Pre-natal and post-natal growth trajectories and childhood cognitive ability and mental health

Transcription

1 Published by Oxford University Press on behalf of the International Epidemiological Association ß The Author 2011; all rights reserved. Advance Access publication 15 July 2011 International Journal of Epidemiology 2011;40: doi: /ije/dyr094 Pre-natal and post-natal growth trajectories and childhood cognitive ability and mental health Seungmi Yang, 1 * Kate Tilling, 2 Richard Martin, 2,3 Neil Davies, 2,3 Yoav Ben-Shlomo 2,3 and Michael S Kramer 1,4 1 Department of Paediatrics, McGill University, Montreal, Canada, 2 Department of Social Medicine, University of Bristol, Bristol, UK, 3 Department of Social Medicine, MRC Centre for Causal Analyses in Translational Epidemiology, University of Bristol, Bristol, UK and 4 Department of Epidemiology, Biostatistics and Occupational Health, McGill University, Montreal, Canada *Corresponding author. The Research Institute of McGill University Health Centre, 4060 Ste-Catherine West (Place Toulon), Montreal, Quebec H3Z 2Z3, Canada. seungmi.yang@mail.mcgill.ca Accepted 17 May 2011 Background Most studies of the associations between pre-natal or post-natal growth and cognitive ability have been based on children with pathologically slow growth measured between two time points only, rather than children with normal growth trajectories estimated from multiple measures of growth. Methods We investigated the associations of pre-natal and post-natal trajectories in both weight and length/height through the first 5 years of life with cognitive ability and mental health at 6.5 years of age among healthy children. Our study is based on children who were born healthy at 537 completed weeks with birth weight g and had up to 13 measures of weight and length/height from birth to age 5 years and cognitive ability and behaviour measured at 6.5 years. Using a linear spline random-effects model with 2 knots at 3 and 12 months, we estimated growth trajectories for each child from birth to age 5 years in weight and length/ height in four periods: gestational age-specific birth weight and length (pre-natal growth ), early infancy (0 3 months), late infancy (3 12 months) and early childhood (1 5 years). We used generalized estimating equations to estimate mean differences in IQ and mental health according to pre-natal and post-natal growth trajectory. IQ was measured using the Wechsler Abbreviated Scales of Intelligence, and mental health was assessed using the Strengths and Difficulties Questionnaire. Results A 1 standard deviation (SD) in birth weight was positively associated with cognitive ability (0.82 IQ points, 95% CI: ) after adjusting for confounders. For post-natal weight gain trajectories, a 1 SD faster weight gain was associated with an increase of 0.77 (95% CI: ) IQ points for early infancy, 0.30 (95% CI: ) points for late infancy, and 0.40 (95% CI: ) for early childhood after adjusting for confounders and for earlier growth. For length/height trajectories, the magnitudes of increase in cognitive ability were similar to each other (0.6 points) across the four periods. Pre-natal and infancy growth, but not early childhood growth, were associated with reduced externalising behaviours. 1215

2 1216 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY Conclusions Although the effect sizes are small and residual confounding cannot be excluded, our results suggest that among healthy children, faster growth from the pre-natal period through age 5 years is positively associated with cognitive ability, whereas faster growth in the pre-natal period and infancy is positively associated with mental health at early school age. Keywords Term birth, pre-natal growth, post-natal growth, growth trajectory, cognitive ability, child behaviour, child mental health Introduction The pre-natal period (particularly the third trimester of pregnancy) and the first few years of life are crucial for brain development. 1,2 During this period, the brain develops rapidly through the processes of neurogenesis, axonal and dendritic growth, synaptogenesis, cell death, synaptic pruning, myelination and gliogenesis. 3 These processes occur sequentially and depend on each other. Although cognitive ability has a heritable component, environmental insults during the early months and years of life are likely to have longlasting adverse effects on cognitive development. 2 Insults with demonstrated adverse effects include preterm birth 4 6 and failure to thrive in infancy. 7 Relatively few studies, however, have examined associations of pre- and post-natal growth with cognitive ability in healthy children born at term without pathological growth restriction Whereas most have reported a positive association of weight gain in infancy and early childhood with later cognitive ability, 8 10,12 some have found no association. 11,13 Not only is the empirical evidence inconclusive, but also most studies have estimated differences in cognitive ability associated with a change in weight between two time points only, rather than with trajectories estimated from multiple measures of growth. Finally, although cognitive ability is closely related to mental health problems and poor pre-natal growth has been associated with later behavioural problems, 14 behavioural sequelae of post-natal growth have rarely been studied. The aim of our study, therefore, was to examine the extent to which pre-natal and post-natal trajectories in both weight and length/height through the first 5 years of life are associated with cognitive ability and mental health at 6.5 years of age among children who had been born at or after term. Methods Study participants Study children were participants in the Promotion of Breastfeeding Intervention Trial (PROBIT). A full description of PROBIT has been published elsewhere. 15 In brief, PROBIT is a cluster-randomized controlled trial of a breastfeeding promotion intervention modelled on the WHO/UNICEF Baby-Friendly Hospital Initiative. A total of mothers and their healthy infants in the Republic of Belarus born at 537 completed weeks of gestation and with birth weight g were recruited from 31 maternity hospitals and affiliated polyclinics during their postpartum stay between June 1996 and December They had scheduled follow-up study visits at 1, 2, 3, 6, 9 and 12 months in the first year and at age 6.5 years. Follow-up thorough the first year was achieved in infants (97%); (81.5%) children were seen and examined at the 6.5-year follow-up. Our study is based on (86%) of the children who had cognitive ability and behaviour scores measured at the 6.5-year visit. The study received approval from the Institutional Review Board of the Montreal Children s Hospital, and participating mothers signed the consent in Russian. Measures Exposures (growth) Infant weight (grams) and length (centimetres) at birth and gestational age in completed weeks were obtained from obstetric records during the postpartum hospital stay. Weight and length during the first year of life were measured by paediatricians at the scheduled PROBIT visits at 1, 2, 3, 6, 9 and 12 months. Weights and lengths/heights between 12 months and 5 years were abstracted from the polyclinic records of routine check-ups. A total of 2095 (15%) children had no routine check-up visits, and on average 4 (range 1 6) measures of weight and height per child were abstracted from the clinic records. Outcomes At the 6.5-year follow-up, cognitive ability was measured by the Wechsler Abbreviated Scales of Intelligence (WASI). 16 The WASI consists of four subtests of vocabulary and similarities to measure verbal IQ and block designs and matrices to measure performance IQ. The WASI was translated from English to Russian and back-translated to ensure comparability of the Russian version. It was administered by the polyclinic paediatricians after extensive

3 PRE-NATAL OR POST-NATAL GROWTH AND CHILD DEVELOPMENT 1217 training and follow-up monitoring by child psychologists and psychiatrists in Minsk, Belarus. Interpaediatrician agreement was high, with Pearson correlation coefficients (95% confidence intervals) of 0.80 ( ) for vocabulary, 0.72 ( ) for similarities, 0.80 ( ) for block designs and 0.79 ( ) for matrices in a convenience sample of 45 children during a 1-week training workshop. 17 The present study used the full-scale IQ from the WASI for the primary measure of general cognitive ability of children. We also assessed the associations with verbal and performance IQ. Additionally, at the 6.5-year follow-up visit, parents provided teachers names if the child had started formal schooling by the time of his/her follow-up visit, and the teachers were asked to evaluate the child s performance in mathematics, reading, writing and other subjects according to a 5-point Likert scale as far below (1), somewhat below, at, somewhat above, or far above (5) his or her grade level, based on items in the Teacher Report Form of the Child Behaviour Checklist. 18 Child mental health was measured using the Strengths and Difficulties Questionnaire (SDQ). 19 The SDQ is a brief behavioural screening questionnaire for mental illness in children and adolescents from ages 4 to 16 years and consists of five subscales (hyperactivity, conduct problems, emotional symptoms, peer problems and prosocial behaviour), each with five items. Each item is rated as not true (0), somewhat true (1) or certainly true (2). The score for each of the five scales is generated by summing the scores for the five items, each ranging from 0 to 10. Scores for hyperactivity, conduct problems, emotional symptoms and peer problems are summed to generate a total difficulties score. A recent study 20 has shown that although the four difficulties subscales tap distinct dimensions of behavioural problems, summarizing them into externalising behaviour (sum of hyperactivity and conduct problem scores) and internalizing behaviour (sum of emotional symptoms and peer problems) is more parsimonious and better able to reduce measurement errors in a general population. Thus, our behaviour measures of the SDQ are presented as total difficulties, externalizing behaviours, internalizing behaviours and prosocial behaviour. The SDQ has been validated not only against other measures of child behaviour problems, including the Child Behaviour Checklist (CBCL) 21 but also clinical diagnostic measures of mental disorder in children. 22 The SDQ has been shown to compare favourably with other measures for identifying hyperactivity and attention problems Several studies have demonstrated the cross-cultural validity of the SDQ in European and developing countries. 23,26 28 The Russian version of the SDQ has previously been used in clinical and research settings. 29 The parent accompanying the child (usually the mother) at the 6.5-year follow-up completed the SDQ in the polyclinic waiting room. Of the total of children, parents of children completed the SDQ. The teachers of those who had begun school also completed the teacher version of the SDQ distributed by the polyclinic paediatricians. The SDQ items are identical in the parent and the teacher versions. Of all children seen at the follow-up, the teacher SDQ was obtained in 87% (n ¼ ); most of the remainder had not yet begun formal schooling at the time of the follow-up. As previously reported, 30 internal consistency and test retest reliability of the parent and teacher SDQ were high in our study; Cronbach s for total difficulties, for example, was 0.82 and 0.73 in the teacher and parent SDQ, respectively. Correlations between the parent and the teacher SDQ scores were modest: 0.28 for total difficulties, 0.39 for externalizing behaviours and 0.21 for internalizing behaviours. Potential confounders Potentially confounding maternal and family characteristics included maternal and paternal age at the time of the child s birth, maternal smoking and alcohol consumption during pregnancy, breastfeeding, marital status, number of other children in the household, parental education and occupation (all based on maternal report at enrolment), and maternal and paternal height and body mass index (based on maternal report at the 6.5-year follow-up). Statistical analysis We examined the association of growth trajectories during the first 5 years to allow a time lag before the outcome ascertainment (at 6.5 years). We estimated growth trajectories for each child from birth to age 5 years in weight and length/height using a linear spline random-effects model 31 with MLwiN version 2.1 ( shtml). 32 Such models estimate mean growth trajectories based on the repeated measures of weight and length/height over time and allow individual growth trajectories to randomly vary around the mean. This approach allows for changes in scale and variance of growth measures over time and uses all available data for each individual. We used fractional polynomials to find the best-fitting growth trajectory for weight and length/ height, and identified the best-fitting spline knots at 3 and 12 months for each sex. The model fits the data well; 95% limits of agreement between observed and expected growth measures were within 10% of the mean values at each planned visit in the first year and at the 6.5-year follow-up. Thus, for both boys and girls, linear random-effects models were fitted for weight and length/height growth trajectory for four periods: pre-natal growth (birth weight and length standardized by gestational age for each sex) and growth between 0 3 months ( early infancy ), 3 12 months ( late infancy ) and 1 5 years ( early

4 1218 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY childhood ). For instance, weight for child i at age j was modelled as: y ij ¼ 0 þ u 0i þð 1 þu 1i Þs i1 þð 2 þu 2i Þs i2 þð 3 þu 3i Þs i3 þ e ij j where 0 is the population average birth weight (intercept) and 1, 2 and b 3 are the population average velocities (slopes) of weight gain between 0 3 months, 3 12 months and months, respectively (the fixed effects). u 0i, u 1i, u 2i and u 3i represent individual-level random effects. u 0i is the deviation from the average intercept for child i, and u 1i, u 2i and u 3i are the deviations for child i from the average velocities of weight gain between 0 3 months, 3 12 months and months, respectively. s i1, s i2 and s i3 are the amount of time spent in each period for child i at age j. Finally, e ij is the deviation from the predicted weight for child i at age j (the occasion-level random effects) and is related to age to allow for the increase in weight (and therefore the increase in the size of the measurement error of weight) with age. The individual-level random effects correlation matrix was unstructured, i.e. each random effect was allowed to be correlated with all other random effects. The occasion-level random effects were independent; thus we assumed that measurement error at one age was unrelated to measurement error for that individual at another age. Occasion- and individuallevel random effects were independent. From these models, individual-specific estimates of growth trajectory (i.e. each child s deviation from the mean) in weight and length/height for the four periods were obtained for each child. 33 These weight and length/height trajectory parameters were then standardized (z-scored) for sex, and additionally for gestational age for pre-natal growth. Mean differences in IQ and SDQ scores by pre-natal and post-natal growth trajectory in weight and length/height were estimated by generalized estimating equations (GEE) regression analysis using STATA 10. We used the GEE approach to account for clustering in the data by hospitals and associated polyclinics and thereby correctly estimate standard errors of the regression coefficients. Associations with each growth parameter were examined in three models: (i) unadjusted; (ii) adjusted for potential confounding factors ( adjusted 1 ); and (iii) adjusted for potential confounders and for earlier growth ( adjusted 2 ) models. Results Table 1 shows the baseline characteristics of the PROBIT children in our study, their growth measures from birth to age 5 years, and the mean IQ, teachers ratings of the children s academic performance, and parent and teacher SDQ scores. Only minimal differences were observed between our analytic sample and those excluded from the original sample of the PROBIT due to loss to follow-up or missing information on exposures or outcomes. Compared with those excluded, children included in our study were similar with respect to birth weight, gestational age, parental age or parental size. Those excluded were slightly more likely to have fathers with university education (16% vs 12%) and non-manual occupation (30% vs 28%) and to be first-born children (61% vs 56%), but slightly less likely to have mothers with non-manual occupation (40% vs 43%). Table 2 shows the association between the growth trajectories and IQ at age 6.5 years. The patterns of association did not vary by sex (all P-values for sex-interaction 40.10), and thus we present sex-adjusted results. For weight gain trajectory, the unadjusted associations show that a 1 SD greater birth weight for gestational age (e.g. 447 g in boys and 434 g in girls at 40 weeks) was associated with a 0.94-point (95% CI: ) increase in full-scale IQ. A 1 SD faster weight gain during early infancy was associated with a 1.25-point (95% CI: ) increase; the corresponding increases were 0.68 (95% CI: ) points for late infancy and 0.99 (95% CI: ) points for early childhood growth. After adjusting for potential confounders (adjusted 1), the associations were attenuated by 15 50%, with greater attenuation for associations with weight gain during late infancy and early childhood. Of the confounding factors included, parental socioeconomic position and size, education and height in particular, most attenuated the associations and were responsible for 30% of the attenuation. Further adjustment for earlier growth (adjusted 2) did not substantially change the patterns of association. Stronger associations with birth weight and early infancy weight gain were also observed when verbal and performance IQ scores were analysed separately. For length/height gain trajectories, faster growth was positively associated with full-scale IQ scores in all four periods, with the largest association observed for height gain during early childhood in unadjusted associations. After adjustment for potential confounders and earlier growth (adjusted 2), the positive associations in the unadjusted model were attenuated by 30 50%, especially for early childhood height gain. As seen with weight gain trajectories, parental education and height were mainly responsible for the attenuation. The attenuated associations between infancy length gain and IQ in adjusted 1 were strengthened with adjustment for earlier growth (adjusted 2), because infancy length gain was negatively correlated with birth length (r ¼ 0.31 for early infancy and r ¼ 0.49 for late infancy). The patterns of association between length/height trajectories and verbal and performance IQ were essentially the same as observed for full-scale IQ. Whereas growth measures in all four periods were positively associated with school performance,

5 PRE-NATAL OR POST-NATAL GROWTH AND CHILD DEVELOPMENT 1219 Table 1 Means (standard deviation) of growth measures, outcomes and covariates in the study sample Growth measures Weight (kg) Birth weight 3.44 (0.42) 3 months 6.11 (0.67) 12 months (1.01) 6.5 years (3.61) Length/height (cm) Birth length 51.9 (2.13) 3 months 60.9 (2.4) 12 months 75.9 (2.7) 6.5 years (5.2) Estimated growth velocity Weight (kg/year) 0 3 months (1.58) 3 12 months 6.38 (0.87) 1 5 years 1.89 (0.36) Length/height (cm/year) 0 3 months 36.0 (6.21) 3 12 months (1.95) 1 5 years 8.88 (0.80) IQ Full-scale IQ (15.8) Verbal IQ (16.9) Performance IQ (15.1) School performance [1 (far below) 5 (far above) the grade level] Mathematics 3.2 (0.8) Reading 3.2 (0.8) Writing 3.2 (0.8) Other 3.3 (0.6) Average 3.2 (0.7) Parental SDQ Total difficulties 11.5 (5.0) Externalizing behaviours 6.3 (3.2) Internalizing behaviours 5.2 (3.0) Prosocial behaviour 8.3 (1.6) Teacher SDQ Total difficulties 9.6 (5.8) Externalizing behaviours 5.3 (4.0) Internalizing behaviours 4.3 (3.0) Prosocial behaviour 7.5 (2.2) Gestational age (weeks) 39.4 (1.0) Mother s pregnancy behaviour [n (%)] Smoking (yes) 249 (2.1) Drinking (yes) 281 (2.4) Mother s age (years) 24.5 (4.9) Father s age (years) 27.4 (5.1) (continued)

6 1220 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY Table 1 Continued Parental body size Mother s weight (kg) 66.2 (12.5) Mother s height (cm) (5.7) Father s weight (kg) 80.2 (11.7) Father s height (cm) (6.6) Mother s marital status [n (%)] Married 10,618 (89.2) Cohabitating 814 (6.8) Unmarried 467 (3.9) No. of children (56.6) (34.7) 2þ 1030 (8.7) Mother s education [n (%)] University degree 1579 (13.3) Partial university 6075 (51.1) Secondary education 3800 (31.9) <Secondary education 445 (3.7) Mother s occupation [n (%)] Non-manual 5134 (43.1) Manual 4041 (34.0) Unemployed 2724 (22.9) Father s education (n (%)) University degree 1467 (12.3) Partial university 5460 (45.9) Secondary education 4297 (36.1) <Secondary education 277 (2.3) Missing 398 (3.3) Father s occupation [n (%)] Non-manual 3294 (27.7) Manual 6506 (54.7) Unemployed 1617 (13.6) Unknown 482 (4.0) associations between pre-natal and infant growth were larger than those for early childhood growth (data not shown). For example, a 1 SD greater birth weight for gestational age was associated with a 0.03-point (95% CI: ) higher teacher rating in mathematics after adjusting for potential confounders. A 1 SD faster weight gain in early infancy, late infancy and early childhood was associated with increased mathematics ratings by 0.02 (95% CI: ), 0.03 (95% CI: ) and 0.01 (95% CI: 0.01 to 0.03) points, respectively, after adjusting for confounders and earlier growth. The corresponding figures for length/height gain were 0.01 (95% CI: ), 0.04 (95% CI: ), 0.04 (95% CI: ) and 0.02 (95% CI: ) points, respectively, in fully adjusted models. Table 3 summarizes the associations between growth trajectories and teacher SDQ scores. In unadjusted analyses, a 1 SD increase in birth weight for gestational age was negatively associated with total difficulties, externalizing behaviours and internalizing behaviours and positively associated with prosocial behaviour. After adjustment for potential confounders, inverse associations between birth weight and total difficulties ( 0.28, 95% CI: 0.40 to 0.15), externalizing behaviours ( 0.16,

7 PRE-NATAL OR POST-NATAL GROWTH AND CHILD DEVELOPMENT 1221 Table 2 Associations between pre-natal and post-natal growth trajectories (z-scores) through 5 years of age and IQ at age 6.5 years Unadjusted Adjusted 1 Adjusted 2 Weight trajectory Full-scale IQ Birth weight 0.94 (0.62, 1.25) 0.82 (0.54, 1.10) Weight gain from 0 to 3 months 1.25 (0.88, 1.62) 0.87 (0.52, 1.21) 0.77 (0.42, 1.11) Weight gain from 3 to 12 months 0.68 (0.44, 0.93) 0.32 (0.06, 0.56) 0.29 (0.01, 0.58) Weight gain from 1 to 5 years 0.99 (0.70, 1.29) 0.54 (0.27, 0.81) 0.40 (0.04, 0.76) Verbal IQ Birth weight 0.89 (0.54, 1.24) 0.81 (0.50, 1.12) Weight gain from 0 to 3 months 1.26 (0.86, 1.66) 0.80 (0.44, 1.65) 0.70 (0.34, 1.05) Weight gain from 3 to 12 months 0.68 (0.41, 0.95) 0.28 (0.001, 0.56) 0.27 (-0.03, 0.58) Weight gain from 1 to 5 years 1.07 (0.76, 1.38) 0.63 (0.31, 0.95) 0.52 (0.08, 0.95) Performance IQ Birth weight 0.77 (0.50, 1.04) 0.65 (0.37, 0.93) Weight gain from 0 to 3 months 0.96 (0.65, 1.27) 0.75 (-0.04, 1.30) 0.67 (0.34, 1.00) Weight gain from 3 to 12 months 0.53 (0.29, 0.76) 0.26 (0.01, 0.51) 0.23 (-0.06, 0.53) Weight gain from 1 to 5 years 0.69 (0.41, 0.98) 0.34 (0.06, 0.63) 0.22 (-0.13, 0.56) Length/Height trajectory Full-scale IQ Birth length 0.89 (0.61, 1.18) 0.62 (0.32, 0.92) Length gain from 0 to 3 months 0.86 (0.48, 1.23) 0.42 (0.04, 0.81) 0.65 (0.24, 1.05) Length gain from 3 to 12 months 0.68 (0.27, 1.09) 0.23 (-0.15, 0.61) 0.68 (0.28, 1.08) Height gain from 1 to 5 years 1.38 (0.94, 1.82) 0.77 (0.39, 1.16) 0.64 (0.16, 1.13) Verbal IQ Birth length 0.80 (0.53, 1.06) 0.53 (0.25, 0.81) Length gain from 0 to 3 months 0.89 (0.51, 1.27) 0.38 (0.02, 0.74) 0.57 (0.17, 0.97) Length gain from 3 to 12 months 0.83 (0.36, 1.30) 0.39 (-0.06, 0.84) 0.84 (0.33, 1.33) Height gain from 1 to 5 years 1.44 (0.97, 1.91) 0.82 (0.40, 1.24) 0.62 (0.10, 1.14) Performance IQ Birth length 0.81 (0.51, 1.11) 0.59 (0.27, 0.91) Length gain from 0 to 3 months 0.62 (0.25, 0.99) 0.35 (-0.01, 0.72) 0.56 (0.18, 0.94) Length gain from 3 to 12 months 0.38 (0.07, 0.69) 0.01 (-0.30, 0.31) 0.37 (0.05, 0.70) Height gain from 1 to 5 years 1.03 (0.67, 1.39) 0.57 (0.23, 0.91) 0.55 (0.14, 0.97) Adjusted 1: adjusted for sex, term status (early term/term/post-term), maternal smoking and drinking during pregnancy, duration of breastfeeding, number of older children, parental marital status, parental education and occupation, and parental height and BMI. Adjusted 2: adjusted for all variables in Adjusted 1 plus earlier growth trajectory. 95% CI: 0.24 to 0.07), internalizing behaviours ( 0.12, 95% CI: 0.18 to 0.05) remained (although were considerably attenuated), but the positive association with prosocial behaviour attenuated towards the null. When potential confounders and (for the post-natal period) earlier growth were adjusted for, weight gain in late infancy was negatively associated with externalising behaviours and positively with prosocial behaviour. Similar patterns of association were observed for length/height gain trajectories. Birth length was negatively associated with total difficulties ( 0.21, 95% CI: 0.33 to 0.08) and externalizing behaviours ( 0.15, 95% CI: 0.23 to 0.06) after adjusting for potential confounders. Length/height gain in later periods (particularly in infancy) was also negatively associated with problem behaviours, showing very similar effect sizes observed in weight gain but with wider confidence intervals after adjusting

8 1222 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY Table 3 Associations between pre-natal and post-natal growth trajectories (z-scores) through 5 years of age and teacher SDQ scores at age 6.5 years Weight trajectory Length/Height trajectory Unadjusted Adjusted a Unadjusted Adjusted a Total difficulties At birth 0.38 ( 0.48, 0.27) 0.28 ( 0.40, 0.15) 0.29 ( 0.40, 0.17) 0.21 ( 0.33, 0.08) Gain from 0 to 3 months 0.15 ( 0.26, 0.04) 0.07 ( 0.19, 0.05) 0.05 ( 0.17, 0.07) 0.07 ( 0.21, 0.06) Gain from 3 to 12 months 0.09 ( 0.20, 0.02) 0.13 ( 0.26, 0.01) 0.01 ( 0.10, 0.12) 0.11 ( 0.25, 0.03) Gain from 1 to 5 years 0.07 ( 0.17, 0.03) 0.05 ( 0.07, 0.18) 0.10 ( 0.20, 0.002) 0.01 ( 0.12, 0.15) Externalizing behaviours At birth 0.23 ( 0.31, 0.16) 0.16 ( 0.24, 0.07) 0.20 ( 0.28, 0.12) 0.15 ( 0.23, 0.06) Gain from 0 to 3 months 0.09 ( 0.17, 0.01) 0.03 ( 0.12, 0.06) 0.04 ( 0.12, 0.04) 0.06 ( 0.16, 0.03) Gain from 3 to 12 months 0.05 ( 0.13, 0.02) 0.09 ( 0.18, 0.01) 0.01 ( 0.07, 0.08) 0.08 ( 0.18, 0.01) Gain from 1 to 5 years ( 0.07, 0.07) 0.07 ( 0.01, 0.16) 0.06 ( 0.13, 0.02) 0.02 ( 0.07, 0.11) Internalizing behaviours At birth 0.14 ( 0.20, 0.08) 0.12 ( 0.18, 0.05) 0.08 ( 0.14, 0.02) 0.05 ( 0.13, 0.01) Gain from 0 to 3 months 0.06 ( 0.12, 0.004) 0.04 ( 0.10, 0.03) 0.01 ( 0.07, 0.05) 0.02 ( 0.09, 0.05) Gain from 3 to 12 months 0.04 ( 0.09, 0.02) 0.04 ( 0.11, 0.02) ( 0.05, 0.06) 0.03 ( 0.11, 0.05) Gain from 1 to 5 years 0.07 ( 0.12, 0.02) 0.02 ( 0.08, 0.05) 0.04 ( 0.09, 0.01) 0.00 ( 0.07, 0.07) Prosocial behavior At birth 0.07 (0.03, 0.11) 0.03 ( 0.01, 0.08) 0.04 (0.002, 0.09) 0.01 ( 0.04, 0.06) Gain from 0 to 3 months 0.02 ( 0.02, 0.07) 0.02 ( 0.07, 0.02) 0.01 ( 0.04, 0.05) 0.02 ( 0.07, 0.04) Gain from 3 to 12 months 0.04 (0.01, 0.08) 0.06 (0.01, 0.11) 0.01 ( 0.03, 0.05) 0.02 ( 0.04, 0.07) Gain from 1 to 5 years 0.02 ( 0.02, 0.05) 0.02 ( 0.06, 0.03) 0.03 ( 0.01, 0.07) 0.01 ( 0.04, 0.07) a Adjusted models include all potential confounding factors and earlier growth trajectory estimates for post-natal growth. additionally for earlier growth. Most of the attenuation from the unadjusted to the fully adjusted associations was due to adjustment for confounding factors rather than for earlier growth. Similar results were observed for parental SDQ scores. Birth weight and early infancy weight gain were negatively associated with total difficulties and externalizing behaviours in unadjusted analyses (data not shown). The associations were small in magnitude, however, and were not observed after adjusting for potential confounding factors, except for the association between birth weight and externalizing behaviours ( 0.07, 95% CI: 0.13 to 0.01). Similar patterns were observed with length/height gain trajectories (data not shown). Discussion In this cohort of healthy children born at or after term, we found small, positive associations of growth trajectories from the pre-natal period through the first 5 years of life with both cognitive ability and mental health at age 6.5 years. However, the pattern of associations across the four periods defined in our study differed somewhat for weight and length/height gain trajectories, in particular for cognitive ability. For weight gain trajectory, mean IQ differences associated with growth during the pre-natal period and early infancy (0 3 months) were larger than with later growth, whereas the contributions of length/height gain trajectories to mean IQ differences across the four periods were similar to one another. These patterns were observed not only with full-scale IQ scores, but also for verbal and performance IQ. Nevertheless, we found that irrespective of earlier growth, faster growth at later ages was associated with increased full-scale IQ scores. For child behaviour, pre-natal growth showed larger associations than did post-natal growth. The large sample size, prospective cohort design and large number of growth measures from birth to age 5 years (up to 13) are strengths of study. Our fractional linear spline random-effects model is an improvement on the approaches used in previous studies. It allowed us to identify important time points at which the rate of growth differs in the first 5 years of life, rather than using arbitrary time points limited by availability of data, as was often the case in previous studies. The time points identified by our linear spline approach were based solely on growth measures to represent distinct biological

9 PRE-NATAL OR POST-NATAL GROWTH AND CHILD DEVELOPMENT 1223 growth patterns over the 5 years and were not influenced by the choice of our outcome measures. Our analytical approach also enabled us to estimate individual trajectories by allowing random variation across individuals and their associations with differences in cognitive ability and behaviour scores. The associations observed in our study are probably generalizable to other developed country settings. First, our findings are not based on children with growth problems but rather on normal, healthy children. Second, Belarus resembles Western developed countries with respect to readily accessible basic health care services, high levels of sanitation, high immunization rates, low incidence of infection and low rates of infant and child mortality. Some limitations in our study also require discussion, however. First, although we adjusted for a wide range of maternal and paternal characteristics, residual confounding by family characteristics (both genetic and environmental factors) cannot be excluded. All coefficients were attenuated (by up to 50%) with adjustment for the confounders included here, and it seems likely that other confounders exist (not measured here) which would attenuate these relationships further. Parental cognitive ability, a strong predictor of cognitive ability of offspring, 34 for example, may also affect growth trajectory via parenting behaviours such as feeding practices and physical activity. The lack of maternal psychological distress data in our data may also explain the observed association for behaviour. However, Alati et al. 35 showed that the association between birth weight and behavioural problems is not affected by the adjustment for maternal anxiety and depression in a relatively large birth cohort. Second, although we deliberately chose a minimum of 1.5 years between the growth and outcome measures, reverse causation could theoretically explain our findings; children with higher cognitive ability and/or better mental health might grow faster via better eating practices. 36 Third, weight and length/height measures in our study were not standardized across hospitals and polyclinics, which is likely to increase measurement errors. These measurement errors are unlikely to vary systematically by child cognitive ability and behaviour across study sites, however, and would therefore tend to attenuate associations toward the null. Measurement errors for IQ and behaviour are also a potential source of bias. Despite our efforts to standardize measurements across paediatricians at 31 polyclinics, mean IQ scores across polyclinics were highly clustered, as indicated by the intra-class coefficient of Since a single paediatrician measured IQ at 24 of the 31 polyclinics (two paediatricians shared the work at the seven busiest polyclinics), this clustering probably reflects differences in strictness or leniency in scoring or timing of responses among paediatricians, rather than true geographic differences in IQ across polyclinics. The clustering of growth measures (which were also measured by the paediatricians) was not as strong, however, as reflected by intra-class coefficients of 0.01 to The clustering of IQ thus does not confound the association but widens CIs of the estimates of the association when statistically accounted for in our analysis. 37 In regard to behaviour measures, having multiple informants both the parent and the teacher is likely to have reduced potential reporting bias of child behaviour. The parent and teacher SDQ scores have been observed to provide complementary information; teachers appear better able to assess externalising behavioural problems, whereas parents appear better at assessing internalizing behaviours. 25 Despite larger effect sizes for the teacher SDQ scores, the direction of associations observed with the parent and the teacher SDQs in our study were consistent with each other. Finally, the Russian version of the WASI has never been formally validated. Nonetheless, WASI scores in PROBIT children were associated strongly with parental education and other family characteristics in the expected directions and were also positively correlated with children s academic performances (correlation coefficients 0.3). 38 Research on long-term effects of post-natal growth has focused mostly on children with very slow growth characterized as growth faltering or failure to thrive. Only recently have researchers directed attention to child growth and its association with cognitive development in general, healthy populations. Wide variations in the growth period examined, growth measures used and ages at which cognitive ability was assessed characterize these studies; most have reported small but positive associations, 8,10,12,39 41 consistent with our results. One exception is the study by Belfort et al., 11 which reported no association between weight gain in the first 6 months and cognitive ability measured at age 3 years. The authors argue that among healthy, term-born children, post-natal growth is not related to cognitive development, but the relatively small sample size (n ¼ 872) is likely to have been underpowered to detect a modest effect. Studies have reported that early infancy is a sensitive period for cognitive development, but most such studies were largely based on children with failure to thrive. 9,42,43 The term sensitive period denotes an age range during which exposure effects are increased, although such effects can be modified or even reversed over time. 44 Among studies from generally healthy populations, the existence of a sensitive period in post-natal growth for later cognitive development remains unclear, mainly owing to limitations in their analytical approach, usually based on available growth data collected at arbitrary limited time points. Our approach to examining growth trajectories suggests that effects may vary somewhat by growth measure and by outcome. Birth weight and early infancy weight gain seem more important for cognitive ability, whereas growth in length/height seem equally

10 1224 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY important across all periods. It is not clear why the patterns of association with cognitive ability differ across weight and length/height trajectories. The different patterns observed in our study suggest that growth throughout the first 5 years contributes to later cognitive ability but does not provide strong evidence for a true sensitive period of cognitive development. The different patterns may merely reflect greater errors in measuring length than weight, especially for infants. We conceptualized family socioeconomic characteristics as potentially confounding the association between growth trajectory and IQ. This view is supported by the substantial attenuation of the observed associations with adjustment for parental socioeconomic position and height, which is also closely related to socioeconomic position. 45 However, it could be hypothesised that the effects of family background on child cognitive ability are mediated by child growth. In an attempt to test this hypothesized pathway, we examined associations of parental education and occupation with IQ and the degree of attenuation after adjustment for child growth. Although the socioeconomic factors were positively associated with IQ independently of growth trajectory, the associations were only trivially attenuated by adjustment for child growth (data not shown). To our knowledge, no previous study has examined the association between post-natal growth and child behaviour in healthy children. Children with experiences of extreme malnutrition and stunting have been reported to show problem behaviours, including aggressiveness, attention deficits, peer problems, depressive symptoms and hyperactivity Our results are consistent in direction, although smaller in magnitude, with those observed in children with severe growth restriction in infancy or early childhood. Given the linear relationship between total difficulties scores of the SDQ and the likelihood of having mental disorder observed in a recent study, 22 our results suggest that faster post-natal growth may also have beneficial effects in mental health among healthy children. A possible biological mechanism underlying the positive association observed in our study is the growth hormone (GH)/insulin-like growth factor (IGF)-I system. GH/IGF-I system plays a key role in somatic growth regulation, particularly for longitudinal growth at pre-pubertal ages. 50 A recent trial has shown that children born small for gestational age who underwent GH treatment showed not only catch-up growth in height, but also improvements in IQ and problem behaviour scores. 51 In addition, serum IGF-I level has been positively associated with cognitive ability in healthy children with normal growth. 52 Concerns and debate have arisen about the adverse effects of rapid weight gain in infancy, including increased long-term risk of obesity, 53,54 high blood pressure and insulin resistance. 58 Although the magnitudes of association are small, we have also observed positive associations of pre-natal and post-natal growth and blood pressure at 6.5 years of age in our study sample. 33 In the present study, however, our results suggest that rapid growth in infancy and early childhood is positively associated with the development of cognition and mental health. Increased susceptibility to infection, 59 smaller adult stature 60 and mental illness 61 are among other adverse health consequences associated with slower infant growth. The overall evidence suggests that the long-term consequences of rapid infant weight gain may vary according to outcome, 62 and clinical and public health practice therefore needs to consider both the risks and benefits of potential interventions. Funding This research was supported by a grant from the Canadian Institutes of Health Research (MOP-53155). Conflict of interest: None declared. KEY MESSAGES Most studies of the associations between pre-natal or post-natal growth and cognitive and behavioural development have been based on children with pathologically slow growth or with growth measured between two time points only. In this large cohort of healthy children who were born at term or later with birth weight of at least 2500 g, children with faster growth from the pre-natal period through the first 5 years of life showed higher scores in IQ at age 6.5 years. For behavioural development, it seems that faster growth in the pre-natal period and infancy but not in early childhood is associated with lower scores in problem behaviours, suggestive of better mental health.

11 PRE-NATAL OR POST-NATAL GROWTH AND CHILD DEVELOPMENT 1225 References 1 Smart JL. Malnutrition, learning and behavior 25 years on from the Mit symposium. Proc Nutr Soc 1993;52: Grantham-McGregor S, Cheung YB, Cueto S, Glewwe P, Richter L, Strupp B. Developmental potential in the first 5 years for children in developing countries. Lancet 2007; 369: Webb SJ, Monk CS, Nelson CA. Mechanisms of postnatal neurobiological development: implications for human development. Dev Neuropsychol 2001;19: Casey PH, Whiteside-Mansell L, Barrett K, Bradley RH, Gargus R. Impact of prenatal and/or postnatal growth problems in low birth weight preterm infants on school-age outcomes: an 8-year longitudinal evaluation. Pediatrics 2006;118: Hack M, Merkatz IR, Gordon D, Jones PK, Fanaroff AA. The prognostic significance of postnatal growth in very low birth weight infants. Am J Obstet Gynecol 1982;143: Ehrenkranz RA, Dusick AM, Vohr BR, Wright LL, Wrage LA, Poole WK. Growth in the neonatal intensive care unit influences neurodevelopmental and growth outcomes of extremely low birth weight infants. Pediatrics 2006;117: Corbett SS, Drewett RF. To what extent is failure to thrive in infancy associated with poorer cognitive development? A review and meta-analysis. J Child Psychol Psychiatry 2004;45: Cheung Y, Ashorn P. Continuation of linear growth failure and its association with cognitive ability are not dependent on initial length-for-age: a longitudinal study from 6 months to 11 years of age. Acta Paediatr 2010;99: Emond AM, Blair PS, Emmett PM, Drewett RF. Weight faltering in infancy and IQ levels at 8 years in the Avon Longitudinal Study of Parents and Children. Pediatrics 2007;120:e Heinonen K, Raikkonen K, Pesonen AK et al. Prenatal and postnatal growth and cognitive abilities at 56 months of age: a longitudinal study of infants born at term. Pediatrics 2008;121:e Belfort MB, Rifas-Shiman SL, Rich-Edwards JW, Kleinman KP, Oken E, Gillman MW. Infant growth and child cognition at 3 years of age. Pediatrics 2008;122: e Corbett SS, Drewett RF, Durham M, Tymms P, Wright CM. The relationship between birthweight, weight gain in infancy, and educational attainment in childhood. Paediatr Perinat Epidemiol 2007;21: Boddy J, Skuse D, Andrews B. The developmental sequelae of nonorganic failure to thrive. J Child Psychol Psychiatry 2000;41: Bhutta AT, Cleves MA, Casey PH, Cradock MM, Anand KJ. Cognitive and behavioral outcomes of school-aged children who were born preterm: a meta-analysis. JAMA 2002;288: Kramer MS, Chalmers B, Hodnett ED et al. Promotion of Breastfeeding Intervention Trial (PROBIT): a randomized trial in the Republic of Belarus. JAMA 2001;285: Wechsler D. Wechsler Abbreviated Scales of Intelligence. San Antonio, TX: Psychological Corp, Kramer MS, Aboud F, Mironova E et al. Breastfeeding and child cognitive development: new evidence from a large randomized trial. Arch Gen Psychiatry 2008;65: Achenback TM, Rescola LA. Manual for the ASEBA School-Age Forms and Profiles. Burlington, VT: University of Vermont, Research Center for Children, Youth, and Families, Goodman R. The strengths and difficulties questionnaire: a research note. J Child Psychol Psychiatry 1997;38: Goodman A, Lamping D, Ploubidis G. When to use broader internalising and externalising subscales instead of the hypothesised five subscales on the Strengths and Difficulties Questionnaire (SDQ): data from British parents, teachers and children. J Abnorm Child Psychol 2010; 38: Goodman R, Scott S. Comparing the strengths and difficulties questionnaire and the child behavior checklist: Is small beautiful? J Abnorm Child Psychol 1999;27: Goodman A, Goodman R. Strengths and Difficulties Questionnaire as a dimensional measure of child mental health. J Am Acad Child Adolesc Psychiatry 2009;48: Goodman R, Renfrew D, Mullick M. Predicting type of psychiatric disorder from Strengths and Difficulties Questionnaire (SDQ) scores in child mental health clinics in London and Dhaka. Eur Child Adolesc Psychiatry 2000;9: Goodman R, Ford T, Richards H, Gatward R, Meltzer H. The development and well-being assessment: description and initial validation of an integrated assessment of child and adolescent psychopathology. J Child Psychol Psychiatry 2000;41: Goodman R, Ford T, Simmons H, Gatward R, Meltzer H. Using the Strengths and Difficulties Questionnaire (SDQ) to screen for child psychiatric disorders in a community sample. Br J Psychiatry 2000;177: Smedje H, Broman J, Hetta J, Von Knorring A. Psychometric properties of a Swedish version of the Strengths and Difficulties Questionnaire. Eur Child Adolesc Psychiatry 1999;8: Klasen H, Woerner W, Wolke D et al. Comparing the German versions of the Strengths and Difficulties Questionnaire (SDQ-Deu) and the Child Behavior Checklist. Eur Child Adolesc Psychiatry 2000;9: Koskelainen M, Sourander A, Kaljonen A. The Strengths and Difficulties Questionnaire among Finnish schoolaged children and adolescents. Eur Child Adolesc Psychiatry 2000;9: Goodman R, Slobodskaya H, Knyazev G. Russian child mental health. A cross-sectional study of prevalence and risk factors. Eur Child Adolesc Psychiatry 2005;14: Kramer MS, Fombonne E, Igumnov S et al. Effects of prolonged and exclusive breastfeeding on child behavior and maternal adjustment: evidence from a large, randomized trial. Pediatrics 2008;121:e Goldstein H. Efficient statistical modeling of longitudinal data. Ann Hum Biol 1986;13: Rabash J, Charlton C, Browne WJ, Healy M, Cameron B. MLwiN Version 2.1. Centre for Multilevel Modeling, University of Bristol, Tilling K, Davies N, Windmeijer F et al. Does accelerated weight gain in infancy increase blood pressure in childhood? An observational analysis of the Promotion of

12 1226 INTERNATIONAL JOURNAL OF EPIDEMIOLOGY Breastfeeding Intervention Trial (PROBIT) cohort, in press. 34 Deary IJ, Strand S, Smith P, Fernandes C. Intelligence and educational achievement. Intelligence 2007;35: Alati R, Najman JM, O Callaghan M, Bor W, Williams GM, Clavarino A. Fetal growth and behaviour problems in early adolescence: findings from the Mater University Study of Pregnancy. Int J Epidemiol 2009;38: Birch LL, Fisher JO. Development of eating behaviors among children and adolescents. Pediatrics 1998; 101(3 Pt 2): Kramer MS, Martin RM, Sterne JAC, Shapiro S, Dahhou M, Platt RW. The double jeopardy of clustered measurement and cluster randomization. BMJ 2009;339: Yang S, Platt RW, Kramer MS. Variation in child cognitive ability by week of gestation among healthy term births. Am J Epidemiol 2010;171: Silva A, Metha Z, O Callaghan FJ. The relative effect of size at birth, postnatal growth and social factors on cognitive function in late childhood. Ann Epidemiol 2006;16: Gale CR, O Callaghan FJ, Bredow M, Martyn CN. The influence of head growth in fetal life, infancy, and childhood on intelligence at the ages of 4 and 8 years. Pediatrics 2006;118: Martorell R, Horta BL, Adair LS et al. Weight gain in the first two years of life is an important predictor of schooling outcomes in pooled analyses from five birth cohorts from low- and middle-income countries. J Nutr 2010;140: Skuse D, Pickles A, Wolke D, Reilly S. Postnatal-growth and mental-development - evidence for a sensitive-period. J Child Psychol Psychiatry 1994;35: Drotar D, Nowak M, Malone CA, Eckerle D, Negray J. Early psychological outcomes in failure to thrive: predictions from an interactional model. J Clin Child Psychol 1985;14: Kuh D, Ben-Shlomo Y, Lynch J, Hallqvist J, Power C. Life course epidemiology. J Epidemiol Commun Health 2003; 2003: Batty GD, Shipley MJ, Gunnell D et al. Height, wealth, and health: an overview with new data from three longitudinal studies. Econ Hum Biol 2009;7: Richardson SA, Birch HG, Grabie E, Yoder K. Behavior of children in school who were severely malnourished in first 2 years of life. J Health Soc Behav 1972;13: Richardson SA, Birch HG, Ragbeer C. Behavior of children at home who were severely malnourished in 1st 2 years of life. J Biosoc Sci 1975;7: Galler JR, Ramsey F. A follow-up-study of the influence of early malnutrition on development - behavior at home and at school. J Am Acad Child Adolesc Psychiatry 1989;28: Walker SP, Chang SM, Powell CA, Simonoff E, Grantham-McGregor SM. Early childhood stunting is associated with poor psychological functioning in late adolescence and effects are reduced by psychosocial stimulation. J Nutr 2007;137: Le Roith D. Insulin-like growth factors. N Engl J Med 1997;336: van Pareren YK, Duivenvoorden HJ, Slijper FS, Koot HM, Hokken-Koelega AC. Intelligence and psychosocial functioning during long-term growth hormone therapy in children born small for gestational age. J Clin Endocrinol Metab 2004;89: Gunnell D, Miller LL, Rogers I, Holly JM. Association of insulin-like growth factor I and insulin-like growth factor-binding protein-3 with intelligence quotient among 8- to 9-year-old children in the Avon Longitudinal Study of Parents and Children. Pediatrics 2005;116:e Baird J, Fisher D, Lucas P, Kleijnen J, Roberts H, Law C. Being big or growing fast: systematic review of size and growth in infancy and later obesity. BMJ 2005;331: Monteiro POA, Victora CG. Rapid growth in infancy and childhood and obesity in later life - a systematic review. Obes Rev 2005;6: Huxley RR, Shiell AW, Law CM. The role of size at birth and postnatal catch-up growth in determining systolic blood pressure: a systematic review of the literature. J Hypertens 2000;18: Belfort MB, Rifas-Shiman SL, Rich-Edwards J, Kleinman KP, Gillman MW. Size at birth, infant growth, and blood pressure at three years of age. J Pediatr 2007;151: Ben-Shlomo Y, McCarthy A, Hughes R, Tilling K, Davies D, Davey Smith G. Immediate postnatal growth is associated with blood pressure in young adulthood: the Barry Caerphilly Growth Study. Hypertension 2008; 52: Soto N, Bazaes RA, Pena V et al. Insulin sensitivity and secretion are related to catch-up growth in smallfor-gestational-age infants at age 1 year: results from a prospective cohort. J Clin Endocrinol Metab 2003;88: Victora CG, Barros FC, Horta BL, Martorell R. Short-term benefits of catch-up growth for small-for-gestational-age infants. Int J Epidemiol 2001;30: Ong KK, Loos RJF. Rapid infancy weight gain and subsequent obesity: systematic reviews and hopeful suggestions. Acta Paediatr 2006;95: Barker DJP, Osmond C, Rodin I, Fall CHD, Winter PD. Low-weight gain in infancy and suicide in adult life. BMJ 1995;311: Victora CG, Adair L, Fall C et al. Maternal and child undernutrition: consequences for adult health and human capital. Lancet 2008;371: