Is sub-nutrition necessary for a poor outcome following early institutional deprivation?


* Correspondence to first author at School of Psychology, University of Southampton, Southampton SO17 1BJ, UK.


Institutional deprivation is multifaceted and includes adverse psychosocial and nutrition-related components. In this study we partitioned these risks in relation to cognitive impairment and mental ill health, and explored the mediating role of reduced head/brain size. There were 138 participants (61 males, 77 females) in the study. Participants were Romanian adoptees who had experienced at least 2 weeks of early institutional deprivation. The sample was stratified on the basis of duration of deprivation (high risk >6mo in institutions) and sub-nutrition (i.e. 1.5 SD below UK age-related norms for weight at UK entry). UK children adopted before 6 months of age and a group of non-institutionally deprived Romanian children constituted the comparison groups. Duration of deprivation was associated with smaller head circumference, lowered IQ, and increased mental heath problems, independently of effects found for sub-nutrition on head circumference and IQ. The mediating role of head circumference was limited to either sub-nourished (IQ) or non-sub-nourished (inattention/overactivity and disinhibited attachment) subgroups. Many negative effects of early deprivation, including stunted brain growth, occur without sub-nutrition: psychosocial deprivation plays a major role in neurodevelopmental effects of deprivation. Further studies of functional and structural neuroanatomy following institutional deprivation are required to delineate the role of brain development in its effects.

The negative effects of early, profound institutional deprivation on development and mental health persist into later life despite radical improvements in social environments following adoption.1,2 Compelling evidence for this comes from the English and Romanian Adoptees (ERA) study which examined the outcomes for a cohort of children who spent their early infancy living in the extremely deprived conditions of the state institutions of the Ceauşescu regime in Romania at the end of the 1980s, and who were subsequently adopted into UK families before 3 years 7 months of age. Despite a remarkable degree of recovery for even some of the most deprived children,1,3 residual intellectual impairment, inattention/overactivity, disinhibited attachment, and quasi-autistic symptoms persisted in a substantial minority, at least up to the age of 11 years.1,2 The persistence of such effects, despite the radical and positive change in the children’s circumstances, suggests that they are mediated by fundamental neurobiological changes;4 a view consistent with the long-term pattern of reduced brain size in Romanian adoptees.4,5

These children were exposed to multiple putative risks in their institutions. Nutritional risk, associated with the very poor quality and quantity of food, was reflected in the large proportion of children with sub-nutrition on entry into the UK.6 Psychosocial risks, associated with social/emotional (e.g. low levels of social contact and emotional support, etc.) and intellectual deprivation (e.g. lack of stimulation because of extended confinement to cots, etc.) were, if anything, even more marked and pervasive.7 In the light of this, the primary aim of this paper was to partition overall institutional deprivation-related risks on cognitive, mental health, and brain-related outcomes, into sub-nutrition-related components as indexed by weight at entry to the UK and psychosocial components – indexed by duration of deprivation.

Our hypotheses were based on the clinical and experimental literature suggesting that both physical and psychosocial components of early deprivation are likely to influence negative outcomes. Sub-nutrition inhibits brain growth and development during the early years8–12 and is associated with both intellectual impairments and general mental health problems.13,14 The cognitive and behavioural effects of psychosocial components of deprivation are also well documented in clinical studies of patient populations4,15,16 and animal models17–19 in which environmental impoverishment is associated with brain size reductions, structure, and function, even when nutrition is adequate.

We hypothesized that both components of deprivation would increase risk across a broad range of outcomes following early institutional care. More specifically, this would mean that whereas sub-nutrition is expected to play an important role in deprivation-related effects, there would also be effects of deprivation duration on all outcomes, even in children who were not sub-nourished. Because sub-nutrition, even that lasting only a few months, may be expected to have major effects on body size (as indexed by height and weight), there is likely to be a concomitant impairment in brain growth (indexed by head circumference). By contrast, the effects of psychosocial deprivation on brain growth are not likely to operate via overall body size. Accordingly, we predicted that effects on brain growth would take longer to operate in the absence of sub-nutrition.

The children in our main sample all experienced some degree of deprivation in the Romanian institutions. However, crucially for the purposes of the current analysis, they differed greatly from one another in terms of the duration of deprivation they had experienced, with the length of time before entry into the UK ranging from 2 weeks to 3 years 7 months. Our published findings on outcomes at age 11,1,2 suggest that there is a threshold of risk associated with duration of deprivation operating in the first year. While children with less than 6 months deprivation appeared at no greater risk of poor outcomes than non-institutionalized UK adoptees, those with between 6 months and 3 years 7 months duration were at a substantially increased risk, though this risk did not increase with increasing time in the institutions within this group. In the absence of direct measures of nutritional care in the institutions or specific biological markers, in the current paper we estimated nutritional status on the basis of weight at the time of entry into UK indexed against population norms.

While sub-nutrition was extreme in many cases and was common in the sample, it was by no means universal; a substantial number of children, even among those suffering the longest periods of deprivation, had weights that were in the normal range and well above the cut-off used in previous studies.20,21 This indicates that these children at least received adequate calorific intake, although details were not available of their full nutritional status. The sample was, therefore, well placed to examine the effects of the contribution of duration of deprivation and sub-nutrition independently of one another and, furthermore, to establish whether in concert, any effects act additively or synergistically.

The question that follows, if we found that duration of deprivation and sub-nutrition were related to both head circumference and intellectual/mental health outcomes, as we predicted, is whether the postulated differences between sub-nutrition and psychosocial deprivation in the timing of effects on overall brain growth would have implications for psychological outcomes. If brain growth was a mediator of these broader outcomes, there might be long-term sequelae in sub-nourished children who left institutional care before the age of 6 months, but no such sequelae in children with similar duration of deprivation who were not sub-nourished. If institutional deprivation, independent of sub-nutrition, has effects on brain growth only when the deprivation lasts longer than 6 months, the question is whether impaired brain growth mediates negative cognitive and psychological effects of deprivation in this group also. In that regard, it is likely that sub-nutrition could operate mainly through effects on brain growth, whereas there is a broader range of possible mediators in the case of psychosocial deprivation. On this basis it might be predicted that brain growth is a stronger mediator in the case of sub-nutrition than where there is psychosocial deprivation in the absence of sub-nutrition.



Details of our sampling frame are reported in previous publications (cf.1). The sample was drawn from 324 children adopted into UK families between February 1990 and September 1992, who were processed through the UK Home Office and/or Department of Health. The initial sample comprised 165 children adopted from Romania before the age of 43 months. The current report focuses on those children who experienced institutional care (n=138: 61 males, 77 females). This was defined in terms of care lasting at least 2 weeks but, in fact, the vast majority had been in institutions for most of their lives by the time they were adopted.7 There were roughly equivalent numbers of children (around 50) adopted before 6 months, between 6 months and 2 years, and more than 2 years. A group of UK children adopted before the age of 6 months were used for comparison in some analyses. There was also a group of 21 Romanian adoptees who had not experienced institutional deprivation according to this definition being; adopted directly from families or after a short stay in an institution. This small group provides a useful control against which we can judge whether the weight of the non-sub-nourished children was equivalent to those that had not experienced institutional deprivation. The agreement to participate for all Romanian children approached was roughly 80%. As the UK sample was a volunteer sample it was harder to estimate the proportion who volunteered, but it was believed to be about 50%.


Outcomes at age 11

 Duration of deprivation. Age on arrival in the UK after leaving the institution was taken as a proxy for duration of deprivation. On the basis of published age 11 outcomes,1,2 which demonstrated a step increase in risk around 6 months, the sample was divided into those who experienced less than 6 months’ deprivation (n=44) and those with more (n=94).

Sub-nutrition. In the absence of more direct indications, weight recorded at the time of entry into the UK was taken as an estimate of overall nutritional status during the institutional period. Children were judged not to be sub-nourished if their weight was no more than 1.5SDs below contemporary weight norms for the UK (n=86). Although somewhat arbitrary, this is a more conservative threshold for defining the non-sub-nourished group than used in earlier studies.20,21 This threshold was chosen because it was crucial for us to identify a group with weight that we were confident was within the normal range. All physical measures were converted into standardized scores based on the British Growth programme based on population norms.22 This metric provides a continuous standardized measure of physical development in terms of SDs above or below the standardization mean for the age and sex. While there may be differences between UK and Romanian ‘norms’, these are not documented and there is no evidence to suggest that if they did exist these can be attributed to any differences in genetic growth patterns. Such differences found between populations within Europe are largely attributable to differences in environment and, therefore, it would be anticipated that under similar environmental conditions there would be no differences in growth patterns.23 This approach enables us to ascertain how much the earlier depriving environment has impeded growth despite the move to an enriched environment.

Brain growth. Head circumference, measured at entry to UK at ages 6 and 11 years, was used as a proxy for brain size. Measurements were replicated three times at each follow-up to ensure accuracy and the result was converted into a continuous standardized measure using contemporary UK norms.22 The use of head circumference as a proxy for brain size is supported by a number of imaging studies.24,25 In a subsample of 14 ERA children, magnetic resonance imaging (MRI) demonstrated a substantial correlation between brain size and head circumference (r=0.74; p<0.0015).

Weight. Weight was measured at ages 6 and 11 years and standardized against UK norms as above.

Pre- and perinatal risk

Birthweight was available for the majority of children and was employed as a measure of prenatal growth-related risk. Children of 2500g or less at birth were designated as low birthweight (n=41). Although a number of the children were believed to be preterm the exact gestational dates were not known. There were three sets of twins in the study from Romania and one set in the UK sample. All of the twins, with the exception of one child, had birthweights below 2500g.

Developmental outcomes

 Intelligence. A short form of the Wechsler Intelligence Scale for Children (WISC-III – UK) was completed at age 11.26 Four subscales of the WISC-III (vocabulary, similarities, block design, object assembly; reliability coefficient = 0.9427) were prorated to form an IQ score. Three children were unable to complete the WISC assessment due to severe cognitive impairment and were allocated a score between 40 and 50 based on other assessments (British Picture Vocabulary Scale28 and Raven’s matrices;29 see1 for further details).

Mental health. The broad-based assessment of behavioural outcomes at age 11 and their psychometric properties is described in previous papers (cf.2). The measures involved in the current analyses were those showing a specific and persistent association with institutional deprivation at ages 4, 6, and 11 years.3,30,33 Disinhibited attachment was assessed from three items from the parental interview (definite lack of differentiation between adults; clear indication that the child would readily go off with a stranger; and definite lack of checking back with the parent in anxiety-provoking situations).34 For the purposes of the current analysis quasi-autistic features were measured using the Social Communication Questionnaire (SCQ; formerly the Autism Screening Questionnaire).35 In addition, 16 children in the sample met diagnostic criteria for autism at age 11 following a more thorough clinical assessment of potential cases. Inattention/overactivity was measured using the relevant items from the Rutter parent and teacher scale, which were combined for the purposes of the current analysis.36 For all mental health measures high scores represent more problems.


The study received ethical approval from the local research ethics committee and informed consent and assent were obtained from the parents and children respectively. Families were visited on separate occasions by different researchers. The first visit involved an intensive interview with the primary caregiver, and a set of behavioural and family relationship questionnaires. During the second visit the developmental assessment was carried out.

Statistical analysis

The Romanian adoptees were divided into four groups on the basis of duration of deprivation (<6mo vs >6mo) and weight at entry (>1.5SDs below norms vs <1.5SDs below norms) for the purposes of analysis.

There were four analytical phases. In phase 1 we explored the validity of the 1.5SD threshold for defining sub-nutrition in order to ensure that those above that threshold were not sub-nourished. We were especially concerned to show that: (1) children in the non-sub-nourished group were close to the UK average according to national norms; (2) they were not lighter at entry than the non-institutionally deprived Romanian children; and (3) their weight did not differ as a function of whether they experienced more or less than 6 months deprivation.

Phase 2 examined the trajectories of growth and physical catch-up as a function of sub-nutrition and duration of deprivation by comparing the different sub-groups of institutionally deprived children with UK norms and UK adoptees using single sample and independent sample t-tests.

In phase 3, analysis of variance was used to assess the independent effects of duration of deprivation and sub-nutrition on cognitive and mental health outcomes and head circumference. Where there was an association with sex and/or birthweight these factors were included as covariates.

In phase 4, mediational analyses were performed to examine whether brain size, as indexed by head circumference at age 6, mediated the effects of duration of deprivation and/or sub-nutrition on outcomes measured at 11 years. In order to allow a longitudinal analysis, head circumference at age 6 was chosen as the mediator variable because it was measured after the experience of deprivation and the measurement of sub-nutrition but before outcomes at age 11. The standard regression approach recommended by Barron and Kenny,37 was employed. First, the univariate associations between sub-nutrition/or deprivation, outcomes at age 11, and head circumference at 6 were established using univariate regression. Second, a stepwise multiple regression was run to test for the effects of controlling head circumference at age 6 on the association between duration of deprivation and/or sub-nutrition and outcomes at age 11. A mediated effect was supported if a significant association between duration of deprivation and/or sub-nutrition and outcomes at 11 years (step 1) was reduced to non-significant levels when head circumference was added to the model (step 2). For all analyses except the mediational analyses, the significance levels for multiple testing were corrected using Bonferroni’s formulae.


Did the threshold for sub-nutrition identify a non-sub-nourished group of children?

The weight cut-off succeeded in picking out a group without sub-nutrition (Table I) with the average weight on arrival of those above it close to population norms; also the sub-nutrition group was indeed severely sub-nourished. In fact the mean weight at entry in our non-sub-nourished group of institutionalized children was marginally higher than for the sample of 21 non-institutionally deprived Romanian children adopted directly from families (t(69)=2.01; p=0.048). Crucially there was no difference in weight at entry for the non-sub-nourished children who experienced more than or less than 6 months of deprivation (t(48)=1.08; p=0.28): both these groups could legitimately be described as equally non-sub-nourished.

Table I.   Effects of sub-nutrition and length of deprivation on physical measures at entry, 6, and 11 years, and IQ and mental health outcomes at 11 years
 UK Adoptees
> 6 months deprivationRomanian adoptees
NoYesAnalysis of variance results
dfSub-nutrition FDuration FInteraction F
  1. Physical measures are standardized against UK norms. p values in italics are significant after correction for multiple testing based on the number of tests per subset of analyses (e.g. three for head circumference and weight and four for age 11 outcomes). Bold italic F values indicate significant effects after controlling for multiple tests. Variations in df are due to case-wise deletion of missing data. HC, head circumference; DA, disinhibited attachment; I/OA, inattention/overactivity; df, degrees of freedom.

Weight at entry- −0.33 (0.96)−3.32 (1.20)−0.60 (0.77)−3.52 (2.10)1,13797.74
Weight at 6y (SD)0.30 (0.90) 0.54 (0.72)−0.36 (0.65)−0.22 (0.86)−0.76 (1.02)1,12917.53
Weight at 11y (SD)0.44 (0.93) 1.14 (0.87)−0.28 (0.92)0.18 (1.01)−0.36 (1.15)1,12322.58
HC at entry (SD)- −0.74 (1.75)−2.74 (1.73)−2.49 (1.37)−3.01 (1.45)1,13318.65
HC at 6y (SD)−0.57 (0.80) −0.65 (0.98) −1.45 (1.01)−1.50 (0.78)−2.03 (1.12)1,12811.64
HC at 11y (SD)−0.24 (0.95) 0.29 (1.07)−0.87 (1.04)−0.69 (1.36)−1.61 (1.33)1,12317.80
IQ at 11y (SD) 105.08 (15.59)  111.88 (15.89)93.76 (15.90)85.64 (17.76)83.48 (15.83)1,1229.58
I/OA at 11y (SD)0.40 (0.49) 0.33 (0.43)0.55 (0.38)0.65 (0.44)0.89 (0.60)1,1315.65
DA at 11y (SD)0.28 (0.85) 0.39 (1.14)0.11 (0.33)1.44 (1.66)1.43 (1.74)1,1340.24
Quasi-autsim at 11y3.16 (3.37) 3.06 (3.11)3.83 (2.97)6.04 (5.67)6.03 (4.32)1,1230.21

Growth trajectories in comparison to UK norms

As predicted, one sample t-tests comparing Romanian adoptees scores with UK norms demonstrated that in the under 6-month group there was substantially reduced head circumference at entry in the sub-nourished group (t(30)=10.12; p<0.001) but not in those with a normal weight at entry (t(16)=1.74; p=0.10; Table I). In striking contrast, in those entering the UK over the age of 6 months, there were equivalent effects in the sub-nourished and normal weight groups with both groups differing significantly from the UK norms (t(25)=8.05; p<0.001; t(62)=16.45; p<0.001 respectively). Table I also illustrates the differential catch-up for weight and head circumference up to the age of 11 years. For those with sub-nutrition entering the UK before 6 months, the catch-up in weight was virtually complete by age 6 (one sampled t-test vs UK norms; t(25)=2.77; p=0.01) and a comparison of weight at 6 and 11 years suggested that there was no further catch-up during this period (t(23)=0.56; p=0.58). By contrast, head circumference was still significantly below UK population norms at 6 years (t(25)=7.32; p<0.001), with further catch-up by 11 (t(25)= 3.68; p=0.001). The pattern for the sub-nourished children entering after 6 months was very similar. By age 6, their weight had caught up considerably (t(60)=10.20; p<0.001), with residual catch-up by age 11 (t(57)=4.19; p<0.001). Head circumference showed much less catch-up; by 11 years children were still significantly below the norm (t(57)=9.24; p<0.0001). The group with normal weight at entry, entering after the age of 6 months, had no need for catch-up in weight but their was significant catch-up in head circumference between entry and 6 years (t(27)=3.75; p=0.001) and further catch-up by 11 (t(25)=3.14; p = 0.004).

Are there independent effects of duration of deprivation and sub-nutrition on cognition, mental health, and brain growth?

The next question is whether sub-nutrition in those entering under 6 months had an effect on intellectual and mental health outcomes. First, there was no difference between the UK comparison group and those without sub-nutrition in this group on any outcomes (t(65–69)s<0.15; ps>0.13). There were differences, however, between the sub-nourished and non-sub-nourished groups in the under 6-month group with respect to IQ (t(50)=3.56; p=0.001), but not the other outcomes (t(39–41)s <1.78; ps >0.08). There was no association between any outcome and birthweight (rs<0.12; ps>0.2). There was an effect of sex on IQ – with females having lower IQ than males (t(128)=2.40; p<0.05; 86.14 (17.10) vs 93.78 (19.19)). There was no effect of sex on other outcomes at age 11 (ts(128–136)<1.35; ps>0.18).

There were large and pervasive effects of duration of deprivation across all outcomes at 11 independent of sub-nutrition status (Table I). Substantial effects of sub-nutrition were restricted to IQ while there was a smaller effect on inattention/overactivity that became non-significant when multiple testing was taken into account; duration of deprivation and sub-nutrition operated largely additively, although the interaction between these two factors with regard to IQ and head circumference at entry were significant before correcting for multiple tests. The effects of sub-nutrition on disinhibited attachment and autistic symptoms were not significant.

Do the effects of brain growth mediate the effects of duration of deprivation and/or sub-nutrition on cognitive and mental health outcomes?

Because there was evidence that the effects of duration of deprivation on head circumference was moderated by sub-nutritional status, and given the difficulty of interpreting mediational analyses in such a situation, we ran the analyses for the sample as a whole and for the sub- and non-sub-nourished samples separately.38

Full sample

Univariate tests demonstrated a small but significant association between age 6 head circumference and IQ and all mental health outcomes at age 11, and confirmed the predicted pattern of direct associations between duration of deprivation and sub-nutrition to these outcomes (Table II). Univariate regression also confirmed that both predictors were associated with head circumference at age 6 (sub-nutrition β=–0.35; p<0.001; duration of deprivation β=0.27; p=0.001). Introducing age 6 head circumference into the multiple regressions, along with the predictors (the final step in the mediational analysis), did not, on the whole, substantially alter their significant direct effects on outcome. The largest reduction (0.12 of the β weight) being in relation to the effect of duration of deprivation on IQ and the shift from significance to non-significance was for the effects of sub-nutrition on IQ.

Table II.   Regression models exploring the mediating effect of head circumference at age 6 years on outcomes for full sample and for sub-nourished and non-sub-nourished samples
Age 11 OutcomePredictorFull sampleSub-nourishedNon-sub-nourished
  1. aChanges between univariate and multivariate regression suggestive of mediational effect for head circumference. Given the exploratory nature of these analyses. No correction was made for multiple tests. β represent a standard coefficient indicating the amount of variance in the outcome accounted for by the predictor. I/OA, inattention/overactivity; DA, disinhibited attachment; HC, head circumference. Figures in brackets represent degree of freedom for the overall multiple regression model.

IQ  F=13.51 (3,118) F=9.45 (2,78) F=9.32 (2,38)
 Duration of Deprivationβ=0.42
 HC at 6yβ=0.40
I/OA  F=8.07 (2,128) F=4.46 (2,79) F=3.79 (2,43)
 Duration of Deprivationβ=0.32
 HC at 6yβ=0.18
DA  F=9.19 (2,130) F=7.48 (1,81) F=5.23 (2,43)
 Duration of Deprivationβ=0.34
 HC at 6yβ=0.25
Quasi-Autism  F=6.09 (2,122) F=2.85 (2,75) F=2.17 (2,41)
 Duration of deprivationβ=0.28
 HC at 6yβ=−0.21
β=−0.20 p=0.180β=−.010

Sub-nourished and non-sub-nourished sub-samples

The results of the various mediational analyses are reported in Table II. In the non-sub-nourished group, duration of deprivation predicted head circumference (β=–0.38; p=0.002) and this effect approached significance in the sub-nourished group (β=–0.020; p=0.053). A number of the findings in relation to the two sub-samples were suggestive of a different pattern of partial mediation of the effect of duration of deprivation by head circumference at age 6 years for the sub-nourished and the non-sub-nourished groups. For IQ the direct effect of duration of deprivation, although statistically significant, was much smaller in the sub-nourished than the non-sub-nourished groups. For the sub-nourished group the initially small, but significant univariate association between these variables became non-significant when head circumference was added. For the sub-nourished group the reduction in β was only 0.2. For disinhibited attachment the evidence for mediation was found in the non-sub-nourished group where the move from a significant effect of duration of deprivation on this outcome was accompanied by a 0.16 drop in the standardized β weight. There was also a suggestive pattern for inattention/overactivity in this group but the effect of head circumference on outcomes was not significant in the final model. The significant mediational analyses are presented in Figure 1.

Figure 1.

 Mediating effects of IQ on the effects of duration of deprivation on age 11 outcomes. *p<0.05.


As hypothesized, both psychosocial deprivation, even in the presence of normal overall nutrition, and, to a much lesser extent, sub-nutrition itself, predisposed to negative outcomes in a group of children with a history of institutional deprivation. These effects were also seen on brain growth as indexed by head circumference, but the timing of effects on head circumference was quite different for each risk factor. Among children whose institutional deprivation ceased at or before 6 months of age, sub-nutrition, nevertheless, was associated with a head circumference that was nearly 3SDs below the mean at the time of leaving institutional care. In sharp contrast, children without sub-nutrition who left institutions before the age of 6 months showed no significant reduction in head growth. On the other hand, if the children remained in institutional care for longer than 6 months, the reduction in head growth was very substantial and much the same in those with and without sub-nutrition. Clearly psychosocial deprivation can have major effects on head growth even in the absence of sub-nutrition.

The next question was whether impaired head growth played a mediating role in the broader range of psychological and psychopathological outcomes. Three findings provided initial indications that this might be the case. First, in the group with sub-nutrition whose deprivation lasted less than 6 months, there was both a major effect on head growth and a significant association with lower IQ. Neither effect was found in children without sub-nutrition. Second, impaired head growth at age 6 was associated with adverse effects on all outcomes at age 11 for the sample as a whole. Third, in the children both with and without sub-nutrition, the recovery in head growth was both limited (in comparison with weight) and more prolonged (with continuing gains made between 6 and 11y of age). Finally, sub-nutrition had a strong effect on head circumference both at the time of leaving institutional care and at age 11; multivariate analyses indicated that duration of deprivation also had a strong effect at both ages, but the effects of sub-nutrition were somewhat greater. In spite of that, whereas duration of deprivation had a strong effect on all psychological and psychopathological outcomes, the effects of sub-nutrition were far weaker and reached statistical significance only in the case of IQ after controlling for multiple testing.

Mediational analyses with the full sample provided very limited support for a role of head circumference mediating the link between duration of deprivation and outcomes at 11 with only a suggestion of an effect in relation to IQ. However, supplementary analysis of the effects within the sub-nutrition and non-sub-nutrition sub-groups suggested that mediation may be occurring but only for certain groups of children. The pattern of results as a whole is complex and the effects in general small so one must be cautious not to over-interpret them, especially given the small cell sizes and associated lack of power. For IQ, duration of deprivation is implicated to a far greater degree where sub-nutrition is not a factor, but head circumference seems to play a bigger mediating role where sub-nutrition is a factor. For inattention/overactivity and disinhibited attachment the results were in the opposite direction with the possible mediating effects being shown in the non-sub-nourished groups. This suggests that the underlying processes linking deprivation to cognitive outcomes might be somewhat different from those linking them to other deprivation specific outcomes.

The findings on head growth (and, by implication, on brain growth) provide some of the strongest clinical evidence to date that psychosocial deprivation, in the absence of sub-nutrition, can have a major long-term effect on brain development. However, as discussed above, rather surprisingly, the findings suggest a limited and complex role for head circumference in mediating deprivation related outcomes. Such findings necessarily raise the question whether head circumference provides an adequate index of brain growth. The evidence from other research suggests that the two are fairly closely associated, and our own findings (on a sub-sample) showed the same: brain growth is the main driver of head growth. Does the rather scant evidence for a mediating role of head circumference for outcomes mean that alterations in brain development are not involved in psychological and psychopathological sequelae of deprivation? Obviously not: overall brain size provides little direct indication of the more subtle alterations in brain structure and function, and underlying neurochemical processes, that are likely to be implicated in the functional deficits associated with institutional deprivation in this sample. Such alterations are perhaps likely to be especially important for the effects of psychosocial deprivation on mental health outcomes. Initial pilot data using magnetic resonance imaging technology have shown structural alterations in the amygdala of the ERA children.5 These findings suggest a potential role of early stress exposure in the developmental pathophysiology of institutional deprivation and a large-scale study of stress reactivity, brain structure, and function and development in the ERA sample is currently being planned. Given the suggestion of differential patterns of mediation as a function of nutritional status, it will be important in future studies to identify whether there are differential patterns of structural and functional brain alteration in relation to IQ in these two groups.

A further query needs to be raised on the apparently relatively minor effect of sub-nutrition on psychological and psychopathological outcomes, despite the strong effect on both body weight and head circumference. There can be no doubt about the reality of nutritional deficiencies in our sub-sample of sub-nourished children. Equally, we have been extremely careful to define sub-nutrition in such a way as to be confident as possible that the overall caloric intake of the group without sub-nutrition was adequate; their near normal body weight attests to that fact. What is much less clear is whether the overall balance of their nutritional intake was satisfactory. It is possible that a nutritional imbalance played some role in the effects of psychosocial deprivation even though overall sub-nutrition did not.

As far as IQ, and to a lesser extent inattention/overactivity, are concerned, the largely independent effects of sub-nutrition and of duration of deprivation suggests that their risk effects may be underpinned by different mechanisms. Candidate mechanisms for nutrition-related effects include reduced overall processing capacity,39 and cognitive-energetic resources,40 whereas psychosocial effects may involve some form of biological programming during a critical period of early development. Rutter and O’Connor4 suggested that either experience-expectant (whereby certain experiences are required for normal development) or experience-adaptive (the brain adapts to the circumstances operating during critical periods) mechanisms could be involved. Distinguishing these effects is beyond the scope of the current study.

The current study had a number of limitations. First, direct indices of social and cognitive impoverishment in the institutions were not available because valid contemporaneous measures of the quality of the institutional environments were usually not possible.7 Second, it was not possible to obtain measures on the quality and quantity of food provided in the institutions; we had to rely on weight at entry as our proxy for sub-nutrition. Third, we know that some children moved from one institution to another and we have no satisfactory indices of the relative qualities of the different institutions and especially how these related to changing patterns of nutritional intake. For instance, it is possible that the presence of sub-nutrition early on in life may be masked in this study if a child moved from an institution with poor physical care to one with adequate care some time before leaving institutional care altogether and entering the UK.

In summary, the current study provides evidence for the effects of both sub-nutrition and psychosocial deprivation on long-term psychological and psychopathological development. However, the strength of effect of duration of psychosocial deprivation was substantially greater than that of sub-nutrition. The most striking finding was the major effect of psychosocial deprivation on head growth even in the absence of sub-nutrition; and the most striking negative finding was that, despite the apparent centrality of head circumference effects, these had no detectable mediating role of head circumference on other outcomes.


We are most grateful to all the families who have generously given their time to participating in this study, and whose comments and suggestions have been very helpful in relation to the interpretation of findings. The data collection phase of the study was supported by grants from the University of Gent, the Helmut Horten Foundation and the UK Department of Health. Ongoing support is provided by grants from the Department of Health, the Nuffield Foundation, and the Jacobs Foundation. We are glad to express our thanks to our external Advisory Group, whose input has been invaluable. The views expressed in this article are ours and do not necessarily represent those of the funders. ES-B had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.