Professional Documents
Culture Documents
Nutritional Epidemiology
Departments of 4Human Nutrition and 5Preventive and Social Medicine, University of Otago, Dunedin, New Zealand; and 6School of
Healthcare Sciences, Faculty of Healthy Sciences, Universiti Kebangsaan Malaysia, Kuala Lumpur, Malaysia
Abstract
Introduction
graphic, nutrient intake, and biomarker data are less clear
Healthful dietary habits are associated with better nutrient intake because few studies allow investigation of these relations.
and higher diet quality, which in turn leads to positive health out- To date, 12 studies of children and adolescents used an index
comes (1–3). Because dietary habits are likely to track into adult- embedding information on dietary habits. They mostly examine
hood (4), adoption of optimal dietary habits during adolescence may the association of the index with food or nutrient intakes (11–
have a protective effect against chronic diseases later in life (5). 16), body composition measures (11–19), and 1 with nutritional
Food habits questionnaires have been used to collect qual- biomarkers (11). They were conducted only in convenience
itative information on food behaviors of adolescents, including samples of children and adolescents aged 9–24 y. None of them
food types, food preparation, cooking practices, snacking pat- examined the relation of a dietary habits index with nutrient
terns, and intake frequency of certain food groups, which can intakes and nutritional biomarker levels in a nationally repre-
then be compared with dietary guidelines (6–9). Compared with sentative sample of adolescents.
quantitative reporting of food intake, usual dietary habits are The aims of this study are 2-fold: 1) to develop Healthy
perceived to be more easily and accurately documented (9). Dietary Habits Scores for Adolescents (HDHS-A)7 based on
Because dietary habits are interrelated and tend to cluster (2,10), dietary habits information; and 2) to examine the validity of the
there is a potential synergetic effect of multiple dietary habits, so HDHS-A based on its internal reliability and associations with
it is important to examine them as a group rather than studying sociodemographic factors, nutrient intakes, and nutritional
them in isolation (10). However, dietary habits information has biomarkers in adolescents aged 15–18 y who participated in
seldom been examined comprehensively by means of diet index the 2008/2009 New Zealand (NZ) Adult Nutrition Survey.
scores. In addition, the association of a constellation of desirable
dietary habits, as assessed by a diet index, with sociodemo-
Methods
1
Supported by grants from the New Zealand Ministry of Health. Study design and sample. The study design included a secondary
2
Author disclosures: J. E. Wong, P. M. L. Skidmore, S. M. Williams, and analysis of the 2008/2009 NZ Adult Nutrition Survey, a cross-sectional
W. R. Parnell, no conflicts of interest. population-based survey in a representative sample of New Zealanders
3
Supplemental Tables 1 and 2 are available from the ‘‘Online Supporting
Material’’ link in the online posting of the article and from the same link in the
7
online table of contents at http://jn.nutrition.org. Abbreviations used: DHQ, dietary habits questionnaire; HDHS-A, Healthy
* To whom correspondence should be addressed. E-mail: winsome.parnell@ Dietary Habits Score for Adolescents; NZ, New Zealand; NZDep2006, 2006 New
otago.ac.nz. Zealand Deprivation Index; NZEO, New Zealand European and others.
Development of the HDHS-A. The HDHS-A index was developed using Statistical analysis. Participants aged 15–18 y who completed a single
information from the DHQ after consultation with 2 expert nutritionists and 24-h diet recall and at least 75% of the DHQ (i.e., 19 of 25 questions) were
a dietitian. The index items were selected on the basis that they were relevant considered eligible for the analysis (n = 695). Data for eligible participants
to and captured the key nutrients that are important in determining the diet with 1–3 missing responses (n = 103), entered either as ‘‘did not answer’’ or
quality of NZ adolescents, as guided by existing scientific evidence and the ‘‘donÕt know,’’ were estimated from other information that the participant
NZ Food and Nutrition Guidelines for Healthy Children and Young People supplied as long as at least 50% of questions related to the items within a
(21). Four index prototypes were initially created. Using 21 questions from cluster were completed. Mean values of nonmissing items in a cluster were
the DHQ, the final prototype included 17 items grouped into 5 clusters: 1) fat estimated and substituted for the missing values (1 participant was
from meat, poultry, and fish; 2) other fats; 3) fruit, vegetables, and bread; 4) excluded because >50% of the data were missing in 1 of the clusters).
sugar sources; and 5) meal habits. These clusters were named after main Because the data were part of a nationwide survey, the Stata survey
nutrients or domains of diet reflected by the healthy dietary habits. procedures, which account for the complex structure of the sample and the
A response that aligned with a more positive dietary habit was sampling weights, were used to analyze the data. These provide the correct
assigned a higher score using a 5-point scoring system ranging from 0 to 4. SE for the estimates. A natural logarithm was applied to transform non-
The total HDHS-A was a summation of scores from the 17 items and normally distributed data. Continuous variables were presented as means
ranged from 0 to 68. A greater total score represents a dietary pattern or geometric means if log-transformed and SEs or 95% CIs. Categorical
reflective of healthier dietary habits. The 17 items derived from the DHQ data were presented as relative frequencies and percentage. Survey post-
and their respective scoring criteria are presented in Supplemental Table 1. estimation tests (lincom command in Stata) were used to compare
continuous variables (e.g., age, BMI Z-score, HDHS-A score) between the
Sociodemographic information. Age was derived from the date of sexes. All analyses were performed using the statistical software package
birth and interview start date. Participants were asked to identify Stata (version 11.2; StataCorp), with statistical significance set at P < 0.05.
$1 ethnic groups to which they belonged, and participants were
classified hierarchically into 3 ethnic groups: 1) Maori; 2) Pacific; and 3) Evaluation of the HDHS-A. Content validity of the HDHS-A was
NZ European and others (NZEO) (25). established through an expert review to ensure that all 17 items of the
Socioeconomic status was estimated using an area-based scale of index were important dietary habits contributing to the diet quality of
deprivation, namely the 2006 NZ Deprivation Index (NZDep2006) NZ adolescents. The index was also tested for internal reliability using
(26). This proxy measures deprivation based on 8 dimensions of correlations and CronbachÕs a coefficient. To examine construct validity,
deprivation for each neighborhood (i.e., a mesh block containing ;87 3 hypotheses were generated with regard to the associations between
people) in NZ, including income and benefit receipt, home ownership, HDHS-A scores and sociodemographic factors, 24-h nutrient intakes,
support for sole-parent families, employment status, qualifications, and nutritional biomarkers. It was hypothesized that higher HDHS-A
HDHS-A scores vs. sociodemographic variables. The mean in-depth and independent evaluation of the HDHS-A. Using a
6 SE HDHS-A was 44 6 0.4 (range, 16 to 66). HDHS-A score nationally representative sample and 2 independent methods for
was associated with sex (P < 0.001), ethnicity (P < 0.001), and validation of the HDHS-A, the results from this study can
NZDep2006 quintile (P < 0.001) (Table 1). Females had significantly therefore be generalized to NZ adolescents aged 15–18 y.
higher total HDHS-A scores than males. NZEO had higher HDHS- For the first time, it was demonstrated by means of a diet
A scores than Maori (P < 0.001) and Pacific (P < 0.01) groups. index that the dietary habits of NZ adolescents differed by sex,
ethnicity, and socioeconomic status. Compared with males, fe-
HDHS-A score vs. nutrient intakes and nutritional biomarkers. males scored higher in the HDHS-A, particularly in the fol-
The associations between HDHS-A scores and 24-h nutrient intakes lowing 3 clusters: 1) fat from meat, poultry, and fish; 2) other
and nutritional biomarker levels are shown in Table 2. Because fats; and 3) fruit, vegetables, and bread (data not shown). This
HDHS-A score was negatively associated with energy intake finding is in agreement with previous studies of adolescents that
(r = 20.18, P < 0.001), all nutrients were adjusted for total energy showed that females compared with males had intakes of fruit
intake (in megajoules). Overall, higher relative intakes of protein, (35) and meat and meat products (36) that were more in line with
dietary fiber, PUFA, and lactose and lower intakes of sucrose were dietary guidelines and better overall diet quality as measured by
associated with increasing thirds of HDHS-A. Associations in the the Healthy Eating Index (14,37) and Mediterranean Dietary
expected directions were also found with most micronutrient Score (38). The mean HDHS-A of the Maori and Pacific participants
intakes, urinary sodium excretion, and whole-blood, serum, and were 12% and 8%, respectively, lower than their NZEO
RBC folate concentrations (P < 0.05). counterparts. In addition, there was a trend for poorer dietary
habits with increasing amount of deprivation (P-trend < 0.001).
The associations of HDHS-A score with ethnicity and NZDep2006
Discussion
were likely explained by the fact that a higher proportion of Maori
This study describes the development and validation of a and Pacific participants experience a higher amount of deprivation
behavioral-based diet index, namely the HDHS-A, derived (P < 0.001; data not shown). In a previous regional study in
from self-reported dietary habits captured in a DHQ. The Auckland, Maori and Pacific adolescents appeared to have less
strength of this simple diet index lies in its practicality, because it satisfactory dietary intakes as marked by higher fat intake and
is calculated based on dietary habits information that is easier to larger-than-standard portions of most food items but fewer daily
obtain than quantitative food and nutrient intake data. The use servings of vegetables and cheese compared with NZ Euro-
of 24-h diet recall and biomarker information allowed an pean and Asian adolescents (39). Considering dietary habits as a
Healthy dietary habits score for adolescents 939
TABLE 2 Adjusted nutrient intakes and nutritional biomarker levels by thirds of HDHS-A1
Thirds of HDHS-A
Low Medium High P-trend2
whole by using a diet index approach, this study confirms calcium, iron, magnesium, phosphorus, potassium, selenium,
findings of previous studies that disparities in diet quality of NZ and vitamins A, B-1 (thiamin), B-2 (riboflavin), and C. Of
adolescents are associated with ethnicity and socioeconomic particular interest was the observed higher lactose but lower
status. This also provides evidence of the construct validity of sucrose intakes. This finding is reflective of the index scoring
the HDHS-A because it distinguishes differences in diet quality system that rewards milk consumption but penalizes intakes of
among ethnic and socioeconomic groups in NZ. sugary foods, such as fruit juice, soft drinks, and confectionery
HDHS-A scores were significantly associated with more (Supplemental Table 1). In addition, the HDHS-A seems to be a
favorable intake profiles of dietary fiber, PUFA, lactose, sucrose, good indicator of types of fat in the diet. Although the total fat
940 Wong et al.
intake was constant within the acceptable macronutrient distri- conform more closely to the NZ Food and Nutrition Guidelines
bution range of 20–35% of total energy (40), the unsaturated- for Healthy Children and Young People (21). However, it must be
to-saturated fat ratio increased significantly across the thirds of noted that the scoring of the HDHS-A remains arbitrary for some
HDHS-A (P < 0.001). This increasing fat ratio reflected a higher items, because not all dietary habits could be ordered on a scale
proportion of PUFA intake relative to SFA intake, i.e., a better according to their healthfulness. The scoring may be less sensitive
fat quality across thirds of HDHS-A. in discerning subtleties in diet quality when a dietary habit was
Four studies examined a diet index in relation to nutritional defined as neither optimum nor poor. Because of the cross-sectional
biomarkers in children and adolescents (11,41–43). Recently, nature of this study, reliability or temporal stability of the HDHS-A
Vyncke et al. (42,43) studied the Flemish food-based Diet was not examined. Future work should examine the predictive
Quality Index for Adolescents in relation to a comprehensive capacity of the HDHS-A in relation to health outcomes in
range of blood biomarkers in a large multicenter cohort of longitudinal studies. Contrary to most diet indices that use
European adolescents [the HELENA study (for Healthy Lifestyle quantitative measures of nutrients and foods, the current index is
in Europe by Nutrition in Adolescence)]. The Diet Quality Index based on 17 key dietary habits arranged in 5 dimensions or clusters.
for Adolescents was strongly associated with 2 nutritional The HDHS-A measures both desirable and nondesirable dietary
biomarkers of long-term intake of vitamin D and vitamin B-12 habits that are of nutritional concern in adolescence. The index also
and weakly associated with some serum FAs, such as EPA and penalizes undesirable dietary habits relevant to youth culture, such as