Clinical Nutrition xxx (2013) 1e9

Contents lists available at ScienceDirect

Clinical Nutrition journal homepage: http://www.elsevier.com/locate/clnu

Original article

Reproducibility and comparative validity of a food frequency questionnaire for Australian adults Clare E. Collins a, b, *, May M. Boggess c, Jane F. Watson a, b, Maya Guest a, Kerith Duncanson a, b, Kristine Pezdirc a, b, Megan Rollo a, b, Melinda J. Hutchesson a, b, Tracy L. Burrows a, b a b c

School of Health Sciences, Faculty of Health, University of Newcastle, Callaghan, NSW, Australia Priority Research Centre in Physical Activity and Nutrition, University of Newcastle, Callaghan, NSW, Australia School of Mathematical and Statistical Sciences, Arizona State University, Tempe, AZ, USA

a r t i c l e i n f o

s u m m a r y

Article history: Received 2 July 2013 Accepted 26 September 2013

Background: Food frequency questionnaires (FFQ) are used in epidemiological studies to investigate the relationship between diet and disease. There is a need for a valid and reliable adult FFQ with a contemporary food list in Australia. Aims: To evaluate the reproducibility and comparative validity of the Australian Eating Survey (AES) FFQ in adults compared to weighed food records (WFRs). Methods: Two rounds of AES and three-day WFRs were conducted in 97 adults (31 males, median age and BMI for males of 44.9 years, 26.2 kg/m2, females 41.3 years, 24.0 kg/m2. Reproducibility was assessed over six months using Wilcoxon signed-rank tests and comparative validity was assessed by intraclass correlation coefficients (ICC) estimated by fitting a mixed effects model for each nutrient to account for age, sex and BMI to allow estimation of between and within person variance. Results: Reproducibility was found to be good for both WFR and FFQ since there were no significant differences between round 1 and 2 administrations. For comparative validity, FFQ ICCs were at least as large as those for WFR. The ICC of the WFR-FFQ difference for total energy intake was 0.6 (95% CI 0.43, 0.77) and the median ICC for all nutrients was 0.47, with all ICCs between 0.15 (%E from saturated fat) and 0.7 (g/day sugars). Conclusions: Compared to WFR the AES FFQ is suitable for reliably estimating the dietary intakes of Australian adults across a wide range of nutrients. Ó 2013 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved.

Keywords: Food frequency questionnaire Nutrition Dietary methods Reproducibility Comparative validity Intra-class correlation coefficient

1. Introduction Accurate assessment of dietary intake is critical to examining associations between food intake, obesity and risk of chronic disease mortality.1 The prevalence of obesity has almost doubled in men (4.8%e9.8%) and women (7.9%e13.8%) over the past 30 years.2 Given that obesity precedes development of many chronic

Abbreviations: FFQ, food frequency questionnaire; AES, Australian Eating Survey; WFR, weighed food record; ICC, intra-class correlation coefficient; ACAES, Australian child and adolescent eating survey; CSIRO, Commonwealth Scientific Industrial Research Organisation; ACCV, Australian Cancer Council of Victoria; BMI, body mass index; %E, percent energy; CI, confidence interval. * Corresponding author. School of Health Sciences, Faculty of Health, HA 12 Hunter Building, The University of Newcastle, University Drive, Callaghan, NSW 2308, USA. Tel.: þ61 2 49215646; fax þ61 2 49217053. E-mail address: [email protected] (C.E. Collins).

conditions, it is important to examine current food patterns, using current and valid tools to assist in monitoring intake. A number of methods have been used to measure usual dietary intake at the population level, however the accurate assessment of diet still presents on-going challenges, including substantial burden for both individuals and researchers, particularly in large population samples.3 Although 24-h recalls and weighed food records (WFRs) have been used successfully, the resource burden and economic constraints of these methods make them unsuitable for most large scale studies.4 Food frequency questionnaires (FFQs) have a lower respondent burden, are relatively inexpensive, do not require trained interviewers and can be semi-automated using technological administration, rendering them practical for large epidemiologic studies.5 Frequency data can explain much of the variation in dietary intake and FFQs can provide sufficient accuracy to rank individuals in terms of risk for subsequent health outcomes.6 FFQs have been used in adults to predict associations

0261-5614/$ e see front matter Ó 2013 Elsevier Ltd and European Society for Clinical Nutrition and Metabolism. All rights reserved. http://dx.doi.org/10.1016/j.clnu.2013.09.015

Please cite this article in press as: Collins CE, et al., Reproducibility and comparative validity of a food frequency questionnaire for Australian adults, Clinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.09.015

2

C.E. Collins et al. / Clinical Nutrition xxx (2013) 1e9

between dietary intake and disease specific mortality and morbidity including, colon cancer, heart disease and diabetes.7 In Australia, the Commonwealth Scientific Industrial Research Organisation (CSIRO) developed a FFQ in the early 1980s8 and the Australian Cancer Council of Victoria (ACCV) developed a FFQ for adults in the late 1980s.9 However, evaluation of the performance of the CSIRO FFQ showed it had limited validity and the ACCV FFQ, also known as the dietary questionnaire for epidemiological studies (DQES) was designed for a specific population of 40e69 year old men and women living in Australia but born in Greece, Italy or Australia.9 Both of these instruments are now over 25 years old and the Australian food supply has changed significantly over that period. While an Australian study reported the reproducibility, of a FFQ in adults with a mean age of 60 years,10 it did not evaluate validity, nor did it consider nutrient intakes. The Australian Eating Survey (AES) was developed in response to these gaps. Therefore the aim was to assess the reproducibility and comparative validity of nutrient intakes derived from a semi-quantitative, selfcompleted FFQ designed for adults residing in Australia. 2. Materials and methods 2.1. Participants and recruitment The population of interest was healthy adults living full-time with at least one child aged 8e10 years, living in the Hunter and Great Lakes regions of New South Wales, Australia, so as to potentially extend the age range and use of a previous FFQ validated for use in children aged 9e16 years.11 Potential participants were recruited through a range of avenues, including newspapers, community notice boards and school newsletters. 2.2. Study design Participants completed the test measure AES FFQ and the reference measure, a 3-day weighed food record, on two occasions, approximately six months apart. Round 1 occurred from September 2010 to July 2011 and Round 2 between January 2011 and February 2012. A brief assessment session was also conducted at both time points to collect anthropometric measurements. 2.3. AES FFQ The AES was self-completed during the assessment sessions in Round 1 and 2. The adult AES FFQ was modified from the Child and Adolescent Eating Survey’ (ACAES) that had been previously validated for youth11 by replacing the portion sizes with adult serve size data. Validation studies using biomarkers in conjunction with the ACAES have been conducted using plasma carotenoids as marker of fruit and vegetable intakes,12 red blood cell membrane fatty acids and dietary fat13 and doubly labelled water to evaluate validity of reporting total energy intake14 highlighting its utility in assessing a range of dietary components. AES is a 120-item semi-quantitative FFQ with 15 supplementary questions regarding age, use of vitamin supplements, food behaviours and sedentary behaviours. Standard portion sizes for adult men and women were determined for each food item using data derived from the most current National Nutrition Survey15 from unpublished data purchased from the Australian Bureau of Statistics and the ’natural’ serving size from standard items such as a slice of bread. AES is designed as a self-administered tool, to collect information about the dietary intake of Australian adults over the previous 6 months. An individual response for each food, or food type, is required, with frequency options ranging from ‘Never’ to ‘4 or more times per day’ and for some beverages up to ‘7 or more

glasses per day’, but varies depending on the item. The FFQ groups food items according to their food group which includes; drinks, breads and cereals, dairy food, main meals, sweets and snacks, fruit and vegetables. The frequency categories for seasonal fruit were calculated by adjusting for the number of months per year the fruit was available. Nutrient intakes were computed using Australian AusNut 1999 database (All Foods) Revision 17 primarily, and AusFoods (Brands) Revision 5 (Australian Government Publishing Service, Canberra). The estimated mean individual daily intake for 20 macro- and micro-nutrients was calculated using FoodWorks (version 3.02.581, Xyris Software Australia, Highgate Hill, Queensland). 2.4. WFR instrument The reference method was two three-day WFRs completed in conjunction with the assessment sessions time periods in Round 1 and 2. At the baseline assessment session participants were advised to maintain usual eating habits and detailed instructions and a demonstration of how to weigh and measure foods was given by Accredited Practising Dietician (APD) research assistants. All food and drink consumed was weighed and recorded, not including medications or supplements, using SOENLE Venezia electronic kitchen scales (accuracy  1 g) (Soehnle-Waagen GmbH & Co, Murrhardt, Germany) which were provided to participants. The brand name of products was recorded and detailed recipes recorded for home prepared dishes. The WFRs were completed over three consecutive days and included at least one weekend day. Record sheets with written instructions were provided for Round 1 and 2 and returned in prepaid envelopes. Daily nutrient intakes were estimated from the WFR by entering the itemised foods and beverages into Foodworks (version 3.02.581, Xyris Software Australia, Highgate Hill, Queensland) using the same database as for the AES FFQ nutrient data and following a standardized protocol by one APD research assistant to reduce error. 2.5. Ethics Procedures followed were in accordance with the Helsinki Declaration of 1975 and as revised in 1983 with the protocol for this study was approved by the University of Newcastle Human Research Ethics Committee (Approval No. H-2010-1170). 2.6. Statistics Medians and interquartile ranges were calculated for all nutrients. Univariate relationships were assessed using Fisher’s exact tests to compare categorical variables by sex within round, and exact symmetry tests to compare categorical variables by sex on paired data. Continuous variables were similarly assessed using Wilcoxon rank-sum tests and Wilcoxon signed-rank tests for paired data. Reproducibility was evaluated using all participant data in both WFR and FFQ by comparing the two rounds using Pearson correlations (r), which were estimated by fitting simple linear regression models to standardized nutrient data and using standard errors clustered on family, given some of the participants were spouses. The normality of the residuals from these models was assessed graphically using normal probability plots and where a lack of normality was observed, square root and cubed root transforms were considered. Comparative validity was assessed using the same methods applied to the difference between FFQ and WFR. To address the issue of a non- random sample and more than one participant per family, correlation coefficients have also been calculated via a linear regression model and using standard errors

Please cite this article in press as: Collins CE, et al., Reproducibility and comparative validity of a food frequency questionnaire for Australian adults, Clinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.09.015

C.E. Collins et al. / Clinical Nutrition xxx (2013) 1e9

3

Table 1 Demographic and anthropometric data on 97 adults (31 male) comprising of 165 observations, over two rounds of data collection, across 68 families). Round 1 Male

Education Year 10 Year 12 Trade Certificate Degree Postgrade Total Smoked within 10 yrsc Yes No Total Current smokerc Yes No Total General healthc Excellent Very good Good Fair/poor Total

Age (years) Height (cm)d Weight (kg)d BMI (kg/m2) d Waist (cm) d a b c d

Round 2 Female

Pa

Male

Female

Pa

N (%)

N (%)

N (%)

N (%)

31

65

20

47

2 (6%) 1 (3%) 5 (16%) 7 (23%) 8 (26%) 8 (26%) 31

5 (8%) 9 (14%) 2 (3%) 16 (25%) 20 (31%) 12 (19%) 64

0.18

1 (5%) 1 (5%) 5 (25%) 2 (10%) 3 (15%) 8 (40%) 20

3 (6%) 7 (15%) 1 (2%) 11 (23%) 14 (30%) 11 (23%) 47

0.03

2 (6%) 29 (94%) 31

4 (6%) 61 (94%) 65

1

3 (15%) 17 (85%) 20

3 (6%) 46 (94%) 49

0.35

1 (3%) 30 (97%) 31

2 (3%) 63 (97%) 65

1

1 (5%) 19 (95%) 20

0 (0%) 49 (100%) 49

0.29

4 (33%) 3 (25%) 5 (42%) 0 (0%) 12

6 (21%) 15 (54%) 7 (25%) 0 (0%) 28

0.25

1 5 1 0 7

8 (35%) 11 (48%) 4 (17%) 0 (0%) 23

0.62

(14%) (71%) (14%) (0%)

Median (minemax)

Median (minemax)

Pb

Median (minemax)

Median (minemax)

Pb

44.9 179 82.6 26.2 92.0

41.3 165 65.6 24.0 81.0

0.01 0 0 0.01 0

44.2 179 81.6 26.8 91.4

41.9 164 65.0 23.5 80.4

0.07 0 0 0.12 0

(36e58) (167e189) (59e130) (20e40) (73e134)

(32e74) (152e180) (46e105) (18e38) (66e133)

(38e53) (167e188) (61e114) (21e36) (76e112)

(33e51) (151e180) (50e100) (18e38) (66e119)

Fisher’s exact test of homogeneity. Wilcoxon rank-sum test for equality of populations. No significant difference by sex in Round 1, Round 2 or in total according to the exact symmetry test of homogeneity for discrete paired data. No significant difference by sex in Round 1, Round 2 or in total according to the Wilcoxon signed-rank test for equality of distributions for continuous paired data.

clustered on family. Intra-class correlation coefficients (ICCs) were used to address the issues related to non-linear relationships between the two measures and variability due to differences within and between participants. It is the ratio of variation between participants and the total variation, meaning that the ICC is the fraction of the variability due to causes other than variability within a participant. Thus, when the ICC is close to 1, this indicates that a single observation suffices, in that if a subsequent observation were taken, it is likely to be similar to the original. The ICC, the total variance and its component parts (the within and the between person variance) are estimated using a linear regression model with a person-level random effect. In this study ICCs were calculated using random effects models. To account for within-family correlation, family-clustered bootstrapped standard errors with 1000 replications were used. Mean difference between FFQ and WFR was estimated using linear regression models with a random effect for family. Demographic and anthropometric variables age, sex, education, general health, body mass index (BMI) and waist were assessed for possible effect on the difference between FFQ and WFR also using linear regression models with a random effect for family. Statistical significance is determined at the 5% level. All statistical analysis was performed using Stata MP version 12. 3. Results A total of 98 participants were recruited to the study, with one excluded as neither FFQ nor WFR data was completed. All 97

participants completed a FFQ at the initial assessment session and of these, 68 completed a second FFQ in round 2. Ninety one of the 97 participants returned completed WFRs in round 1 and 66 in round 2, 65 of whom had completed a round 1 WFR. Table 1 reports round 1, round 2 and change in demographic and anthropometric variables of the study participants, by sex. Thirty one participants were male and 65 were female from 68 families in round 1 and of these 20 males and 47 females from 50 families remained for round 2. There were no significant differences by sex or by round in education, smoking habits and general health according to Fisher’s exact tests. While there were some significant sex differences in weight, height, BMI and waist, there were no significant differences in these variables between round 1 and round 2. 3.1. WFR reproducibility The median and first and third quantiles for WFRs are presented in Table 2. Good reproducibility was demonstrated based on the lack of significant differences between round 1 and 2 WFRs using the Wilcoxon signed-rank tests (except for water, greater consumption in round 1, p-value ¼ 0.03). The median correlation r between round 1 and 2 WFRs was 0.59 (saturated fats), with all r between 0.32 (cholesterol) and 0.74 (sugars). Only the residuals from the models involving alcohol were found to lack normality, so a cubed root transform was used for that variable. All correlations were significantly different from zero, other than niacin equivalent (p-values ¼ 0.08).

Please cite this article in press as: Collins CE, et al., Reproducibility and comparative validity of a food frequency questionnaire for Australian adults, Clinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.09.015

4

C.E. Collins et al. / Clinical Nutrition xxx (2013) 1e9

Table 2 Descriptive statistics for Weighed Food Records (WFR): Median (1ste3rd quartile), percentage change and Pearson correlation coefficient. The % Change is Round 2 subtracted from Round 1, as a percentage of Round 1. Nutrients/day

Energy Energy (MJ) Protein (g) Total fat (g) Saturated fat (g) Polyunsat. Fat (g) Monounsat. Fat (g) Cholesterol (mg) Carbohydrate (g) Sugars (g) Alcohol (g) Nutrients Fiber (g) Thiamin (mg) Riboflavin (mg) Niacin (mg) Vitamin C (mg) Folate (mg) Vitamin A (mg) Retinol (mg) Beta-carotene (mg) Sodium (mg) Potassium (mg) Magnesium (mg) Calcium (mg) Phosphorus (mg) Iron (mg) Zinc (mg) Water (L) Percent energy Protein Carbohydrate Total fats Saturated fat Alcohol Percent fat Saturated Polyunsaturated Monounsaturated a b

Round 1 N ¼ 91

Round 2 N ¼ 66

% Change N ¼ 65

Correlation

Median (Q1eQ3)

Median (Q1eQ3)

Median (Q1eQ3)

r

95%CIa

8.36 89.4 64.4 25.8 9.56 22.7 227 224 95.8 0.43

(7e10) (76e112) (49e82) (17e34) (7e13) (18e29) (171e359) (172e278) (76e127) (0e13)

8.17 83.6 66.6 25.0 9.60 22.9 226 210 97.5 2.91

(7e9) (72e106) (49e81) (18e32) (7e13) (17e30) (155e290) (165e279) (66e126) (0e10)

1 2 0 7 1 1 7 6 7 10

(18e13) (21e24) (21e29) (35e32) (17e28) (23e31) (29e44) (19e7) (24e16) (69e298)b

0.61 0.46 0.49 0.59 0.49 0.48 0.32 0.71 0.74 0.57

(0.4, (0.3, (0.2, (0.4, (0.2, (0.2, (0.1, (0.5, (0.5, (0.3,

0.8) 0.6) 0.7) 0.8) 0.8) 0.8) 0.5) 0.9) 0.9) 0.8)

23.8 1.71 2.41 45.8 101 426 0.80 270 2.94 2.15 3.35 381 0.86 1.61 12.5 12.1 2.60

(19e32) (1e2) (2e3) (39e58) (66e133) (329e535) (1e1) (182e399) (2e5) (2e3) (3e4) (316e473) (1e1) (1e2) (9e17) (9e14) (2e3)

22.3 1.67 2.37 47.6 83.0 423 0.85 283 3.27 2.34 3.15 372 0.91 1.49 11.7 11.6 2.39

(17e31) (1e2) (2e3) (38e59) (59e138) (310e562) (1e1) (151e386) (2e5) (2e3) (3e4) (291e454) (1e1) (1e2) (9e15) (9e14) (2e3)

6 4 1 10 0 10 3 4 11 8 5 6 1 4 6 4 6

(26e14) (27e23) (23e19) (18e27) (37e46) (20e41) (17e31) (37e30) (29e60) (16e30) (20e13) (21e17) (26e16) (21e14) (25e21) (20e25) (23e13)

0.69 0.66 0.64 0.37 0.63 0.68 0.42 0.59 0.43 0.54 0.65 0.63 0.65 0.49 0.65 0.52 0.55

(0.5, (0.4, (0.4, (0.0, (0.4, (0.3, (0.1, (0.4, (0.1, (0.4, (0.4, (0.4, (0.5, (0.3, (0.5, (0.4, (0.4,

0.9) 0.9) 0.9) 0.8) 0.8) 1.0) 0.7) 0.8) 0.7) 0.7) 0.9) 0.8) 0.8) 0.7) 0.8) 0.7) 0.8)

18.8 44.2 30.1 10.9 0.16

(17e22) (39e48) (24e34) (8e14) (0e6)

18.0 45.0 30.7 11.5 0.98

(16e21) (38e49) (26e35) (8e14) (0e4)

(8e11) (10e6) (10e16) (25e25) (71e349)b

0.56 0.66 0.52 0.67 0.45

(0.3, (0.5, (0.3, (0.5, (0.0,

0.8) 0.8) 0.7) 0.8) 0.9)

0.74 0.67 0.52

(0.6, 0.9) (0.3, 1.0) (0.3, 0.8)

41.5 (36e49) 17.4 (13e21) 40.0 (36e44)

2 3 2 4 11

1 (9e8) 3 (12e18) 2 (6e10)

41.1 (36e48) 16.8 (14e21) 40.5 (37e44)

95% confidence intervals using standard errors clustered on family. N ¼ 44 as % change does not include participants with zero grams of alcohol in Round 1.

3.2. AES FFQ reproducibility The median and first and third quantiles for FFQs are presented in Table 3. Good reproducibility was demonstrated based on the lack of significant differences between round 1 and 2 in FFQs using Wilcoxon signed-rank tests (except for vitamin C, greater consumption in round 1, p-value ¼ 0.04). The median correlation r between round 1 and 2 FFQs was 0.72 (%E saturated fats), with all r between 0.49 (%E protein) and 0.85 (alcohol). A cubed root transform was again used for alcohol. All correlations were significantly different from zero. 3.3. FFQ to WFR comparative validity Comparative validity results are shown in Table 4. When subtracting the WFR from the FFQ, the median difference was significant and positive for the vast majority of nutrients, in both rounds. When comparing the FFQ-WFR difference between rounds, Wilcoxon signed-rank test showed that there was no significant difference between the rounds, meaning that in both rounds the median FFQ was the same amount greater than the WFR. The correlation between the two rounds of the FFQ and WFR difference

was analysed next. The median r value was 0.46 (mono-unsaturated fats), with all r between 0.10 (cholesterol) and 0.78 (alcohol), with all significantly different from zero except for cholesterol, beta-carotene, %E alcohol and niacin equivalent, (p-values ¼ 0.4, 0.13, 0.16, 0.08, respectively). Table 5 lists the ICCs for the FFQs, WFRs and the difference between the two. The ICCs for the FFQs, WFRs are displayed in Fig. 1 along with their 95% CIs. The median ICC for WFR was 0.59 (%E alcohol), with all ICC between 0.36 (cholesterol) and 0.73 (sugars) and all are significantly different from zero. The median ICC for FFQ was 0.74 (niacin equivalent), with all ICC between 0.52 (%E protein) and 0.88 (alcohol gram/day) and all are significantly different from zero. The FFQ ICCs are significantly greater than those for WFR (seen as non-overlapping confidence intervals in Fig. 1) for total energy, total fats, cholesterol, sodium, carbohydrate, water, niacin equivalent, beta-carotene, vitamin C, folate, zinc, phosphorus, magnesium, alcohol and %E alcohol. Figure 2 displays ICCs for the difference between FFQ and WFR. The median ICC was 0.49 (fiber), with all ICC between 0.14 (cholesterol) and 0.74 (sugars), and all other than cholesterol were significantly different from zero.

Please cite this article in press as: Collins CE, et al., Reproducibility and comparative validity of a food frequency questionnaire for Australian adults, Clinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.09.015

C.E. Collins et al. / Clinical Nutrition xxx (2013) 1e9

5

Table 3 Descriptive statistics for Food Frequency Questionnaire (FFQ): Median (1ste3rd quartile), percentage change and Pearson correlation coefficient. The % Change is Round 2 subtracted from Round 1, as a percentage of Round 1. Nutrients/day

Energy Energy (MJ) Protein (g) Total fat (g) Saturated fat (g) Polyunsat. fat (g) Monounsat. fat (g) Cholesterol (mg) Carbohydrate (g) Sugars (g) Alcohol (g) Nutrients Fiber (g) Thiamin (mg) Riboflavin (mg) Niacin (mg) Vitamin C (mg) Folate (mg) Vitamin A (mg) Retinol (mg) Betacarotene (mg) Sodium (mg) Potassium (mg) Magnesium (mg) Calcium (mg) Phosphorus (mg) Iron (mg) Zinc (mg) Water (L) Percent energy Protein Carbohydrate Total fats Saturated fat Alcohol Percent fat Saturated Polyunsaturated Monounsaturated a b c

Round 1 N ¼ 96

Round 2 N ¼ 68

% Change N ¼ 67

Correlation

Median (Q1eQ3)

Median (Q1eQ3)

Median (Q1eQ3)

r

95%CIa

9.60 101 75.5 30.1 9.70 27.8 283 262 141 12.0

(8e12) (82e125) (63e85) (25e36) (8e11) (23e32) (224e360) (217e341) (100e182) (2e20)

9.16 97.7 73.7 30.7 9.27 27.3 252 245 122 7.45

(7e11) (78e125) (54e92) (22e35) (7e12) (20e35) (212e331) (196e339) (98e170) (2e14)b

3 1 2 2 5 4 1 3 7 1

(13e10) (17e13) (17e17) (21e14) (19e22) (15e20) (21e20) (18e9) (21e15) (40e35)c

0.81 0.65 0.71 0.67 0.76 0.73 0.66 0.83 0.82 0.85

(0.7, (0.5, (0.5, (0.4, (0.6, (0.5, (0.4, (0.7, (0.7, (0.7,

1.0) 0.8) 0.9) 0.9) 0.9) 0.9) 0.9) 1.0) 1.0) 1.0)

30.5 1.77 2.59 45.3 184 372 1.23 297 5.32 2.27 3.88 450 1.20 1.74 15.1 13.9 3.47

(24e37) (1e2) (2e3) (39e56) (140e235) (288e455) (1e2) (227e410) (4e7) (2e3) (3e5) (371e531) (1e1) (1e2) (11e18) (11e17) (3e4)

29.9 1.75 2.55 44.0 168 359 1.23 324 5.10 2.33 3.81 431 1.21 1.71 14.6 13.6 3.39

(24e36) (1e2) (2e3) (37e55) (136e212) (281e461) (1e2) (222e488) (4e7) (2e3) (3e5) (344e543) (1e2) (1e2) (11e18) (11e17) (3e4)

2 3 1 1 4 3 1 0 1 1 1 3 1 1 1 2 1

(17e10) (19e21) (16e16) (15e13) (16e11) (17e15) (15e24) (19e26) (16e18) (15e21) (14e14) (11e10) (14e17) (13e12) (19e12) (18e10) (10e7)

0.76 0.72 0.72 0.7 0.81 0.78 0.62 0.69 0.61 0.76 0.73 0.83 0.72 0.73 0.75 0.71 0.8

(0.6, (0.5, (0.5, (0.5, (0.6, (0.6, (0.4, (0.4, (0.3, (0.6, (0.6, (0.7, (0.5, (0.6, (0.6, (0.5, (0.6,

0.9) 0.9) 0.9) 0.9) 1.0) 0.9) 0.9) 1.0) 0.9) 0.9) 0.9) 1.0) 0.9) 0.9) 0.9) 0.9) 1.0)

18.0 47.5 30.0 12.0 4.00

(16e20) (44e53) (27e33) (11e14) (1e6)

18.0 48.0 30.0 12.0 2.00

(16e20) (43e52) (28e34) (11e14) (1e5)b

0 2 3 0 0

(9e11) (8e6) (7e12) (8e14) (50e19)c

0.49 0.68 0.64 0.64 0.57

(0.2, (0.5, (0.5, (0.4, (0.3,

0.8) 0.9) 0.8) 0.9) 0.8)

0.72 0.8 0.57

(0.6, 0.9) (0.6, 1.0) (0.4, 0.8)

45.0 (42e49) 14.0 (12e16) 41.0 (39e43)

2 (7e4) 0 (7e14) 0 (3e5)

45.0 (42e48) 14.5 (13e16) 41.0 (39e42)

95% confidence intervals using standard errors clustered on family. N ¼ 60 for alcohol in round 2. N ¼ 48 as % change does not include participants with zero grams of alcohol in Round 1.

3.4. Mean differences in WFR and FFQ and factor effects Linear regression models with a random effect for adults related in a family unit were used to estimate the mean difference between FFQ and WFR. There was no significant mean difference for cholesterol, thiamin, riboflavin, sodium, %E fats, %E saturated fats, % fat from mono-unsaturated fats. For the remainder the majority had positive significant differences between mean FFQ and WFR, with FFQ amounts larger of the two. Those with mean FFQ amounts less than WFR were polyunsaturated fats, %fat from polyunsaturated fats, %E protein, niacin equivalent and folate. Similar models were used to estimate the effect of demographic and anthropometric variables, age, sex, university education, general health, BMI and waist, on the mean difference between FFQ and WFR. These models found very few significant effects of these variables. Most significant was that mean sugars were higher on the FFQ, and that the difference for males was significantly more than that for females (p-value < 0.0001): males 64 g/day (95% CI 48e80) compared to females 27 g/day (95% CI 165e39). In other words both sexes report more sugars on their FFQ compared to their WFR, but for men this difference between FFQ sugars and WFR sugars is

much greater. Next most significant was carbohydrate, again higher on the FFQ but the difference for males was significantly greater than for females (p-value ¼ 0.001): males 75 g/day (95% CI 49e101) compared to females 33 g/day (95% CI 143e53). For %E from carbohydrate the result was similar: (p-value ¼ 0.004): males 5.9% (95% CI 4.1e7.8%) compared to females 3.1% (95% CI 1.8e4.4%). Both genders had a lower %E from protein on the FFQ and that difference was greatest for men: (p-value ¼ 0.003): males 2.3% (95% CI 3.5 to 1.1%) compared to females 0.5% (95% CI 1.2 to 0.4%). No other significant effects of demographic or anthropometric on the difference between FFQ and WFR were found. 4. Discussion The reproducibility and comparative validity of the AES FFQ were assessed in the current study using intra-class correlation coefficients (ICC) estimated by comparing nutrient data generated from three-day WFRs with the AES FFQ data over two administration rounds. The analysis methods utilized all collected data and accommodates for probable correlation of observations within the same family.

Please cite this article in press as: Collins CE, et al., Reproducibility and comparative validity of a food frequency questionnaire for Australian adults, Clinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.09.015

6

C.E. Collins et al. / Clinical Nutrition xxx (2013) 1e9

Table 4 Descriptive statistics for differences between Food Frequency Questionnaire and Weighed Food Records (FFQ-WFR) by round: (1ste3rd quartile), percentage change, and Pearson correlation coefficient. The % Change is Round 2 subtracted from Round 1, as a percentage of Round 1. Nutrients/day

Energy Energy (MJ) Protein (g) Total fat (g) Saturated fat (g) Polyunsat. fat (g) Monounsat. fat (g) Cholesterol (mg) Carbohydrate (g) Sugars (g) Alcohol (g) Nutrients Fiber (g) Thiamine (mg) Riboflavin (mg) Niacin (mg) Vitamin C (mg) Folate (mg) Vitamin A (mg) Retinol (mg) Beta-carotene (mg) Sodium (mg) Potassium (mg) Magnesium (mg) Calcium (mg) Phosphorus (mg) Iron (mg) Zinc (mg) Water (L) Percent energy Protein Carbohydrate Total fats Saturated fat Alcohol Percent fat Saturated Polyunsaturated Monounsaturated a b c

Round 1 N ¼ 91

Round 2 N ¼ 65

% Change N ¼ 64

Correlation

Median (Q1eQ3)

Median (Q1eQ3)

Median (Q1eQ3)

r

95%CIa

1.57 (1.0e3) 10.3 (12e34) 8.63 (12.8e26) 4.18 (5.9e13) 0.43 (3.9e1) 3.41 (4.0e10) 39.2 (77e119) 50.4 (17e104) 37.8 (3e80) 2.34 (0.0e12)

1.27 (0.1e3) 10.5 (1e31) 6.63 (5.5e24) 4.49 (2.0e13) 0.79 (4.1e2) 1.91 (1.7e) 53.8 (65e108) 38.0 (10e89) 35.3 (3e54) 2.26 (0.0e9)

55 44 70 47 49 51 49 36 11 56

(104e14) (101e63) (168e3) (97e75) (172e28) (113e24) (133e97) (105e34) (55e29) (113e0)b

0.6 0.39 0.45 0.45 0.32 0.46 0.1 0.65 0.73 0.78

0.4, 0.3, 0.2, 0.2, 0.1, 0.2, 0.0, 0.4, 0.5, 0.6,

0.8 0.5 0.7 0.7 0.6 0.7 0.3 0.9 0.9 1.0

4.40 (1.2e13) 0.02 (0.5e0) 0.13 (0.5e1) 0.53 (13.1e9) 78.7 (36e124) 70.52 (153e52) 0.37 (0.0e1) 18.3 (89e133) 2.15 (0.2e3) 0.08 (0.7e1) 0.67 (0.5e1) 58.3 (32e142) 0.29 (0.0e1) 0.17 (0.2e1) 2.50 (0.9e5) 2.43 (1.5e6) 0.95 (0.2e2)

6.29 (1.8e12) 0.06 (0.5e0) 0.16 (0.7e1) 1.91 (11.1e5) 83.5 (31e124) 32.98 (189e68) 0.33 (0.0e1) 86.2 (53e176) 1.56 (0.3e3) 0.10 (0.7e1) 0.86 (0.1e1) 66.5 (14e158) 0.35 (0.0e1) 0.17 (0.1e1) 2.42 (1.0e5) 2.43 (0.3e4) 0.92 (0.4e1)

39 (121e58) 61 (152e89) 48 (147e36) 46 (109e90) 4 (44e52) 56 (145e61) 20 (106e38) 25 (104e42) 32 (119e14) 67 (152e32) 24 (116e35) 24 (119e62) 26 (123e36) 61 (118e39) 49 (142e16) 60 (136e4) 7 (61e52)

0.51 0.5 0.58 0.29 0.63 0.65 0.4 0.46 0.22 0.37 0.66 0.61 0.57 0.51 0.39 0.41 0.58

0.3, 0.2, 0.3, 0.0, 0.4, 0.2, 0.1, 0.2, 0.0, 0.2, 0.5, 0.4, 0.4, 0.3, 0.1, 0.2, 0.4,

0.7 0.8 0.9 0.6 0.8 1.0 0.7 0.7 0.5 0.6 0.8 0.8 0.8 0.7 0.7 0.6 0.8

0.90 (3.1e2) 3.23 (0.1e7) 0.28 (3.8e4) 0.47 (1.5e3) 1.00 (0.0e3)

0.63 (2.8e1) 3.33 (1.2e8) 0.24 (2.2e5) 0.71 (1.2e3) 0.51 (0.0e2)

51 50 57 63 57

0.3 0.3 0.3 0.41 0.22

0.1, 0.0, 0.0, 0.2, 0.0,

0.5 0.5 0.6 0.7 0.5

3.58 (2.5e8) 2.50 (6.3e0) 1.01 (2.4e5)

3.16 (2.9e6) 2.17 (5.9e0) 0.40 (1.8e3)

47 (121e16) 35 (95e66) 77 (130e6)

0.46 0.38 0.46

0.2, 0.7 0.0, 0.7 0.2, 0.7

(123e73) (167e40) (141e29) (172e10) (104e31)c

95% confidence intervals using standard errors clustered on family. N ¼ 61 for alcohol in round 2. N ¼ 55 as % change does not include participants with zero difference in grams of alcohol in Round 1.

The reproducibility of the AES FFQ was confirmed as shown by the ICCs for each nutrient assessed as being at least as large as that for the three-day WFR. Indeed, for total energy, total fats, polyunsaturated fats, cholesterol, carbohydrate, sodium, water, niacin equivalent, beta-carotene, vitamin C, alcohol and %E alcohol, the ICC for the FFQ was significantly greater than that for the WFR. The comparative validity of the AES FFQ with WFR was confirmed by the ICCs for the differences between the AES FFQ being significantly different from zero for all nutrients except cholesterol. These results confirm that the AEQ FFQ can be used to reliably estimate nutrient intakes of adults residing in Australia. Food frequency questionnaires have been reported to overestimate nutrient intakes when compared to 24-h recalls and WFRs.16 This was partially corroborated in the current study, with the AES FFQ providing higher estimates of intakes for 23 of the 35 nutrients examined compared to WFRs. The correlation coefficients from this study are comparable to those of Subar et al.17 with their values ranging from 0.41 to 0.83 for the Diet History Questionnaire, 0.19e0.80 for the Block FFQ and 0.28e0.83 for the Willett FFQ using measurement error models adjusted for energy with 24 h recall as the reference

method. Kesse-Guyot et al. have demonstrated acceptable validity and reproducibility of an FFQ for assessment of energy, nutrient and food group intakes in an adult French population with crude Pearson correlation coefficients for relative validity ranging from 0.28 to 0.67 in men and from 0.25 to 0.55 in women and median Pearson correlation coefficients for reproducibility of 0.70 and 0.65 for men and women respectively.18 While the current study found poor agreement for cholesterol, a review of the use of FFQs highlighted that others have also shown this.19 Results in the current study are similar to those in the ACAES comparative validation study,11 which was anticipated given that AES was modified from ACAES and both studies had a similar design. When the comparative validity of the AES FFQ is compared to those from the ACAES study the correlation for carbohydrate is equivalent (r ¼ 0.47), while the correlations for fat, saturated fat, folate, iron and riboflavin are marginally lower. The correlations for beta-carotene, magnesium and zinc were higher for the AES FFQ compared to ACAES, while correlations for all other nutrients were either lower (energy, protein, monounsaturated fat, fiber, thiamine, vitamin C, vitamin A, calcium), not significant (polyunsaturated fat,

Please cite this article in press as: Collins CE, et al., Reproducibility and comparative validity of a food frequency questionnaire for Australian adults, Clinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.09.015

C.E. Collins et al. / Clinical Nutrition xxx (2013) 1e9 Table 5 Reproducibility and Comparative Validity for AES Food Frequency Questionnaire and Weighed Food Records (FFQ-WFR): Intra-class correlation coefficient (ICC). For WFRs, there were 157 observations on 92 subjects in 67 families. For FFQs, there were 163 observations on 97 subjects in 68 families. For the difference, there were 156 observations on 92 subjects in 67 families.

Energy Energy (MJ) Protein (g) Total fat (g) Saturated fat (g) Polyunsat. fat (g) Monounsat. fat (g) Cholesterol (mg) Carbohydrate (g) Sugars (g) Alcohol (g) Nutrients Fiber (g) Thiamine (mg) Riboflavin (mg) Niacin (mg) Vitamin C (mg) Folate (mg) Vitamin A (mg) Retinol (mg) Beta-carotene (mg) Sodium (mg) Potassium (mg) Magnesium (mg) Calcium (mg) Phosphorus (mg) Iron (mg) Zinc (mg) Water (L) Percent energy Protein Carbohydrate Total fats Saturated fat Alcohol Percent fat Saturated Polyunsaturated Monounsaturated

WFR Reproducibility

FFQ Reproducibility

Difference (FFQ-WFR) Comparative validity

ICC

95%CI

ICC

95%CI

ICC

95%CI

0.63 0.51 0.46 0.55 0.46 0.49 0.36 0.7 0.73 0.64

(0.54, (0.39, (0.33, (0.45, (0.30, (0.33, (0.20, (0.64, (0.66, (0.53,

0.72) 0.63) 0.59) 0.65) 0.63) 0.65) 0.53) 0.77) 0.81) 0.74)

0.85 0.69 0.69 0.66 0.69 0.72 0.7 0.85 0.83 0.87

(0.80, (0.62, (0.61, (0.56, (0.61, (0.64, (0.61, (0.81, (0.78, (0.82,

0.89) 0.77) 0.78) 0.76) 0.77) 0.79) 0.79) 0.89) 0.88) 0.91)

0.6 0.4 0.4 0.39 0.66 0.41 0.14 0.65 0.74 0.26

(0.50, (0.28, (0.25, (0.22, (0.57, (0.25, (0.00, (0.55, (0.67, (0.07,

0.70) 0.51) 0.54) 0.56) 0.74) 0.57) 0.32) 0.75) 0.81) 0.45)

0.66 0.6 0.62 0.39 0.61 0.6 0.44 0.51 0.42 0.52 0.62 0.62 0.64 0.46 0.65 0.55 0.52

(0.57, (0.47, (0.54, (0.19, (0.48, (0.47, (0.27, (0.37, (0.23, (0.37, (0.52, (0.53, (0.52, (0.33, (0.57, (0.43, (0.37,

0.75) 0.72) 0.70) 0.60) 0.74) 0.74) 0.60) 0.64) 0.61) 0.68) 0.73) 0.71) 0.76) 0.60) 0.74) 0.67) 0.66)

0.8 0.76 0.74 0.74 0.88 0.81 0.69 0.68 0.74 0.8 0.78 0.85 0.71 0.75 0.78 0.75 0.87

(0.73, (0.70, (0.67, (0.68, (0.83, (0.76, (0.55, (0.59, (0.64, (0.74, (0.73, (0.81, (0.62, (0.70, (0.72, (0.69, (0.83,

0.87) 0.82) 0.80) 0.80) 0.92) 0.86) 0.82) 0.76) 0.83) 0.85) 0.83) 0.89) 0.80) 0.81) 0.85) 0.81) 0.91)

0.49 0.34 0.49 0.73 0.71 0.53 0.67 0.59 0.43 0.37 0.67 0.63 0.53 0.5 0.46 0.46 0.68

(0.36, (0.16, (0.34, (0.67, (0.64, (0.33, (0.53, (0.47, (0.25, (0.20, (0.59, (0.54, (0.41, (0.37, (0.33, (0.32, (0.57,

0.62) 0.52) 0.63) 0.80) 0.77) 0.73) 0.81) 0.71) 0.61) 0.54) 0.76) 0.72) 0.66) 0.63) 0.59) 0.59) 0.78)

0.56 0.61 0.47 0.61 0.59

(0.39, (0.49, (0.34, (0.52, (0.47,

0.72) 0.74) 0.60) 0.70) 0.70)

0.52 0.66 0.59 0.63 0.83

(0.38, (0.59, (0.49, (0.53, (0.78,

0.66) 0.74) 0.69) 0.72) 0.88)

0.53 0.26 0.29 0.35 0.32

(0.36, (0.09, (0.12, (0.20, (0.15,

0.69) 0.44) 0.45) 0.51) 0.49)

0.72 (0.63, 0.80) 0.73 (0.65, 0.80) 0.49 0.62 (0.48, 0.77) 0.77 (0.68, 0.86) 0.49 0.63 (0.54, 0.72) 0.85 (0.80, 0.89) 0.6

7

size.25,26 Sex-specific portion sizes have been found to better distinguish sex differences27 and were used in the AES FFQ. Adding open-ended questions regarding portion size can reduce FFQ validity due to sources of error in conceptualising serve sizes and also large within-person variations in serving sizes when the same food is consumed on different occasions.7 However, the portion sizes were derived from nationally representative adult data from the most recent National Nutrition Survey of adults conducted in Australia. A general limitation of validation studies is that the results are not necessarily transferable to other populations due to regional variations such as local foods.28 A sample size of at least 50 is desirable for each demographic group29 and ideally between 10 and 20 participants. Although the sample size in the present study was adequate at the group level it was inadequate to confirm validity and reproducibility for sub-sets for subpopulations, such as smokers and those of varying education and ethnicity. Further, the parents included may not be representative of adults in this age range generally. The day of the week may also influence the results of validation studies. In the present study, the proportion of records was collected on weekdays (75%) and weekends (25%). It is possible that this is due to a perceived usual intake being during the week and

(0.37, 0.61) (0.35, 0.63) (0.50, 0.70)

niacin and retinol) or not common to both studies (sugars, niacin equivalent). Results of prior FFQ validation studies that have examined sex differences have been mixed. Sex differences have been observed in some studies, but these differences have been inconsistent.20e23 Unlike our study which indicates greater differences between the FFQ and WFR for males, one study observed higher correlations in males compared to females.24 For a dietary instrument to be appropriate for use in detecting associations between diet and disease, it is suggested that correlations between it and WFRs need to in the range of at least 0.3 or 0.4. The present study had intra-class correlations greater than 0.3 between FFQ and WFR for all nutrients other than cholesterol. Although the gold standard method to assess validity of total energy intake would have been the use of doubly labelled water, the cost was prohibitive. The reference dietary method of choice for FFQ validation studies is reported to be weighed food records.19 Although 24-h recalls have less respondent burden, their sources of error tend to be more correlated with the error in an FFQ due to reliance upon memory, conceptualisation of portion sizes.19 Standard portion sizes were applied to the AES FFQ because frequency has been found to be more discriminatory than portion

Fig. 1. Reproducibility of Australian Eating Survey (AES) Food Frequency Questionnaires (FFQ) (97 subjects) and Weighed Food Records (WFR): Intra-class correlations (ICC), with 95% confidence intervals (92 subjects).

Please cite this article in press as: Collins CE, et al., Reproducibility and comparative validity of a food frequency questionnaire for Australian adults, Clinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.09.015

8

C.E. Collins et al. / Clinical Nutrition xxx (2013) 1e9

Statement of authorship CEC, JW, TB and MG designed the study. CEC and TB oversaw the study. KD, KP, CEC and TB, collected the data. JW, CEC and MB drafted the manuscript. MMB and MG undertook the statistical analysis. All authors contributed to critically reviewing, interpreting the results and approved the final manuscript. CEC had primary responsibility for the final content. Funding sources This research project was funded by Meat and Livestock Australia Human Nutrition Research Program grant. CE Collins is supported by a National Health and Medical Research Council, Career Development Fellowship. Conflict of interest statement No authors declare a conflict of interest. CE Collins was a member of the Meat and Livestock Australia Dietary Petterns Advisory group 2008–2009. A scannable version of the AES FFQ available for research applications from the University of Newcastle, Australia. Acknowledgements This research project was funded by a competitive grant from Meat and Livestock Australia Human Nutrition Research Program (G1000577). The views expressed in this manuscript are those of the authors. CEC is funded by a National Health and Medical Research Council Career Development Fellowship. The authors acknowledge the families who participated in the study as well as the student research assistants for data collection and data entry. References

Fig. 2. Comparative Validity of Australian Eating Survey (AES) Food Frequency Questionnaires (FFQ) and Weighed Food Records (WFR): Intra-class correlations (ICC), with 95% confidence intervals (N ¼ 92 subjects).

greater variation on weekends. Participants reported a usual intake for 64% of the records collected. Of the remaining 36% of records, half were reported as less than usual with the other half reported as more than usual. This large proportion of more or less than usual intakes may be typical of the intakes of this age group, but is likely to contribute to reduced agreement with the FFQ, where they record their perceived usual intake. Agreement may also be affected by the reporting of items in WFRs that were not on the FFQ (e.g. oysters, popcorn, slurpee, added chocolate topping, and sherbet). This study evaluated the reproducibility and comparative validity of a self-administered FFQ developed for use in Australian adults. The AES FFQ demonstrates a high intra-class correlation between nutrient intakes with repeated administrations and in comparison with weighed food records. The AES has results that are comparable to other FFQs for validity and reliability across a wide range of the nutrients evaluated. It provides an important contribution to the tools available for assessing usual intakes in Australian adults and is a useful tool for estimating the dietary intakes of total fat, saturated fat, carbohydrate, sugars, fiber, vitamin C, folate, beta-carotene, calcium and iron in clinical practice, epidemiologic research and public health interventions.

1. WHO. Expert consultation on diet, nutrition and the prevention of chronic diseases. Geneva: WHO; 2003. 2. Finucane MM, Stevens GA, Cowan MJ, Danaei G, Lin JK, Paciorek CJ, et al. National, regional, and global trends in body-mass index since 1980: systematic analysis of health examination surveys and epidemiological studies with 960 country-years and 9$1 million participants. The Lancet 2011;377(9765):557e67. 3. Grandjean AC. Dietary intake data collection: challenges and limitations. Nutr Rev 2012;70(s2):S101e4. 4. Buzzard M. 24-hour dietary recall and food record methods. In: Willett W, editor. Nutritional epidemiology. 2nd ed. Oxford: Oxford University Press; 1998. 5. Willett W. Food-frequency methods. In: Willett W, editor. Nutritional epidemiology. 2nd ed. Oxford: Oxford University Press; 1998. 6. Willett W, Lenart E. Reproducibility and validity of food-frequency questionnaires. In: Willett W, editor. Nutritional epidemiology. 2nd ed. Oxford: Oxford University Press; 1998. 7. Willett WC. Invited commentary: comparison of food frequency questionnaires. Am J Epidemiol 1998;148(12):1157e65. 8. Ambrosini GL. Comparison of an Australian food-frequency questionnaire with diet records: implications for nutrition surveillance. Public Health Nutr 2003;64(4):415e22. 9. Ireland P, Jolley D, Giles G, O’Dea K, Powels J, Rutishausee I, et al. Development of the Melbourne FFQ: a food frequency questionnaire for use in an Australian prospective study involving an ethnically diverse cohort. Asia Pac J Clin Nutr 1994;3:19e31. 10. Ibiebele TI, Parekh S, Mallitt KA, Hughes MC, O’Rourke PK, Webb PM. Reproducibility of food and nutrient intake estimates using a semi-quantitative FFQ in Australian adults. Public Health Nutr 2009 Dec;12(12):2359e65. 11. Watson JF, Collins CE, Sibbritt DW, Dibley MJ, Garg ML. Reproducibility and comparative validity of a food frequency questionnaire for Australian children and adolescents. Int J Behav Nutr Phys Act 2009;6:62. 12. Burrows TL, Warren JM, Colyvas K, Garg ML, Collins CE. Validation of overweight children’s fruit and vegetable intake using plasma carotenoids. Obesity 2009;17(1):162e8. 13. Burrows T, Berthon B, Garg ML, Collins CE. A comparative validation of a child food frequency questionnaire using red blood cell membrane fatty acids. Eur J Clin Nutr 2012 Feb 29;66(7):825e9. 14. Burrows TL, Truby H, Morgan PJ, Callister R, Davies PS, Collins CE. A comparison and validation of child versus parent reporting of children’s energy intake

Please cite this article in press as: Collins CE, et al., Reproducibility and comparative validity of a food frequency questionnaire for Australian adults, Clinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.09.015

C.E. Collins et al. / Clinical Nutrition xxx (2013) 1e9

15. 16.

17.

18.

19.

20.

21.

using food frequency questionnaires versus food records: who’s an accurate reporter? Clin Nutr 2013 Aug;32(4):613e8. ABS. National nutrition survey: nutrient intakes and physical measurements. Canberra: Australian Bureau of Statistics; 1998. Report No.: 4805.0. Block G, Thompson FE, Hartman AM, Larkin FA, Guire KE. Comparison of two dietary questionnaires validated against multiple dietary records collected during a 1-year period. J Am Diet Assoc 1992 Jun;92(6):686e93. Subar AF, Thompson FE, Kipnis V, Midthune D, Hurwitz P, McNutt S, et al. Comparative validation of the Block, Willett, and National cancer Institute food frequency questionnaires the eating at America’s Table Study. Am J Epidemiol 2001;154(12):1089e99. Kesse-Guyot E, Castetbon K, Touvier M, Hercberg S, Galan P. Relative validity and reproducibility of a food frequency questionnaire designed for French adults. Ann Nutr Metab 2010;57(3e4):153e62. Cade J, Thompson R, Burley V, Warm D. Development, validation and utilisation of food-frequency questionnaires e a review. Public Health Nutr 2002 Aug;5(4): 567e87. Brunner E, Stallone D, Juneja M, Bingham S, Marmot M. Dietary assessment in Whitehall II: comparison of 7 d diet diary and food-frequency questionnaire and validity against biomarkers. Br J Nutr 2001 Sep;86(3): 405e14. Marks GC, Hughes MC, van der Pols JC. The effect of personal characteristics on the validity of nutrient intake estimates using a food-frequency questionnaire. Public Health Nutr 2006;9(03):394e402.

9

22. Ocke MC, Bueno-de-Mesquita HB, Pols MA, Smit HA, van Staveren WA, Kromhout D. The Dutch EPIC food frequency questionnaire. II. Relative validity and reproducibility for nutrients. Int J Epidemiol 1997;26(Suppl. 1):S49e58. 23. Resnicow K, Odom E, Wang T, Dudley WN, Mitchell D, Vaughan R, et al. Validation of three food frequency questionnaires and 24-hour recalls with serum carotenoid levels in a sample of AfricaneAmerican adults. Am J Epidemiol 2000 Dec 1;152(11):1072e80. 24. Bohlscheid-Thomas S, Hoting I, Boeing H, Wahrendorf J. Reproducibility and relative validity of energy and macronutrient intake of a food frequency questionnaire developed for the German part of the EPIC project. European Prospective Investigation into Cancer and Nutrition. Int J Epidemiol 1997;26(Suppl. 1):S71e81. 25. Hunter DJ, Sampson L, Stampfer MJ, Colditz GA, Rosner B, Willett WC. Variability in portion sizes of commonly consumed foods among a population of women in the United States. Am J Epidemiol 1988 Jun;127(6):1240e9. 26. Flegal KM. Evaluating epidemiologic evidence of the effects of food and nutrient exposures. Am J Clin Nutr 1999 Jun;69(6):1339Se44S. 27. Molag ML, de Vries JH, Ocké MC, Dagnelie PC, van den Brandt PA, Jansen MC, et al. Design characteristics of food frequency questionnaires in relation to their validity. Am J Epidemiol 2007;166(12):1468e78. 28. Plummer M, Kaaks R. Commentary: an OPEN assessment of dietary measurement errors. Int J Epidemiol 2003 Dec;32(6):1062e3. 29. Cade JE, Burley VJ, Warm DL, Thompson RL, Margetts BM. Food-frequency questionnaires: a review of their design, validation and utilisation. Nutr Res Rev 2004;17:5e22.

Please cite this article in press as: Collins CE, et al., Reproducibility and comparative validity of a food frequency questionnaire for Australian adults, Clinical Nutrition (2013), http://dx.doi.org/10.1016/j.clnu.2013.09.015

Reproducibility and comparative validity of a food frequency questionnaire for Australian adults.

Food frequency questionnaires (FFQ) are used in epidemiological studies to investigate the relationship between diet and disease. There is a need for ...
936KB Sizes 0 Downloads 0 Views