Am J Psychiatry 161:11, November 2004 2005
Article
http://ajp.psychiatryonline.org
Malnutrition at Age 3 Years and Externalizing Behavior
Problems at Ages 8, 11, and 17 Years
Jianghong Liu, Ph.D.
Adrian Raine, D.Phil.
Peter H. Venables, Ph.D., D.Sc.
Sarnoff A. Mednick, Ph.D.,
Objective: Poor nutrition is thought to predispose to externalizing behavior problems, but to date there appear to have been no prospective longitudinal studies testing this hypothesis. This study assessed whether 1) poor nutrition at age 3 years predisposes to antisocial behavior at ages 8, 11, and 17 years, 2) such relationships are independent of psychosocial adversity, and 3) IQ mediates the relationship between nutrition and externalizing behavior problems.
Method: The participants were drawn from a birth cohort (N=1,795) in whom signs of malnutrition were assessed at age 3 years, cognitive measures were assessed at ages 3 and 11 years, and antisocial, aggressive, and hyperactive behavior was assessed at ages 8, 11, and 17 years.
Results: In relation to comparison subjects (N=1,206), the children with malnutrition signs at age 3 years (N=353) were more aggressive or hyperactive at age 8 years, had more externalizing problems at age 11, and had greater conduct disorder and excessive motor activity at age 17. The results were independent of psychosocial adversity and were not moderated by gender. There was a dose-response relationship between degree of malnutrition and degree of externalizing behavior at ages 8 and 17. Low IQ mediated the link between malnutrition and externalizing behavior at ages 8 and 11.
Conclusions: These results indicate that malnutrition predisposes to neurocognitive deficits, which in turn predispose to persistent externalizing behavior problems throughout childhood and adolescence. The findings suggest that reducing early malnutrition may help reduce later antisocial and aggressive behavior.
(Am J Psychiatry 2004; 161:2005–2013)
Despite decades of research into social and biological
risk factors for childhood aggression, surprisingly little is
known about the role of malnutrition in contributing to
the development of childhood externalizing behavior (1,
2). Although deficiency in nutrition has been rarely studied
in relation to externalizing behavior, several studies
have demonstrated the effects of related factors, including
food additives, hypoglycemia, and, more recently, cholesterol
(2–4), on human behavior. In addition, epidemiological
studies have shown associations between increased
aggressive behavior and vitamin and mineral deficiency
(5, 6). Several authors have also claimed links between
iron-deficient anemia or low zinc level and externalizing
behavior in childhood (7, 8). More recently, the male offspring
of nutritionally deprived pregnant women were
found to have 2.5 times the normal rate of antisocial personality
disorder in adulthood (9). Effects were found for
severe malnutrition during the first and second trimesters
of pregnancy, but not the third trimester. Despite these
findings, the research literature on malnutrition and externalizing
behavior problems remains both limited and
controversial (2, 10).
If malnutrition is linked to antisocial behavior, as some
suggest, a key question concerns the mechanism of action.
One possible, but so far untested, hypothesis is that early
malnutrition predisposes to antisocial behavior because
malnutrition impairs neurocognitive functioning, which in
turn predisposes to externalizing behavior problems. There
is now compelling evidence that externalizing behavior
problems are characterized by lower IQ (11–14), and furthermore,
there is evidence that these early neurocognitive
deficits predict the onset of antisocial behavior (1, 15, 16).
In addition, there is increasing evidence that malnutrition
predisposes to lower IQ (17, 18). We recently observed that
malnutrition at age 3 was related to lower IQ at age 3 and
age 11 in a large longitudinal study in Mauritius (19), while
similar findings have also been observed by others (17, 20).
Despite the plausibility of the malnutrition-externalizing
hypothesis, we know of no empirical tests of it to date.
Research on nutrition and antisocial behavior has five
important limitations. First, there are simply very few
studies that have assessed the relationship between nutritional
deficits and externalizing behavior. Second, with
the exception of the prenatal study by Neugebauer et al.
(9), studies have not investigated prospectively the effect
of early nutrition on later aggressive behavior; it is possible,
for example, that antisocial behavior in the child could
produce parental neglect and malnutrition, rather than
vice versa. Third, there have been few, if any, studies on the
relation between nutritional deficits and aggression in fe2006
Am J Psychiatry 161:11, November 2004
MALNUTRITION AND EXTERNALIZING BEHAVIOR
http://ajp.psychiatryonline.org
males as well as males. Fourth, to our knowledge, the possibility
of mediating mechanisms, such as neurocognitive
deficits, have not been tested. Fifth, prior studies have
rarely controlled for the confounding effects of psychosocial
adversity.
In this study we used longitudinal data from Mauritius
in order to begin to address these gaps in the literature by
testing the following hypotheses: 1) poor nutrition at age 3
years predisposes to antisocial behavior at ages 8, 11, and
17 years, 2) such relationships are independent of early
psychosocial adversity, and 3) IQ mediates the nutritionantisocial
relationship. If controlling for childhood intelligence
abolishes the relationship between early malnutrition
and later externalizing behavior problems, this would
provide evidence for the role of intelligence as a significant
mediating mechanism (21).
Method
Participants
Participants were drawn from a birth cohort of 1,795 children
from the island of Mauritius, which lies off the coast of Africa. On
the basis of vaccination records, all children born in 1969 and 1970
in two main towns on the island were recruited into the study between
September 1972 and August 1973 when they were 3 years
old. The two towns (Vacoas and Quatre Bornes) were chosen because
they were representative of the ethnic distribution of the
whole island. The sample consisted of both boys (51.4%) and girls
(48.6%). The ethnic distribution was as follows: Indian, 68.7%;
Creole (African origin), 25.7%; and other (Chinese, English, and
French), 5.6%. Census data for the island as a whole indicated 66%
Indian, 29% Creole, and 5% other, indicating that the study largely
achieved its goal of representing the ethnicity of the population. In
keeping with previous work on this population (14, 22), the data
analyses were restricted to Indians and Creoles because of the
small number of subjects in the “other” ethnic category. Oral informed
consent was obtained from the mothers of the participants
in the early phases and from the participants themselves in
the age 17 phase. Early research activities were conducted according
to the principles outlined in the Declaration of Helsinki (23),
which prevailed in 1972, when the research was initiated, while research
activities in later years were conducted according to principles
outlined in the Belmont Report (24). Institutional review
board approval for the later research phases and retrospective
data analyses was obtained from the University of Southern California
and from the University of California, Los Angeles.
Signs of Malnutrition at Age 3
At age 3 years, four early signs of malnutrition (19) were assessed
in a clinical examination of 1,559 of the children. Assessments
of the children were conducted with a structured protocol
by local pediatricians who had received their medical training in
Europe. All assessments were conducted at the research unit. The
four signs were as follows:
Angular stomatitis. Cracking in the lips and corners of the
mouth is predominantly a sign of riboflavin deficiency (vitamin
B2) but also reflects niacin deficiency (25, p. 97). The base rate for
angular stomatitis in the sample was 7.0%.
Hair dyspigmentation. This condition reflects protein malnutrition
(26) and is found in tropical regions, particularly in Africa
(26), where children’s hair takes on a reddish-orange color. The
base rate for red hair in the sample was 6.8%.
Sparse, thin hair. This indicator is a sign of protein-energy
malnutrition in particular (27, 28) and malnutrition in general
(29). Protein reduction impairs hair growth, while zinc and iron
deficiency can also lead to thin hair. The base rate for this symptom
in the sample was 5.8%.
Anemia. Anemia was indicated by a low hemoglobin level,
which reflects iron deficiency. Hemoglobin level was assessed
from a laboratory test of blood drawn from the child. Anemia was
defined as a hemoglobin level below 8.5 g/dl. This definition of
anemia was consistent with medical practice in Mauritius in the
early 1970s. The base rate in this sample was 17.0%.
Definition of malnutrition. A participant was defined as suffering
from nutritional deficits if at least one of the four preceding
indicators was present; 22.6% of the assessed children met this
definition. A participant with no indicator present was classified
as having relatively normative nutrition; 77.4% fit this category.
To assess for a dose-response relationship between malnutrition
and externalizing behavior, each subject for whom behavior data
were available was categorized into one of four groups: no malnutrition
(N=766 at age 8, N=807 at age 11, N=422 at age 17), one indicator
of malnutrition (N=160 at age 8, N=172 at age 11, N=90 at
age 17), two indicators (N=45 at age 8, N=50 at age 11, N=25 at age
17), or three indicators (N=10 at age 8, N=13 at age 11, N=4 at age
17). Because only two individuals had all four nutrition indicators,
this category could not be included in the dose-response
analyses. Children thought to have potentially significant medical
problems of any kind (including malnutrition, scabies, and parasitic
worm) were referred to appropriate agencies for treatment
(30), but neither these referrals nor the treatment outcomes were
recorded or coded.
Intelligence and Cognitive Ability at Ages 3 and 11
Age 3. Measures of total cognitive ability were derived from six
subtests of the Boehm Test of Basic Concepts—Preschool Version
(14, 31, 32), which assesses basic verbal and visual-spatial concepts
that are fundamental for early school achievement. Full details
of measurement, factor structure, reliability, and validity in
this sample are given elsewhere (14, 31). Data were available for
1,260 subjects.
Age 11. Estimates of full-scale IQ were assessed at age 11 years
by using seven subtests of the WISC (33). The similarities and
digit span subtests were used to form an estimate of verbal IQ,
while the block design, object assembly, coding, mazes, and picture
completion subtests were used to form an estimate of performance
IQ. Data were available on 1,260 subjects for the preceding
three measures.
Psychosocial Adversity at Ages 3 and 11
The age 3 index of psychosocial adversity (14, 22) was based on
nine psychosocial variables assessed by social workers who visited
the homes of the children at age 3 years (see reference 34 for
full details). The index was created along lines similar to those described
by Rutter (35) and Moffitt (36). A total adversity score was
created by adding one point for each of the following nine variables:
father uneducated, mother uneducated, semiskilled or unskilled
parental occupation, single parent status, separation from
parents, large family size, poor health of mother, teenage mother,
and overcrowded home. Complete data for this construct were
available for 1,795 participants.
The age 11 psychosocial adversity index (14, 19, 22) was based
on 14 variables assessed by social workers who visited the homes
of the children at age 11 years. A total adversity score was created
by adding one point for each of the following 14 variables: living
in rented accommodation, house without electricity or water,
child with neither good toys nor good books, no television, poor
housing, father uneducated, mother uneducated, parent psychiAm
J Psychiatry 161:11, November 2004 2007
LIU, RAINE, VENABLES, ET AL.
http://ajp.psychiatryonline.org
atrically ill, parent physically ill, teenage mother (age 19 or
younger when child was born), single parent status, separation
from both parents, five or more siblings, and overcrowded home
(five or more family members per room). Complete data were
available on 1,272.
Externalizing Behavior Problems
Age 8. Externalizing behavior was assessed by teacher ratings
with the Children’s Behavior Questionnaire (37); see earlier reports
(14, 22, 38) for full details of reliability and validity. Briefly, a
factor analysis of this scale produced a factor of “aggression-hyperactivity”
(38), with the four-item aggression scale having an
internal reliability (coefficient alpha) of 0.79 and the six-item hyperactivity
scale having a reliability of 0.64. Complete data were
available for 1,130 participants on the aggression measure and for
1,128 on the hyperactivity measure.
Age 11. Externalizing behavior was assessed from parental ratings
using the three externalizing subscales (aggression, delinquency,
hyperactivity) of the Child Behavior Checklist (39), with
item content common across boys and girls (22). The internal reliability
values (coefficient alpha) for the scales were as follows: 0.72
for aggression, 0.66 for delinquency, 0.57 for hyperactivity, and 0.84
for total externalizing problems. Construct validity data are provided
elsewhere (22, 34). Data were available for 1,206 children.
Age 17. Externalizing behavior was assessed by parent and
teacher ratings using the Revised Behavior Problem Checklist
(40); full details of reliability and validity have presented elsewhere
(14, 22). All four checklist subscales that reflect externalizing
behavior problems were included in the analyses: conduct
disorder, socialized aggression, excessive motor activity, and attention
problems. Complete data were available for 608 subjects.
Representativeness of Groups
Complete data on both the malnutrition and externalizing behavior
variables were available on 982 subjects at age 8, 1,044 at age
11, and 541 at age 17. Those with and without complete data at
each age were compared on gender and ethnicity, variables that
were available on all subjects at age 3. Results of these analyses are
shown in Table 1. There was a statistically significant overrepresentation
of Indians at all ages among the participants with complete
data and an overrepresentation of boys at age 11. Consequently,
ethnicity and gender were entered into the subsequent analyses in
order to assess their roles as moderator effects or confounds.
Statistical Analyses
In the comparisons of the malnourished and normal groups,
separate analyses were conducted for each of the three ages (8, 11,
and 17 years). To test for overall effects of malnutrition on externalizing
behavior problems, we conducted a multivariate analysis
of variance (MANOVA) on all dependent variables for each age
(e.g., hyperactivity, aggression, and delinquency at age 11). Univariate
F tests were then used to assess which specific subcomponents
of externalizing behavior were associated with malnutrition.
IQ and social adversity at ages 3 and 11 were identified as
potential mediators, i.e., variables accounting for a significant
portion of the relationship between the predictor (malnutrition)
and the outcome variable (externalizing behavior) (21). To test for
the mediating effects of cognition and psychosocial adversity,
these variables were each entered separately as covariates in the
MANOVAs. Gender and ethnicity were identified as potential
moderators, i.e., factors that may reduce or enhance the influence
of malnutrition on externalizing behavior, as indicated by an interaction
between the independent variable (malnutrition) and a
factor (gender, ethnicity) (21). To assess for moderating effects of
gender and ethnicity, we entered these variables as factors in the
MANOVA alongside the nutrition grouping. Any interactions were
broken down by stratifying on the moderator variable and conducting
two-way MANOVAs to test for simple interactions (41).
These in turn were broken down by tests of simple main effects.
To test for a dose-response relationship between degree of malnutrition
and degree of externalizing behavior problems, we constructed
four levels of the grouping factor (none, one, two, and
three malnutrition indicators), repeated the preceding MANOVAs
and univariate ANOVAs, and conducted tests for linear trends to
assess for a linear fit between degree of malnutrition and degree
of externalizing behavior. Two-tailed tests of significance with an
alpha set at 0.05 were used throughout.
Results
Detailed results of both the multivariate and univariate
F tests of the effects of malnutrition on externalizing behavior
at all three ages, including moderator and mediator
effects, are shown in Table 2.
Externalizing Behavior Problems at Age 8
Effect of malnutrition. A MANOVA on the two dependent
variables (aggression and hyperactivity) indicated a
main group effect (Table 2), demonstrating that the malnourished
children had higher overall externalizing behavior
scores. Univariate F tests indicated that the malnourished
group had significantly higher scores on both
hyperactivity and aggression (Figure 1, Table 2).
Mediators. The malnourished children were more likely
to have lower cognitive ability at age 3 than the normal
children and were more likely to suffer psychosocial adversity
at age 3 than the normal children (Table 3). Consequently,
it is possible that poor cognition or greater
psychosocial adversity could mediate the main effect of
malnutrition on externalizing behavior. This possibility
was tested by entering cognitive and adversity measures
separately as covariates in the preceding MANOVA.
The main effect of malnutrition was abolished after we
controlled for cognitive ability, indicating that cognitive
ability mediates the link between malnutrition and externalizing
behavior. In contrast, after we controlled for age 3
psychosocial adversity, the main effect of malnutrition remained
significant (Table 2), indicating that the relation-
TABLE 1. Ethnicity and Gender of Subjects With and Without
Complete Follow-Up Data Among 1,559 Children in
Mauritius Whose Nutritional Status Was Assessed at Age 3
Age at Follow-Up,
Ethnicity,
and Gender
Percent With Indian Ethnicity
or Male Gendera Analysis
Subjects With
Complete Data
Subjects With
Missing Data
÷2
(df=1) p
Age 8
Indian ethnicity 74.5 67.6 9.98 0.002
Male gender 51.2 52.5 0.30 0.59
Age 11
Indian ethnicity 74.6 66.9 12.37 0.001
Male gender 54.1 48.7 5.11 0.03
Age 17
Indian ethnicity 76.4 69.2 9.32 0.002
Male gender 53.1 51.2 0.55 0.46
a Percentages are based on the number of subjects at each follow-up
with complete or missing data.
2008 Am J Psychiatry 161:11, November 2004
MALNUTRITION AND EXTERNALIZING BEHAVIOR
http://ajp.psychiatryonline.org
ship between malnutrition and externalizing behavior
was independent of psychosocial adversity.
Creoles were more likely to be malnourished than Indians
(Table 3). After we controlled for ethnicity by entering
it as a covariate, the effect of malnutrition on externalizing
behavior remained significant (Table 2).
Moderators. There were no interactions between nutrition
grouping and gender. However, there was a significant
interaction between ethnicity and nutrition grouping. In
order to clarify this two-way interaction, the sample was
stratified by race and one-way MANOVAs were conducted
separately on Creoles and Indians. The multivariate results
indicated that in both groups, malnutrition was significantly
associated with externalizing problems. For Creoles
there was a significant effect of malnutrition on externalizing
behavior (F=3.67, df=2, 236, p<0.03). Univariate F tests
indicated that malnourished Creoles had significantly
higher scores than normal Creoles on aggression (F=7.37,
TABLE 2. Results of Multivariate and Univariate Tests of the Effect of Malnutrition at Age 3 on Externalizing Behavior at
Ages 8, 11, and 17 Among Children in Mauritius
Age at Follow-Up, Type of Analysis,
and Behavior Variablea
Moderator Effects
Main Effect of Malnutrition
Gender-by-Malnutrition
Interaction
Race-by-Malnutrition
Interaction
N F df p F df p F df p
Age 8 982
Multivariate 5.63 2, 979 0.004 1.62 2, 975 0.24 3.47 2, 932 0.04
Univariate
Aggression 4.68 1, 980 0.04 1.63 2, 975 0.24 4.58 2, 933 0.04
Hyperactivity 11.28 1, 980 0.001 0.42 2, 976 0.52 0.06 2, 933 0.82
Age 11 1,044
Multivariate 4.18 3, 1042 0.006 1.19 3, 1037 0.31 1.50 3, 994 0.21
Univariate
Aggression 2.67 1, 1042 0.11 1.95 1, 1039 0.16 4.25 1, 996 0.04
Delinquency 2.10 1, 1042 0.15 0.18 1, 1039 0.67 3.22 1, 996 0.08
Hyperactivity 12.53 1, 1042 0.001 1.88 1, 1039 0.17 0.81 1, 996 0.37
Age 17 541
Multivariate 5.35 4, 536 0.001 1.34 4, 533 0.25 0.53 4, 514 0.72
Univariate
Conduct disorder 10.07 1, 539 0.002 0.12 1, 536 0.73 0.81 1, 517 0.37
Motor excess 6.11 1, 539 0.02 3.75 1, 536 0.053 0.12 1, 517 0.73
Attention problems 0.31 1, 539 0.58 0.50 1, 536 0.48 0.96 1, 517 0.06
Socialized aggression 0.63 1, 539 0.43 0.00 1, 536 0.98 0.01 1, 517 0.91
a At age 8, externalizing behavior was assessed with the Children’s Behavior Questionnaire. At age 11, externalizing behavior was assessed with
the Child Behavior Checklist. At age 17, externalizing behavior was assessed with the Revised Behavior Problem Checklist.
b Analyses of the follow-up at age 8 used the ratings of psychosocial adversity and cognitive ability at age 3. Analyses of the follow-ups at ages
11 and 17 used the ratings at age 11.
FIGURE 1. Scores for Externalizing Behaviors at Ages 8, 11, and 17 Among Children in Mauritius Who Were or Were Not
Malnourished at Age 3
a Number of subjects assessed at age 3.
4.0
3.0
2.0
1.0
3.5
2.5
1.5
0.5
0.0
Aggression Hyperactivity Aggression Hyperactivity Delinquency Conduct
disorder
Motor
excess
Attention
problems
Socialized
aggression
Score on Scale From Children’s
Behavior Questionnaire
Age 8 Age 11 Age 17
Comparison
subjects (N=1,206)a
Malnourished
children (N=353)a
6
4
2
5
3
1
0
Score on Scale From Child Behavior Checklist
12
8
6
4
2
10
11
9
7
5
3
1
0
Score on Scale From Revised
Problem Behavior Checklist
Am J Psychiatry 161:11, November 2004 2009
LIU, RAINE, VENABLES, ET AL.
http://ajp.psychiatryonline.org
df=1, 237, p<0.008), but the effect for hyperactivity was not
significant (F=2.92, df=1, 237, p=0.09). For Indians, there
was a significant effect of malnutrition on externalizing behavior
(F=3.43, df=2, 695, p<0.04), with higher scores for
externalizing behavior among the malnourished Indians
than among the normal Indians. Univariate F tests indicated
that malnourished Indians had significantly higher
scores than normal Indians on hyperactivity (F=4.85, df=1,
696, p<0.03), but the effect for aggression was not significant
(F=0.15, df=1, 696, p=0.70). Consequently, these analyses
indicate that malnutrition is more likely to predispose
Creole children to aggression at age 8 but predispose Indian
children to hyperactivity at age 8.
Externalizing Behavior Problems at Age 11
Effect of malnutrition. As shown in Table 2, the overall
MANOVA indicated a main effect of malnutrition on externalizing
behavior scores at age 11 years. The univariate F
tests showed a significant effect of malnutrition on hyperactivity
but not on aggression or delinquency (Figure 1).
Mediators. As shown in Table 3, the malnourished children
had lower cognitive ability at age 11 than the nonmalnourished
children. After we controlled for cognitive
ability, the main effect of nutritional status was abolished,
indicating a mediating effect of low cognitive ability. Although
the malnourished and comparison children did
not differ significantly on adverse psychosocial backgrounds
at age 11, the difference approached significance.
Consequently, to ensure that adversity did not mediate the
findings, the age 11 total psychosocial adversity score was
entered as a covariate in the preceding MANOVA. The
main group effect remained significant.
Moderators. Multivariate results indicated that there were
no significant moderator effects for gender at all ages and no
moderator effects for ethnicity at ages 11 and 17, although
ethnicity was a significant moderator at age 8 (Table 2).
Externalizing Behavior Problems at Age 17
Effect of malnutrition. A MANOVA conducted on conduct
disorder, motor excess, attention problems, and socialized
aggression showed a significant overall main effect
of malnutrition (Table 2). Univariate F tests showed that
the malnourished group had significantly higher scores on
conduct disorder and motor excess but not on attention
problems or socialized aggression (Figure 1, Table 2).
Mediators. After age 11 cognitive ability was entered as a
covariate, the main effect of malnutrition remained significant.
The main group effect also remained significant after
we controlled for age 11 psychosocial adversity (Table 2).
Moderators. There was no significant moderator effect
for gender or ethnicity (Table 2).
Dose-Response Relationships
Dose-response relationships between malnutrition and
externalizing behavior are depicted in Figure 2. MANOVAs
indicated a significant main effect of the degree of malnutrition
on externalizing behavior at age 8 (F=2.53, df=6,
1954, p=0.02), and the linear term was also significant for
both aggression (F=5.92, df=1, 977, p<0.02) and hyperactivity
(F=11.96, df=1, 977, p<0.001). At age 17, there was
also a main effect of the degree of malnutrition (F=2.44,
df=12, 1608, p<0.004) and significant linear trends for conduct
disorder (F=10.14, df=1, 537, p<0.002) and motor excess
(F=8.58, df=1, 537, p<0.004). For age 11, the main effect
of nutrition was nonsignificant (F=1.84, df=9, 3114, p=
0.06), although the linear term was significant for hyperactivity
(F=9.19, df=1, 1038, p<0.002). Univariate F tests
showed significant group differences for age 8 hyperactivity
(F=4.60, df=3, 977, p<0.003), age 11 hyperactivity (F=
4.36, df=3, 1038, p<0.005), age 17 conduct disorder (F=
3.53, df=3, 537, p<0.02), and age 17 motor excess (F=2.95,
df=3, 537, p<0.04). The results of all other univariate tests
(i.e., on aggression at age 8, aggression and delinquency at
age 11, and attention problems and socialized aggression
at age 17) were nonsignificant (p>0.11).
In order to assess whether poor cognition mediated the
preceding dose-response relationships, the tests were repeated
after we entered the cognitive measures as covariates.
All effects of nutrition were abolished (age 8: F=1.14,
df=6, 1488, p=0.34; age 17: F=1.69, df=12, 1401, p=0.07), indicating
a mediating role of poor neurocognitive functioning.
Discussion
Key Findings
One key finding of this study is that malnutrition at age
3 years is associated with higher scores for externalizing
behavior problems at ages 8, 11, and 17. A second key find-
Mediator Effects
Psychosocial Adversity
at Age 3 or 11b
Cognitive Ability
at Age 3 or 11b
F df p F df p
4.49 2, 976 0.02 1.58 2, 746 0.21
4.05 1, 977 0.05 1.37 1, 747 0.24
8.97 1, 977 0.003 3.15 1, 747 0.08
4.04 3, 1002 0.007 1.40 3, 973 0.24
2.84 1, 1004 0.10 1.32 1, 975 0.25
2.32 1, 1004 0.13 0.35 1, 975 0.55
12.14 1, 1004 0.001 3.79 1, 975 0.05
3.91 4, 480 0.004 3.69 4, 467 0.006
5.59 1, 483 0.02 4.45 1, 470 0.21
3.55 1, 483 0.06 1.25 1, 470 0.26
0.53 1, 483 0.47 1.16 1, 470 0.28
1.56 1, 483 0.21 1.64 1, 469 0.20
2010 Am J Psychiatry 161:11, November 2004
MALNUTRITION AND EXTERNALIZING BEHAVIOR
http://ajp.psychiatryonline.org
ing is that the relationship between malnutrition and externalizing
behavior was not found to be an artifact of psychosocial
adversity but was instead mediated by cognitive
ability, indicating that malnutrition predisposes children
to a lower IQ, which in turn predisposes them to externalizing
behavior problems. These conclusions are supported
by the finding of dose-response relationships between degree
of malnutrition and degree of externalizing behavior
problems at ages 8 and 17, relationships that were again
found to be mediated by low IQ. To our knowledge, these
are the first findings to show prospectively that malnutrition
assessed in the early postnatal years is associated with
externalizing behavior problems from childhood to late
adolescence and also to show the mediating effects of cognitive
ability. These findings in turn have potential implications
for public health attempts to prevent the occurrence
of externalizing behavior problems in children and
adolescents.
The robustness of the findings is indicated in several
ways. First, malnutrition predisposed to externalizing behavior
problems across several ages from childhood (age 
to late adolescence (age 17). Second, externalizing behavior
problems were measured by three different instruments
at the three different ages, indicating that the findings were
replicated across sources and were largely invariant to the
nature of measurement (42). Third, the fact that gender
and ethnicity did not moderate the multivariate findings at
ages 11 and 17 indicates that the nutrition-externalizing relationship
is not specific to one gender or ethnic grouping,
although it should be noted that at age 8 malnutrition was
more likely to predispose to aggression in Creoles and to
hyperactivity in Indians. It is possible that malnutrition
predisposes to a general disinhibitory tendency and that
broad cultural differences influence the precise manifestations
of such disinhibition at a behavioral level. Fourth, the
fact that dose-response relationships were found at ages 8
and 17 confirms and extends the findings based on comparisons
of the malnourished and nonmalnourished children,
although it is noted that the dose-response effect at
age 11 was not statistically significant (p=0.06). For these
reasons, we believe that the findings cannot be easily attributed
to chance and that, instead, they reflect a reliable
relationship between early malnutrition and later externalizing
behavior problems.
TABLE 3. Demographic and Cognitive Measures at Ages 8, 11, and 17 Among Children in Mauritius Who Were or Were Not
Malnourished at Age 3
Age at Follow-Up
and Variable
Malnourished Not Malnourished
Percent of Ethnic
or Gender Group Mean SD N
Percent of Ethnic
or Gender Group Mean SD
Analysis
N ÷2 t df p
Age 8 216 766
Ethnicity 7.67 1 0.006
Creole 28.0 72.0
Indian 19.5 80.5
Gender 0.82 1 0.40
Male 23.1 76.9
Female 20.7 79.3
Psychosocial adversity
score at age 3a 2.21 1.35 1.83 1.31 3.80 978 0.001
Cognitive ability score
at age 3b 96.51 14.20 100.44 15.25 2.84 748 0.005
Age 11 1,044 762
Ethnicity 0.45 1 0.54
Creole 24.0 76.0
Indian 22.0 78.0
Gender 0.10 1 0.78
Male 22.3 77.7
Female 23.2 76.8
Psychosocial adversity
score at age 11c 2.02 1.49 1.80 1.52 1.87 1005 0.07
Cognitive ability score
(IQ) at age 11 95.88 15.58 101.27 14.36 4.80 976 0.001
Age 17 541 1,265
Ethnicity 0.07 1 0.80
Creole 22.8 77.2
Indian 21.6 78.4
Gender 0.13 1 0.76
Male 22.3 77.7
Female 23.2 76.8
Psychosocial adversity
score at age 11 2.13 1.52 1.87 1.54 1.55 484 0.12
Cognitive ability score
(IQ) at age 11 94.37 16.57 99.46 15.56 2.92 471 0.004
a Range=0–9; 9=greatest severity.
b Higher scores indicate greater cognitive ability.
c Range=0–14; 14=greatest severity.
Am J Psychiatry 161:11, November 2004 2011
LIU, RAINE, VENABLES, ET AL.
http://ajp.psychiatryonline.org
Nutrition, Brain Development,
and Externalizing Behavior
A critical question concerns the mechanism by which
malnutrition predisposes to later externalizing behavior
problems. We hypothesize that early malnutrition negatively
affects brain growth and development and that
brain impairments predispose to antisocial and violent
behavior by affecting cognitive functions. The indicators
of malnutrition in this study reflect deficits in protein (red
hair, sparse/thin hair), iron (low hemoglobin level), and
zinc (red hair, sparse/thin hair). There is extensive experimental
evidence in animals both that zinc and protein deficiency
impairs brain development (8, 43–45) and that
protein, iron, and zinc deficiency predisposes to aggression
(45–47). In humans, zinc deficiency during pregnancy
has been linked to impaired DNA, RNA, and protein synthesis
during brain development as well congenital brain
abnormalities (48). There is also evidence in humans that
antisocial behavior is related to protein deficiency (9) and
iron-deficient anemia (7). Consequently, protein, iron,
and zinc deficiencies may contribute to the brain impairments
that have been found in aggressive adult offenders
and that in turn are thought to predispose to aggressive
antisocial behavior (3, 15, 46, 49).
While early malnutrition could thus relatively directly predispose
to externalizing behavior problems by impairing
brain mechanisms such as those in the prefrontal cortex
that are thought to regulate emotion and inhibit impulsive
aggressive behavior (for example, see reference 50), malnutrition
could also predispose to externalizing behavior problems
more indirectly by impairing cognitive functioning,
which in turn predisposes to externalizing behavior problems.
The findings from the present study provide partial
(but not total) support for this possibility. Cognitive functioning
was established as a mediator for the malnutritionexternalizing
relationship at ages 8 and 11 years in that controlling
for the effect of IQ on externalizing behavior abolished
the malnutrition-externalizing relationship. Poor cognitive
ability has been found consistently to predispose to
externalizing behavior problems (51). Nevertheless, support
for this cognitive explanation of the malnutrition-externalizing
relationship is not entirely complete. While mediating
effects were observed at ages 8 and 11 and while poor cognition
mediated the dose-response relationship at age 17, it
did not mediate overall differences between the malnourished
and comparison groups in age 17 externalizing behavior,
possibly because of the 6-year gap between the assessments
of IQ (age 11) and externalizing behavior (age 17).
Clinical Implications and Limitations
Externalizing behavior problems are important predisposing
factors in later adult violent offenses (52), and violence
prevention and protection from victimization have
become two of the most pressing issues facing society today
(52, 53). One recent double-blind, placebo-controlled,
randomized experimental trial showed that supplementation
of adult prisoners’ diet with vitamins, minerals, and
essential fatty acids significantly reduced antisocial and
violent behavior in prison (54). Although dietary interventions
for adults may prove helpful in reducing antisocial
and violent behavior, identification of early risk factors for
childhood and adolescent aggression is a critically important
first step for developing successful prevention
programs for such adult violence. Because nutrition is a
FIGURE 2. Dose-Response Relationships Between Number of Malnutrition Indicators at Age 3 and Externalizing Behaviors
at Ages 8, 11, and 17 Among Children in Mauritiusa
a Four indicators of malnutrition were assessed: angular stomatitis, hair dyspigmentation, sparse/thin hair, and anemia.
4.0
2.0
1.0
1.5
0.5
3.5
2.5
3.0
0.0
Score on Scale From Children’s
Behavior Questionnaire
Age 8 (N=982) Age 11 (N=1,044) Age 17 (N=541)
12
11
10
8
6
4
2
9
7
5
3
1
0
Score on Scale From Revised
Problem Behavior Checklist
0 1 2 3 0 1 2
Number of Malnutrition Indicators
3 0 1 2 3
6
4
2
5
3
1
0
Score on Scale From Child Behavior Checklist
Motor excess
Conduct disorder
Delinquency
Hyperactivity
Aggression
2012 Am J Psychiatry 161:11, November 2004
MALNUTRITION AND EXTERNALIZING BEHAVIOR
http://ajp.psychiatryonline.org
malleable factor, it may be more practical and easier to
prevent externalizing behavior through better early nutrition
targeting at-risk populations than more complex and
expensive psychosocial manipulations. The fact that several
studies that have included the promotion of early nutrition
as part of a larger prevention program have met
with success (55, 56) argues for further attention to early
malnutrition as a predisposing factor in externalizing behavior
among children. It is possible, however, that malnutrition
at age 3 years is correlated with poor prenatal
malnutrition and that prenatal, rather than postnatal, interventions
for malnutrition may be most effective (9, 55).
Finally, three potential limitations of the study should
be acknowledged. First, the findings suggest but do not
prove that early malnutrition predisposes to later externalizing
behavior. Randomized, controlled designs that
manipulate nutritional intake and evaluate diet in children
are needed to support the role of nutrition as an etiological
factor in externalizing behavior. Nevertheless, the
fact that interventions that indirectly manipulate nutrition
have been found to reduce later conduct disorder (55,
56) suggests that a possible etiological role of malnutrition
in externalizing behavior should not be discounted. Second,
because nutrition was assessed at only one time
point (age 3 years) this study could not ascertain whether
the effects on externalizing behavior were produced by
transient malnutrition (i.e., the first 3 years only), by more
sustained malnutrition, extending beyond the third year,
or by maternal malnutrition during pregnancy (9). Third,
there are cultural, ethnic, and social differences between
Mauritius and the United States that highlight the importance
of replicating the present findings in Europe and the
United States. Nevertheless, the current findings from
Mauritius may be a good model for externalizing problems
in underserved subpopulations of American society,
particularly since food insufficiency is relatively common
in poor rural areas of the United States and has been associated
with poor behavioral functioning in low-income
children in American inner cities (57, 58).
Received June 10, 2003; revision received Dec. 2, 2003; accepted
Jan. 19, 2004. From the Social Science Research Institute and the Department
of Psychology, University of Southern California; and the Department
of Psychology, University of York, York, U.K. Address reprint
requests to Dr. Raine, Department of Psychology, University of Southern
California, Los Angeles, CA 90089-1061; raine@usc.edu (e-mail).
Supported by a predoctoral fellowship award (F31 NR-07518) and
postdoctoral fellowship (F32 NR-08661) from the National Institute of
Nursing Research to Dr. Liu; an NIMH Independent Scientist Award
(K02 MH-01114), a grant from the Borchard Foundation, and a grant
from NIMH (RO1 MH-46435) to Dr. Raine; grants from the Medical Research
Council, Leverhulme Trust, and Mental Health Foundation to
Dr. Venables; an NIMH Research Scientist Award to Dr. Mednick (5
K05 MH-00619); and a grant from the Ministry of Health of the Mauritian
government.
The authors thank Marie-Clare Calambay, Meena Calinghen,
Athene Chiriaca, Cyril Dalais, Fazila Dinally, Devi Jaganathen, Goorah
Rajah, and Charles Yip Tong for help in data collection and scoring.
References
1. Fishbein DH, Pease SE: Diet, nutrition, and aggression. J Offender
Rehabilitation 1994; 21:117–144
2. Rutter M, Giller H, Hagell A: Antisocial Behavior by Young People.
New York, Cambridge University Press, 1998
3. Raine A: The Psychopathology of Crime: Criminal Behavior as
a Clinical Disorder. San Diego, Academic Press, 1993
4. Kaplan JR, Muldoon MF, Manuck SB, Mann JJ: Assessing the observed
relationship between low cholesterol and violence-related
mortality: implications for suicide risk. Ann NY Acad Sci
1997; 836:57–80
5. Breakey J: The role of diet and behaviour in childhood. J Paediatr
Child Health 1997; 33:190–194
6. Werbach MR: Nutritional influences on aggressive behavior. J
Orthomolecular Med 1992; 7:45–51
7. Rosen GM, Deinard AS, Schwartz S, Smith C, Stephenson B, Grabenstein
B: Iron deficiency among incarcerated juvenile delinquents.
J Adolesc Health Care 1985; 6:419–423
8. Peters DP: Effects of prenatal nutritional deficiency on affiliation
and aggression in rats. Physiol Behav 1978; 20:359–362
9. Neugebauer R, Hoek HW, Susser E: Prenatal exposure to wartime
famine and development of antisocial personality disorder
in early adulthood. JAMA 1999; 4:479–481
10. McGee R, Williams S, Anderson J, McKenzie-Parnell JM, Silva PA:
Hyperactivity and serum and hair zinc levels in 11-year-old
children from the general population. Biol Psychiatry 1990;
28:165–168
11. Moffitt TE: Juvenile delinquency and attention-deficit disorder:
developmental trajectories from age three to fifteen. Child Dev
1990; 61:893–910
12. McCall RB, Beach SR, Lau S: The nature and correlates of underachievement
among elementary schoolchildren in Hong
Kong. Child Dev 2000; 71:785–801
13. Noterdaeme M, Amorosa H: Evaluation of emotional and behavioral
problems in language impaired children using the Child Behavior
Checklist. Eur Child Adolesc Psychiatry 1999; 8:71–77
14. Raine A, Yaralian P, Reynolds C, Venables P, Mednick S: Spatial
but not verbal cognitive deficits at age 3 years in persistently
antisocial individuals. Dev Psychopathol 2002; 14:25–44
15. Moffitt TE: The neuropsychological studies of juvenile delinquency:
a critical review, in Crime and Justice: An Annual Review
of Research, vol 12. Edited by Tonry M, Morris N. Chicago,
University of Chicago Press, 1990, pp 99–169
16. Fishbein DHE: Biobehavioral Perspectives in Criminology:
Wadsworth Series in Criminological Theory. Belmont, Calif,
Wadsworth/Thomson Learning, 2001
17. Grantham-McGregor S, Ani C: A review of studies on the effect
of iron deficiency on cognitive development in children. J Nutr
2001; 131 (2 suppl 2):649S–666S
18. Lozoff B, Jimenez E, Hagen J, Mollen E, Wolf AW: Poorer behavioral
and developmental outcome more than 10 years after
treatment for iron deficiency in infancy. Pediatrics 2000; 105:E51
19. Liu J, Raine A, Venables PH, Dalais C, Mednick SA.: Malnutrition
at age 3 years and lower cognitive ability at age 11 years: independence
from psychosocial adversity. Arch Pediatr Adolesc
Med 2003; 157:593–600
20. Grantham-McGregor SM, Walker SP, Chang S: Nutritional deficiencies
and later behavioural development. Proc Nutr Soc
2000; 59:47–54
21. Baron RM, Kenny DA: The moderator-mediator variable distinction
in social psychological research: conceptual, strategic,
and statistical considerations. J Pers Soc Psychol 1986; 51:
1173–1182
22. Raine A, Reynolds C, Venables PH, Mednick SA, Farrington DP:
Fearlessness, stimulation-seeking, and large body size at age 3
Am J Psychiatry 161:11, November 2004 2013
LIU, RAINE, VENABLES, ET AL.
http://ajp.psychiatryonline.org
years as early predispositions to childhood aggression at age
11 years. Arch Gen Psychiatry 1998; 55:745–751
23. Declaration of Helsinki: Recommendations Guiding Doctors in
Clinical Research Adopted by the 18th World Medical Assembly.
Helsinki, World Medical Association, 1964
24. National Commission for the Protection of Human Subjects of
Biomedical and Behavioral Research: The Belmont Report:
Ethical Principles and Guidelines for the Protection of Human
Subjects of Biomedical and Behavioral Research. Washington,
DC, US Department of Health and Human Services, 1979
25. Wong DL, Hess CS: Clinical Manual of Pediatric Nursing. St
Louis, Mosby, 2000
26. van den Biggelaar I, van den Broeck J: Nutrition-related hair
signs in Zairian preschool children and associations with anthropometry.
Trop Geogr Med 1995; 47:248–251
27. Rushton DH: Nutrition factors and hair loss. Clin Exp Dermatol
2002; 27:396–404
28. Friedman M, Orraca-Tetteh R: Hair as an index of protein malnutrition.
Adv Exp Med Biol 1978; 105:131–154
29. Spencer LV, Callen JP: Hair loss in systemic disease. Dermatol
Clin 1987; 5:565–570
30. Bell B: A longitudinal psychophysiological study of 3-year-old
Mauritian children: the Joint Child Health Project, in Primary
Prevention of Schizophrenia in High-Risk Groups: Report on a
Working Group, 9–12 June, 1975. Copenhagen, World Health
Organization, 1977
31. Boehm A: Boehm Test of Basic Concepts—Preschool Version.
San Antonio, Tex, Psychological Corp, 1986
32. Raine A, Reynolds C, Venables PH, Mednick SA: Stimulation
seeking and intelligence: a prospective longitudinal study. J
Pers Soc Psychol 2002; 82:663–674
33. Wechsler D: Wechsler Preschool and Primary Scale of Intelligence.
San Antonio, Tex, Psychological Corp, 1967
34. Raine A, Venables PH, Mednick SA: Low resting heart rate at
age 3 years predisposes to aggression at age 11 years: evidence
from the Mauritius Child Health Project. J Am Acad Child Adolesc
Psychiatry 1997; 36:1457–1464
35. Rutter M: Family, area, and school influences in the genesis of
conduct disorder, in Aggression and Anti-Social Behavior in
Childhood and Adolescence. Edited by Hersov LA, Shaffer D.
Oxford, UK, Pergamon Press, 1978, pp 95–114
36. Moffitt TE: Juvenile delinquency and attention-deficit disorder:
developmental trajectories from age three to fifteen. Child Dev
1990; 61:893–910
37. Rutter M: A children’s behaviour questionnaire for completion
by teachers: preliminary findings. J Child Psychol Psychiatry
1967; 8:1–11
38. Venables PH, Fletcher RP, Dalais JC, Mitchell DA, Schulsinger F,
Mednick SA: Factor structure of the Rutter Children’s Behavior
Questionnaire in a primary school population in a developing
country. J Child Psychol Psychiatry 1983; 24:213–222
39. Achenbach TM, Edelbrock C: Manual for the Child Behavior
Checklist and Revised Child Behavior Profile. Burlington, University
of Vermont, Department of Psychiatry, 1983
40. Quay HC, Peterson DR: Manual for the Revised Behavior Problem
Checklist. Coral Gables, Fla, University of Miami, Department
of Psychology, 1987
41. Maxwell SE, Delaney HD: Designing Experiments and Analyzing
Data: A Model Comparison Approach. Belmont, Calif, Wadsworth,
1990
42. Moffitt TE: Measuring children’s antisocial behavior. JAMA
1996; 275:403–404
43. Oteiza PI, Hurley L, Lonnerdal B, Keen C: Effects of marginal
zinc deficiency on microtubule polymerization in the developing
rat brain. Biol Trace Elem Res 1990; 23:13–23
44. Bennis-Taleb N, Remacle C, Hoet JJ, Reusens B: A low-protein
isocaloric diet during gestation affects brain development and
alters permanently cerebral cortex blood vessels in rat offspring.
J Nutr 1999; 129:1613–1619
45. Halas ES, Reynolds GM, Sandstead HH: Intra-uterine nutrition
and its effects on aggression. Physiol Behav 1977; 19:653–661
46. Tikal K, Benesova O, Frankova S: The effect of pyrithioxine and
pyridoxine on individual behavior, social interactions, and
learning in rats malnourished in early postnatal life. Psychopharmacologia
1976; 46:325–332
47. Munro N: A three year study of iron deficiency and behavior in
rhesus monkeys. Int J Biosoc Res 1987; 9:35–62
48. Pfeiffer CC, Braverman ER: Zinc, the brain and behavior. Biol
Psychiatry 1982; 17:513–532
49. Volavka J: Electroencephalogram among criminals, in The
Causes of Crime: New Biological Approaches. Edited by Mednick
SA, Moffitt TE, Stack S. Cambridge, UK, Cambridge University
Press, 1987, pp 137–145
50. Raine A, Lencz T, Bihrle S, LaCasse L, Colletti P: Reduced prefrontal
gray matter volume and reduced autonomic activity in
antisocial personality disorder. Arch Gen Psychiatry 2000; 57:
119–127
51. Donnellan MB, Ge X, Wenk E: Cognitive abilities in adolescentlimited
and life-course-persistent criminal offenders. J Abnorm
Psychol 2000; 109:396–402
52. Farrington DP, Hawkins JD: Predicting participation, early onset
and later persistence in officially recorded offending. Crim Behav
Ment Health 1991; 1:1–33
53. Comerci GD: Efforts by the American Academy of Pediatrics to
prevent and reduce violence and its effects on children and adolescents.
Bull NY Acad Med 1996; 73:398–410
54. Gesch CB, Hammond SM, Hampson SE, Eves A, Crowder MJ: Influence
of supplementary vitamins, minerals and essential
fatty acids on the antisocial behaviour of young adult prisoners:
randomised, placebo-controlled trial. Br J Psychiatry 2002;
181:22–28
55. Olds D, Henderson CJ, Cole R, Eckenrode J, Kitzman H, Luckey
D, Pettitt L, Sidora K, Morris P, Powers J: Long-term effects of
nurse home visitation on children’s criminal and antisocial behavior:
15-year follow-up of a randomized controlled trial.
JAMA 1998; 280:1238–1244
56. Lally JR, Mangione PL, Honig AS: The Syracuse University Family
Development Research Program: long-range impact of an early
intervention with low income children and their families, in Parent
Education as Early Childhood Intervention: Emerging Directions
in Theory, Research, and Practice. Edited by Powell DR,
Sigel IE. Norwood, NJ, Ablex Publishing Corp, 1988, pp 79–104
57. Murphy JM, Wehler CA, Pagano ME, Little M, Kleinman RE, Jellinek
MS: Relationship between hunger and psychosocial functioning
in low-income American children. J Am Acad Child Adolesc
Psychiatry 1998; 37:163–170
58. Smith J, Lensing S, Horton JA, Lovejoy J, Zaghloul S, Forrester I,
McGee BB, Bogle ML: Prevalence of self-reported nutritionrelated
health problems in the Lower Mississippi Delta. Am J
Public Health 1999; 89:1418–1421
