The determination of energy and nutrient intakes during infancy and childhood is important for the evaluation of the impact of diet on growth, development and long-term health. Various methods have been used to assess the dietary intakes of infants and toddlers(Reference Serdula, Alexander, Scanlon and Bowman1). FFQ are particularly useful for establishing generalised patterns of food intake but with the exception of a few recent studies(Reference Klohe, Clarke, George, Milani, Hanss-Nuss and Freeland-Graves2–Reference Blum, Wei, Rockett, Langliers, Leppert, Gardner and Colditz5) have not been widely used to assess dietary intakes of infants and/or young children.
We have previously described an FFQ for dietary assessment at 6 months of age(Reference Marriott, Robinson, Poole, Borland, Godfrey, Law and Inskip6). In the present paper, we evaluate the use of an interviewer-administered FFQ to assess the diets of infants aged 12 months, for use in a large population survey (the Southampton Women’s Survey (SWS)), that would describe dietary patterns, and would rank infants in terms of their nutrient intake. In a sample of SWS infants, we compare energy and nutrient intakes assessed by the FFQ with intakes determined from 4d weighed diaries (WD).
Subjects and methods
Subjects
The study was set within the SWS(Reference Inskip, Godfrey, Robinson, Law, Barker and Cooper7), in which non-pregnant women were recruited, then followed through their subsequent pregnancy and the offspring followed up. The SWS study population of infants born between 1998 and 2003 comprised 1973 infants. Of these, 1618 (82 %) were visited within 2 weeks before and 3 weeks after their first birthday when diet was assessed. After stratification by current breast-feeding status and infant sex, a random number system was used to select ninety-eight families to participate in a weighed intake study between September 2001 and April 2003. In total, sixty-four (65 %) families approached agreed to participate in the study; fifty (51 %) successfully completed the WD.
The FFQ
Foods were included in the FFQ(Reference Robinson, Marriott, Poole, Crozier, Borland, Lawrence, Law, Godfrey, Cooper and Inskip8) following review of data from a UK survey of children aged 18 months(Reference Gregory, Collins, Davies, Hughes and Clarke9), 24 h recalls collected from the mothers attending a baby clinic, and the food diaries of 12-month-old preterm infants(Reference Marriott, Foote, Bishop, Kimber and Morgan10). The FFQ assesses intakes of meat, fish, vegetables, fruit, cereals, breads, cheese, eggs, vegetarian foods, puddings, spreads and snack foods (fifty-eight categories), ten categories of commercial baby foods (sub-divided by brand), and ten types of non-milk drinks. The frequency of consumption and the amount consumed over the previous 28 d of each food in the seventy-eight-item FFQ are recorded. Prompt cards are used to show the foods included in each food group, to ensure standardised responses to the FFQ. Portion size is described using household measures and food models. At the end of the FFQ, an open section in the same format is included, to record frequencies of consumption and amounts of any foods that were not listed on the FFQ, if consumed once per week or more. Human milk, baby formula and other milk intakes are recorded in a separate section.
The weighed diary
A 4d WD was used to assess the relative validity of the 12-month FFQ, to be consistent with the diary lengths chosen for recent dietary intake studies(Reference Lanigan, Wells, Lawson and Lucas11–Reference Heinig, Nommsen, Peerson, Lonnerdal and Dewey13), and because calculations of inter-individual and intra-individual variation from energy intake data from a previous 7d WD study for infants aged 12 months(Reference Marriott, Foote, Bishop, Kimber and Morgan10) showed that 4 d would classify more than 75 % of infants into the correct quintile of intake at this age. All non-human milks, foods and drinks consumed by the infant on each of 4 d were recorded by weight and cooking method. Each WD was completed after the FFQ. The methods for FFQ and WD collection resembled those reported for the evaluation of the 6-month FFQ(Reference Marriott, Robinson, Poole, Borland, Godfrey, Law and Inskip6).
Dietary analysis
For the FFQ and WD, human milk intake was estimated using an algorithm based on length of suckling, derived from published intake data(14, Reference Paul, Black, Evans, Cole and Whitehead15). Nutritional composition information for infant formulas and commercial baby foods was obtained from manufacturers and/or by calculation from ingredients; and for non-baby products was taken from The Composition of Foods, 5th edition and supplements(Reference Holland, Welch, Unwin, Buss, Paul and Southgate16). Nutrient intakes from dietary supplements were excluded.
Statistical methods
Spearman’s rank correlation coefficients were used to summarise the association between the two assessments of each dietary variable. All nutrients were then adjusted for energy intake using the method of Willett(Reference Willett17) and the correlation coefficients recalculated. The Bland–Altman plots(Reference Bland and Altman18) were produced to assess the level of agreement between the two methods. As the distributions of all nutrients were skewed, logarithmic transformations were used prior to the Bland–Altman analysis; the Bland–Altman limits of agreement are expressed as symmetric percentages, which summarise the percentage differences between the FFQ and the WD intakes(Reference Cole19).
Ethical approval
Ethical approval was granted by Southampton & South West Hants Local Research Ethics Committee.
Results
The characteristics of the fifty mother–infant pairs studied were comparable with the remainder of the SWS cohort(Reference Robinson, Marriott, Poole, Crozier, Borland, Lawrence, Law, Godfrey, Cooper and Inskip8) (P > 0·05 for all comparisons; Table 1). The median (interquartile range, IQR) interval between FFQ completion and the beginning of the WD was 7 (4–10) d; 72 % of diaries were completed on four consecutive days.
Table 1 Characteristics of mother–infant pairs in weighed diary sub-sample and remainder of the cohort born by end of 2003(Reference Robinson, Marriott, Poole, Crozier, Borland, Lawrence, Law, Godfrey, Cooper and Inskip8)
IQR, interquartile range; GCSE, general certificate of secondary education; HND, higher national diploma.
Comparison of FFQ with WD intake
In general, intakes assessed by the FFQ were higher than WD-assessed intakes, but the correlation coefficients (r) demonstrated comparable rankings of intakes from the two methods. The coefficients ranged from r = 0·25 (vitamin E) to 0·66 (Na) (Table 2). The range of correlation coefficients for micronutrients tended to be wider than that for energy and macronutrients (0·36 (fat) to 0·52 (protein)). Adjustment for energy intake had little effect on the levels of correlation, range, r a = 0·24–0·75 (Table 2).
Table 2 Energy and nutrient intakes from all foods and drinks, FFQ and WD, the Spearman rank correlation coefficients (r), energy-adjusted correlation coefficients (r a) and Bland–Altman statistics
WD, weighed diary; IQR, interquartile range.
*All P < 0·01 except fat P = 0·01 and vitamin E P = 0·08.
**All P < 0·01 except fat P = 0·01, vitamin E P = 0·06, vitamin B12P = 0·079, folic acid P = 0·03.
In the comparison of absolute intakes of energy and nutrients assessed by the FFQ and the WD, all mean differences (MD) were positive, and within the range of 5 % to 60 % except for vitamin D (106 %) (Table 2). The median energy/kg body weight was 450 kJ (FFQ) and 387 kJ (WD).
Exploration of differences between FFQ and weighed diary
At 1 year, milk still comprises a substantial part of the diet for most infants; the median (IQR) energy intake derived from milks in the diet was 37 % (27–45 %) from the FFQ and 38 % (25–45 %) from the WD. Although the energy derived from milks was very similar for both methods, there were changes in the patterns of milk feeding between the periods assessed by the FFQ and the WD (Table 3). Eleven infants (22 %) changed from consuming a baby milk formula as their main milk to consuming only cow’s milk during WD collection. We defined these eleven infants as ‘major milk changers’.
Table 3 Types of milk used at the FFQ and the WD assessments
WD, weighed diary.
Note: Total > 100 % because many infants consumed more than one type of milk.
Since there are substantial differences in the nutritional composition between formula milks and cow’s milk (including much higher levels of Fe, Cu, vitamins C, D, E and niacin in formula milks), we compared nutrient intakes assessed by the FFQ and the WD from foods and drinks other than milks (referred to as ‘non-milk’ foods and drinks). Comparison of nutrients from non-milk foods and drinks assessed by the FFQ and the WD (Table 4) generally yielded higher correlation coefficients (range 0·44–0·70) than those found for the whole diet, with the exceptions of Ca and Cu (Table 4). Similarly, the range of MD was narrowed to 9–43 %, and for those six nutrients present at high levels in formulas compared with cow’s milk, the MD were lower than those for the whole diet. For the eleven infants denoted as major milk changers, the differences in composition between cow’s milk and formulas were associated with a wider distribution of points in the Bland–Altman plots for the six nutrients specified above. The Bland–Altman plots for energy and Fe are shown in Figs 1 and 2. The effect of a major change in milk feeding on intakes is illustrated in the plot for Fe, a mineral frequently cited as present in sub-optimal quantities in the diets of older infants(Reference Hopkins, Emmett, Steer, Rogers, Noble and Emond20).
Table 4 Energy and nutrient intakes from foods and drinks other than milks, FFQ and WD, the Spearman rank correlation coefficients (r) and Bland–Altman statistics
WD, weighed diary; IQR, interquartile range.
*All P < 0·001 except fat P = 0·0013, vitamin E P = 0·007 and vitamin B12P = 0·0015.
Fig. 1 The Bland–Altman plot for energy intake, according to whether there was a major milk change or not (●, no major milk change; ×, major milk change). WD, weighed diary
Fig. 2 The Bland–Altman plot for iron intake, according to whether there was a major milk change or not (●, no major milk change; ×, major milk change). WD, weighed diary
Discussion
We have compared the dietary intakes assessed by a newly developed FFQ with those assessed by WD at around 12 months of age. Although the FFQ-assessed intakes tended to be higher than those from the WD, there was reasonable to good ranking of intakes for most nutrients.
The WD was completed after the FFQ. Because of the changes in milk feeding around 12 months of age, we considered nutrient intakes from non-milk foods and drinks as well as nutrients from the whole diet. The rank correlation coefficients for nutrients from non-milk foods and drinks were generally higher than those for nutrients from the whole diet, with the majority of correlation coefficients above 0·5. Given the importance of milk as a component of the diet for most infants and the very different nutrient compositions of formulas v. cow’s milk, this may indicate that the differences in nutrient intakes between methods for those micronutrients present at high levels in formulas that we observed, were greater than they would have been if the FFQ and the WD had assessed diet over the same period. Similarly, the range of Bland–Altman MD (%) was smaller for nutrients from non-milk foods and drinks than nutrients from the whole diet, suggesting that the differences were influenced by changes in the pattern of milk consumption between the periods assessed by the FFQ and the WD.
Comparison with other studies
We are not aware of any other FFQ validation studies of energy and nutrient intakes at 1 year of age from the UK. However, our correlation coefficients were comparable with those from other FFQ relative validation studies for young children from countries which might be expected to have similar feeding practices to those in the UK(Reference Andersen, Lande, Arsky and Trygg3–Reference Blum, Wei, Rockett, Langliers, Leppert, Gardner and Colditz5, Reference Williams and Innis21). Although our FFQ measures of intake were higher than those from the WD, the MD (%) between the FFQ and the WD for energy and macronutrients were all less than 25 %, and both the FFQ and the WD produced median energy intakes (450 v. 387 kJ/kg, respectively) that were higher than the energy requirements/kg for 12-month-old infants derived from doubly labelled(Reference Butte22) studies (334 kJ). Higher FFQ-assessed intakes in infants have been noted by Andersen(Reference Andersen, Lande, Arsky and Trygg3), who reported a difference of 25 % in energy intakes between a semi-quantitative FFQ and 7d weighed record in Norwegian children at 1 year. These differences in intakes between methods may arise from difficulties in describing portion size and/or frequency in response to the FFQ(Reference Andersen, Lande, Arsky and Trygg3).
Strengths and weaknesses
We studied a stratified random sample of infants, who were representative of the SWS population, which in turn is comparable to the UK population(Reference Inskip, Godfrey, Robinson, Law, Barker and Cooper7). The quality of the information in our FFQ was enhanced by collecting data over a short recall period, recording both brand information and amount and frequency of consumption of all foods and drinks, and the use of trained personnel to administer the questionnaire.
A weakness of our study is that the WD was completed after the FFQ was administered. Although the interval between dietary assessments was short there were significant changes in patterns of milk consumption over this time that may have led to an underestimation of the level of agreement between the FFQ and the WD. In the absence of an alternative feasible method and in common with other UK infant dietary studies(14, Reference Ong, Emmett, Noble, Ness and Dunger23), both the FFQ and WD estimated human milk intake using an algorithm relating length of suckling to intake; this could have also contributed to the overestimation in intakes in our study.
Conclusion
The FFQ is appropriate for use in large epidemiological studies and is a useful instrument for the assessment of infant diet at around 1 year of age. The comparison of nutrient intakes assessed by the FFQ and the WD showed that the ranking of infants according to nutrient intake was comparable for both methods. While there were differences between methods in estimates of energy and nutrient intakes, the magnitudes of the differences were small to moderate for energy and macronutrients. The level of agreement between methods was influenced by the effect of a change in infants’ milk consumption in the period between assessments, and may be underestimated in our study.
Acknowledgements
Funding support: The SWS has received support from the UK Medical Research Council, University of Southampton, Dunhill Medical Trust, the British Heart Foundation and the Food Standards Agency.
Conflict of interest: None of the authors has any conflicting interests.
Authors’ contributions: S.E.B. devised the FFQ; H.M.I., K.M.G., C.M.L., S.M.R. and L.D.M. planned the diary study; L.D.M. carried out the fieldwork and, with S.M.R., was responsible for the nutritional analysis; H.M.I. performed the statistical analysis. L.D.M. wrote the first draft of the manuscript with contributions from all individual authors.
Acknowledgements: We are grateful to the SWS families who have given information to the survey. We thank the SWS nurses for data collection, Ken Cox, Patsy Coakley and Vanessa Cox for managing the data, and the SWS Study Group for their help.
