RESEARCH

Original Research

Relative Validity and Reproducibility of a Food Frequency Questionnaire for Identifying the Dietary Patterns of Toddlers in New Zealand Virginia C. Mills, MSc; Paula M.L. Skidmore, PhD; Emily O. Watson, MSc; Rachael W. Taylor, PhD; Elizabeth A. Fleming, MSc, RD*; Anne-Louise M. Heath, PhD ARTICLE INFORMATION Article history: Accepted 8 September 2014

Keywords: Food frequency questionnaire Validity Reproducibility Dietary pattern 2212-2672/Copyright ª 2014 by the Academy of Nutrition and Dietetics. http://dx.doi.org/10.1016/j.jand.2014.09.016 *

Certified in New Zealand.

ABSTRACT Background Dietary patterns provide insight into relationships between diet and disease. Food frequency questionnaires (FFQs) can identify dietary patterns in adults, but similar analyses have not been performed for toddlers. Objective The aim of the Eating Assessment in Toddlers study was to evaluate the relative validity and reproducibility of dietary patterns from an FFQ developed for toddlers aged 12 to 24 months. Design/setting Participants were 160 toddlers aged 12 to 24 months and their primary caregiver who completed an FFQ twice, approximately 5 weeks apart (FFQ1 and FFQ2). A 5-day weighed food record was collected on nonconsecutive days between FFQ administrations. Statistical analysis Principal component analysis identified three major dietary patterns similar across FFQ1, FFQ2, and the 5-day weighted food record. Results The sweet foods and fries pattern was characterized by high intakes of sweet foods, fries and roast potato and kumara (sweet potato), butter and margarines, processed meat, sweet drinks, and fruit or milk drinks. The vegetables and meat pattern was characterized by high intakes of vegetables, meat, eggs and beans, and fruit. The milk and fruit pattern was characterized by high intakes of milk and milk products and fruit, and low intakes of breastmilk and infant and follow-up formula. The FFQ (FFQ1) correctly classified 43.1% to 51.0% of toddlers into the same quartile of pattern score as the 5-day weighted food record, and Pearson correlations ranged from 0.56 to 0.68 for the three patterns. Reliability coefficients ranged from 0.71 to 0.72 for all three dietary patterns. Conclusions the Eating Assessment in Toddlers study FFQ shows acceptable relative validity and high reproducibility for identifying dietary patterns in toddlers. J Acad Nutr Diet. 2014;-:---.

H

IGHER BODY WEIGHT AND RAPID GROWTH BETWEEN birth and age 2 years are associated with an increased risk of obesity in childhood and adulthood.1-3 Identifying dietary patterns contributing to weight and rapid growth in infancy is therefore essential. Although food frequency questionnaires (FFQs) are a cost-effective method for dietary assessment in large population groups,4,5 the relative validity of each new FFQ requires testing.4,6 Validation studies in young children are particularly important because assessing diet during growth presents unique challenges that are not present in older populations, including reliance on surrogate reporters and the presence of multiple caregivers who may provide food and drinks. Few FFQs have been validated for use in young children, particularly in those aged 12 to 24 months, and the focus of these has been food and nutrient intakes rather than dietary patterns.7-18 The analysis of dietary patterns that describe overall diets rather than individual nutrient intakes is an alternative, yet

ª 2014 by the Academy of Nutrition and Dietetics.

complementary, approach that can account for interactions between nutrients and food components eaten together, and allow relationships between dietary patterns and risk of disease to be explored.19,20 Although several adult FFQs have been evaluated for their ability to determine dietary patterns using principal component analysis (PCA),21-28 such research in young children is scarce. The aim of our study was to evaluate the relative validity and reproducibility of dietary patterns defined by PCA from the Eating Assessment in Toddlers (EAT) FFQ, an FFQ developed for use in toddlers aged 12 to 24 months.

SUBJECTS AND METHODS Study Population One hundred sixty parents of toddlers aged 12 to 24 months living in three regions of New Zealand were recruited from September 2011 to April 2012 by newspaper advertisements, JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

1

RESEARCH social media, flyers, and word of mouth. Parents were eligible to participate if their child was born
36 weeks gestation, had no diagnosed medical condition that affected feeding or growth, and was aged between 12 and 24 months. Ethical approval was obtained from the University of Otago Ethics Committee and all parent participants provided written informed consent.

Study Methods Participants attended two appointments, held either at the child’s home or at the University research clinics, approximately 5 weeks apart. At the first appointment, demographic data were collected from the parentechild pair, anthropometric measurements of the child were undertaken, and the FFQ was administered to the parent. The FFQ was repeated at the second appointment. Between appointments, parents completed a weighed food record (ie, the 5DWFR) for their child on 5 assigned randomized nonconsecutive days. At the first appointment, measurements of weight were made to the nearest 0.1 kg (Alpha Model 770, Seca) and length to the nearest 0.1 cm (Rollameter 100, Harlow Healthcare) in duplicate with the infant wearing a dry, preweighed diaper, and a singlet.

identified and corrected. Data were the daily frequency with which each food item was eaten.

Food Record. Participants completed a 5DWFR for their child over a 5-week period.31 Nonconsecutive days were randomly assigned to account for day-of-the-week effects, including weekdays and weekend days. Participants were shown how to complete the food record and provided with calibrated Salter Vista electronic kitchen scales. Additional instructions were provided specifically for early childhood education staff, or other caregivers, to use when the child was away from his or her parents. The researchers reviewed completed food records and clarified any illegible or incomplete data with the participants. Food records were entered into dietary assessment software Kai-culator (version 0.85, 2013, University of Otago). Data were entered by qualified nutritionists and checked for accuracy by a single qualified dietitian, and corrections made where necessary. The daily frequency of food group intake for each participant was obtained from Kai-culator, and weighted so that weekdays made up five-sevenths of the contribution to weekly intake to account for variation in dietary intake between weekdays and weekend days.

Food Groupings. The FFQ contained 91 food items that

Dietary Assessment Food Frequency Questionnaire. The EAT FFQ is an interviewer-administered questionnaire with 91 food items. It is designed to describe dietary intake over the previous 4 weeks and to rank toddlers aged 12 to 24 months by nutrient intake29 and dietary pattern score (the subject of this article). The questionnaire exhibits good validity and high reproducibility for intake of macronutrients and key micronutrients in New Zealand toddlers.29 It contains 10 frequency response options: zero times per month; less than once a week; one, two, three, four, five, six, or seven times a week; plus an open-ended category for recording frequencies greater than once per day. Portion size data were collected to enable the relative validity of the FFQ for determining nutrient intake to be evaluated.29 The frequency data alone were used to determine dietary patterns in the present study. The use of frequency data (rather than grams of intake) allowed inclusion of breastmilk, the portion size for which could neither be estimated by parents, nor measured by weighed diet record. The EAT FFQ was adapted from the Southampton Women’s Survey FFQ, which was designed to assess dietary intakes of UK children aged 12 months over the previous 28 days.14 The food list was tailored to New Zealand toddlers by including foods that at least 10% of New Zealand toddlers eat,30 replacing UK infant foods with comparable New Zealand foods, replacing UK food names with New Zealand equivalents (eg, the drink “squash” was changed to “cordial,” a nonalcoholic sweetened fruit drink), and removing foods not commonly eaten in New Zealand (eg, “gammon”). The EAT FFQ was administered twice: at the first appointment (FFQ1) before food record completion, and at the second appointment (FFQ2), after food record completion. An online interface to enter the FFQ data was created using the LimeSurvey open-source survey application (version 1.90þ, 2010, LimeSurvey.org). The data entry interface was duplicated so that each FFQ was entered twice and errors 2

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

were collapsed into 16 food groups. The sample size of 160 allowed for examination of 16 food groups using PCA, because 10 participants per variable are required for robust results.32 Foods were allocated to the food groups based on nutrient profile and similarity of use (see the Figure). The same 16 food groups were created for the 5DWFR data. In total, 1,480 different food items were entered into Kai-culator from the food records. Of these, 1,339 (90.5%) were matched to the 16 food groups. The remaining items were excluded from the analysis (eg, herbs, dressings, and baking ingredients such as flour).

Statistical Analysis To identify dietary patterns, separate PCAs were conducted using Stata (version 12.1, 2013, StataCorp) on the FFQ1, FFQ2, and 5DWFR frequency data for the 16 food groups using orthogonal varimax rotation. Eigenvalues >1, the elbow of scree plots, and the component interpretability were considered when identifying the number of components retained in the solution. The percentage variation explained by each component depends on the number of variables analyzed, so this was not used as a decision criterion.33 As recommended by Newby and Tucker,20 dietary patterns were named quantitatively, with patterns named based on the first two food groups that loaded highly and positively on each component. Factor loadings of 0.3 were considered significant.20 The relative validity of the dietary patterns was determined by comparing FFQ1 dietary pattern scores to those from the 5DWFR. This was to avoid potential memory effects associated with FFQ2 because it was administered after the 5DWFR had been completed. Cross-classification, Pearson’s correlation coefficients, and intraclass correlation coefficients were used to assess the relative validity of the FFQ. For crossclassification, the dietary pattern scores were categorized into quartiles separately for FFQ1 and for the 5DWFR. The --

2014 Volume

-

Number

-

RESEARCH Food group

Food frequency questionnaire food item

Baby and toddler foods

Little Kids cheesy ravioli (Wattie’s), other Little Kids meals (Wattie’s), other toddler meals, Simply Create meat pouches (Wattie’s), baby rice, baby muesli from packet, vegetable-based meals, meat-based meals, pasta- or rice-based savory meals, rice- and other cereal-based desserts, custard and other milkbased desserts, fruit purée, fruit-based desserts, junior fruit drink

Bread, pasta, rice, low-sugar cereal

White bread, buns (not iced), crumpets, wholemeal or wholegrain bread or buns, rice cakes, rice wheels, crispbreads, Weet-bix (Sanitarium Health Food Company), Fruity-bix (Sanitarium Health Food Company), porridge (not instant porridge in individual servings), cornflakes, rice bubbles, other breakfast cereals (0.4 indicate good agreement.4

neighborhoods and 10% (rather than the expected 30%) from the most deprived neighborhoods).35

RESULTS AND DISCUSSION

Dietary Patterns

One hundred fifty-three parentechild pairs completed the study (95.6%). Child participants (51.3% boys) had a median age of 16.8 months, median length of 81.9 cm, and median weight of 11.1 kg. Children were predominantly European (77.1%) and from higher socioeconomic groups; that is, 42.1% (rather than the expected 30%) were from the least deprived

Dietary data for all 153 participants were included in the PCA, giving a ratio of 9.6 participants to each food group, which is sufficient to produce stable PCA results.32 The PCA produced three patterns that explained 34.1% (for FFQ1), 34.8% (for FFQ2), and 38.2% (for 5DWFR) of the total variation. The three patterns produced (Table 2) were sweet foods and fries,

Food Group Intakes Food group intake frequencies from FFQ1 were comparable to the 5DWFR, with nine food groups having a mean difference in intake frequency of 0.1 daily servings or fewer (Table 1). The biggest differences were observed for vegetables and milk and milk products, where FFQ1 underestimated vegetable intake frequency by 0.551.94 daily servings, and milk and milk products by 0.281.19 daily servings. Intake frequencies were similar for FFQ1 and FFQ2. Fruit had the largest difference between FFQ1 and FFQ2, with FFQ1 estimating fruit intake to be 0.251.25 of a daily serving higher than FFQ2.

Table 1. Frequency of intake of the 16 food groups according to the food frequency questionnaire (FFQ) administered twice, 5 weeks apart (FFQ1 and FFQ2), and the 5-day weighed food record (5DWFR), and differences between FFQ1 and the 5DWFR, and between FFQ1 and FFQ2 for the Eating Assessment in Toddlers Study (N¼153) Frequency of Intake Per Day Food group

FFQ1

FFQ2

5DWFR

Difference FFQ1 vs 5DWFR

FFQ1 vs FFQ2

ƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒmeanstandard deviationƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒƒ!

4

Baby and toddler foods

0.520.81

0.370.69

0.400.70

0.120.45

0.140.43

Bread, pasta, rice, low-sugar cereal

3.080.92

2.900.87

3.281.05

0.211.07

0.170.89

Meat

0.730.35

0.710.35

0.740.50

0.010.45

0.020.35

Processed meat

0.430.32

0.470.31

0.580.45

0.150.41

0.040.22

Eggs and beans

0.460.37

0.440.34

0.510.47

0.050.46

0.020.27

Vegetables

3.071.39

2.981.62

3.622.10

0.551.94

0.090.10

Fruit

3.761.65

3.511.40

3.511.72

0.241.73

0.251.25

Milk and milk products

3.381.58

3.371.52

3.661.79

0.281.19

0.010.90

Breastmilk

0.831.76

0.741.60

0.731.52

0.100.72

0.080.44

Infant and follow-up formula

0.631.07

0.520.96

0.591.05

0.040.48

0.110.59

Butter and margarines

0.770.59

0.780.59

0.880.67

0.110.62

0.010.48

Sweet foods (cakes, cookies, puddings, confectionary, sweet snacks, sweet cereals)

1.471.03

1.481.02

1.471.02

0.000.79

0.010.74

Sweet drinks

0.040.16

0.060.22

0.050.20

0.020.16

0.030.17

Fries, roast potato and kumara (sweet potato)

0.140.13

0.140.15

0.100.15

0.040.17

0.000.13

Savory snacks

0.740.66

0.830.70

0.780.55

0.040.67

0.090.65

Fruit or milk drinks

0.100.25

0.140.33

0.140.33

0.050.27

0.050.26

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

--

2014 Volume

-

Number

-

RESEARCH Table 2. Factor loadings for the three dietary patterns determined by principal component analysis (N¼153)a for the Eating Assessment in Toddlers study based on food frequency questionnaires (FFQs) administered at the first clinic visit (FFQ1) and at the second visit 5 weeks later (FFQ2), and the 5-day weighed food record (5DWFR) FFQ1

FFQ2

Milk Sweet foods Vegetables and and fries and meat fruit

Food group

5DWFR

Milk Sweet foods Vegetables and and fries and meat fruit

Milk Sweet foods Vegetables and and fries and meat fruit

Sweet foods (cakes, cookies, puddings, confectionary, sweet snacks, sweet cereals)

0.47a

0.06

0.12

0.53a

0.05

0.10

0.51a

0.05

0.05

Fries, roast potato and kumara (sweet potato)

0.40a

0.17

0.07

0.42a

0.01

0.28

0.37a

0.07

0.16

Savory snacks

0.36a

0.03

0.17

0.06

0.05

0.16

0.11

0.16

0.26

0.43

Baby and toddler foods Fruit or milk drinks

0.34

0.35

0.08

0.12

0.12

0.01

0.31a

0.02

0.10

0.29

0.02

0.13

0.36a

0.13

0.12

0.01

0.21

0.25

0.04

0.25

a

Butter and margarines

0.12

a

0.30

a

0.03

0.15

0.35

a

a

0.20

Vegetables

0.00

0.54a

0.02

0.00

0.55a

0.06

0.02

0.51a

0.09

Meat

0.14

a

0.07

0.07

a

0.15

0.12

0.53

a

0.08

Fruit

0.01

0.44a

0.21

0.05

0.18

0.36a

0.01

0.13

0.57a

Milk and milk products Breast milk Processed meat

0.49

0.37a

0.01

0.03

0.02

0.12

0.20

0.19

0.38a

0.05

a

0.04

0.63

0.49 0.04

0.45a 0.04 0.19

0.40

a

0.60

0.38a —

—

0.00

0.14

0.37a

0.17

0.15

0.12

a

0.09

0.08 a

a

—

0.20

0.03

0.03

0.38

0.12

0.36

0.04

Bread, pasta, rice, 0.03 low-sugar cereal

0.19

0.23

0.06

0.43a

0.07

0.09

0.22

Eggs and beans

0.27

0.29

0.06

0.01

0.15

0.39a

0.07

0.28

0.21

0.11

0.01

0.27

0.11

0.11

0.21

0.12

0.35a

Sweet drinks

Infant and followup formula Eigenvalue

2.22

Explained variance (%)

12.8

1.80 11.1

1.42 10.2

1.96 12.1

2.06 12.7

0.25 1.54 10.0

2.23 13.8

2.17 12.5

0.42a

1.34 11.9

a Absolute values
0.30. High, positive loadings on a food group demonstrate that the food group is strongly positively associated with a dietary pattern, and negative loadings demonstrate that the food group is inversely associated with a pattern.

which loaded positively for sweet foods, fries, and roast potato and kumara (sweet potato), butter and margarines; processed meat, sweet drinks, and fruit or milk drinks; vegetables and meat, which loaded positively for vegetables, meat, eggs and beans, and fruit, and negatively on baby and toddler foods; and milk and fruit, which loaded positively for --

2014 Volume

-

Number

-

milk and milk products and fruit, and negatively on breastmilk and infant and follow-up formula. All pattern scores were normally distributed. The three patterns that explained the greatest variance were similar for FFQ1, FFQ2, and the 5DWFR, although these were generated in a different order for FFQ2 compared with FFQ1 and the 5DWFR. JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

5

RESEARCH sweetened beverage was consumed by 63% to 81% of toddlers on at least one occasion, salty snacks by 10% to 24% of toddlers, and fries and other fried potatoes by 12% to 19% of toddlers. Compared with other New Zealand data,30 a higher percentage of our sample was consuming white bread, potatoes, and bananas, whereas more children from the United States were consuming soft drinks and chicken. Fruit and vegetable consumption was below recommended levels in both countries. The three patterns identified in our study accounted for a maximum of 38% of the total variation in the data, which is higher than most previous dietary pattern work in adults (14%27 to 34%28 for two to three dietary patterns).21,22,24,25,27,28 The correlation coefficients between the dietary patterns identified in FFQ1 and the 5DWFR were similar to or higher than the correlations observed for similar “Western” and “Sweet” patterns produced in previous validation studies of adults (0.3524 to 0.5821). Correlations >0.70 in FFQ validation studies are rare,34 and in that context, the EAT FFQ (correlations of 0.56 to 0.68) shows good relative validity. There may be several reasons why the EAT FFQ showed good validity compared with the 5DWFR. First, this FFQ was modified from the Southampton Women’s Survey FFQ,14 which had already demonstrated good ability to rank individuals by intake for selected nutrients in a sample of UK toddlers.14 Second, the food list for the EAT FFQ was specifically tailored to New Zealand toddlers based on published data.30 Third, the FFQ was interviewer-administered, so there were minimal missing data. Fourth, 5 days of weighed food records were collected on random nonconsecutive days over 5 weeks, providing a good estimate of habitual intake, making it more comparable to the habitual dietary data collected by the FFQ. The reproducibility of all three dietary patterns was high, with intraclass and Pearson’s correlation coefficients >0.70 for all three patterns, suggesting strong agreement between FFQ administrations. Only four other dietary pattern validation studies have examined reproducibility, reporting correlation coefficients ranging from 0.55 to 0.77.21,23,27,28 Only one

Validity and Reproducibility of Dietary Patterns The cross-classification results, Pearson correlation coefficients, and intraclass correlation coefficients for FFQ1 compared with the 5DWFR are presented in Table 3. Correct classification was highest for the sweet foods and fries pattern, where 51.0% of toddlers were classified into the same quartile of pattern score for FFQ1 as for the 5DWFR (25% would be expected by chance). Across the three patterns, FFQ1 grossly misclassified 0.7% to 3.3% of the toddlers into the opposite quartile from the 5DWFR (12.5% would be expected by chance). Pearson correlation coefficients between the dietary pattern scores for FFQ1 and the 5DWFR were higher than 0.50 for all three patterns. The sweet foods and fries pattern had the highest Pearson correlation coefficient of 0.68. Intraclass correlation coefficients ranged from 0.56 for vegetables and meat to 0.69 for sweet foods and fries. Comparable results were observed when analyses were conducted comparing FFQ2 and 5DWFR (correlations ranged from 0.47 to 0.69) or comparing the average of FFQ1 and FFQ2 with 5DWFR (correlations ranged from 0.56 to 0.74) (data not shown). For reproducibility, strong Pearson correlation coefficients of above 0.70 and strong intraclass correlation coefficients of above 0.70 between the dietary pattern scores for FFQ1 and FFQ2 were observed for all three dietary patterns (data not shown).

Implications of Results and Comparison with Other Studies The food choices of New Zealand toddlers have been reported to be similar in many ways to those of US toddlers.30 Certainly both our study and the Feeding Infants and Toddlers Study36 suggest that intake of energy-dense and nutrient-poor foods is of concern. The frequency of consumption of sweet foods (1.47 times per day) and savory snacks (0.78 times per day) ranked fifth and seventh of 16 food groups in our study, and sweet foods and fries was one of the three dietary patterns identified. Similarly, in the Feeding Infants and Toddlers Study,36 dessert, sweet, or

Table 3. Validity of the food frequency questionnaire administered during the first clinic visit compared with the 5-day weighed food record for the Eating Assessment in Toddlers Study (N¼153) Cross-Classification Proportion Classified (%)

6

Correlation

Dietary pattern

Classified into same quartile

Classified into same or adjacent quartiles

Grossly misclassified into opposite quartiles

Chance

25

62.5

12.5

Sweet foods and fries

51.0

86.9

Vegetables and meat

45.1

Milk and fruit

43.1

Pearson correlation coefficients

Intraclass correlation coefficients

0.7

0.68

0.69

81.7

2.6

0.56

0.56

84.3

3.3

0.58

0.59

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

--

2014 Volume

-

Number

-

RESEARCH study27 had reproducibility correlations consistently above 0.70 for all of the dietary patterns, similar to our study. Both our study and the study by Beck and colleagues27 had much shorter time intervals between the first and second administrations of the FFQ (1 month compared with 1 year in the other three studies21,23,28), which may have contributed to the high reproducibility observed. However, this interval is longer than, or comparable to, studies that tested FFQ reproducibility for food and nutrient intake in preschool populations.10,12,37-40 The sweet foods and fries pattern included cakes, cookies, puddings, confectionary, sweet snacks and sweet cereals, fries, roast potato and kumara (sweet potato), butter and margarines, sweet drinks, fruit or milk drinks, and processed meat. The “Western” and “Sweet” patterns in the literature are associated with intakes of similar foods, including processed meat, butter and margarine, refined grains, fries or roast or fried potato, confectionary, cakes, cookies, sweets, and sweet drinks. It is interesting that such a consistent dietary pattern has been established in adult populations in several different countries (the United States,21 Sweden,23 Denmark,22 United Kingdom,24 and Australia26). It also resembles an “Adult Foods” dietary pattern identified in infants aged 6 and 12 months in the Southampton Women’s Survey.41 Although not as strikingly similar, the vegetable and meat pattern bears some resemblance to “Healthy” and “Prudent” patterns established in the literature, although these tend to also include whole grains or wholemeal bread, and low-fat dairy.21,23,24,26 Differences between adult and toddler healthy patterns are to be expected because New Zealand parents are discouraged from giving toddlers foods that are very low in fat or high in fiber.42 The milk and fruit pattern in the EAT study included milk and fruit intake, but also had an inverse relationship with breastmilk, and infant and follow-up formula. The milk and fruit pattern describes what may be a diet in transition from infant to more adult foods, with cows’ milk replacing previously high intakes of breastmilk or infant formula, and fruit a convenient snack acceptable to the taste preferences of toddlers but also commonly given to infants during the complementary feeding period.

Strengths and Limitations The EAT study has a number of strengths. An appropriate number of participants were recruited to allow for analyses (n¼160) (ie, 9.6 participants per food group), and the completion rate was exceptionally high (96%). Further strengths are the use of the gold standard reference method of 5 nonconsecutive days of weighed food records. The EAT study also has some limitations. The sample was not random, so those who volunteered for the study are likely to be motivated, possibly increasing the apparent validity and reproducibility of the FFQ. The small number of food groups used may also lead to increased estimates of validity.43 The sample was not representative of the New Zealand populaori and Pacific tion, and requires further examination in Ma people. Because the FFQ only covers the past month it is unable to capture seasonal variations in intake; however, in high-income countries such as New Zealand, this is more likely to affect types of fruit and vegetables consumed rather than total frequency of consumption.43 Although weighed --

2014 Volume

-

Number

-

diet records are recognized as the gold standard,43 errors in recording as well as changes in dietary habits as a result of keeping a record are inevitable. This noise may have decreased the apparent validity of the FFQ. Finally, in common with other studies, a number of subjective decisions had to be made during the PCA, including how foods were grouped, eigenvalue cut points and method of rotation, the number of components to retain in the solution, the interpretation of the factor loading matrix, and the naming of the dietary patterns. To account for this subjectivity, the decisions made were documented and reported.

CONCLUSIONS The EAT FFQ shows good relative validity and high reproducibility for determining dietary patterns in toddlers. Possible future research applications might include using the EAT FFQ to evaluate whether toddlers’ dietary patterns predict dietary patterns in later childhood, are related to their risk of obesity and nutrition-related diseases later in life, and are amenable to early intervention for the prevention of obesity and nutrition-related disease.

References 1.

Baird J, Fisher D, Lucas P, Kleijnen J, Roberts H, Law C. Being big or growing fast: Systematic review of size and growth in infancy and later obesity. BMJ. 2005;331(7522):929.

2.

Monteiro POA, Victora CG. Rapid growth in infancy and childhood and obesity in later life—a systematic review. Obes Rev. 2005;6(2): 143-154.

3.

Druet C, Stettler N, Sharp S, et al. Prediction of childhood obesity by infancy weight gain: An individual-level meta-analysis. Paediatr Perinat Epidemiol. 2012;26(1):19-26.

4.

Gibson RS. Principles of Nutritional Assessment, 2nd ed. New York, NY: Oxford University Press; 2005.

5.

Willett W. Nutritional Epidemiology. 2nd ed. New York, NY: Oxford University Press; 1998.

6.

Willett W, Lenart E. Reproducibility and validity of food-frequency questionnaires. Nutritional Epidemiology. 2nd ed. Oxford, UK: Oxford University Press; 1998:101-147.

7.

Andersen L, Lande B, Trygg K, Hay G. Validation of a semiquantitative food-frequency questionnaire used among 2-year-old Norwegian children. Public Health Nutr. 2004;7(6):757-764.

8.

Andersen LF, Lande B, Arsky GH, Trygg K. Validation of a semiquantitative food-frequency questionnaire used among 12-monthold Norwegian infants. Eur J Clin Nutr. 2003;57(8):881-888.

9.

Blum RE, Wei EK, Rockett HRH, et al. Validation of a food frequency questionnaire in Native American and Caucasian children 1 to 5 years of age. Matern Child Health J. 1999;3(3):167-172.

10.

D’Ambrosio A, Tiessen A, Simpson JR. Development of a food frequency questionnaire for toddlers of low-German-speaking Mennonites from Mexico. Can J Diet Pract Res. 2012;73(1):40-44.

11.

Iannotti RJ, Blyer EM, O’Brien RW, Finn J, Spillman DM. Comparison of dietary intake methods with young children. Psychological Rep. 1994;74:883-889.

12.

Klohe DM, Clarke KK, Chacko George G, Milani TJ, Hanss-Nuss H, Freeland-Graves J. Relative validity and reliability of a food frequency questionnaire for a triethnic population of 1-year-old to 3year-old children from low-income families. J Am Diet Assoc. 2005;105(5):727-734.

13.

Marriott LD, Robinson SM, Poole J, et al. What do babies eat? Evaluation of a food frequency questionnaire to assess the diets of infants aged 6 months. Public Health Nutr. 2008;11(7):751-756.

14.

Marriott LD, Inskip HM, Borland SE, et al. What do babies eat? Evaluation of a food frequency questionnaire to assess the diets of infants aged 12 months. Public Health Nutr. 2009;12(7):967-972.

15.

Marshall TA, Eichenberger Gilmore JM, Broffitt B, Levy SM, Stumbo PJ. Relative validation of a beverage frequency questionnaire

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

7

RESEARCH in children ages 6 months through 5 years using 3-day food and beverage diaries. J Am Diet Assoc. 2003;103(6):714-720.

Old [master’s thesis]. Dunedin, New Zealand: Department of Human Nutrition, University of Otago; 2012.

16.

Parrish LA, Marshall JA, Krebs NF, Rewers M, Norris JM. Validation of a food frequency questionnaire in preschool children. Epidemiology. 2003;14(2):213-217.

30.

Szymlek-Gay E, Ferguson E, Heath A-L, Fleming E. Quantities of foods consumed by 12- to 24-month-old New Zealand children. Nutr Diet. 2010;67:244-250.

17.

Collins CE, Burrows TL, Truby H, et al. Comparison of energy intake in toddlers assessed by food frequency questionnaire and total energy expenditure measured by the doubly labeled water method. J Acad Nutr Diet. 2013;113(3):459-463.

31.

Stram DO, Longnecker MP, Shames L, et al. Cost-efficient design of a diet validation study. Am J Epidemiol. 1995;142(3):353-362.

32.

Floyd F, Widaman K. Factor analysis in the development and refinement of clinical assessment instruments. Psychol Assess. 1995;7(3):286-299.

18.

Williams PL, Innis SM. Food frequency questionnaire for assessing infant iron nutrition. Can J Diet Pract Res. 2005;66(3):176-182.

33.

19.

Hu FB. Dietary pattern analysis: A new direction in nutritional epidemiology. Curr Opin Lipidol. 2002;13(1):3-9.

Michels K, Schulze MB. Can dietary patterns help us detect dietdisease associations? Nutr Res Rev. 2005;18:241-248.

34.

20.

Newby PK, Tucker KL. Empirically derived eating eating patterns using factor or cluster analysis: A review. Nutr Rev. 2004;62(5):177-203.

Willett W. Invited commentary: A further look at dietary questionnaire validation. Am J Epidemiol. 2001;154(12):1100-1102.

35.

21.

Hu FB, Rimm E, Smith-Warner SA, et al. Reproducibility and validity of dietary patterns assessed with a food-frequency questionnaire. Am J Clin Nutr. 1999;69:243-249.

Salmond C, Crampton P, Atkinson J. New Zealand Index of Deprivation, 2006. Wellington, New Zealand; 2007.

36.

Togo P, Heitman BL, Sorensen TI, Osler M. Consistency of food intake factors by different dietary assessment methods and population groups. Br J Nutr. 2003;90:667-678.

Siega-Riz AM, Deming DM, Reidy KC, Fox MK, Condon E, Briefel RR. Food consumption patterns of infants and toddlers: Where are we now? J Am Diet Assoc. 2010;110(12 suppl):S38-S51.

37.

Huybrechts I, De Backer G, De Bacquer D, Maes L, De Henauw S. Relative validity and reproducibility of a food-frequency questionnaire for estimating food intakes among Flemish preschoolers. Int J Environ Res Public Health. 2009;6:382-399.

38.

Huybrechts I, De Bacquer D, Matthys C, De Backer G, De Henauw S. Validity and reproducibility of a semi-quantitative food-frequency questionnaire for estimating calcium intake in Belgian preschool children. Br J Nutr. 2006;95(4):802-816.

39.

Metcalf P, Scragg R, Sharpe S, Fitzgerald E, Schaaf D, Watts C. Shortterm repeatability of a food frequency questionnaire in New Zealand children 1-14 y. Eur J Clin Nutr. 2003;57:1498-1503.

40.

Treiber FA, Leonard SB, Frank G, et al. Dietary assessment instruments for preschool children: Reliability of parental responses to the 24-hour recall and a food frequency questionnaire. J Am Diet Assoc. 1990;90(6):814-820.

41.

Robinson S, Marriott L, Poole J, et al. Dietary patterns in infancy: The importance of maternal and family influences on feeding practice. Br J Nutr. 2007;98:1029-1037.

42.

Ministry of Health. Food and Nutrition Guidelines for Healthy Infants and Toddlers (Aged 0-2): A Background Paper e Partially Revised December 2012. Wellington, New Zealand: Ministry of Health; 2008.

43.

Cade J, Thompson R, Burley V, Warm D. Development, validation and utilisation of food-frequency questionnaires—A review. Public Health Nutr. 2002;5(4):567-587.

22.

23.

Khani BR, Ye W, Terry P, Wolk A. Reproducibility and validity of major dietary patterns among Swedish women assessed with a foodfrequency questionnaire. J Nutr. 2004;134:1541-1545.

24.

Crozier S, Inskip H, Godfrey K, Robinson S. Dietary patterns in pregnant women: A comparison of food-frequency questionnaires and 4 d prospective diaries. Br J Nutr. 2007;99:869-875.

25.

Okubu H, Murakami K, Sasaki S, et al. Relative validity of dietary patterns derived from a self-administered diet history questionnaire using factor analysis among Japanese adults. Public Health Nutr. 2009;13(7):1080-1089.

26.

Ambrosini GL, O’Sullivan TA, de Klerk NH, Mori TA, Beilin LJ, Oddy WH. Relative vailidity of adolescent dietary patterns: A comparison of a FFQ and 3 d food record. Br J Nutr. 2010;105:625-633.

27.

Beck K, Kruger R, Conlon C, et al. The relative validity and reproducibility of an iron food frequency questionnaire for identifying iron-related dietary patterns in young women. J Acad Nutr Diet. 2012;112(8):1177-1187.

28.

Nanri A, Shimazu T, Ishihara J, et al. Reproducibility and validity of dietary patterns assessed by a food frequency questionnaire used in the 5-year follow-up survey of the Japan Public Health Center-Based Prospective Study. J Epidemiol. 2012;22(3):205-215.

29.

Watson E. Validation of a Multi-Nutrient Food Frequency Questionnaire to Determine Nutrient Intakes of New Zealand Toddlers 12-24 Months

AUTHOR INFORMATION V. C. Mills and E. O. Watson are nutritionists; at the time of the study, they were graduate students, Department of Human Nutrition, University of Otago, Dunedin, New Zealand. P. M. L. Skidmore and A.-L. M. Heath are senior lecturers, and E. A. Fleming is an assistant research fellow, Department of Human Nutrition, and R. W. Taylor is a research associate professor, Department of Medicine, all at the University of Otago, Dunedin, New Zealand. Address correspondence to: Paula M. L. Skidmore, PhD, Department of Human Nutrition, University of Otago, PO Box 56, Dunedin 9054, New Zealand. E-mail: [email protected]

STATEMENT OF POTENTIAL CONFLICT OF INTEREST No potential conflict of interest was reported by the authors.

FUNDING/SUPPORT Funding for the study was provided by the University of Otago.

8

JOURNAL OF THE ACADEMY OF NUTRITION AND DIETETICS

--

2014 Volume

-

Number

-