE)

Prioritised Ethnicity

Maori 585 (18.7%) 380 (18.9%) 153 (21.3%) 52 (13.0%) 343.1 (25.7) 28.4 (2.4)

Pacific 383 (12.3%) 267 (13.3%) 83 (11.6%) 33 (8.3%) 301.6 (28.1) 24.5 (2.7)

NZEO 2158 (69.0%) 1363 (67.8%) 481 (67.1%) 314 (78.7%) 497.5 (18.0) 42.2 (1.7)

P valuee 0.045 0.117 <0.0001 <0.0001 <0.0001

NZ Deprivation quintile

1 (least deprived) 485 (15.5%) 311 (15.5%) 104 (14.5%) 70 (17.5%) 511.5 (34.4) 42.8 (3.2)

2 583 (18.7%) 398 (12.7%) 122 (17.0%) 63 (115.8%) 482.8 (31.0) 40.5 (3.0)

3 529 (16.9%) 347 (17.3%) 112 (15.6%) 70 (17.5%) 492.4 (43.1) 42.1 (4.1)

4 690 (22.0%) 448 (22.3%) 159 (22.2%) 83 (20.8%) 440.2 (34.8) 37.9 (3.4)

5 (most deprived) 839 (26.8%) 506 (25.2%) 220 (30.7%) 113 (28.3%) 421.9 (28.0) 35.7 (2.7)

P valuef 0.035 0.091 0.402 0.2492 0.4454 a Percentage of those in each age, BMI, Sex, Ethnicity or NZdep index group, for each CVD risk category b For differences between age groups. c For differences between BMI categories. d For differences between genders. e For differences between ethnic groups f For differences between NZdep quintiles

! 58!

5.5 Main Food sources of Flavonoids and Flavan-3-ol

As seen in Table 5.2, the main food groups contributing to total flavonoid intake and

flavan-3-ol intake include sugar and sweets, non-alcoholic beverages, alcoholic beverages,

fruit and vegetables. Sugar and sweets included chocolate, which is known to be high in

flavan-3-ols, however that it is outside the scope of the thesis to investigate each individual

food, therefore it is unknown how much individual foods contributed to intake.

Table 5.2 Main food groups contributing to total flavonoid intake and flavan-3-ol intake*

Food group Examples of food items included

Total flavonoid intake by food group (% total

intake)

Total Flavan-3-ol intake by food group (% total

intake)* Sugar and sweets

Sugars, syrups, confectionery, chocolate, jam, honey, jelly,

sweet toppings and icing, ice-blocks, artificial sweeteners

64% 81%

Non-alcoholic beverages

All teas, coffee and substitutes, hot chocolate drinks, juices, cordial, soft drinks, water,

powdered drinks, sports and energy drinks

25% 17%

Fruit

All fruit, fresh, canned, cooked and dried 6% 1%

Vegetables

All vegetables (fresh, frozen, canned) including mixes, coleslaw, tomatoes, green

salads, legumes and pulses, legume products and dishes

(baked beans, hummus, tofu), vegetable dishes

2% 0%

Alcoholic beverages

Red wine, white wine, dessert wine, cider, beer 1% 0%

*See appendix for a full list of food groups

** Values may not add up to 100 due to rounding.

! 59!

5.6 Intake of Flavonoids and Flavan-3-ols for each CVD risk category

As seen in Table 5.3, the majority of participants were in the mild CVD risk category

(64.3%) with a minority in the high CVD risk category (12.7%). Those in the high CVD

risk category had the highest flavonoid intake, with those in the mild CVD risk category

consuming the least. Those in the mild CVD risk category had the lowest flavan-3-ol

intake, with those in the moderate to high CVD risk category consuming the most.

Table 5.3 Mean total flavonoid and flavan-3-ol intake by CVD risk category in unadjusted

analyses

Cardiovascular risk

Total n (%) Mean total flavonoid

intake (mg/day) (SE)

Mean flavan-3-ol

intake (mg/day)

(SE)

Mild CVD risk 2010 (64.3%) 430.9 (17.1) 36.0 (1.7)

Moderate to High CVD

risk 717 (22.9%) 612.8 (32.0) 53.6 (3.0)

High CVD risk 399 (12.7%) 622.6 (48.0) 55.3 (4.6)

Total 3126 (100%) 471.1 (15.6) 39.9 (1.5)

5.7 Total Flavonoid Intake and CVD risk

As seen in Table 5.4, in the unadjusted analyses, those with a low moderate and high

flavonoid intake were less likely to have a low CVD risk score than those with a low

flavonoid intake. Those with a high flavonoid intake were more likely to be in the high

CVD category compared to those with a low flavonoid intake. Adjusted results show that

the only significant relationship seen was that those in the highest CVD risk category were

almost half as likely (OR=0.052, CI:0.29,0.91) to be in the highest category for total

flavonoid intake compared to the other groups (p=0.023).

! 60!

Table 5.4 Relationships between groups of total flavonoid intake and CVD risk categories. Results are presented as Odds Ratio (CI) for being in a particular total flavonoid intake group for each CVD risk category, compared to the low total flavonoid intake group

Mild CVD risk score Moderate to high

CVD risk score High CVD risk score

Unadjusted results as Odds Ratio (CI) for being in a particular total flavonoid

intake group for each CVD risk category, compared to the low total flavonoid

intakea group

Moderate total flavonoid

intakeb 0.68 (0.52,0.89) 1.52 (1.11,2.11) NS!

P value 0.007 0.010 NS!High total flavonoid

intakec 0.38 (0.30,0.50) 2.60 (1.94,23.48) 1.85 (1.24,2.78)

P value <0.0001 <0.0001 0.003

Adjusted resultsd as Odds Ratio (CI) for being in a particular total flavonoid

intake group for each CVD risk category, compared to the low total flavonoid

intakea group

Moderate total flavonoid

intakeb 0.93 (0.60,1.45) 1.15 (0.80,1.69) 0.74 (0.40,1.37)

P value NS! NS! NS!High total flavonoid

intakec 0.95 (0.63,1.44) 1.26 (0.88,1.80) 0.52 (0.29,0.91)

P value NS! NS! 0.023 a Low < 71.32mg b Moderate intake = 71.32mg – 580.73mg c High intake > 580.73mg d Adjusted for age, BMI, gender, NZdep index and prioritised ethnicity NS = not significant

As seen in Table 5.5, in the unadjusted analyses, those in the lowest category of total

flavonoid intake were twice as likely to be in the lowest risk category for CVD, compared

to those in the medium and high flavonoid intake groups. Adjusted results show those with

a higher total flavonoid intake are around 40% less likely (OR=0.62 CI:0.41,0.92) to have

a high CVD risk score (p=0.017).

! 61!

Table 5.5 Relationships between groups of total flavonoid intake and CVD risk categories. Results are presented as Odds Ratio (CI) for being in a particular total flavonoid intake group for each CVD risk category!

Mild CVD risk score Moderate to high

CVD risk score High CVD risk score

Unadjusted results as Odds Radio (CI) for being in a particular total flavonoid

intake group for each CVD risk category

Low total flavonoid

intakea 1.94 (1.54,2.46) 0.50 (0.38,0.65) 0.66 (0.46,0.96)

P value 0.0001 <0.0001 0.029

Moderate total flavonoid

intakeb 1.16 (0.92,1.46) 0.88 (0.67,1.15) 0.86 (0.62,1.18)

P value NS! NS! NS!High total flavonoid

intakec 0.48 (0.38,0.60) 2.05 (1.60,2.63) 1.67 (1.22,2.30)

P value 0.0001 0.0001 0.001

Adjusted resultsd as Odds Ratio (CI) for being in a particular total flavonoid

intake group for each CVD risk category

Low total flavonoid

intakea 1.06 (0.72,1.56) 0.82 (0.60,1.15) 1.67 (0.96,2.87)

P value NS! NS! NS

Moderate total flavonoid

intakeb 0.96 (0.67,1.37) 1.00 (0.73,1.36) 1.17 (0.76,1.80)

P value NS! NS! NS!High total flavonoid

intakec 0.99 (0.71,1.39) 1.14 (0.85,1.53) 0.62 (0.41,0.92)

P value NS! NS! 0.017 a Low < 71.32mg b Moderate intake = 71.32mg – 580.73mg c High intake > 580.73mg d Adjusted for age, BMI, gender, NZdep index and prioritised ethnicity NS = not significant

5.8 Flavan-3-ol intake and CVD risk

As seen in Table 5.6, in the unadjusted results, those with a moderate and high flavan-3-ol

intake were less likely to have a low CVD risk score than those with a low flavan-3-ol

intake. Those with a high flavonoid intake were more likely to be in the high CVD

! 62!

category compared to those with a low flavan-3-ol intake. In the adjusted analyses, the

only significant relationship seen was that those in the highest CVD risk category were

around half as likely (OR=0.48, CI:0.29,0.80) to be in the highest category for flavan-3-ol

intake compared to those in the lowest flavan-3-ol intake category (reference group)

(p=0.004)

Table 5.6 Relationships between groups of flavan-3-ol intake and CVD risk categories. Results are presented as Odds Ratio (CI) for being in a particular flavan-3-ol intake group for each CVD risk category, compared to the low flavan-3-ol intake group

Mild CVD risk score

Moderate to high

CVD risk score High CVD risk score

Unadjusted results as Odds Ratio (CI) for being in a particular flavan-3-

ol intake group for each CVD risk category, compared to the low flavan-

3-ol intakea group

Moderate Flavan-3-ol

intakeb 0.74 (0.57,0.97) 1.36 (1.00,1.87) 1.20 (0.80,1.78)

P value 0.032 NS NS

High Flavan-3-ol

intakec 0.40 (0.31,0.51) 2.50 (1.86,3.36) 1.89 (1.28,2.80)

P value <0.0001 <0.0001 0.002

Adjusted resultsd as Odds Ratio (CI) for being in a particular flavan-3-ol

intake group for each CVD risk category, compared to the low flavan-3-ol

intakea group

Moderate Flavan-3-ol

intakeb 1.01 (0.66,1.56) 1.10 (0.75,1.61) 0.69 (0.39,1.20)

P value NS NS NS

High Flavan-3-ol

intakec 1.04 (0.70,1.56) 1.19 (0.83,1.70) 0.48 (0.29,0.80)

P value NS NS 0.004

a Low intake <1.82mg/day b Moderate intake = 1.82mg-49.84mg/day

c High intake > 49.84mg/day

d Adjusted for age, BMI, gender, NZdep index and prioritised ethnicity

! 63!

As seen in Table 5.7, in the unadjusted analyses, those in the lowest category of flavan-3-

ol intake were almost twice as likely to be in the lowest risk category for CVD, compared

to those in medium and high flavn-3-ol intake groups. Adjusted results show those in the

lowest flavan-3-ol intake category (<1.82mg/day) are 80% more likely (OR=1.80,

CI:1.12,2.90) to have a high CVD risk score (p=0.016). Those in the highest flavan-3-ol

intake category (>49.84mg/day) are 40% less likely (OR=0.6, CI:0.40,0.89) to have a high

CVD risk score (p=0.011).

! 64!

Table 5.7 Relationships between groups of flavan-3-ol intake and CVD risk categories. Results are presented as Odds Ratio (CI) for being in a particular flavan-3-ol intake group for each CVD risk category

Mild CVD risk score

Moderate to high

CVD risk score High CVD risk score

Unadjusted results as Odds Radio (CI) for being in a particular flavan-3-

ol intake group for each CVD risk category

Low Flavan-3-ol

intakea 1.85 (1.47,2.33) 0.54 (0.41,0.70) 0.66 (0.45,0.94)

P value 0.0001 0.0001 0.021

Moderate Flavan-3-ol

intakeb 1.25 (0.99,1.57) 0.81 (0.62,1.06) 0.84 (0.61,1.15)

P value NS! NS! NS!High Flavan-3-ol

intakec 0.47 (0.37,0.58) 2.0 (1.64,2.69) 1.71 (1.25,2.35)

P value 0.0001 0.0001 0.001

Adjusted resultsd as Odds Ratio (CI) for being in a particular total

flavonoid intake group for each CVD risk category

Low Flavan-3-ol

intakea 0.97 (0.67,1.41) 0.87 (0.63,1.21) 1.80 (1.12,2.90)

P value NS! NS! 0.016

Moderate Flavan-3-ol

intakeb 0.98 (0.69,1.41) 0.98 (0.72,1.34) 1.15 (0.75,1.76)

P value NS! NS! NS!High Flavan-3-ol

intakec 1.03 (0.74,1.45) 1.12 (0.84,1.50) 0.6 (0.40,0.89)

P value NS! NS! 0.011 a Low intake <1.82mg/day b Moderate intake = 1.82mg-49.84mg/day

c High intake > 49.84mg/day d Adjusted for age, BMI, gender, NZdep index and prioritised ethnicity NS not significant

! 65!

6 Discussion and Conclusions

The present study was carried out to develop a flavonoid database in order to determine the

flavonoid intake of participants in the NZANS. This database was used to investigate

whether higher flavonoid intake was associated with lower CVD risk. This discussion

focuses on the relationship between flavan-3-ol and CVD in New Zealand adults. It will

also discuss the accuracy of the flavonoid database that was developed as the main

component of this thesis.

6.1 Main Findings

The unadjusted results of the current study indicated that higher intakes of flavan-3-ol were

associated with a higher CVD risk. However, when adjusted for age, sex, BMI, ethnicity

and NZdep, results indicate that those with a higher intake of flavan-3-ol have a

significantly lower chance of having a high CVD risk score. Those in the lowest flavan-3-

ol intake category had a significantly higher chance of being in the high CVD risk

category. It is not surprising that these results differed after adjustment for these factors,

which are known to influence CVD risk. Results were similar for total flavonoid intake,

showing that those in the highest intake group for total flavonoids had a significantly lower

chance of being in the high CVD risk category. These results are consistent with studies

carried out overseas (4,6), which have largely found those with a higher flavonoid intake,

particularly flavan-3-ol, have a lower CVD risk. It has been demonstrated that higher

intakes of flavan-3-ol decreases risk of CVD by various mechanisms. Firstly it has been

suggested that high flavan-3-ol intake is involved in the pathophysiology of CVD by

means of the NOS system. It has been suggested in vitro, that flavan-3-ol helps to activate

endothelial NOS, which helps to enhance the relaxation of the endothelium, thus

! 66!

modulating vascular tone and improving vascular health (7). It has also been suggested

high intakes can reduce the susceptibility of LDL cholesterol oxidation, which in turn

reduces risk endothelial dysfunction and vascular disease (5,7). These functions help to

explain why a higher intake of flavan-3-ols would reduce the risk of CVD.

!

In the present study it was found the mean intake of total flavonoids was 471.1mg/day.

The Spanish EPIC cohort by Zamora-Ros et al (2010) carried out on 40,683 subjects aged

35 to 64 years from northern and southern regions of Spain, found participants had a mean

total flavonoid intake of 313.26mg/day (5). A Finnish study on 2007 Finnish adults found

participants had a mean intake of 208.9mg/day (57). A US study assessing flavonoid intake

in a total of 8809 individuals from the NHANES 1999-2002, reported participants had a

mean flavonoid intake of 189.7mg/day (58). This shows total flavonoid intake was slightly

higher in the present study than several overseas studies. This may be due to different

methods used for dietary assessment. The Spanish EPIC study used information on usual

food intake during the preceding year, which took into account seasonal variation, with

data collected on cooking methods, frequency of consumption per week, and usual portion

size, along with a detailed diet history questionnaire (5). Therefore it would have taken

into consideration total flavonoid intake year round. The present study used 24-hour recalls

to assess dietary intake, therefore it may not show participant’s habitual year round intake

of flavonoids. However, the Finnish study by Ovaskainen et al. (2008), used 48-hour

dietary recalls, and the US study by Chun et al. (2007) used 24-hour recalls similar to the

present study, therefore there may be other reasons for the higher intakes observed in the

New Zealand population (57,58). Differences may be due to dietary habits in these

different populations, which are often dictated by culture and thus affect the intake of

flavonoids (58).

! 67!

In the present study it was found the main food group contributing to total flavonoid intake

was sugar and sweets. This was followed (in quantity) by non-alcoholic beverages, fruit,

vegetables and alcoholic beverages. The main foods contributing to total flavonoid intake

in the Spanish EPIC study were fruit, alcoholic beverages, chocolate, and vegetables (5,6).

Berries and other fruit were the main sources of flavonoids in the Finnish study (57), and

tea, citrus fruit juices, wine and citrus fruits were the main sources of flavonoids in the US

study (58). Different dietary patterns in these populations may be reasons for differences

seen in the main foods contributing to flavonoids, as well as different mean intakes.

Similar to the present study, the Spanish EPIC cohort by Zamora-Ros et al. (2010) used the

USDA database to estimate flavonoid intake, applied retention factors to cooked foods,

and developed a number of recipes (5). This may be a reason the findings from the present

study are most comparable to this study than other overseas studies. The US study by Chun

et al. (2007) did use the USDA database to estimate flavonoid intake, however they did not

adjust cooked foods using retention factors (58). This may have resulted in an

overestimation of flavonoid intake for this particular study. They did not however compile

composites such as in the present study, which may have resulted in an underestimation of

intake. In the Finnish study by Ovaskainen et al. (2008), 110 food items were individually

analysed for flavonoid content, and this was compiled with data from other sources to

update a Finnish FCDB (57). This method may have resulted in some foods consumed by

the Finnish population being excluded from the study, thus leading to potential

underestimation of intake. They did consider retention factors, and also considered

flavonoids in recipes, therefore results are comparable to the present study.

!

! 68!

The mean flavan-3-ol intake for the present study was 39.9mg/day. This was very similar

to that found in the Spanish EPIC cohort (2010), which found participants had a mean

flavan-3-ol intake of 32.47mg/day (5). The Finnish study by Ovaskainen et al. (2008),

found participants had an intake of 12mg/day, and the US study by Chun et al. (2007)

reported an intake of 156.6mg/day (5,57,58). It has been suggested people in the US drink

a lot of tea, which may be a reason for the higher observed intake of flavan-3-ols in this

study (5). The Spanish EPIC cohort (2010) used similar methods to the present study,

which may be why intakes were very similar. Differences in flavan-3-ol intake between the

present study compared to the Finnish (57) and US (58) studies may be due to differences

in methodology mentioned earlier. In the present study, the main food group contributing

to flavan-3-ol intake was sugar/sweets. Although no values were produced for individual

foods, it is likely that the individual food that contributed to this was chocolate, which is

known to contain high amounts of flavan-3-ols. Non-alcoholic beverages (such as tea and

chocolate flavored drinks) were the second highest contributing food group, followed by

fruit then alcoholic beverages. Similarly, in the EPIC study, tea, red wine, fruit (mainly

apples) and chocolate were observed as the main contributors of flavan-3-ol intake (5).

Further research is needed to identify the individual foods that contribute most to total

flavonoid and flavan-3-ol intake in this population.

!

6.2 Strengths and Limitations of the Present Study

6.2.1 Strengths

The present study was the first in New Zealand to investigate the relationship between

flavonoid intake and CVD risk. The participant data used was from the NZANS, which

was a large and national population based sample (16). Therefore an accurate

representation on the effects of flavan-3-ol intake and CVD risk score at the New Zealand

! 69!

population level was possible. As the sample size was large, it gave a large sample for each

CVD risk category, and flavonoid/flavan-3-ol intake category, enabling accurate

comparisons to be made. It was a strength to use a CVD risk score profile, as it takes into

account the CVD risk factors TC:HDL ratio, SBP, ethnicity, tobacco use, diabetes, and

age; which is preferable to using a single endpoint for CVD risk (26). This is because

single risk factors that fall above the ‘normal’ range may not necessarily increase ones risk

of CVD. Those with many marginal risk factors are at increased risk, as they act

synergistically to have potentially harmful overall effects (26). The CVD risk profile used

was also specific to New Zealand and considered higher risk of certain ethnic groups (27).

There are various strengths in the methodology of the database development. A

conservative approach was taken, in line with other studies investigating flavonoid

consumption (5). The present study used the same protocol for matching foods as that used

in the Spanish EPIC cohort by Zamora-Ros et al. (2010) (5). This method ensured that

foods were matched as accurately as possible.

Several methods were used to ensure flavonoid intake was not overestimated or

underestimated whilst developing the database. It has been demonstrated that postharvest

handling and manufacturing processes of plant foods can lead to reduction, and/or

elimination of flavonoids (5,6). Thus, applying retention factors to cooked and

manufactured products meant that flavonoid intake from cooked foods was not

overestimated. To ensure flavonoid intake was not underestimated, composites were

developed. That is, packet ingredients and/or recipes were used to make a match to the

food as closely as possible. If this step in the development of the database was excluded,

! 70!

foods that contained flavonoids but did not have a direct match to a food in either database,

would have been given a flavonoid value of 0mg/100g. Thus, flavonoid intake would have

been greatly underestimated. Although recipes could not be entirely accurate, developing a

composite is preferable to leaving particular foods out altogether.! As in the Spanish EPIC

Cohort by Zamora-Ros et al. (2010), we used both the USDA and Phenol explorer

databases, calculated flavonoid values for composite foods, and used recognised retention

factors (5) This ensured as many foods as possible in FOODfiles were given an accurate

flavonoid value. Flavonoid underestimation is a problem with various studies; therefore

intake of total flavonoids and flavan-3-ols may be more accurate than has been

demonstrated in other overseas studies. However, it must be acknowledged that 24-hour

recalls were used, and this is not a measure of habitual diet, thus food intake methods

could be improved to more accurately determine flavonoid intake in New Zealand.

6.2.2 Limitations

Similar to the Spanish EPIC study (2010), when developing the new FCDB, foods were

matched as closely as possible (5). In some cases the USDA database and Phenol Explorer

did not contain all the exact foods on the FOODfiles list. Exceptions were made if foods

were not an exact match, but were very close and contained a similar amount of

flavonoids. For example the beverage port in FOODfiles was matched to a red wine from

USDA database, as the flavonoid content would be expected to be similar given the

composition of both products. Further, there was no data for certain foods that are likely to

be high in flavonoids, for which consumption in New Zealand is also high. For example,

there was no match for flaxseed oil, but there was a high consumption of this food in the

NZANS. We were unable to substitute this for another type of oil, therefore this may result

in underestimation of flavonoid intake. Although it was a strength of the study to develop

! 71!

composites, there were challenges in determining the exact proportions of ingredients.

Furthermore, as there were limited cooked foods in the flavonoid databases, adjustments

were made to raw foods to account for potential loss of flavonoids during the cooking

process. The retention factors applied were based on the reported cooking methods, and

may not be an exact representation of the flavonoid content of each of these cooked foods.

There is limited data about retention factors for flavonoids for cooked foods in the

literature. As the timeframe for this project was been limited, the flavonoid dataset

generated provides a picture of flavonoid values. Each recipe or composite, once

calculated, was checked by a trained nutritionist and, not all the suggested modifications

were able to be made during the time period of this research. This has implications for the

flavonoid values produced, and may be an additional reason why calculated total flavonoid

values found in this study were higher than in previous research.

It was found that fruit, vegetables and red wine contributed highly to flavnon-3-ol intake. It

has been demonstrated that high consumption of fruit, vegetables and moderate

consumption of red wine is likely to reduce ones risk of CVD (2). Therefore, compounds

other than flavonoids in these foods may have somewhat contributed to the observed

results. However, these were not confounders that could be adjusted for.

6.3 Implications for Future Research

There are difficulties with all studies investigating the relationship between flavonoids and

CVD, due to difficulties with dietary exposure assessment. That is, the FCDB used needs

to provide accurate quantitative information on the specific compound in foods that is

being investigated in order to estimate exposure of participants in the studies (59). The

! 72!

flavonoid database produced in the present study is still a work in process, and refinements

still need to be made to give a more accurate representation of flavonoid content of New

Zealand foods. It must be noted it was outside the scope of this thesis to complete the

database, and analyses should be completed once all necessary modifications to the

database are made.

From the results of the present study it can be seen that there is a relationship between

flavonoid intake and CVD risk. These analyses should be repeated in the completed

flavonoid database, thus providing a more accurate quantitative assessment of flavonoid

content in the NZFCDB. Further RCTs and observational studies carried out in New

Zealand will help to further clarify the relationship and amounts of flavonoids required to

decrease ones risk of CVD.

6. Conclusions

In summary, the present study demonstrated a relationship between high total

flavonoid/flavan-3-ol intake, and mild/low CVD risk in New Zealand. Those with a

habitual high total and flavan-3-ol intake may reduce their risk of developing CVD.

Furthermore, it may reduce the likelihood of a cardiovascular event occurring in those

already at risk of CVD.

! 73!

7 Application to Dietetic Practice

As high consumption of flavonoids/flavan-3-ols is likely associated with lower risk of

CVD, we may see a decline in rates of CVD if consumption of foods high in

flavonoids/flavan-3-ols is promoted. It must be considered that the food that contributes

highly to flavan-3-ol intake is chocolate. While consuming chocolate with a high

percentage of cocoa (that is dark chocolate) may provide benefits this must be in the

context of a well balanced diet. It would not be recommended to consume large amounts of

chocolate, as this may impact energy balance and cause weight gain, a known risk factor

for CVD. Along with flavan-3-ols, a higher total flavonoid intake appears to reduce risk of

CVD, thus it would be recommended to increase consumption of other foods high in total

flavonoids such as tea, fruit and vegetables, as these foods contain other nutrients that are

cardio-protective, and may help reduce the risk of other chronic diseases.

!

!

! 74!

8 References

1. Levenson JW, Skerrett PJ, Gaziano JM. Reducing the global burden of cardiovascular disease: The role of risk factors. Preventive Cardiology. 2002;5(4):188–99.

2. Mann J, Chisholm A. Cardiovascular disease. Essentials of Human Nutrition. 4th ed. Oxford: Oxford University Press; 2012. p. 326–58.

3. Sacks FM, Svetkey LP, Vollmer WM, Appel LJ, Bray GA, Harsha D, et al. Effects on blood pressure of reduced dietary sodium and the dietary approaches to stop hypertension (DASH) diet. New England Journal of Medicine. 2001;344(1):3–10.

4. Hooper L, Kay C, Abdelhamid A, Kroon PA, Cohn JS, Rimm EB, et al. Effects of chocolate, cocoa, and flavan-3-ols on cardiovascular health: a systematic review and meta-analysis of randomized trials. American Journal of Clinical Nutrition. 2012;95(3):740–51.

5. Zamora-Ros R, Andres-Lacueva C, Lamuela-Raventos RM, jakszyn TBP, Barricarte A, Ardanaz E, et al. Estimation of dietary sources and flavonoid intake in a Spanish adult population (EPIC-Spain). Journal of the American Dietetic Association. 2010;110:390–8.

6. McCullough ML, Peterson JJ, Patel R, Jacques PF, Shah R, Dwyer JT. Flavonoid intake and cardiovascular disease mortality in a prospective cohort of US adults. American Journal of Clinical Nutrition. 2012;95:454–64.

7. Engler MB, Engler MM, Chen CY, Malloy MJ, Browne A, Chiu EY, et al. Flavonoid-rich dark chocolate improves endothelial function and increases plasma epicatechin concentrations in healthy adults. Journal of the American College of Nutrition. 2004;23(3):197–204.

8. Sies H, Schewe T, Heiss C, Kelm M. Cocoa polyphenols and inflammatory mediators. American Journal of Clinical Nutrition. 2005;81:304S–12S.

9. Heart Foundation. Statistics [Internet]: Heart Foundaiton; 2012 [cited 2012Sep.17]; Available from: http://www.heartfoundation.org.nz/know-the-facts/statistics

10. Deaton C, Froelicher ES, Wu LH, Ho C, Shishani K, Jaarsma T. The global burden of cardiovascular disease. Journal of Cardiovascular Nursing. 2011;26(4S):S5–S14.

11. Cameron VA, Faatoese AF, Gillies MW, Robertson PJ, Huria TM, Doughty RN, et al. A cohort study comparing cardiovascular risk factors in rural Maori, urban Maori and non-Maori communities in New Zealand. British Medical Journal. 2012;2:e000799.

12. Ministry of Health. Cardiovascular Disease (35+ years) [Internet]: Ministry of Health; 2008 [cited 2012Sep.17]. Available from: http://www.health.govt.nz/nz-health-statistics/health-statistics-and-data-sets/maori-health-data-and-stats/tatau-kahukura-maori-health-chart-book/nga-mana-hauora-tutohu-health-status-

! 75!

indicators/cardiovascular-disease-35-years

13. World Health Organisation. Noncommunicable disease country profiles 2011. Geneva: World Health Organisation;2011.

14. Ministry of Health. Report on New Zealand cost-of-illness studies on long-term conditions. Wellington: Ministry of Health;2009.

15. Scott WG, White HD, Scott HM. Cost of coronary heart disease in New Zealand. New Zealand Medical Journal. 1993;106(962):347–9.

16. Ministry of Health. A Portrait of Health: key results of the 2006/07 New Zealand health survey. Wellington: Ministry of Health; 2012.

17. Mink PJ, Scrafford CG, Barraj LM, Harnack L, Hong C-P, Nettleton JA, et al. Flavonoid intake and cardiovascular disease mortality: a prospective study in postmenopausal women. American Journal of Clinical Nutrition. 2007;85:895–909.

18. Siontis GCM, Tzoulaki I, Siontis KC, Ioannidis JPA. Comparisons of established risk prediction models for cardiovascular disease: systematic review. British Medical Journal. 2012;344:e3318.

19. D’Agostino RB, Vasan RS, Pencina MJ, Wolf PA, Cobain M, Massaro JM, et al. General Cardiovascular Risk Profile for Use in Primary Care. Circulation. 2008;117(6):743–53.

20. World Health Organisation. Global status report on noncommunicable diseases 2010. Geneva: World Health Organisation; 2011.

21. Williams M. Risk assessment and management of cardiovascular disease in New Zealand. The New Zealand Medical Journal. 2003;116(1185):1–3.

22. Arruda H, editor. Framingham Heart Study [Internet]: Framingham Heart Study Organisation; 2012 [cited 2012Sep.12]. Available from: http://www.framinghamheartstudy.org/risk/gencardio.html

23. Truett J, Cornfield J, Kannel W. A multivariate analysis of the risk of coronary heart disease in Framingham. Journal of Chronic Diseases. 1967;20(7):511–24.

24. Brindle P, Beswick A, Fahey T, Ebrahim S. Accuracy and impact of risk assessment in the primary prevention of cardiovascular disease: a systematic review. Heart. 2006;92:1752–9.

25. Kris-Etherton P, Eckel RH, Howard BV, Jeor SS, Bazzarre TL. Lyon diet heart study. Circulation. 2001;103:1823–5.

26. Milne R, Gamble G, Whitlock G, Jackson R. Discriminative ability of a risk-prediction tool derived from the Framingham Heart Study compared with single risk factors. The New Zealand Medical Journal. 2003;116(1185):1-3.

27. New Zealand Guidelines Group. New Zealand Primary Care Handbook 2012. 3rd ed. Wellington: New Zealand Guidelines Group; 2012.

! 76!

28. Jalil AMM, Ismail A. Polyphenols in cocoa and cocoa products: is there a link between antioxidant properties and health? Molecules. 2008;13:2190–219.

29. Hjermann I, Holme I, Byre KV, Leren P. Effect of diet and smoking intervention on the incidence of coronary heart disease. International Journal of Cardiology. 1981;318(8259):1303–10.

30. Heiss C, Kleinbongard P, DeJam A, Perre S, Schroeter H, Sies H, et al. Acute consumption of flavanol-rich cocoa and the reversal of endothelial dysfunction in smokers. Journal of the American College of Cardiology. 2005;46(7):1276–83.

31. Faridi Z, Njike VY, Dutta S, Ali A, Katz DL. Acute dark chocolate and cocoa ingestion and endothelial function: a randomized controlled crossover trial. American Journal of Clinical Nutrition. 2008;88:58–63.

32. Sesso HD, Gaziano JM, Liu S, Buring JE. Flavonoid intake and the risk of cardiovascular disease in women. American Journal of Clinical Nutrition. 2003;77:1400–8.

33. Schroeter H, Heiss C, Balzer J, Kleinbongard P, Keen CL, Hollenberg NK, et al. (–)-Epicatechin mediates beneficial effects of flavanol-rich cocoa on vascular function in humans. PNAS. 2006;103(4):1024–9.

34. Gordon T, Kannel WB. Premature Mortality From Coronary Heart Disease: The Framingham Study. The Journal of the American Medical Association. American Medical Association; 1971;215(10):1617–25.

35. Kannel WB, McGee D, Gordon T. A General Cardiovascular Risk Profile: The Framingham Study. The American Journal of Cardiology. 1976;38:46–51.

36. Wilson PWF, Castelli WP, Kannel WB. Coronary risk prediction in adults (The Framingham Heart Study). The American Journal of Cardiology. 1987;59(14):91–4.

37. Assmann G, Cullen P, Schulte H. Simple scoring scheme for calculating the risk of acute coronary events based on the 10-Year follow-Up of the prospective cardiovascular Münster (PROCAM) study. Circulation. 2002;105:310–5.

38. Conroy RM, Pyörälä K, Fitzgerald AP, Sans S, Menotti A, De Backer G, et al. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project. European Heart Journal. 2003;24:987–1003.

39. Woodward M, Brindle P, Tunstall-Pedoe H. Adding social deprivation and family history to cardiovascular risk assessment: the ASSIGN score from the Scottish Heart Health Extended Cohort (SHHEC). Heart. 2007;93:172–6.

40. Hippisley-Cox J, Coupland C, Vinogradova Y, Robson J, May M, Brindle P. Derivation and validation of QRISK, a new cardiovascular disease risk score for the United Kingdom: prospective open cohort study. British Medical Journal. 2007;335(7611):136–6.

41. Pietta PG. Flavonoids as Antioxidants. Journal of Natural Products. American Chemical Society; 2000;63(7):1035–42.

! 77!

42. Havsteen B. Flavonoids, a class of natural products of high pharmacological potency. Biochemical Pharmacology. 1983;32(7):1141–8.

43. Janszky I, Mukamal KJ, Ljung R, Ahnve S, Ahlbom A, Hallqvist J. Chocolate consumption and mortality following a first acute myocardial infarction: the Stockholm Heart Epidemiology Program. Journal of Internal Medicine. 2009;266(3):248–57.

44. Curtis PJ, Sampson M, Potter J, Dhatariya K, Aedin C. Chronic ingenstion of flavan-3-ols and isoflavones improves insulin sensitivity and lipoprotein status and attenuates estimated 10-year CVD risk in medicated postmenopausal women with type 2 diabetes: a 1-year, double blind, randomized, controlled trial. Diabetes Care. 2012;35(2):226–32.

45. Wedick NM, Pan A, Cassidy A, Rimm EB, Sampson L, Rosner B, et al. Dietary flavonoid intakes and risk of type 2 diabetes in US men and women. American Journal of Clinical Nutrition. 2012;95(4):925–33.

46. Ministry of Health. A Focus on Nutrition: Key findings of the 2008/09 New Zealand Adult Nutrition Survey. Wellington: Ministry of Health; 2012.

47. Wang-Polagruto JF, Villablanca AC, Polagruto JA, Lee L, Holt RR, Schrader HR, et al. Chronic consumption of flavanol-rich cocoa improves endothelial function and decreases vascular cell adhesion molecule in hypercholesterolemic postmenopausal women. Journal of Cardiovascular Pharmacology. 2006;47:S177.

48. Bhagwat S, Haytowitz DB, Holden JM. USDA Database for the Flavonoid Content of Selected Foods. Beltsville, Maryland; 2011. p. 1–159.

49. Ellinger S, Reusch A, Stehle P, Helfrich H-P. Epicatechin ingested via cocoa products reduces blood pressure in humans: a nonlinear regression model with a Bayesian approach. American Journal of Clinical Nutrition. 2012;95:1365–7.

50. McCullough ML, Chevaux K, Jackson L, Preston M, Martinez G, Schmitz HH, et al. Hypertension, the Kuna, and the Epidemiology of Flavanols. Journal of Cardiovascular Pharmacology. 2006;47:S103-S109.

51. Davison K, Coates AM, Buckley JD, Howe PRC. Effect of cocoa flavanols and exercise on cardiometabolic risk factors in overweight and obese subjects. International Journal of Obesity. 2008;32(8):1289–96.

52. Heiss C, DeJam A, Kleinbongard P, Schewe T, Sies H, Kelm M. Vascular effects of cocoa rich in flavan-3-ol. The Journal of the American Medical Association. 2003;290(8):1030–1.

53. Njike VY, Faridi Z, Shuval K, Dutta S, Kay CD, West SG, et al. Effects of sugar-sweetened and sugar-free cocoa on endothelial function in overweight adults. International Journal of Cardiology. 2011;149(1):83–8.

54. Grassi D, Lippi C, Necozione S, Desideri G, Ferri C. Short-term administration of dark chocolate is followed by a significant increase in insulin sensitivity and a decrease in blood pressure in healthy persons. American Journal of Clinical

! 78!

Nutrition. 2005;81:611–4.

55. University of Otago, Ministry of Health. Methodology Report for the 2008/09 NZ Adult Nutrition Survey. Wellington: Ministry of Health; 2011.

56. Neveu V, Perez-Jiménez J, Vos F, Crespy V, du Chaffaut L, Mennen L, Knox C, E isner R, Cruz J, Wishart D, Scalbert A. (2010) Phenol-Explorer: an online comprehensive database on polyphenol contents in foods. Database, doi: 10.1093/database/bap024.

57. Ovaskainen ML, Törrönen R, Koponen JM, Sinkko H, Hellström J, Reinivuo H, et al. Dietary intake and major food sources of polyphenols in Finnish adults. The Journal of Nutrition. 2008;138(3):562–6.

58. Chun OK, Chung SJ, Song WO. Estimated dietary flavonoid intake and major food sources of U.S adults. The Journal of Nutrition. 2007;137:1244–52.

59. Li G, Zhu Y, Zhang Y, Lang J, Chen Y, Ling W. Estimated daily flavonoid and stilbene Intake from fruits, vegetables, and nuts and associations with lipid profiles in Chinese adults. Journal of the Academy of Nutrition and Dietetics. 2013;113(6):786–94.

!!

!

! 79!

!""#

$%"&'

(")*+,-+

*."/

"0#,0/)*12,34

$5("

)*+,!+*.

*/676

"+8,0/

)*12,34

$5,9:

,-""#

,$%"&

'8("

)*+,;

/)<"

=:*/

0/8,0/

)*12,34

$5("

)*+,>

:#%"92

/?"0=

,*=0#8

,0/)*12,34

$5!"

#$%&'"

("&)'*+,

-,.%/

,01213456

7893

:;:265:

678

62;25:1

6864

15:841734;:

<.=)>

7172593

:83:

:37859;

:816

9623531

7168

:;568

:99616

?&'"(

"&)'*+

,-,.%/,0

:324564

;736

3;;57

898;93

8:154

6:7193

353396;4;79

@=/%.A0

B,,>0

7427956

3368

7:61159

6383

97456

9:39:9

:8537

933::6

C,/,>%+

&,0397953;

:978

7;56;

482926

36159

236;86

25167661:16

D">%>

",0E*F=G%

.%*%#

H*>%."

31358

371178

25213:6:32:

;2592

77;8;6

:5:;619343;

I%).J

*K."H

='>0

;:589

92723;

:1513

34;7:

11563

186449

25:24727181

!=>0*%#

H*@,,H0

:6528

6434:3

65816;2:148

752;9837;37

25227367;43

@%-"=.J*0

%=',0*%

#H*'"

#H)G

,#>0

:93:524

8:15;

26342;;

;5;46779413

85117181;84

L%F,0*%

#H*G

=MM)#0

:8588

894788

958194;8124

258837;:7:1

25371;4;:18

@#%'F*+

%.035;

316;;17

:516::43399

257;1;798:9

2529448:843

N.,%FM%

0>*',.,%&0

4457;

949144

35641913371

256;;879697

25464;2787:

O.%)#

0*%#H

*D%0>%

64:51

36:39;

45379378;48

2597:344391

31566

7893

N,,M*

%#H*C

,%&

;5818716319

2521:6;4;;7

25289;4:;76

25::96;8:87

D=HH

)#/0AH

,00,.>0

:5793792194

256324999:4

2529;334219

252:726679:

N)0'=)>0

185;7

267:;:

6574188;689

25237173891

2586;;62:73

N.,%H*P

)#'&=H

,0*."

&&0*%#

H*0K,

')%&)>J

*+.,%H0Q

:8;56

42:619

:5;29211:33

252:;86;996

95833294;91

N.,%H*+

%0,H

*H)0(,0

;6569

78;8;7

2521923:7:4

2258

681;6284

R)&F

9513733:;;9

1579;211116

225:

8;691:88

L(,,0,

22

22

N=>>,

.*%#H

*R%./

%.)#,

22

22

<%>0*%#

H*")&0

2522923123

22

252231::92;

S//0*%#

H*,//*H)0(,0

22

22

T%G+

AR=>>"#

25783:6941

22

25222391317

D".F

:578;229;87

252249171:3

2252

2169:474

D"=&>

.J34532

322218

253236:267

2251

7662;2:8

U>(,

.*G,%>

22

22

@%=0%/,0*%#

H*K."',00,

H*G,%>0

259861;;661

22

2522888:628

D),0*%

#H*K%

0>),0

95:6:;79268

22

2527:834767

<)0(A@,%M"

"H;52

88;43284

2526;893139

2252

4:;71488

@#%'F*M""

H0257

41186;;6

25222462;87

2252

:6994:6:

@"=K

0*%#H

*0>"'F0

75173:98321

25:3;423964

225:

2:341439

I),>%

.J*0=KK

&,G,#

>0252

9:19;;41

252:96;386

22

@2*/,0/)*12,-

"%,;A

B,'*%)0=0'*

/)8,3/C

7DEF5,G

<",'%

".0#2

#,*,="

4'+2)

2,82),"-

,#*)*,)<

*),*++"G

,HIJ,%081,2

8)04*

)0"/

9 Appendix

Full list of food groups contributing to total Flavonoid and Flavan-3-ol intake

! 80!

! 81!

!""#

$%"&

'()"

*+,-.,+/

"0"1#-1

0*+23-4

5$6

)"*+,-!,+/

+0787",(-10*+23-45

$6-9:

-.""#

-$%"&

'()"

*+,-;

0*<"

=:+0

10(-10*+23-4

5$6

)"*+,->

:#%"93

0?"1=-+

=1#(-10*+23-4

5$6

!"#$%

&'#%(

')#&*#

+,-.+

/01-00

2./0,3/+

4.3420+0042

03.30

301-13

5"6#('7

$%89:(

6&'";

99&'#%

('&<6

8$#9$*='>

"6#(

&?@/-./

301/,0

@.-/2--@133

4.4@3/414/0

-./4,+344/-

5"6#AB#

&*'86"6#9&

23.-@

3-2@/3

0.320-1/,@3

4.1,@0/,2/

4.2,32,4+-@

5$&8:$*

&,+.4/

220,01

1.,140004--

4.40311,4-1

4./430--2,/

C#A6&'#

%('D

:BB$%

&@/.40

-+-0/@

-.1/-,-@11

4./--,11-20

4.01403@4@

5"6#('>

#&6(

'($&E6&

[email protected]

0-/0@-

4.4,1/030-/

[email protected]

3042-0/1

):((

$%F&G(6

&&6"*&

@.+,113@22

4.1@21+3-03

4.4-34+4403

4.4@+44+@@1

H$9A

-.041+10131

,.1-4+2@421

44.@

/4@/1/22

I#$"=

'<";(

:8*&

-2.-4

023/3-

@0.-@

-3/422

,@.,,

@+4-@-

4.@4@0/1-01

CE66&6

44

44

5:**6

"'#%(

'H#"F#"$%

64

44

4J#*&'#%

(';$9&

4.44,+/@+,-

44

4.440@+230

KFF&'#%

('6FF'($&E6&

44

44

566B'#%

('L6#9

1.412,+--/1

4.4,0,--@/-

4.4/110//30

4.@@+@-0,//

M#D>GH:**;%

4.23@@4-4@+

44

4.44401@,,3

);"A

@./-/+0+/,1

4.44/@1-32

44.4

4,/33+/1

);:9*

"=,@.33

11433@

4.001+@@4-3

44.-

0@-4@22@

N*E6

"'D6#*

44

44

O#:&#F6&'#%

('<";86&&6(

'D6#*&

4.-/12,+//-

44

4.4@444@,@/

)$6&'#

%('<#

&*$6&

-.-13,,3-+

44

4.4++@+2/2+

J$&EGO6#B;;

(3.0

+,+0@1/2

4.412@/-@@@

44.4

2-,3/-+

L6F6*#>96

&0,30.12

3//

+@./-

@1@,3

034.-

1@4///

4.,-/3//40+

);*#*;6&P'A:D

#"#'#

%('*#

";0,/.3

3/@/30

4.4,,4+1222

[email protected]

0/32-/

@.@2/013/+

O%#8A'B;;

(&4.+

1140-,13

4.4442-@3@,

44.4

@142,-@3

J":$*

+@[email protected]

3/3+

@0-,.@@

0+01

[email protected]

,2-/

@3.0-

3++301

Q:*&'#%

('O66(&

@3.4+

+-@-,3

1.-/+0@04-/

1.142@,20-+

4.44124+,03

O:F#"G&

R66*&

1@,20.+

2320

31,@2.+

4,,0

,+1.,

@4/@1,

@3.@3

,304-0

O;:<

&'#%(

'&*;8A&

+.@1-0@4++/

4.@0003-4/@

44.4

/2-0+4@2

O#S;:"='&

#:86&'#

%('8;

%($D

6%*&

@,3,.--

-413

,.,,/,+21-2

3.,042,/13,

2.2/0/032@,

Q;%T#98

;E;9$8'>

6S6"#F6&

,4044.4

4243

@34-0.4

+/4@

140-.-1

22+-

,.@2-24,3-/

U98;E;

9$8'>6

S6"#F6&

@@1-.3@

230-1.-

-+2@30

224.-

1+202@

0.@-,,11@-1

I$6*#"='&

:<<96

D6%

*&4.4

3@3043-+

4.4@2032/22

44

O%#8A'>

#"&

0.0-11@2@3/

@.0@@+312+3

4.1+4234,-/

4.4-,30-242

@3+0-10*+23-.

"%-;AA-;BC

-'+%*1=1'

+0*(-40DEFGH6

! 82!