THE INFLUENCE OF BIRTH WEIGHT AND SOCIOECONOMIC POSITION

ON COGNITIVE DEVELOPMENT: DOES THE EARLY HOME AND

LEARNING ENVIRONMENT MODIFY THEIR EFFECTS?

CHRIS POWER, PHD, BARBARA J. M. H. JEFFERIS, MSC, ORLY MANOR, PHD, AND CLYDE HERTZMAN, MD

Objective To establish whether effects of birth weight and socioeconomic position on cognition are explained or modified

by home or learning environments.

Study design Prospective birth cohort (n = 13,980) with math tests at 7, 11, and 16 years of age and qualifications by

33 years of age.

Results For 1 kg increase in birth weight, 7-year math Z score increased 0.23 (0.19 adjusted for parental interest in child’s

progress) and adult qualifications increased 0.22 (on a 5-point scale). Maternal reading benefited math less among lower than

higher birth weights (p < .05). The birth weight effect remained unchanged 7 to 16 years of age. For each increment in social class

(4 categories; IV&V to I&II), 7-year math increased 0.19 (0.12 adjusted for parental interest). Benefits of mother’s reading and

father’s interest were greatest in classes IV&V (interaction p < .05). The difference in Z scores between classes I&II to IV&V was

0.57 at 7 years; 1.12 at 16 years of age. Estimates were little affected by home and school factors. Adult qualifications increased

0.40 per unit social class (0.33 adjusted for parental interest). Maternal interest reduced the chances of those from unskilled

manual origins gaining few qualifications (p < .05). Similarly, interactions were seen for maternal reading and paternal interest.

Conclusion Influences in the home partly underlie associations between social background and cognition, but they do little

to explain a birth weight/cognition association. (J Pediatr 2006;148:54-61)

B irth weight and socioeconomic position are both known to influence cognitivedevelopment.1-3 We, and others, have shown that birth weight is associated withtrajectories of cognitive development and that the relationship is not confined to

a low-weight (<2500 g) group but applies to the whole distribution.1,4 In a large Britishcohort, the association between birth weight and math skills was evident at 7 years ofage, the earliest at which cognition was tested.4 This effect remained constant throughoutchildhood to 16 years of age, and it was largely unaffected after allowing for gestational age,maternal age, social class, parity, sex, breast feeding, and parental education. There was aseparate influence of socioeconomic position on cognitive development, which had agreater effect than birth weight. Moreover, in contrast to birth weight, the social classdifference in math skills increased over time to 16 years of age, suggesting that socialconditions played an increasingly important role in cognitive development. The persistinginfluence of birth weight and its effect across the full distribution, together with thestrengthening effect of social class over time are remarkable and highlight the need toconsider whether the factors underlying these associations are modifiable.

The literature provides evidence for several potential factors in the home and learningenvironment that influence the trajectory of an individual child’s cognitive development,including, for example, parenting style,5,6 verbal fluency in the home,7 parental ‘‘invest-ment’’ in the child’s learning,2 cognitive stimulation,8,9 ‘‘quality’’ child care,10 mentorship,11

and neighborhood characteristics.12,13 Yet, the role of home and learning environments inexplaining effects of factors important at the population level, such as birth weight and so-cial class, is not well understood. Our overall objective is to establish whether characteristicsof the home and learning environment explain, or have the capacity to modify, the effects ofbirth weight and social class. We should emphasize that our concern here is not to demon-strate the independent effects of home and learning environments on cognitive develop-ment, but rather to understand how they influence birth weight and social class effects.

54

From the Institute of Child Health,Centre for Peadiatric Epidemiologyand Biostatistics, London, United King-dom; the School of Public Health andCommunity Medicine, The HebrewUniversity, Hadassah Medical Organi-zation, Jerusalem, Israel; and theHuman Early Learning Partnership,University of British Columbia, Van-couver, British Columbia, Canada.Barbara Jefferis is supported by a JointMRC/DH Special Training Fellowshipin Health of the Public Research.

Submitted for publication Feb 14,2005; last revision received Jun 29,2005; accepted Jul 19, 2005.

Reprint requests: Dr Chris Power, In-stitute of Child Health, Centre forPaediatric Epidemiology and Biostatis-tics, 30 Guilford Street, London,WC1N 1EH, United Kingdom E-mail:c.power@ich.ucl.ac.uk.0022-3476/$ - see front matter

Copyrightª 2006 Elsevier Inc. All rightsreserved.

10.1016/j.jpeds.2005.07.028

We examine these issues using longitudinal data from the1958 British birth cohort regarding math skills duringchildhood to 16 years of age, as well as adult educationalqualifications.

METHOD

Sample

The 1958 birth cohort comprises persons born in 1 weekin March 1958 in England, Scotland, and Wales. Details ofthe study are published elsewhere.14,15 In brief, the cohorthas been followed-up in childhood at 7, 11, and 16 years ofage and in adulthood, with data in the present study extendingto 33 years of age. Response rates were 91%, 91%, and 87%,respectively, at 7, 11, and 16 years of age. The sample usedhere (n = 13,980) is singletons (390 multiple births are ex-cluded) with data on birth weight and a gestational age 32to 44 weeks (2843 without a recorded gestational age wereexcluded). Sample size varied for different cognitive measuresand was smallest (n = 8825) for analysis of adult qualifications.We compared this sample (ie, 8825) to the total maximum(ie, 13,980 singletons). Both samples had a similar weight atbirth: in males 3.437 kg versus 3.425 kg, respectively; infemales 3.290 kg in both samples. The proportions in socialclasses IV&V were similar in both samples: for males 20.6%and 23.0%, respectively; for females 22.4% and 23.5%. Thus,with respect to weight at birth and social class, the analysissample for adult qualifications is similar to the larger birthsample.

Measure

Participants took age-appropriate tests of math at school(7, 11, 16 years of age). The arithmetic test at 7 years of agecomprises 10 problems with graded levels of difficulty (range0-10); teachers read the questions to poor readers. At 11 yearsof age, the math test was constructed by the National Founda-tion for Educational Research in England and Wales (range0-40). At 16 years of age a math comprehension test (range0-31) was constructed at the University of Manchester. High-est qualifications by 33 years of age were in five groups: none,<O level (or equivalent), O level (or equivalent), A level (orequivalent) or higher qualification; broadly equivalent to UScategories: no qualifications; < grade 10; grade 10; high schooldiploma/grade 12; and > high school diploma. Weight at birthwas recorded in pounds and ounces and converted into kilo-grams. Duration of gestation was estimated using date oflast menstrual period reported by the mother at birth andchecked against General Practitioner records. Social class atbirth was based on father’s occupation in 1958, classified usingthe Registrar General’s scale: ranging from class I (profes-sional) to V (unskilled manual). Groups I and II were com-bined as were groups IV and V to give larger numbers inextreme groups. Households with no male head of householdwere included with group IV&V.

We selected several factors to characterize the home andschool environment at 7 years of age. The home environment

The Influence Of Birth Weight And Socioeconomic Position On CognitiveDevelopment: Does The Early Home And Learning Environment ModifyTheir Effects?

was represented by: (1) the parents’ report of the frequencythat they read to or with their child (at least weekly, occa-sionally, never or hardly ever); (2) teachers’ assessments ofhow much interest each parent took in the child’s educationalprogress (very interested, some interest, little interest, overconcerned). Data were grouped as very interested versusothers (‘‘over concerned’’ were <2% of the sample); and (3)teachers’ assessments of whether or not the parents took ini-tiative to discuss the child. The school environment was rep-resented by teacher reports of whether or not the followingwere available: (1) parent-teacher association; (2) meetingsarranged for parents, by the school or the parent teacherassociation, on educational matters; and (3) social functionsarranged for parents.

Analysis

Math scores were converted to standard deviationscores for comparability between ages (7, 11 and 16 years)separately for males and females, approximating a continuousnormally distributed variable at each age. Preliminary analy-ses of the association between the math scores and birthweight and social class were conducted separately by sex.Relationships were similar for males and females; thereforesubsequent analyses were based on the combined sample,with adjustment for sex. Next, the associations between birthweight or social class and math scores were examined usinggrowth trajectory multilevel models with math score as a re-peated outcome (3 occasions, 7, 11, and 16 years of age) andbirth weight and social class as continuous variables. (For thelatter, classes IV&V = 1, IIIm = 2, IIInm = 3, I&II = 4).These models provide estimates of the intercept and slopeof math score trajectories associated with birth weight or so-cial class: the intercept representing the effect of the factor onoccasion one, that is, 7 years of age, with the slope represent-ing the estimate of linear change over the three occasions. Totest whether the associations with math for birth weight orsocial class were explained by the home and school factors,we adjusted for each factor separately (treating categoricalvariables as continuous). To test whether home and schoolfactors modified the effects of birth weight or social class,we included interaction terms in the multilevel models when-ever a significant effect on the intercept or slope was found.We reported previously for this study population that associ-ations for birth weight with math Z score were unaffected byexcluding births at #37 weeks.4 From the 13,980 partici-pants in our study sample, 1081 were born at #37 weeks;498 were born at <37 weeks. In analyses for the present arti-cle, models for math included an adjustment for exact age attest, with further adjustment for gestational age in analyses ofbirth weight. Finally, effects of home and school factors onthe relationship of birth weight or social class on qualifica-tions at 33 years of age were examined using linear regres-sion. Interactions with home and school factors were alsotested. Multivariate analyses are based on subjects with dataon birth weight, gestational age, cognitive measure (and exactage of measure for math score), and home or school factors:thus sample sizes vary.

55

Table 1. Mean z-scores for maths at 7 years by class of origin and birth weight (boys and girls)

Boys Girls

ny % Mean z-Score ny % Mean z-Score

Maths

Birth weight (kg) (6216) (5908),2.5 232 3.7 20.21 289 4.9 20.312.5-3 957 15.4 20.15 1258 21.3 20.103-3.5 2202 35.4 0.00 2402 40.7 0.003.5-4 2073 33.4 0.06 1526 25.8 0.10.4 752 12.1 0.10 433 7.3 0.14

Class at birth (6072) (5781)I&II 1060 17.5 0.30 1037 17.9 0.31IIINM 589 9.7 0.15 568 9.8 0.14IIIM 3044 50.1 20.03 2835 49.1 20.04IV&V# 1379 22.7 20.21 1341 23.2 20.18

yn is the total available with data on each exposure and score separately: from a maximum sample of 13, 980 with information on birth-weight, 12,124 alsohad information on maths score; for social class the sample with data on maths score was 11,853.#includes those from single-parent (no male head) households.

RESULTSTrends in mean math Z scores were observed at 7 years

of age by birth weight and by social class, in both boys and girls(Table I). For birth weight, mean Z scores for math increasedfrom those with the lowest to the highest weight at birth: forexample, from 20.21 in boys weighing <2.5 kg to 10.10among those weighing >4 kg. The trend for social class wasan increase in Z score from unskilled manual classes (IV&V)to professional and managerial classes (I&II): for example,from 20.18 to 0.31 in girls (Table I). With few exceptions,home and school factors identified for inclusion in the study,and described earlier, were associated (p < .05) with math Zscores at 7, 11, and 16 years of age and with highest qualifica-tions achieved by 33 years of age (data not presented). Forthree factors (frequency of mother reading, parent-teacherassociation, and school meetings for parents) the direction ofassociation with math score at 7 years of age was consistent withother results, although it failed to reach statistical significance.

Math Skills

The relationship between birth weight and math scoresat 7, 11 and 16 years of age was tested in a multilevel model.Estimates from these models indicate that for a 1-kg increasein birth weight the math result at 7 years of age increased by0.23 of a Z score (intercept in Table II). This birth weighteffect reduced to 0.19 after adjustment for social class at birth.Separate adjustments for each home and school factor had anegligible effect, although adjustment for mother’s level ofinterest in the child’s education also reduced the estimate forbirth weight to 0.19. The modest impact of these home andschool factors suggests that they did not play a major role inexplaining the birth weight effect on math at 7 years of age.

56 Power et al

However, statistically significant (p < .05) interactions on theintercept were found for birth weight by three factors:mother’s and father’s reading and mother’s interest in thechild; effects that were largely unaffected by excluding births<37 weeks (data not shown). We found a greater benefit formother’s reading and her interest in the child’s progress amongchildren with heavier rather than lighter weight at birth. Toillustrate, mother’s reading weekly showed a benefit to mathZ score of 0.25 among the heaviest (>4 kg) but little benefit,0.05, among the lightest (<2.5 kg). No discernable patternemerged for father’s reading. Birth weight was not associatedwith the rate of change in math score over time, as indicatedby the estimate of linear change (Table II).

For social class, Table II shows that there was an effecton math for both the intercept and the rate of linear changeover time. For each unit change in social class (from 1, classesIV&V to 4, classes I&II), there was an increase of 0.19 Z scoremath at the intercept (7 years of age). Adjustment for birthweight did not affect this association. Adjustment for mother’sand father’s level of interest in the child’s education reducedthe estimate for social class moderately (from 0.19 to 0.12 Zscore), whereas adjustment for other home and school factorshad little effect. Interactions (p < .05) on the intercept werefound for social class by two factors (mother’s reading andfather’s interest in the child): for both factors benefits weregreatest in unskilled manual classes (IV&V). For example,the gain in math associated with the mother reading weeklywas 0.10 of a Z score in classes IV&V compared with 20.14in classes I&II (Figure 1).

There was also an effect of social class on the linearchange in math Z score, such that the relationship strength-ened over time (Table II). The estimates for math from themodel equate to a higher Z score in classes I&II of 0.57

The Journal of Pediatrics � January 2006

Table II. Estimates# for intercept and linear change over 3 ages (7, 11 and 16 years) of maths z-scores for (i) birthweight and (ii) social class at birth, adjusted for home and school factors in childhood (age 7)

Maths (z-score)

n = 13,293;

Birth weighty Intercept at age 7 co-efficient (se) Linear change from age 7 (z-score/year)Unadjusted 0.23 (0.02)* 0.003 (0.002)Adjusted for social class 0.19 (0.02)*

Adjusted separately for:Home factors:Frequency of mother reading 0.22 (0.02)*

Frequency of father reading 0.22 (0.02)*

Mothers interest in child’s progress 0.19 (0.02)*

Fathers interest in child’s progress 0.21 (0.02)*

Parents initiative 0.22 (0.02)*

School factors:Parent-teacher association 0.22 (0.02)*

School meetings for parents 0.22 (0.02)*

Social functions for parents 0.22 (0.02)*

Class at birthz n = 12,998;

Unadjusted 0.19 (0.01)* 0.020 (0.001)*

Adjusted for birth weight 0.19 (0.01)* 0.020 (0.001)*

Adjusted separately for:Home factors:Frequency of mother reading 0.18 (0.01)* 0.020 (0.001)*

Frequency of father reading 0.18 (0.01)* 0.020 (0.001)*

Mothers interest in child’s progress 0.12 (0.01)* 0.021 (0.001)*

Fathers interest in child’s progress 0.12 (0.01)* 0.022 (0.001)*

Parents initiative 0.19 (0.01)* 0.020 (0.001)*

School factors:Parent-teacher association 0.19 (0.01)* 0.020 (0.001)*

School meetings for parents 0.19 (0.01)* 0.021 (0.001)*

Social functions for parents 0.16 (0.01)* 0.020 (0.001)*

*p < 0.05.yincludes adjustments for gender, gestational age (32-44 weeks) and exact age of test.zincludes adjustments for gender and exact age of test.#estimates are from a multi-level model with maths z-score as a repeated outcome (3 occasions, 7-16y).;the analysis sample (n) includes participants with information on maths for at least one follow-up (7, 11, 16y) and with information on gender, exact ageof maths test and for birth-weight, gestational age. These are the numbers in the unadjusted analyses. Ns are larger than those for table 1, because for inclusionin the latter, information was required only for the 7y test score. N for unadjusted analyses in this table is the maximum available: Ns reduce with adjustmentfor home and school factors.

relative to classes IV&V at 7 years of age (3 increments acrosssocial classes, ie, 0.19 3 3 = 0.57) and 1.12 at 16 years of age(0.023 9 years3 3 social classes, plus the difference at the in-tercept). With adjustment for each home and school factorthe effect of social class on the linear change in math scorewas little changed. However, there was a significant interac-tion (p < .05) between social class, frequency of mother readingto her child at 7 years of age, and math progress between 7 and16 years of age (Figure 1). Whereas children whose mothersread to them weekly or more had a faster rate of gain inmath score compared with others; this benefit was greatestamong classes I&II with a gain for weekly versus hardly everof 0.31 of a Z score and least among those from classesIV&V for whom the benefit in Z score was only 0.07.

The Influence Of Birth Weight And Socioeconomic Position On CognitiveDevelopment: Does The Early Home And Learning Environment ModifyTheir Effects?

Adult Qualifications

Birth weight was associated with highest qualificationsby 33 years. To illustrate the association using categories ofbirth weight, the proportion of those with the highest qualifi-cations increased from 20.4% in those weighing #2.5 kg atbirth to 31.5% in those weighing >4 kg. From linear regressionanalyses (with birth weight as a continuous variable), we esti-mate that for a 1-kg increase in birth weight there was an in-crease of 0.22 on the 5-point scale of qualifications (Table III).The coefficient for birth weight reduced after adjustment forsocial class at birth, from 0.22 to 0.16. Separate adjustmentsfor each home and school factor had little impact on the asso-ciation between birth weight and highest qualificationsachieved, although modest reductions in the effect were seen

57

after adjustment for mother’s and father’s interest. No interac-tions between birth weight and home and school factors werefound.

Social position at birth was also associated with adultqualifications: for each unit change in social class (across 4points, from IV&V to I&II) there was a 0.4 increase in thequalifications scale (Table III). Thus, qualification levelswere highest among those from class I&II at birth. The socialclass–qualifications association was little affected by adjust-ment for each home and school factor, although reductionsin effect estimates, from 0.40 to 0.33, were seen after adjust-ment for mother’s and father’s interest. These home factorspartly explain the effect of class at birth on qualifications.However, an interaction effect on qualification level was foundfor social class and mother’s interest in her child’s educationalprogress: having a mother rated by the teacher as ‘‘very inter-ested’’ lessened the chances of those from classes IV&V havinglower qualifications (none or <O level) more than it did forthose from classes I&II (Figure 2). Interactions with socialclass were also found for frequency of mother’s reading andfor father’s interest in the child (patterns were similar to Fig-ure 2), suggesting that parental interest and maternal readingcan to some extent buffer the effect on adult qualifications oflower class background in early life. No interactive effects wereobserved for other home and school factors.

DISCUSSIONIn this British population, birth weight and social class

of origin both influenced subsequent cognitive development,with the lowest birth weights and those from unskilled manualbackgrounds having the poorest outcomes. An influence ofbirth weight at 7 years of age persisted throughout childhood,although it did not affect the person’s change in cognitive

Figure 1. Trajectories# of maths z-scores (7-16 years) according tosocial class at birth and frequency of mother reading to her child at 7years. # Based on estimates from a multi-level model which includesan interaction of social class at birth by frequency of mother reading toher child by age (covariates are gender and exact age of test). Read =mother reads at least weekly; read less = mothers read less frequentlythan ‘‘at least weekly’’.

58 Power et al

score; whereas, the social class effect strengthened over theperiod of follow-up. Both birth weight and social class wereassociated with adult qualifications. As expected, we foundthat the home and learning environment influenced cognitivedevelopment. Figures 1 and 2 illustrate these effects for themother’s frequency of reading and teachers assessments of herinterest in the child’s educational progress at 7 years of age.Although further details are not presented, our findings areconsistent with the literature suggesting that parental invest-ment in the child’s learning environment has a positive impacton cognitive test scores.2,16,17

There are several ways in which this article adds impor-tant information to previous work. First, for birth weight, weshow that its effect on measures of cognition at 7 years of agedid not appear to be a result of differences in the home andlearning environment. There was, however, some evidenceto suggest that frequency of parental reading increased math

Table III. Linear regression coefficients (StandardError) for effects of birth weight and social class oforigin on highest qualifications achieved by 33 years,adjusted for home and school factors in childhood(age 7)

Birth weight (kg)y

(n = 9024;)co-efficient (SE)

Social classz

(n = 8825;)co-efficient (SE)

UnadjustedAdjusted for social class 0.22 (0.03)* 0.40 (0.01)*

Adjusted for birth weight 0.16 (0.03)*

Adjusted separately for: 0.40 (0.01)*

Home factors:Frequency of

mother reading0.21 (0.03)* 0.38 (0.01)*

Frequency offather reading

0.21 (0.03)* 0.38 (0.01)*

Mothers interest inchild’s progress

0.17 (0.03)* 0.33 (0.01)*

Fathers interest inchild’s progress

0.19 (0.04)* 0.33 (0.02)*

Parents initiative 0.22 (0.03)* 0.38 (0.01)*

School factors:Parent-teacher

association0.22 (0.03)* 0.39 (0.01)*

School meetingsfor parents

0.22 (0.03)* 0.40 (0.01)*

Social functionsfor parents

0.22 (0.03)* 0.40 (0.01)*

Home and school factors baseline categories: mother and father reading =infrequently; mother and father interest = little/some or over; parent’sinitiative = did not discuss progress; parent-teacher association, schoolmeetings and functions = none.*p < 0.05.yadjusted for gender and gestational age.zadjusted for gender with social class as a continuous variable (1 = IV&V).;numbers vary: for ‘fathers interest’ teachers often gave no information(n = 5300 in birth weight; 5182 in social class analyses).

The Journal of Pediatrics � January 2006

(and reading) scores at 7 years of age, less so for those with alower weight at birth than for others. This finding is counter toour expectation at the outset that those with lower weight atbirth would gain more with exposure to cognitive stimulation.Second, for social class, part of the association with cognitivemeasures at 7 years of age and also with qualifications achievedby adulthood may be because of levels of parental interest ineducation. Yet surprisingly, parental interest in educationdid not contribute to the influence that social class had onthe change in cognitive scores over time. Also noteworthywas the lack of any substantial influence of other home andschool factors examined here, on the association betweensocial class and cognitive outcome.

Third, by examining both the level at 7 years of age andchange to 16years of age in cognitive scores, our study allowsus to better understand the influences of family factors in dif-ferent social classes. In particular, mother’s reading to the childinfluences the level and change in math scores differentlyacross the social classes. Mother’s reading had increasing ben-efits for classes I & II and III nonmanual between 7 years ofage and 16 years of age but much less so for manual socialgroups. Paradoxically, maternal reading showed greater bene-fits for manual than nonmanual social classes in the child’smath score initially at 7 years of age. Finally, our findings sug-gest that high levels of parental interest and greater frequencyof maternal reading to the child can to some extent buffer theeffect on adult qualifications of lower-class background inearly life. These findings for social class contrast with thosefor low birth weight for which no buffering effects were ob-served. Identification of factors that can ameliorate the effectsof early life adversity was a major objective of this article be-cause such factors can have important implications for policy,and they can inform research that seeks to identify causalprocesses through which health risk in adulthood becomesestablished.

Strengths and Limitations

Two main limitations to the study need to be acknowl-edged. First, the factors used to represent home and schoolenvironments may fail to adequately capture the underlyingfactors that impact on the relationships with cognitivedevelopment for birth weight and socioeconomic position.Moreover, the teacher’s assessment of parental interest inthe educational progress of the child at 7 years of age may inpart reflect the teacher’s knowledge of the child’s readingand math skills. Thus, the teacher’s perception of parental in-terest may be a surrogate marker for cognitive ability. Second,several of the home and school factors examined, such as thefrequency with which parents read to their child, have changeddramatically over historical time and, correspondingly, themeaning of these characteristics may have altered. Amongthe strengths of this study are the prospective data from birthonward in a representative population-based sample,18 withinformation on reading and math trajectories as core basicskills in modern societies. As in our previous study, we exam-ined cognition across the range of birth weight with adjust-ment for gestational age,4 thereby providing a measure of

The Influence Of Birth Weight And Socioeconomic Position On CognitiveDevelopment: Does The Early Home And Learning Environment ModifyTheir Effects?

fetal growth. The present study has the additional strengththat it addresses some of the complexity of how family andschool factors can modify the effects of birth weight andsocioeconomic position on cognitive development.

Birth Weight, Social Class, and Cognitive Outcome:The Role of Home and School Factors

The cognitive deficits of low birth weight and prematureinfants are widely recognized and research programs have beeninitiated to establish how such deficits might be reduced.19

Whereas early research focussed primarily on cognitive out-come in the low weight group,20 recent studies have showneffects across the entire birth weight distribution.1 One reviewhas commented on the differential effect of early environmentson low and normal birth weight infants, highlighting theimportance of assessing interactions between birth weightand early environmental circumstances.21 Likewise, the im-portance of developmental trajectories has been illustrated ina lower birth weight cohort, among whom birth weight influ-enced mental development at 3 months of age but not thechange from 3 to 30 months.22 This finding is consistentwith our results, but at an earlier age.

Unlike our study and with few exceptions,23 moststudies have not considered how birth weight (across the fulldistribution) and early environmental circumstances (socio-economic position, home and school factors) might act incombination to influence cognitive trajectories throughoutchildhood. Nonetheless, our findings agree with the literatureshowing consistent social class effects on both math and read-ing, and with long-term associations with cognitive measureslater in childhood and adulthood. A biological manifesta-tion of this finding comes from a study measuring head cir-cumference (which indicates brain growth) in the prenatalperiod and postnatally up to 9 years of age, showing important

Figure 2. Interaction between social class at birth and mother’s in-terest# in her child’s educational progress at age 7, and qualificationsachieved by age 33y. # Mother’s interest assessed by the teacher: lessinterested (includes ‘some’ and ‘little’ interest and ‘over concerned’)and very interested. Of the five categories of qualifications used inanalyses only 1 group is included in the figure. Lower qualificationscombines the two lowest categories, ‘none’ and ‘<O level’. Based ondata rather than estimates from statistical models, with n = 7521.

59

effects of the early postnatal environment on cognitive devel-opment.24 Between 9 months and 9 years of age, head growthwas poorer in children whose mothers were from unskilledmanual social classes and, in turn, greater head growth was re-lated to higher IQ scores at 9 years of age. Our study showsthat effects of early social origins persist beyond childhoodto influence educational qualifications in adulthood, withthose from nonmanual origins having a long-term benefit.This finding is consistent with other studies demonstratinglong-lasting benefits of social origins to cognition; in some in-stances still evident during the sixth and seventh decade of lifefor those with more favorable socioeconomic circumstances inchildhood.25,26 Because of the persistence of early social envi-ronment, studying the pathways through which origins affectcognitive outcome is important. Previous studies have shown,for example, how socioeconomic disadvantage influences levelsof cognitive stimulation in the home and parenting style, whichin turn affect cognitive development of offspring.2,3,27 Our re-sults are consistent with this evidence. Importantly, our find-ings also suggest that stimulating home environments (asrepresented here by parental reading and interest in their child’seducational progress) afford greater benefits, in general, to cog-nitive outcome among those from unskilled manual groups.

Implications for Later Health and for Policy

There are at least three ways whereby early cognitioncould influence adult health. First, it affects social trajectoriesand locations in adulthood that, in turn, confer health (dis)ad-vantage. Second, cognitive trajectories are likely to influencereceptivity to health promotion and positive behaviors, andmay contribute to the well-established associations betweeneducation level and health behaviors. Third, cognitive trajec-tories may be a marker for central nervous system developmentthat may affect perception and regulation of stress reactivity,which in turn directly influence adult health.28 Recent studiesconfirm that childhood cognition predicts mortality.29,30 Inthe 1958 cohort, childhood ability and influences on cogni-tion, notably parental interest in the child’s education and fre-quency of reading, appear to contribute to social inequalitiesin adult psychological distress31 and self-rated health.32 Asothers have argued, parental interest in the child’s educationis likely to be influenced by parents’ own school experiences,thereby forming a transmission mechanism of (dis)advantageacross generations.33 In this context it is important to notethat all home and school factors examined in our study, aswell as social class, were related to cognitive outcome. Thissupports the view that there are many potential modifiablecharacteristics of the early environment that may have a life-long impact on health as well as cognitive status. Thus the em-phasis of recent policy initiatives across Western countries toimprove the conditions of early childhood is promising.

Data acknowledgement: Centre for Longitudinal Studies, Institute ofEducation. National Child Development Survey Composite File in-cluding selected Perinatal Data and sweeps one to five [computer file].National Birthday Trust Fund, National Children’s Bureau, CityUniversity, Social Statistics Research Unit (original data producers).The Data Archive distributor, Colchester, Essex. SN:3148. 1994.

60 Power et al

REFERENCES1. Shenkin SD, Starr JM, Deary IJ. Birth weight and cognitive ability in

childhood: a systematic review. Psychol. Bull 2004;130:989-1013.

2. Yeung WJ, Linver MR, Brooks-Gunn J. How money matters for young

children’s development: parental investment and family processes. Child Dev

2002;73:1861-79.

3. Guo G, Harris KM. The mechanisms mediating the effects of poverty

on children’s intellectual development. Demography 2000;37:431-47.

4. Jefferis BJ, Power C, Hertzman C. Birth weight, childhood socioeco-

nomic environment, and cognitive development in the 1958 British birth

cohort study. BMJ 2002;325:305-10.

5. Stanton WR, McGee R, Silva PA. Indices of perinatal complications,

family background, child rearing, and health as predictors of early cognitive

and motor development. Pediatrics 1991;88:954-9.

6. Chao RK, Willms JD. The effects of parenting practices on children’s

outcomes. In: Willms JD, ed. Vulnerable Children: Canada’s National Lon-

gitudinal Survey of Children and Youth. Edmonton: University of Alberta

Press; 2002. p. 149-65.

7. Hart B, Risley TR. Meaningful Differences in the Everyday experience

of Young American Children. London: Paul.H.Brookes Publishing Com-

pany; 1995.

8. SchweinhartLJ, BarnesHV,WeikartDP. Significant benefits: theHigh/

Scope Perry preschool study through age 27. Monographs of the High/Scope

Educational Research Foundation; 1993;10. Ypsilanti High/Scope Press.

9. Ramey CT, Ramey SL. Which children benefit the most from early

intervention? Pediatrics 1994;94:1064-6.

10. Kohen D, Hertzman C, Willms JD. The importance of quality child-

care. In: Willms JD, ed. Vulnerable Children: Canada’s National Longitudi-

nal Survey of Children and Youth. Edmonton: University of Alberta Press;

2002. p. 261-76.

11. Werner E, Smith RS. Vulnerable but Invincible: Longitudinal Study of

Resilient Children and Youth. London: McGraw Hill Book Company; 1982.

12. Kohen D, Hertzman C, Brooks-Gunn J. Neighbourhood effluents and

school readiness. Educational Quarterly Review 1999;6:44-52.

13. Kohen DE, Brooks-Gunn J, Leventhal T, Hertzman C. Neighborhood

income and physical and social disorder in Canada: associations with young

children’s competencies. Child Dev 2002;73:1844-60.

14. Butler NR, Alberman E. Perinatal Problems. The Second Report of the

British Perinatal Mortality Survey. Edinburgh and London: E & S Living-

stone Ltd; 1969.

15. Ferri E, ed. Life at 33 The Fifth Follow-up of the National Child

Development Study. London: National Children’s Bureau; 1993.

16. Levenstein P, O’Hara J, Madden J. The mother-child home programme

of the verbal interaction project. Consortium for Longitudinal Studies, As the

Twig is Bent. New Jersey: Lawrence Erlbaum Associates; 1983. p. 237-63.

17. Bromwich RM, Parmelee AH. An intervention programme for

pre-term infants. In: Field TM, Sosteck AM, Goldberg S, Shumann HH,

eds. Infants Born at Risk: Behaviour and Development. New York: Medical

and Scientific Books; 1979. p. 389-411.

18. Goldstein H. A study of the response rates of 16-year-olds in the

National Child Development Study. In: Fogelman K, ed. Growing up in

Great Britain. London and Basingstoke: Macmillan Press; 1983.

19. Berlin LJ, Brooks-Gunn J, McCarton C, McCormick MC. The

effectiveness of early intervention: examining risk factors and pathways to

enhanced development. Prev Med 1998;27:238-45.

20. Bhutta AT, Cleves MA, Casey PH, Cradock MM, Anand KJ. Cogni-

tive and behavioral outcomes of school-aged children who were born preterm:

a meta-analysis. JAMA 2002;288:728-37.

21. Grantham-McGregor SM. Small for gestational age, term babies, in the

first six years of life. Eur J Clin Nutr 1998;52(suppl 1):S59-64.

22. Carter RL, Resnick MB, Ariet M, Shieh G, Vonesh EF. A random

coefficient growth curve analysis of mental development in low-birth-weight

infants. Stat Med 1992;11:243-56.

23. Stanton WR, McGee RO, Silva PA. A longitudinal study of the inter-

active effects of perinatal complications and early family adversity on cognitive

ability. Aust Paediatr J 1989;25:130-3.

The Journal of Pediatrics � January 2006

24. Gale CR, O’Callaghan FJ, Godfrey KM, Law CM, Martyn CN. Crit-

ical periods of brain growth and cognitive function in children. Brain 2004;

127:321-9.

25. Kaplan GA, Turrell G, Lynch JW, Everson SA, Helkala EL, Salonen

JT. Childhood socioeconomic position and cognitive function in adulthood.

Int J Epidemiol 2001;30:256-63.

26. Richards M, Wadsworth MEJ. Long term effects of early adversity on

cognitive function. Arch Dis Child 2004;89:922-7.

27. Baharudin R, Luster T. Factors related to the quality of the home envi-

ronment and children’s achievement. J Family Issues 1998;19:375-403.

28. Hertzman C, Frank J. Biological pathways linking the social environ-

ment, development and health. In: Heymann SJ, Hertzman C, Barer ML,

Evans RG, eds. Creating Healthier Societies. New York: OUP; 2005.

29. Whalley LJ, Deary IJ. Longitudinal cohort study of childhood IQ and

survival up to age 76. BMJ 2001;322:819-24.

The Influence Of Birth Weight And Socioeconomic Position On CognitiveDevelopment: Does The Early Home And Learning Environment ModifyTheir Effects?

30. Osler M, Andersen AM, Due P, Lund R, Damsgaard MT, Holstein

BE. Socioeconomic position in early life, birth weight, childhood cognitive

function, and adult mortality: a longitudinal study of Danish men born in

1953. J Epidemiol Community Health 2003;57:681-6.

31. Power C, Stansfeld SA,Matthews S, Manor O, Hope S. Childhood and

adulthood risk factors for socio-economic differentials in psychological dis-

tress: evidence from the 1958 British birth cohort. Soc Sci Med 2002;55:

1989-2004.

32. Hertzman C, Power C, Matthews S, Manor O. Using an interactive

framework of society and lifecourse to explain self-rated health in early adult-

hood. Soc Sci Med 2001;53:1575-85.

33. Due P, Lynch J, Holstein B, Modvig J. Socioeconomic health inequal-

ities among a nationally representative sample of Danish adolescents: the

role of different types of social relations. J EpidemiolCommunityHealth 2003;

57:692-8.

61