Accepted Manuscript The Development, Application and Validation of a Healthy Eating Index for Australian (HEIFA – 2013) Adults Rajshri Roy, Lana Hebden, Anna Rangan, Margaret Allman-Farinelli PII:
S0899-9007(15)00408-6
DOI:
10.1016/j.nut.2015.10.006
Reference:
NUT 9627
To appear in:
Nutrition
Received Date: 28 April 2015 Revised Date:
16 September 2015
Accepted Date: 1 October 2015
Please cite this article as: Roy R, Hebden L, Rangan A, Allman-Farinelli M, The Development, Application and Validation of a Healthy Eating Index for Australian (HEIFA – 2013) Adults, Nutrition (2015), doi: 10.1016/j.nut.2015.10.006. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Title Page The Development, Application and Validation of a Healthy Eating Index for Australian (HEIFA – 2013) Adults
RI PT
Running Title: Healthy Eating Index for Australians
Lana Hebden
1
Anna Rangan
1
Margaret Allman-Farinelli
Author Addresses 1
M AN U
Rajshri Roy (Corresponding Author)
1
SC
Authors 1
School of Molecular Bioscience, Department of Nutrition and Metabolism, Charles Perkins
Centre, The University of Sydney, Sydney, New South Wales, Australia
Corresponding author details: Ms Rajshri Roy, B.Sc. (Nutrition) Hons.
TE D
School of Molecular Bioscience, Department of Nutrition and Metabolism Charles Perkins Centre, Level 4, E40 The University of Sydney,
Australia.
EP
Sydney 2006, New South Wales,
AC C
M: +61 0432 744 202
E-mail:
[email protected];
[email protected]
ACCEPTED MANUSCRIPT Title Page Abstract Objective: Diet quality indices are used to assess dietary behaviour and adherence to dietary guideline recommendations. The aim of this study was to develop, apply and validate a Healthy Eating Index for Australian Adults (HEIFA - 2013) based on the updated Dietary Guidelines for
RI PT
Australians 2013.
Research Methods and Procedures: The HEIFA - 2013 used an eleven component system of 5 food groups, 4 nutrients, water intake and a measure of dietary variety. The total possible index
SC
score was 100. The HEIFA - 2013 was applied to Weighed Food Record (WFR) and Food Frequency Questionnaire (FFQ) data of a sample (n=100) of young adults. The HEIFA -
M AN U
2013 was assessed using principal component analysis (PCA), Cronbach's coefficient and correlation coefficient with nutrient intakes. Scores for HEIFA - 2013 components were compared between methods using means, frequencies, 95% limits of agreement, Bland and Altman methods and weighted Kappa.
Results: PCA indicated multiple underlying dimensions comprise the index and Cronbach's
TE D
coefficient α was 0.41. A higher HEIFA - 2013 was associated with higher dietary quality, including a low intake of saturated fat and sodium; a high intake of selected vitamins and minerals. Low correlations with energy were observed. The mean HEIFA - 2013 score (+/-SE) for WFR was 53.84 (+/-1) and for FFQ were 54.82 (+/- 0.9). The total mean scores were 54.33
EP
(+/-0.1). Young adults had the lowest mean scores for sodium 2.9 (+/-0.2), fat 3.0 (+/-0.0), vegetables 4.7 (+/-0.1) and grains 5.1 (+/-0.1). The WFR and FFQ scored individual components
AC C
differently but at the group level the differences were not significant.
Conclusion(s): The HEIFA - 2013 will need to be catered for different diet collection methods. It is a useful index of overall diet quality and can be used to monitor changes in dietary intake of adults over time. The findings infer that even a highly educated affluent group of young adults fail to meet recommended dietary guidelines.
Keywords: healthy eating index; dietary assessment; diet quality; dietary guidelines; young adults.
ACCEPTED MANUSCRIPT
Abstract
2
Objective: Diet quality indices are used to assess dietary behaviour and adherence to dietary guideline
3
recommendations. The aim of this study was to develop, apply and validate a Healthy Eating Index for
4
Australian Adults (HEIFA – 2013) based on the updated Dietary Guidelines for Australians 2013.
5
Research Methods and Procedures: The HEIFA – 2013 used an eleven-component system of 5 food
6
groups, 4 nutrients, water intake and a measure of dietary variety. The total possible index score was
7
100. The HEIFA – 2013 was applied to Weighed Food Record (WFR) and Food Frequency
8
Questionnaire (FFQ) data of a sample (n=100) of young adults. The HEIFA – 2013 was assessed using
9
principal component analysis (PCA), Cronbach’s coefficient and correlation coefficient with nutrient
10
intakes. Scores for HEIFA – 2013 components were compared between methods using means,
11
frequencies, 95% limits of agreement, Bland and Altman methods and weighted Kappa.
12
Results: PCA indicated multiple underlying dimensions comprise the index and Cronbach’s coefficient
13
α was 0.41. A higher HEIFA - 2013 was associated with higher dietary quality, including a low intake of
14
saturated fat and sodium; a high intake of selected vitamins and minerals. Low correlations with energy
15
were observed. The mean HEIFA – 2013 score (+/-SE) for WFR was 53.84 (+/-1) and for FFQ were
16
54.82 (+/-0.9). The total mean scores were 54.33 (+/-0.1). Young adults had the lowest mean scores for
17
sodium 2.9 (+/-0.2), fat 3.0 (+/-0.0), vegetables 4.7 (+/-0.1) and grains 5.1 (+/-0.1). The WFR and FFQ
18
scored individual components differently but at the group level the differences were not significant.
19
Conclusion(s): The HEIFA – 2013 will need to be catered for different diet collection methods. It is a
20
useful index of overall diet quality and can be used to monitor changes in dietary intake of adults over
21
time. The findings infer that even a highly educated affluent group of young adults fail to meet
22
recommended dietary guidelines.
23
Keywords: healthy eating index; dietary assessment; diet quality; dietary guidelines; young adults.
AC C
EP
TE D
M AN U
SC
RI PT
1
1
ACCEPTED MANUSCRIPT
Introduction
25
Over-consumption of energy-dense nutrient-poor foods has become an international public health
26
concern within the context of the obesity epidemic 1. However, in order to measure the impact of over-
27
consumption of these foods on health outcomes and associated risk factors, a number of scoring
28
mechanisms or indices have been developed to provide a measure of the overall diet quality of an
29
individual or population. These are a collection of scores applied to dietary components to represent a
30
healthy balanced diet 2. These indices aim to categorise individuals according to whether their diets are
31
‘healthy’ by allocating scores for consuming a variety of foods or ‘recommended’ foods or nutrients 3.
32
Diet quality indices fall into three categories; those based on dietary guidelines, those based on
33
recommended foods, and those based on dietary variety 4. Several indices have been validated for their
34
ability to predict all cause and cause specific mortality 5. Among the most used diet indices are the
35
Healthy Eating Index (HEI), most recently updated to accommodate the 2010 US Dietary Guidelines
36
(HEI-2010) which are based on recommended foods accounting for individual nutrient and non-nutrient
37
components within foods 6. As the consumption of a greater variety of foods is considered beneficial for
38
nutritional adequacy, many investigators have used dietary variety to evaluate food consumption
39
retain the nutritional complexity of foods 4. Examples include the Dietary Diversity Score (DDS) and the
40
Recommended Food Score (RFS), both having been used to examine associations between dietary
41
quality and mortality in adults
42
have been modelled on previously developed indices and are based on national food based dietary
43
guidelines
44
updated diet quality index to reflect these new guidelines 18.
7-13
to
AC C
EP
TE D
M AN U
SC
RI PT
24
5,14-17
14, 15
. Within Australia, a number of dietary quality indices and scores
. However, these national guidelines were recently revised warranting the need for an
2
ACCEPTED MANUSCRIPT
This paper presents the development and application of a new Healthy Eating Index for Australian
46
Adults (HEIFA – 2013), based on the updated Dietary Guidelines for Australians Adults (DGAA) and
47
Australian Guide to Healthy Eating (AGHE) 18.
48
Methods
49
Scoring criteria for HEIFA – 2013 were developed reflecting compliance with the updated 2013 DGAA
50
(e.g. intake of recommended vegetable serves) and recent changes including greater emphasis on types
51
of fat, whole-grains, increasing vegetable variety and limiting sugar-sweetened drinks. The HEIFA –
52
2013 was then applied to Weighed Food Records (WFR) and Food Frequency Questionnaires (FFQ)19
53
data from a sample (n=100) of young adults with scores for the latter compared with scores derived from
54
the WFR, considered the gold standard of dietary data collection 20.
55
Structure of Healthy Eating Index for Australians - 2013 (HEIFA – 2013)
56
Scoring criteria for HEIFA – 2013 consisted of eleven components based on different aspects of a
57
healthy diet. These components included the five core food groups; i.e. vegetables, fruits, grain (cereal)
58
foods, milk and milk alternatives, meat and protein food alternatives. The components also included
59
discretionary foods high in saturated fat and/or added sugars, added salt or alcohol that do not fit into the
60
five core food groups, and should be limited because they are unnecessary for a healthy diet
61
component measured the degree to which a person’s diet conforms to the serving recommendations for
62
each food group. The scoring components represented the suggestions in the guidelines such as fruit
63
intake to exclude fruit juice and include different variety of fruits, vegetables to include different
64
varieties, fifty percent intake of total grains be whole grains, adequate water and limited alcohol intake.
65
The HEIFA – 2013 also includes components for specific nutrients, which are not included in the food
AC C
EP
TE D
M AN U
SC
RI PT
45
18
. Each
3
ACCEPTED MANUSCRIPT
groups such as fats (mono-and polyunsaturated fats and saturated fats), added sugars and sodium to
67
capture the guidelines regarding these nutrients.
68
Individual Scoring for each component and total HEIFA – 2013 score
69
HEIFA – 2013 scores ranged from 0-100, comprising nine components (core foods, discretionary foods,
70
and negative nutrients such as saturated fat, added sugar and sodium) each allocated a score from 0-10
71
and two components (water and alcohol) allocated a score from 0-5. The score a person received for a
72
food group category was determined by how closely an individual’s intake matched the appropriate
73
number of servings recommended for their age and gender. A person who complied with the
74
recommended number of servings from any food group would receive a maximum score i.e. either 10 or
75
5 for that group; conversely, a person not consuming recommended servings within a food group would
76
receive a minimum score of 0. For example, the recommended number of discretionary food serves18, 21
77
for 19 to 60 year old males is ≤ 3 and ≤ 2.5 for females; thus an individual consuming more than this
78
amount would receive a score of 0, while an individual meeting this recommended serving received a
79
score of 10. Between the maximum and minimum score for a category, composite sub-scores were
80
developed based on factors derived from the serving size recommendations given in the AGHE
81
(supplementary file) (Table 1). For example, a person who consumed five serves of grains/cereals, but
82
needed to have consumed six for a maximum score would get a score of 4.17. Food serving amounts
83
were computed using factors derived from the serving size assumptions given in the AGHE
84
(supplementary file). In calculating the HEIFA – 2013, legumes were assigned to the meat category; any
85
additional legumes beyond the optimum serve were assigned to the vegetable group. The one exception
86
was soy products, which are usually used as meat substitutes and thus were always assigned to the meat
87
group except for soymilk, which belonged to the dairy and dairy alternatives group. Existing national
AC C
EP
TE D
M AN U
SC
RI PT
66
4
ACCEPTED MANUSCRIPT
recommendations for food-based nutrition indicators for saturated fat, added sugars, and sodium were
89
utilized. In the 2013 Dietary Guidelines, the recommendation for saturated fat is not expressed as a
90
single value, but rather as “less than 10% of energy intake.” A maximum score was achieved when total
91
saturated fat was 1, indicate at least four factors exist. The standardized
175
Cronbach’s coefficient α was 0.40.
176
The relationship between the index and the nutrient intake levels was confirmed by the Spearman’s
177
correlation coefficients (Table 2). For each of the essential micronutrients (vitamin C, thiamine, calcium,
178
iron and zinc), there was a positive correlation between both the WFR and FFQ HEIFA – 2013
179
(continuous data) and the intake of nutrients. Table 3 shows the trends across tertiles for HEIFA-2013
180
for both WFR and FFQ. Energy intake decreased, total protein and fat increased but saturated fat and
181
sodium decreased across tertiles. The amounts of each of the five core food groups consumed increased
182
across tertiles, as did micronutrients such as vitamins and minerals.
183
The total mean intakes of the components were analysed separately for each method of dietary intake
184
measurement (Table 4). The mean HEIFA – 2013 score (+/- standard error) for WFR was 53.84 (+/-1)
185
and for FFQ were 54.82 (+/-0.9). The majority of component scores from the FFQ were higher than
186
those from the WFR. The total means WFR and FFQ HEIFA – 2013 scores were 54.33 (+/-0.1). Few
187
people scored very high or very low on the HEIFA – 2013. Only 2% of the sample had a mean score
188
below 40 and 10% of the sample had a score higher than 60 in the frequency distribution of mean scores.
189
No one category contributed disproportionately to the mean score. This sample of young adults had the
190
lowest total ([WFR Mean + FFQ] / 2) mean (+/- SE) component scores for sodium 2.9 (+/-0.2), fat 3.0
191
(+/-0.0), vegetables 4.7 (+/-0.1) and grains 5.1 (+/-0.1), especially whole-grains.
AC C
EP
TE D
M AN U
SC
RI PT
173
9
ACCEPTED MANUSCRIPT
The 95% LOA indicated that at the individual level the differences between the methods were quite
193
large, for example the FFQ scored higher for fruit components while it scored lower for the vegetable
194
component (Table 4). Weighted Kappa agreement between the scores from the two methods ranged
195
between -0.11 for dairy (poor agreement) and 1 for added sugars and alcohol (perfect agreement).
196
Grains, vegetables, dairy, water and sodium had large variations in scores and the two methods
197
demonstrated poor agreement 32.
198
Through visual inspection of the Bland and Altman plots (i.e. the difference in scores between methods
199
(FFQ–WFR Mean) plotted against the average [(WFR Mean + FFQ)/2] scores, it can be observed
200
(Figure 2) that the average discrepancy or the bias (0.932) in scores between the two methods of dietary
201
data collection is not large enough to be clinically important. The limits of agreement (-21.28 – 23.15)
202
and the bias are small (0.932), showing that the overall HEIFA-2013 scores from the two methods are
203
essentially equivalent (P=0.29). The differences in scores between the methods tend to get larger as the
204
average score increases. The scatter around the bias line gets larger as the average score gets higher
205
indicating poorer agreement with higher HEIFA scores as reflected in the regression line of best fit.
206
Discussion
207
The HEIFA-2013 application results shows that it is an acceptable tool for assessing diet quality at the
208
group level in the sample of young adults studied. It was developed using the most current scientific
209
food-based dietary guidelines that were informed by systematic literature review of evidence and diet
210
modelling. The HEIFA – 2013 was based on food groups (and some deleterious nutrients) rather than on
AC C
EP
TE D
M AN U
SC
RI PT
192
10
ACCEPTED MANUSCRIPT
nutrients only because people select foods and this tool is a basis on which to build individual and
212
population education. It is being recognised that taking a reductionist nutrient-centric approach to study
213
diet-disease relationships is less useful than a food based approach 22.
214
Fruit (r=0.59, P
