Other topics

Are dietary patterns in childhood associated with IQ at 8 years of age? A population-based cohort study
s Kate Northstone,1 Carol Joinson,1 Pauline Emmett,1 Andy Ness,2 Toma Paus3,4,5
< Additional tables are

published online only. To view these les please visit the journal online (http://jech.bmj. com/content/66/7.toc).
1

School of Social and Community Medicine, University of Bristol, Bristol, UK 2 Department of Oral and Dental Science, University of Bristol, Bristol, UK 3 Rotman Research Institute, University of Toronto, Toronto, Canada 4 Brain & Body Centre, University of Nottingham, Nottingham, UK 5 Montreal Neurological Institute, McGill University, Montreal, Canada Correspondence to Dr Kate Northstone, Department of Social Medicine, University of Bristol, Oakeld House, Oakeld Grove, Clifton, Bristol BS8 1TQ, UK: kate.northstone@bristol.ac.uk Accepted 21 December 2010 Published Online First 7 February 2011

ABSTRACT Background Little is known about the effects of overall diet in childhood and intelligence later in life. Methods The current study, based on the Avon Longitudinal Study of Parents and Children, uses data on childrens diet reported by parents in food-frequency questionnaires at 3, 4, 7 and 8.5 years of age. Dietary patterns were identied using principal-components analysis and scores computed at each age. IQ was assessed using the Wechsler Intelligence Scale for Children at 8.5 years. Data on a number of confounders were collected, and complete data were available for 3966 children. Results After adjustment, the processed (high fat and sugar content) pattern of diet at 3 years of age was negatively associated with IQ assessed at 8.5 years of ageda 1 SD increase in dietary pattern score was associated with a 1.67 point decrease in IQ (95% CI 2.34 to 1.00; p<0.0001). The health-conscious (salad, rice, pasta, sh, fruit) pattern at 8.5 years was positively associated with IQ: a 1 SD increase in pattern score led to a 1.20 point increase in IQ (95% CI 0.52 to 1.88; p0.001). Conclusion There is evidence that a poor diet associated with high fat, sugar and processed food content in early childhood may be associated with small reductions in IQ in later childhood, while a healthy diet, associated with high intakes of nutrient rich foods described at about the time of IQ assessment may be associated with small increases in IQ.

INTRODUCTION
Studies investigating the long-term effects of nutrition on intelligence are sparse and conicting. In particular, with the exception of studies of infant feeding, there appears to be little known about the effects of the diet in early childhood on general intelligence later in life. Several studies have examined the association between breastfeeding and later child intelligence. A meta-analysis of 20 such studies found that breastfeeding was associated with higher cognitive development with an increase of 3.16 points after adjustment for covariates (95% CI 2.35 to 3.98).1 In addition, several studies have examined the effects of vitamin supplementation on IQ in children with mixed results (see Bellisle2 for a review). It has been suggested that micronutrient supplementation is benecial only to children with poor nutritional status.3 Only one study could be found that examined the effects of general dietary intake on total IQ in childhood: Nelson et al did not demonstrate any association between current diet and total IQ during childhood in a UK study of
624

227 children aged 7e12 years.4 Another study investigated the association between dietary intake and two subtests of the Wechsler Intelligence Scale for Children (WISC-R)5: Zhang et al, in a study of 3666 children (6e16 years old) in the USA, showed that lower cholesterol intake and higher polyunsaturated fatty acid intake were associated with higher scores on the digit span, a subtest assessing short-term and working memory.5 More recently, Theodore et al examined the intake of specic food groups at 3.5 and 7 years of age.6 They reported positive associations between the consumption of margarine, breads/cereals and sh and higher IQ. This study was restricted, however, to children who were born small for gestational age. Additionally, it was cross-sectional in design and did not appear to test for any independent effects of individual food groups, thus ignoring the intercorrelations between groups. Assessing dietary patterns as opposed to individual foods or nutrients allows to us take into account these intercorrelations, which may otherwise be overlooked, We do not eat foods in isolation, rather consuming combinations of foods in meals and snacks. Principal-components analysis (PCA), the most popular data-driven method of obtaining dietary patterns, uses the correlations between a large number of variables to identifying underlying dimensions in the data. These dimensions can then be used as exposure variables and are complementary to the traditional methods of examining dietehealth relationships. The current study, based on the Avon Longitudinal Study of Parents and Children (ALSPAC), aims to examine the association between dietary patterns obtained using PCA through early and mid-childhood (3 to 8.5 years) and IQ assessed at 8.5 years of age. We hypothesised that a more healthy diet would be associated with increased IQ scores.

METHODS Sample
ALSPAC is an ongoing longitudinal cohort study that was designed to investigate the determinants of development, health, and disease during childhood and beyond.7 Pregnant women with an expected date of delivery between 1 April 1991 and 31 December 1992, resident in the former Avon health authority area in Southwest England, were eligible to participate in the study. A cohort of 14 541 pregnant women was established, resulting in 13 988 children who were alive at 12 months of age. Ethical approval for the study was obtained from the ALSPAC Law and Ethics committee and the three local research-ethics committees.

J Epidemiol Community Health 2012;66:624e628. doi:10.1136/jech.2010.111955

Other topics
Measures
Primary data collection was via self-completion questionnaires answered by the primary care givers. From the age of 7 years, all children were invited to attend an annual research clinic during which various tests, both physical and psychological, were completed. Composition of Foods,11 based on standard portion sizes appropriate to each age.

Potential confounders/mediating factors

A wide variety of factors were considered as potential confounders or mediating factors in the relationship between diet and IQ. The following variables were taken into account: gender; age at WISC assessment; the WISC administrator; the number of stressful life-events experienced by the child; breastfeeding duration (ascertained at 6 months of age), estimated energy intake at each time point, a measure of parenting (HOME score) assessed at 18 months of age,12 maternal education, housing tenure and social class recorded during pregnancy and maternal age at birth of the study child. Finally, maternal consumption of oily sh during pregnancy was included, as this has been shown to be associated with IQ in this cohort.13

IQ
At 8 years of age, a short form of the Wechsler Intelligence Scale for Children (WISC-III),8 was used to assess IQ. Alternate items from each subtest were administered, with the exception of the coding subtest, which was administered in full.9 Raw scores were calculated by summing the individual items within each subtest and multiplying by 2 for picture completion, information, arithmetic, vocabulary, comprehension and picture arrangement; multiplying by ve thirds for similarities; and multiplying by 3 halves for object assembly and block design. This made the raw scores comparable with those that would have been obtained had the full test been administered (the raw score for the coding subtest was calculated in the standard way as the full subtest was administered). Using look-up tables provided in the WISC-III Manual, age-scaled scores were obtained from the raw scores for each subtest, and total, performance and verbal IQ scores were calculated. A total of 7044 children had complete IQ scores available. The mean age of assessment was 8.5 years (SD 0.3 years).

Statistical methods
PCA with varimax rotation14 15 was performed on the standardised food items. The methods have been described in detail elsewhere.16e18 Briey, the number of components best representing the data at each time point was primarily chosen on the basis of the scree plot19 and the interpretability of the components. Children were excluded from each PCA if they had more than 10 dietary items missing from the respective questionnaire. If 10 or fewer items were missing, we made the assumption that the child did not consume those items, and they were given a value of 0. The number of food items/groups included in the PCA respectively was 34, 35, 41 and 41 at 3, 4, 7 and 8.5 years of age. Foods with loadings above 0.3 on a component were considered to have a strong association with that component and were deemed to be the most informative in describing the dietary patterns. Labels were given to each component at each time point. While these do not perfectly describe each underlying pattern, they aid in the reporting and discussion of the results. For each child, a score was created for each component identied at each time point by multiplying the factor loadings by the corresponding standardised value for each food and summing across the food items. All component scores were approximately normally distributed and had a mean of 0 and an SD of 1. The associations between each dietary pattern at each time point and IQ score were assessed using general linear models. Two models were used for each comparison. The rst included adjustment for gender, age at WISC assessment, WISC administrator and the other dietary pattern scores from that particular time point. The second model made further adjustment for all the potential confounding/mediating factors described above, together with all other dietary pattern scores from all time points. We chose to adjust for other dietary pattern scores at the same point in time, as, although the pattern scores are uncorrelated at a population level, it is possible that an individual could score high or low on more than one pattern. Adjusted parameter estimates and 95% CIs are presented. In each fully adjusted model, complete data were available for 3966 cases. Secondary analyses examined the associations between dietary patterns and performance IQ and verbal IQ separately. All analyses were performed using SPSS for Windows v.15.0 and STATA v.10.0.

Diet
Dietary information was collected via food-frequency questionnaire (FFQ) at the ages of 3, 4, 7 and 8.5 years. All four questionnaires contained a set of questions enquiring about the frequency of consumption of a wide range of foods and drinks. The mother or main carer was given the following options to indicate how often her child was currently consuming a variety of food types nowadays: (1) never or rarely; (2) once in 2 weeks; (3) 1e3 times a week; (4) 4e7 times a week; (5) more than once a day. Mothers were also asked to record the number of cups of tea or coffee, the number of glasses of cola and the number of slices of bread consumed daily. The usual type of bread (white or other) and milk (full-fat or other) usually consumed was also recorded. The data on frequency of consumption were numerically transformed into times consumed per week, in order to apply quantitative meaning to the frequency categories, as follows: (1) 0, (2) 0.5, (3) 2, (4) 5.5 and (5) 10 times per week. All data were standardised by subtracting the mean and dividing by the SD for each variable; this was necessary because tea, coffee, cola and bread were measured on a different scale to the other variables. The FFQs were adapted from one used to assess maternal diet at 32 weeks of pregnancy.10 The questionnaires were modied slightly over time in the light of analyses of dietary data as they became available and due to changes in the availability of various foods. For example, separate categories were created from the age of 4 years for coated poultry and sh products, vegetarian pies and tuna, which had previously been encompassed in other categories. Therefore, as time progressed, additional foods/drinks were included in the questionnaires (at 3 years, there were no separate questions enquiring about ice cream, milk-based puddings/deserts, cold meats, meat and vegetarian pies, tinned pasta and nuts, and at 4 years a question specically about cold meats was not included). Energy intake was estimated at each time point from the FFQs using the 5th edition of McCance and Widdowsons The
J Epidemiol Community Health 2012;66:624e628. doi:10.1136/jech.2010.111955

RESULTS
Children with IQ data available (n7044), as compared with the rest of the ALSPAC cohort (n6944), were more likely to be
625

Other topics
girls, to have been breastfed, to have mothers with higher levels of education, to be of a higher social class and to be older (all p<0.0001, table 1). These children were also more likely to live in owner-occupied housing, to have experienced fewer stressful life-events and to have mothers who consumed oily sh during pregnancy (all p<0.001). But note that children with IQ data had a lower mean birth weight, as compared with the rest of the cohort (p<0.0001). The mean (SD) IQ score for the sample was 104 (16). Three consistent dietary patterns were obtained via PCA at each time point: the processed, traditional and healthconscious patterns were stable over time.16e18 The processed pattern was described by foods with high fat and sugar content and by processed and convenience foods, while the traditional pattern was associated with consumption of meat, poultry, potato and vegetable. Salads, fruit, vegetables, sh, pasta and rice were the foods loading highly on the health-conscious Table 1 Differences in characteristics of children with and without IQ data at 8.5 years of age
Without IQ data (n[6944) Gender Boy (51.7%) 3708 Girl (48.3%) 3234 c217.14 (p<0.0001) Breast-feeding duration Never (24.8%) 1569 <6 months (44.9%) 2030 6 months+ (30.3%) 1047 c2419.8 (p<0.0001) Maternal education <O level (30.0%) 2258 O level (34.6%) 1933 >O level (35.4%) 1441 c2626.3 (p<0.0001) Maternal social class I/II (37.3%) 1278 III NM (42.8%) 1840 III M (7.8%) 399 IV/V (12.1%) 663 c2204.0 (p<0.0001) Maternal age at birth of study child <20 (4.6%) 526 20e29 (57.9%) 4387 30+ (37.5%) 2031 c2570.9 (p<0.0001) Housing tenure Owner/occupied (76.0%) 3862 Council/housing association (16.5%) 1470 Rented/other (7.5%) 581 c2722.5 (p<0.0001) No of life events 0 (30.6%) 718 1 (30.5%) 721 2 (20.1%) 487 3+ (18.8%) 527 c216.8 (p0.001) Maternal oily-sh consumption in pregnancy Never (42.2%) 2612 Less than once a week (33.1%) 1683 Once a week or more (24.7%) 1190 c2123.4 (p<0.0001) Mean (SD) birth weight (g) 3366.0 t5.19 (p<0.0001) (53.4%) (46.6%) With IQ data (n[7044) 3516 (49.9%) 3528 (50.1%)

pattern. An additional pattern was extracted at 3 years of age: the snack pattern that was associated with nger foods such as fruit, biscuits, bread and cakes. Table 2 presents the associations between the dietary patterns at each age and IQ at 8.5 years of age. It can be seen that, on minimal adjustment, all dietary pattern scores were associated with IQ with the exception of the traditional pattern at 4 and 7 years of age. The processed patterns were negatively associated with IQ at all ages, while the snacks pattern at 3 years and the health-conscious pattern at all ages were positively associated with IQ. After adjustment for a wide variety of factors, many associations were lost. However, relationships remained (after full adjustment) for the processed pattern at 3 years such that a 1 SD in this score resulted in an almost 2-point decrease in IQ at 8.5 years (95% CI 2.34 to 1.00; p<0.0001). The snacks pattern at 3 years was associated with a 1-point increase (95% CI 0.39 to 1.42; p<0.0001). Finally, the health-conscious pattern at 8.5 years was associated with an increased in IQ of 1.20 points per 1 SD of pattern score (95% CI 0.52 to 1.88; p0.001). Virtually identical patterns were seen with Verbal IQ score (web table 1). Similar associations were evident with Performance IQ (web table 2) after minimal adjustment. However, after full adjustment, the majority of associations between Performance IQ and dietary pattern scores were lost, and those that remained (processed and snacks patterns at 3 years and health-conscious patterns at 8.5 years) were markedly attenuated.

(33.8%) (43.7%) (22.5%)

1197 (18.4%) 2974 (45.7%) 2334 (35.9%)

DISCUSSION
In this paper, we report weak but novel associations between dietary patterns in early childhood, and current diet, with general intelligence assessed at 8.5 years of age. After adjustment, higher scores on the processed pattern at 3 years of age were associated with a lower IQ 5 years later, while higher scores on the health-conscious pattern at about the time of IQ assessment were associated with higher IQ scores. After full adjustment, there was no evidence to suggest that dietary patterns measured between these ages were associated with IQ. These results are in line with previous studies we have performed in the ALSPAC cohort: overall dietary patterns in early childhood are associated with both later child behaviour, in particular hyperactivity20 and school performance.21 This suggests that any cognitive/behavioural effects relating to eating habits early in childhood may well persist into later childhood, despite any subsequent changes (including improvements) to dietary intake. To our knowledge, only one other study has examined any longitudinal effects of dietary patterns obtained using PCA on IQ in childhood: Gale et al22 reported that dietary patterns obtained in infants at 6 and 12 months of age had some effect on cognitive development. They found that higher scores on the infant guidelines dietary pattern (associated with fruit, vegetables and home prepared foods) were associated with a higher full scale and verbal IQ at the age of 4. These results do not take into account current diet and are on a different timescale from the data presented here. There has been some interest in the short- to medium-term effects of dietary intake on cognitive ability and behaviour in children. For example, in short-term experiments, skipping breakfast has been shown to be detrimental23 24 and taking vitamin or iron supplements in malnourished groups can be benecial.3 25 Little is known, however, about the long-term effects of nutrition on cognitive abilities in a potentially wellnourished population. It is likely that children scoring highly on
J Epidemiol Community Health 2012;66:624e628. doi:10.1136/jech.2010.111955

(40.1%) (34.3%) (25.6%)

1473 (21.7%) 2370 (34.9%) 2953 (43.5%)

(30.6%) (44.0%) (9.5%) (15.9%)

2482 2469 386 551

(42.2%) (41.9%) (6.6%) (9.4%)

(7.6%) (63.2%) (29.2%)

124 (1.8%) 3788 (52.6%) 3212 (45.6%)

(65.3%) (24.9%) (9.8%)

5707 (85.5%) 612 (9.2%) 352 (5.3%)

(29.3%) (29.4%) (19.9%) (21.5%)

1875 1866 1220 1066

(31.1%) (31.0%) (20.2%) (17.7%)

(47.6%) (30.7%) (21.7%) (566.1)

2511 (37.8%) 2332 (35.1%) 1808 (27.2%) 3145.4 (554.1)

626

Other topics
Table 2 Associations between dietary scores at various ages and IQ at 8 years
Minimally adjusted* model Coefcient (95% CI) Dietary patternsd3 years Processed Traditional Health-conscious Snacks Dietary patternsd4 years Processed Traditional Health-conscious Dietary patternsd7 years Processed Traditional Health-conscious Dietary patternsd8.5 years Processed Traditional Health-conscious 4.17 0.48 2.23 1.78 (4.57 to 3.77) (0.86 to 0.10) (1.85 to 2.60) (1.40 to 2.17) p Value <0.0001 0.014 <0.0001 <0.0001 <0.0001 0.461 <0.0001 <0.0001 0.096 <0.0001 <0.0001 <0.0001 <0.0001 Fully adjustedy model Coefcient (95% CI) 1.67 0.11 0.64 0.90 (2.34 to 1.00) (0.70 to 0.48) (0.05 to 1.33) (0.39 to 1.42) p Value <0.0001 0.716 0.068 0.001 0.841 0.732 0.416 0.063 0.668 0.528 0.758 0.085 0.001

2.53 (2.94 to 2.12) 0.15 (0.55 to 0.25) 2.46 (2.07 to 2.85) 3.09 (3.51 to 2.68) 0.34 (0.75 to 0.06) 1.86 (1.46 to 2.26) 0.87 (0.47 to 1.27) 3.13 (3.54 to 2.72) 1.99 (1.60 to 2.38)

0.08 (0.81 to 0.66) 0.12 (0.57 to 0.81) 0.34 (1.15 to 0.48) 0.70 (1.43 to 0.04) 0.15 (0.54 to 0.85) 0.25 (1.04 to 0.53) 0.11 (0.61 to 0.84) 0.62 (1.32 to 0.09) 1.20 (0.52 to 1.88)

For minimally adjusted models, n varies between 5828 and 6313. For the fully adjusted model, n3966. *Adjusted for gender, age at testing and WISC administrator, dietary pattern scores at that time point. yAs above and breastfeeding duration, energy intake, maternal education, maternal social class, maternal age, housing tenure, life events, HOME score and all other dietary pattern scores.

the processed pattern have diets with poorer nutrient content and higher intakes of fat and sugar, as we have shown in the mothers and their partners26 (Northstone K, submitted). Furthermore, high scores on the health-conscious patterns in adults were associated with better nutrient proles26 (Northstone K, submitted). Such dietary habits that have been present from an early age could persist throughout childhood and even into adulthood. An interesting aspect of the research we have presented is the association evident between dietary patterns at 3 years of age and IQ at 8.5 years, which is present even after adjustment for diet at later ages. A possible explanation for this is that the brain grows at its fastest rate during the rst 3 years of life.27 Studies have shown that head growth during this time is associated with cognitive outcome,28 29 and it is possible that good nutrition during this early period may encourage optimal brain growth. It is interesting to note that the associations seen with the dietary patterns were stronger for verbal than for performance IQ. Previous studies of vitamin/mineral supplementation have shown improvements in non-verbal intelligence only.25 30 Performance IQ relates to an individuals innate intellectual ability, while verbal IQ more reects the impact of education, which in turn is affected by inuences such as parenting and environment.31 In comparison with the existing evidence, our results may therefore be somewhat surprising; as Benton explains, Non-verbal tests tap the biologically based functioning that could be expected to reect the adequacy of the diet.32 As our results suggest, however, a more long-term effect of diet on the childs ability to learn, this may be partly explained by favourable growth of the brain in early childhood as discussed above. The strengths of the current study include the large sample size and repeat measures of diet. In addition, a large number of potential confounders have been taken into account. However, given the levels of attenuation seen in the effect sizes in the adjusted analyses, we cannot exclude the possibility of residual confounding. The primary limitation is the bias in attrition and
J Epidemiol Community Health 2012;66:624e628. doi:10.1136/jech.2010.111955

the reduction in the number of children with complete data in the fully adjusted models; children included in the study and who had complete data were more socially advantaged than the remainder of the cohort for whom IQ was not measured. It could be argued, however, that we have underestimated any effect sizes as a result Inclusion of children with learning disabilities could also explain the results we have reported: neurological impairment in children is associated with an increased risk of feeding and nutritional problems.33 It is therefore possible that the observed reduction in IQ associated with poor diet might have been due to a relatively large proportion of children with learning disabilities in the sample. We repeated our analyses excluding those children with an IQ score of <70 (n135); however, the effect sizes were virtually identical. A further potential limitation is the use of a shortened form of the WISC-III. This was done primarily because of the large number of children being tested and to reduce the length of the session. It could be argued that using this shortened version could result in a loss of reliability in the derived IQ measures; however, shortened versions have been used successfully in other studies, and are felt to be a valid substitute.34 35 A further potential limitation of this work is that we could not adjust for maternal intelligence. However, studies have shown that education level, which we did adjust for, is closely associated with intelligence.36 37 The use of dietary patterns obtained using PCA has been criticised for the subjectivity required in identifying the underlying patterns;38 researchers are required to make a number of decisions through the process such as how many patterns to extract, the interpretation of the patterns based on the size of the factor loadings and the labelling of the patterns. It has also been argued that such patterns are population-specic, and therefore cannot be reproduced in other samples. However, the use of dietary pattern scores obtained from PCA is now widely used in nutritional epidemiology and, used as exposures, can provide a different perspective to that using individual nutrients or food groups. Exploiting dietary patterns could help to unravel the complex relationships between the foods we ingest and our
627

Other topics
4. Nelson M, Naismith DJ, Burley V, et al. Nutrient intakes, vitaminemineral supplementation and intelligence in British schoolchildren. Br J Nutr 1990;64:13e22. Zhang J, Hebert JR, Muldoon MF. Dietary fat intake is associated with psychosocial and cognitive functioning of school-aged children in the United States. J Nutr 2005;135:1967e73. Theodore FT, Thomson JMD, Waldie KE, et al. Dietary patterns and intelligence in early and middle childhood. Intelligence 2009;37:506e13. Golding J, Pembrey M, Jones R; ALSPAC Study Team: ALSPAC. The Avon Longitudinal Study of Parents and Children. I. Study methodology. Paediatr Perinat Epidemiol 2001;15:74e87. Wechsler D, Golombok S, Rust J, et al. Wechsler Intelligence Scale for Children. Sidcup, UK: Psychological Corp, 1992. Joinson C, Heron J, Butler R, et al. A United Kingdom population-based study of intellectual capacities in children with and without soiling, daytime wetting and bed-wetting. Pediatrics 2007;120:e308e16. Rogers IS, Emmett PM; ALSPAC Study Team. Diet during pregnancy in a population of pregnant women in South West England. Eur J Clin Nutr 1998;52:246e50. The Royal Society of Chemistry and MAFF (1991a). The Composition of Foods. 5th edn. London: HMSO, 1991. Caldwell BM, Bradley RH. Home Observation For Measurement of the Environment. Little Rock, AR: University of Arkansas at Little Rock, 1979. Hibbeln JR, Davis JM, Steer C, et al. Maternal seafood consumption in pregnancy and neurodevelopmental outcomes in childhood (ALSPAC Study): an observational cohort study. Lancet 2007;369:578e85. Gorsuch RL. Factor Analyses. Philadelphia: WB Saunders, 1974. Kline P. An Easy Guide To Factor Analysis. London: Routledge, 1994. North K, Emmett P; the ALSPAC Study Team. Multivariate analysis of diet among three-year old children and associations with socio-demographic characteristics. Eur J Clin Nutr 2000;54:73e80. Northstone K, Emmett P; the ALSPAC Study Team. Multivariate analysis of diet in children at four and seven years of age and associations with socio-demographic characteristics. Eur J Clin Nutr 2005;59:751e60. Northstone K, Emmett PM. Are dietary patterns stable throughout early and mid-childhood? A birth cohort study. Br J Nutr 2008;100:1069e76. Cattell RB. The screen test for the number of factors. Multivariate Behav Res 1966;1:245e76. Wiles NJ, Northstone K, Emmett P, et al. Junk food diet and childhood behavioural problems: Results from the ALSPAC cohort. Eur J Clin Nutr 2009;63:491e8. Feinstein L, Sabates R, Sorhaindo A, et al. Dietary patterns related to attainment in school: the importance of early eating patterns. J Epidemiol Community Health 2008;62:734e9. Gale CR, Martyn CN, Marriott LD, et al. Dietary patterns in infancy and cognitive and neuropsychological function in childhood. J Child Psychol Psychiatry 2009;50:816e23. Simeon T, Grantham-McGregor S. Effects of missing breakfast on the cognitive functions of school children with differing nutritional status. Am J Clin Nutr 1989;49:649e53. Wyon DP, Abrahamsson L, Jartelius M, et al. An experimental study of the effects of energy intake at breakfast on the test performance of 10-year-old children in school. Int J Food Sci Nutr 1997;48:5e12. Benton D. Micro-nutrient supplementation and the intelligence of children. Neurosci Biobehav Rev 2001;25:297e309. Northstone K, Emmett PM, Rogers I. Dietary patterns in pregnancy and associations with nutrient intakes. Br J Nutr 2008;99:406e15. Paus T, Collins DL, Evans AC, et al. Maturation of white matter in the human brain: A review of magnetic resonance studies. Brain Res Bull 2001;54:255e66. Gale CR, OCallaghan FJ, Godfrey KM, et al. Critical periods of brain growth and cognitive function in children. Brain 2004;127:321e9. Gale CR, OCallaghan FJ, Bredow M, et al. The inuence of head growth in fetal life, infancy and childhood on intelligence at the later ages of 4 and 8 years. Pediatrics 2006;118:1486e92. Benton D, Roberts G. Effect of vitamin and mineral supplementation on intelligence in a sample of schoolchildren. Lancet 1988;1:140e3. Kaufman A. Intelligent Testing with the WISC-III. New York: John Wiley & Sons, 1994. Benton D, Buts JP. Vitamin/mineral supplementation and intelligence. Lancet 1990;335:1158e60. Sullivan PB, Lambert B, Rose M, et al. Prevalence and severity of feeding and nutritional problems in children with neurological impairment: Oxford Feeding Study. Dev Med Child Neurol 2000;42:674e80. Minshew NJ, Turner CA, Goldstein G. The application of short forms of the Wechsler Intelligence scales in adults and children with high functioning autism. J Autism Dev Disord 2005;35:45e52. Donders J. Using a short form of the WISC-III: sinful or smart? Child Neuropsychol 2001;7:99e103. Mackintosh NJ. IQ and Human Intelligence. Oxford: Oxford University Press, 1998. Gottfredson LS, Deary IJ. Intelligence predicts health and longevity, but why? Curr Direct Psychol Sci 2004;13:1e4. Martinez ME, Marshall JR, Sechrest L. Invited commentary: Factor analysis and the search for objectivity. Am J Epidemiol 1998;148:17e19.

What is already known on this subject

< Breastfeeding and other nutritional factors are associated with

5. 6. 7. 8. 9. 10.

increased IQ in childhood. < Previous research investigating associations between nutrition and IQ in children has tended to focus on supplements or nutrient intakes. < Little is known about the long-term effects of diet during early childhood on general intelligence later in life.

What this study adds

< Dietary patterns were obtained using principal-components

11. 12. 13. 14. 15. 16. 17. 18.

analysis, which provide overall summaries of dietary intake to examine the association between diet and IQ. < We found a negative association between a processed dietary pattern at 3 years of age and IQ at 8.5 years. < We also report a positive association between a healthconscious pattern at 8 years of age and IQ.

mental (and physical) well-being. With the exception of breastfeeding, we have not examined diet prior to 3 years in this study. FFQs have been completed for the children at 6 and 15 months and 2 years of age, but we have not established any meaningful dietary patterns owing to the nature of the diet at this age. It is likely that the infant diet, beyond the type of milk consumed, plays an important part in cognitive development, and this will be explored elsewhere.

19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38.

CONCLUSION
In this population of contemporary British children, a poor diet, associated with increased intake of processed foods, fat and sugar, in early childhood may be associated with lower IQ at the age of 8.5 years. In addition, a concurrent healthy diet may be associated with higher IQ. Dietary patterns between the ages of 4 and 7 years were not predictive. Further research is required to help determine the true effects of early diet on intelligence.
Acknowledgements We are extremely grateful to all the families who took part in this study, the midwives for their help in recruiting them, and the whole ALSPAC team, which includes the interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists and nurses. Funding The UK Medical Research Council, the Wellcome Trust and the University of Bristol currently provide core support for ALSPAC. Competing interests None Ethics approval Ethics approval was provided by the ALSPAC Law and Ethics committee and the three local research-ethics committees. Provenance and peer review Not commissioned; externally peer reviewed.

REFERENCES
1. 2. 3. Anderson JW, Johnstone BM, Remley DT. Breast-feeding and cognitive development: a meta-analysis. Am J Clin Nutr 1999;70:525e35. Bellisle F. Effects of diet on behaviour and cognition in children. Br J Nutr 2004;92 (Suppl 2):S227e32. Schoenthaler SJ, Bier ID. Vitaminemineral intake and intelligence: a macrolevel analysis of randomised controlled trials. J Altern Complement Med 1999;5:125e34.

628

J Epidemiol Community Health 2012;66:624e628. doi:10.1136/jech.2010.111955