Journal of Child Psychology and Psychiatry 45:4 (2004), pp 779–788

Psychosocial determinants of behaviour problems

in Brazilian preschool children
Luciana Anselmi,1 Cesar A. Piccinini,1 Fernando C. Barros,2 and Rita S. Lopes1
1
Institute of Psychology, UFRGS, and Post-Graduate Programme in Epidemiology, UFPel, Brazil; 2PAHO/WHO, Latin
American Center for Perinatology, Uruguay

Background: This investigation aimed at examining, through an epidemiological study, the social
demographic, family and individual determinants of behaviour problems in preschool chil-
dren. Method: Six hundred and thirty-four children, age four years, and their mothers, belonging to a
cohort of 5,304 children being followed from birth, took part in the study. During a home visit, the
child’s behaviour problems and IQ were examined, as well as the prevalence of maternal psychiatric
disorder, the quality of the home environment, and other social demographic, family and individual
factors. Results: The results revealed a prevalence of children’s behaviour problems of 24% (clinical
and borderline groups). Regression analysis showed that maternal psychiatric disorder, education and
age, number of younger siblings and quality of the home environment explained 28% of the variance of
the child’s behaviour problems. Conclusions: The results point to the multi-determination of child
behaviour problems. Keywords: Behavioural problems, externalising disorders, longitudinal studies,
preschool children, prevalence, Third World children.

Young children frequently show behaviours that the relationship with the child. Parents with higher
worry adults but only recently have researchers education tend to adopt values, attitudes and beliefs
examined the clinical significance of these behav- that promote child development. These parents tend
iours in children of less than six years of age. The to have a conception of development as a complex
literature does not offer a precise concept of beha- process and not as determined by unique factors,
viour problems, confining it to operational defini- constitutional or environmental. They also have ex-
tions based on the instrument used to measure the pectations of more independent and self-directed
construct. For Achenbach (1991), behaviour prob- behaviour for their children (Sameroff, Seifer, Zax, &
lems in young children involve symptom patterns of Barocas, 1987) and hold greater knowledge of the
internalisation and externalisation. Internalising child’s developmental needs (Lewis, 1993). The
problems involve withdrawal, somatic complaints, family’s socioeconomic status has been widely
sadness, fear, depression and anxiety. Externalising investigated as an important determinant of beha-
problems refer to aggressive behaviours, hyperac- viour problems. Low family income affects the child’s
tivity, disobedience, low impulse control, displays of social life, leaving him/her in geographically and
anger and delinquent behaviour. socially isolated communities and, as a result,
The prevalence of behaviour problems in preschool depriving him/her of support networks. The imme-
children has been of approximately 10% to 15% diate environment is also of influence, affecting their
(Campbell, 1994), even though studies have indic- daily routines, the caregiversÕ roles and responsibil-
ated an increase in this prevalence in North Ameri- ities and the overall home interpersonal environment
can children and adolescents (Achenbach & Howell, (Halpern, 1993).
1993). Studies show a greater frequency of these Since the home environment is the main develop-
problems in boys (Murray, 1992) and in children of mental context for children, several family factors
non-white ethnic groups (Dubow & Luster, 1990). have been investigated, as they contribute to chil-
Several social demographic factors have been dren’s behaviour problems, such as family cohesion
associated to behaviour problems in children. Par- or environment quality and maternal affective dis-
ental age, and especially younger maternal age, is orders. Maternal affective disorder increases signific-
associated with a higher probability of physical, antly the risk of a child developing psychopathology
emotional and cognitive problems in children. For from preschool age onwards (Caplan et al., 1989).
example, there is evidence that younger mothers There is ample epidemiological evidence suggesting
show a distinct interaction pattern with their child, that mental illness is intergenerational (Weissman
being less responsive, showing less tendency to en- et al., 1987). However, it has been difficult to specify
gage in affective interactions and giving less lin- the precise nature of this association between par-
guistic stimulation to the child (Osofsky, Hann, & entsÕ and children’s mental disorder (Murray, 1992).
Peebles, 1993). Another important factor is parental The mediating mechanisms through which this
education, since it is a family social and cultural transmission occurs are considered more powerful
indicator, as well as having affective implications for than parental pathology itself and have received

Association for Child Psychology and Psychiatry, 2004.
Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA
780 Luciana Anselmi et al.

attention in the literature (Rutter, 1996). For exam- 1990), and experience a feeling of loss of the perfect
ple, attachment theory (Bowlby, 1982) has contri- baby (Minde, 1993).
buted towards the understanding of the mechanisms The child’s sex is another individual factor which
through which maternal psychopathology may be has been associated with behaviour problems. Males
associated with poor infant development through have been considered a risk group in studies of child
disorders in the dyad’s relationship. Studies have mental health. In a recent review on attention deficit
indicated that maternal mental illness is associated disorder, Rohde et al. (1998) explains that the ratio
with poor affective interaction and an insecure of boys and girls affected is around 2:1 in popula-
attachment relationship (Goodyer, 1990). Large-size tion-based studies, and up to 9:1 in clinical trials.
families have also been considered as a risk factor for This difference in ratio in study designs is probably
the development of behaviour problems. This due to the fact that girls present fewer behaviour
association has been attributed, in particular, to symptoms in comorbidity, causing less discomfort to
financial and educational resources worn down to families and school, and therefore receiving less
the point of impairing parental functions (Dubow & consideration for treatment. There is evidence
Luster, 1990) or to the imitation of deviant beha- showing that boys are more exposed to family hos-
viours among siblings (Lewis, 1995). The birth order tility than girls – who are more protected from these
among siblings has also been widely studied, the situations – and have more biological vulnerability to
firstborn being considered as a risk group for beha- developing physical illnesses (Rutter, 1996). This
viour problems. There is evidence that firstborn greater propensity to physical illness may be an
children are more frequently referred by parents for important factor in the development of boys, espe-
psychological assistance (Gimeniz & Silvares, 1993). cially when the illness requires hospitalisation.
They not only receive greater care from the parents Hospitalisations during childhood have been shown
but also more negative responses which may be a to be a risk factor for child development, particularly
result of more severe patterns and greater parental in the case of children unable to understand the
expectations regarding the firstborn child (Baskett, need for hospitalisation and separation from the
1984). The birth of a sibling is a difficult transition in parents (Minde, 1993). Recurrent hospital admis-
a child’s development and has also been considered sions are associated with an increase in the risk of
a stressful event for the child, who usually reacts emotional problems, since the first hospitalisation
with anxiety, feelings of abandonment and anger predisposes the child to react in an adverse manner
(Winnicott, 1964/1966). These reactions have been on the next occasion (Rutter, 1981). Some studies
seen, together with the reduction of attention and of have suggested that between the first and fourth
maternal individual care, an increase in negative and years of life the child may be especially sensitive to
controlling interactions directed towards the child hospitalisation (Mrazek, Anderson, & Strunk, 1985),
and a decrease in the quality of attachment to the given his/her cognitive limitations to deal with
child who has gained a brother or sister (Teti, Sakin, stressful events. Children belonging to ethnic mi-
Kucera, Corns, & Eiden, 1996). Concern with the norities tend to have more behaviour problems than
new baby has been frequently associated with the those belonging to dominant ethnic groups (Velez,
decrease in attachment behaviour towards the other Johnson, & Cohen, 1989), even in countries where
children (Goodyer, 1990). non-white groups are not minorities (Bird et al.,
Finally, individual factors of the child him/herself 1988). Stress due to stigmatisation created by racial
(i.e., prematurity, sex, illnesses, ethnic background prejudice plays an important role in determining this
and IQ) are always investigated as being associated prevalence (Miranda, 1996). Clinical experience in-
with child behaviour problems. Studies on perinatal dicates a similarity among races, but differences
predictors of behaviour problems in children have among different socioeconomic levels, being more
mainly investigated birth weight and gestational age frequent in less privileged classes and in countries
but their consequences have also been attributed to where the population falls predominantly into the
psychological factors related to the infant’s interac- lower SES groups. Nevertheless, a low SES may not
tion with the environment and not only to biological necessarily be the determining factor, since other
risk itself. Therefore, parental care practices are determinants of psychopathology may also be more
different when the infant is preterm or of low weight prevalent in families of low SES (Bird et al., 1988).
and the mother–preterm infant interaction is im- The child’s IQ has also been associated with child
paired. Mothers of preterm babies show greater behaviour problems. The literature has revealed
anxiety and little confidence in their ability to care that, in general, children with an IQ above the
for the baby (Crnic, Greenberg, Ragosin, Robinson, average have lower rates of psychiatric morbidity
& Basham, 1983). They do not know how to stimu- (Rutter, 1981), whereas children with an IQ below
late the babies properly since they change rapidly the average are more vulnerable to the development
from a state of hypo- to hyperactivity (Barnard, Bee, of behaviour problems, showing a greater prevalence
& Hammond, 1984), they perceive their babies as (Grizenko, Cvejic, Vida, & Sayegh, 1991). A possible
less physically developed, less active, less responsive explanation for these differences is that children
and less intellectually competent (Stern & Karraker, with a high IQ have more ability to solve problems
 Psychosocial determinants of behaviour problems in Brazilian preschool children 781

and to cope with environmental stressors (Dubow & located and visited. Approximately half of this sub-
Luster, 1990). However, children who are cognitively sample (n ¼ 634) was randomly selected in order to
more advanced than their peers may cope with investigate behaviour problems in the child. Out of the
frustrations which also generate behaviour problems total number of children selected there was only one
(Achenbach, 1991). loss (due to residential mobility) and one refusal to
carry on with the study. The mean age of the children
The first aim of the present study was to in-
was 4 years and 5 months (SD ¼ 3.6 months). As low
vestigate the prevalence of behaviour problems in birth-weight children were over-sampled at 6 and
Brazilian preschool children. Studies using the Child 12 months, they were down-weighted on every analysis
Behavior Checklist (CBCL) have described the pre- to correct the over-sampling.
valence of behaviour problems in developed coun- Table 1 shows the social demographic characteristics
tries (i.e., Achenbach & Howell, 1993), but there are of the children from the sample and their families. As
few studies on their prevalence in developing coun- can be seen, the sample is representative of the target
tries. Furthermore, several of the existing studies population recruited when the infant was born.
involve school-age children and fewer studies have
investigated prevalence in preschool children. The
Design and procedures
second aim was to investigate which of the factors
comprising each of the three domains suggested by A longitudinal design of an epidemiological nature was
Bronfenbrenner (1994/1996) – social, family and used in which the participants were followed prospec-
individual organisation – are significant predictors of tively from birth to the child’s fourth year of life. During
behaviour problems in children. this period, several visits to the familiesÕ homes were
carried out when data were collected on growth, mor-
bidity, development, and feeding habits, as well as so-
cial demographic and family data (Victora et al., 1996).
Method In Phase I of the study, the mothers and the new-
Participants borns (5304) were visited in the hospital when mothers
answered a questionnaire, and the babies were sub-
Six hundred and thirty-four children and their mothers mitted to the evaluation of gestational age. In the in-
took part in the study which involved 5304 hospital fant’s sixth and twelfth month of life a sub-sample of
births occurring during the calendar year 1993 in families (n ¼ 1363) were visited at home. On both
Pelotas, Brazil (population 300,000), where over 99% of occasions mothers answered a questionnaire evaluat-
city births take place in hospitals. The history of the ing social demographic and infant health and develop-
study and of the sample was described by Victora et al. ment factors. The data collection in the twelfth month
(1996). In Phase I of the study, sub-samples of the co- also included an anthropometric evaluation and an
hort were followed up at 1, 3, 6 and 12 months. A total assessment of neuropsychomotor problems.
of 655 children, chosen by systematic sampling with a In Phase II, 1273 families were located and visited.
fixed weekly quota, were scheduled to be visited at 1 The mother answered a questionnaire assessing social
and 3 months. Of these, 649 (99.1%) were located demographic and infant health and development fac-
successfully at 1 month and 644 (98.3%) at 3 months. tors and a later visit was agreed for a randomly selected
At the age of 6 and 12 months, 1460 infants were half of the sample (n ¼ 634 children). One week later, a
selected to be visited, representing a 20% systematic psychologist made another visit at a time when the
sample of the whole cohort (including the 655 children mother and the child were present. Initially the mothers
studied at 1 and 3 months) plus all children born with a answered a questionnaire evaluating family composi-
birth weight below 2500 g. Of these 1460 children, tion and an assessment of the home environment was
1414 (96.8%) were located at 6 months and 1363 also carried out. Next, the child’s IQ was assessed and
(93.4%) at 12 months. the mother evaluated the child’s behaviour problems.
In Phase II, when children were around four years Finally, the mother answered a questionnaire on
old, 1273 (93.6% of the sub-sample of 12 months) were maternal psychiatric disorders.

Table 1 Social demographic characteristics of the children and families from the sample

Characteristics of the children Birth 4th year

and their families at birth (N ¼ 5304) (±SD) (N ¼ 634) (±SD)

Girls 50.3 50.5%

White children 77.4% 76.4%
Family income (minimum wage) 4.34 (5.8) 4.45 (6.8)
Maternal education (total years) 7 (3.6) 7 (3.7)
Maternal age* 26 (6.4) 27 (6.4)
Paternal education (total years) 7 (3.6) 7 (6.6)
Paternal age 30 (7.7) 30 (6.6)
Father absence 12.1% 10.5%
Number of siblings 1 (1.5) 1 (1.6)

*13.4% of mothers were adolescents (<20 years old).

Chi-square for proportions and analysis of variance for means.
782 Luciana Anselmi et al.

A pilot study included a home visit by interviewers to and gross motor abilities, of language and aspects of the
30 preschool children from the community (around 5% child’s socialisation.
of the sample size) in order to test the administering of
the instruments concerning the child’s fourth year of Phase II (fourth year of the child). Family composi-
life. For the purpose of quality control, all instruments tion: a structured questionnaire was administered.
were re-administered to 5% of the sample, randomly Information regarding number of younger siblings and
selected. Most of the instruments described below re- whether the mother’s partner lived with her were used.
quired judgements by raters. In all cases the Home environment: the quality of the home environment
researchers were trained for evaluation until they was assessed through the Home Observation for
reached minimum reliability criteria (for categorical Measurement of the Environment – HOME (Bradley &
judgements, K ‡ .70 coefficient; for continuous meas- Caldwell, 1984). The version for preschool children
ures, interclass coefficients ‡ .80). Raters did not have comprises 45 items arranged in eight subscales which
knowledge of the results of their colleagues or of the evaluate the socio-emotional and cognitive support
aims of the study. available in the home environment. They include:
stimulation with toys, games and reading material,
Phase I (from birth to 12 months). Evaluation of linguistic stimulation, physical environment, mother–
perinatal factors: a structured questionnaire was child affective interaction, stimulation for learning and
administered to the mother of the newborn in order to academic behaviour, model of conduct, variety of
examine her reproductive history, as well as social stimulation and experiences, discipline and physical
demographic, socioeconomic and environmental fac- punishment. All items were rated in terms of good or
tors. The infant’s birth conditions were complemented poor family environment. IQ assessment: the WPPSI
by information from the hospital record. For the present (Wechsler Preschool Intelligence Scale) intelligence
study, information concerning family income (sum of all scale for preschool children, developed by Wechsler
earnings of the people living in the household), parental (1963/1991), was used. As the administration of the
age and education, father presence, birth weight and WPPSI is long and therefore not adequate for epidemi-
the child’s perinatal complications was used. Gesta- ological studies, a brief form of the test developed by
tional age: through the Dubowitz method (Dubowitz & Kaufman (1972) was used. It comprises two verbal
Goldberg, 1970), in the first day of life. The examination subtests (comprehension and arithmetic) and two exe-
consisted in evaluating the newborn against neurolo- cution subtests (figure completion and construction
gical and somatic criteria, which are assessed accord- with cubes). It was adapted to Portuguese by Cunha
ing to a previously established scale. Social (1992). Maternal psychiatric disorders: the Self-Report
demographic and infant development factors: this Questionnaire of Minor Psychiatric Disorders – SRQ-20
questionnaire aimed to obtain information on social (Harding et al., 1980), adapted to Portuguese by Mari &
demographic factors, infant growth and morbidity, Williams (1986) was used. It is a scale comprising 20
hospitalisations, non-home care of the child, medical items aiming to detect minor psychiatric disorders
sanitary assistance, feeding habits and early weaning. through the investigation of non-psychotic symptoms.
Similar versions of this questionnaire were answered by Assessment of behaviour problems: the Child Behaviour
the mothers in the infant’s sixth and twelfth months of Checklist – CBCL/4-18 – Achenbach (1991) was used.
life, as well as in the data collection in the fourth year. The 112 items consist of behaviour descriptions refer-
From the data collection in the sixth month, informa- ring to a list of childhood psychopathological symp-
tion was obtained concerning maternal work (number toms. The following syndromes are investigated in nine
of daily hours worked outside the house). In the data subscales: withdrawal, somatic complaints, anxiety/
collection in the twelfth month, information was gath- depression, problems with social contact, thought
ered regarding paternal care (sum of the points ob- problems, attention problems, delinquent behaviour,
tained from a list of care given to the child by the father aggressive behaviour and sexual problems. Mothers
in the previous week: e.g., played, fed the child, put the were interviewed and the interviewer circled a 0 if the
child to sleep, changed diapers, gave bath, cared for, item is not true, a 1 if the item is somewhat or some-
went out together). From the data collection in the times true, and a 2 if it is often true. A total problem
fourth year, information on the child’s bed-sharing with score is computed by adding up all 0s, 1s and 2s. Raw
parents, ethnic group and hospitalisations (sum of all scores are treated mathematically, from statistics
hospitalisations of the child from the first to the fourth extracted from the general population in North
year of life) was used. The child’s skin colour was ob- American standardisation, in order to get ÔT-scoresÕ. A
served by the interviewer and the non-white group cut-off point is used for these scores in order to obtain
included blacks and mixed race. Anthropometric evalu- a clinical judgement. Two categories are derived:
ation: children were weighed in portable scales with ÔclinicalÕ and Ônon-clinicalÕ. The sum of the raw scores
100 g precision, which were calibrated weekly with obtained in all subscales corresponds to the total of
standard weights. The child’s length was measured in behaviour problems. Specific scores on externalising
the supine position with standard equipment, an and internalising problems can also be derived. The
AHRTAG infanthometer (Victora, Barros, & Vaughan, CBCL was adapted and validated in Brazil by Bordin,
1988). To assess the child’s nutritional state in the Mari, and Caeiro (1995). Data on a CBCL validation
twelfth month his/her length for his/her age was con- study for the Brazilian population testified to a good
sidered. Neuropsychomotor assessment: the Denver II sensibility level (87.0%–75.0% for tenuous cases,
screening test (Frankenburg, Dodds, Archer, & Bres- 95.4% for moderate and 100% for severe) using the
nick, 1990) was administered when the child was twelve cut-off point recommended for American population,
months old. The test consists of the assessment of fine but this study could not measure specificity due to the
 Psychosocial determinants of behaviour problems in Brazilian preschool children 783

small number of healthy children drawn from the Table 2 Correlation between social demographic, family and
sample. As an instrument calibration has not been individual factors with CBCL T scores (n ¼ 634)
performed in Brazil using a large population-based
Factors r
sample of children, the cut-off point used on the
ÔT-scoreÕ is the same as that derived from the North 1 CBCL/behaviour problems –
American population. As it is a widely and interna- Social demographic
tionally used instrument and already validated in dif- 2 Family income/birth ).01
ferent countries, its application on this study is also 3 Maternal age/birth ).19**
explained by the multicultural validity of the instru- 4 Maternal education/birth ).08*
ment (Bird, 1996). 5 Paternal age/birth ).13**
6 Paternal education/birth ).03
7 Father presence/birth .11**
8 Mother’s partner/4th yr ).06
Results 9 Maternal work/6 months .03
Family
Based on the CBCL (T-scores from 26 to 89), the 10 Paternal care/1 year ).10*
prevalence of children with behaviour problems in 11 SRQ/Maternal Psych Disorder/4 yr .48**
12 Number of younger siblings/4 yr .20**
the sample, summing up the clinical (T-score >63)
13 Bed sharing with parents/4 yr ).01
and borderline groups (T-score ‡60 and £63), was 14 Home/family envrionment/4 yr ).21**
24% (n ¼ 153), and 15% (n ¼ 99) considering only 14.1 /Affective interaction ).22**
the clinical group. Examining the two scales that 14.2 /Play ).14**
comprise the CBCL, it is found that externalising 14.3 /Experiences ).11**
Individual
problems show greater prevalence (clinical and bor-
15 Birth weight ).08*
derline groups ¼ 31.8%) than internalising problems 16 Gestational age ).06
(clinical group and borderline ¼ 15.2%). 17 Perinatal complications ).01
In the present study, a chi-square test revealed 18 Neuropsychomotor problem/1 yr .03
no significant differences regarding children’s sex 19 Number of hospitalisations/1–4 yr .18**
20 Nutritional state/1 yr ).05
(boys ¼ 21.9%; girls ¼ 26.2%). Moreover, no signi-
21 WPPSI/QI/4 yr ).16**
ficant differences were found in the prevalence of
behaviour problems as far as the child’s ethnic *p £ 05.
group is concerned (white ¼ 23.6%; non-white ¼ **p £ 01.
25.7%).
One of the aims of the present study was to this scale were associated with a lower number of
investigate the association between social demogra- behaviour problems.
phic, family and individual factors and the child’s Among the factors associated with the child him/
behaviour problems. Table 2 shows Pearson corre- herself, the number of hospitalisations (none to 5
lations between the factors investigated. hospitalisations) was shown to be significantly cor-
Among the social demographic factors, maternal related with the CBCL. The greater the number of
age (14 to 46 years), paternal age (15 to 75 years) hospitalisations, the greater the number of beha-
and maternal education (0 to 17 years) were signifi- viour problems. The child’s birth weight (960 to
cantly correlated with the CBCL scores, indicating 4.700 g) and his/her IQ (53 to 144 points) were
that the greater the parentsÕ age and maternal edu- shown to be negatively correlated with the CBCL,
cation, the lower the scores of behaviour problems. indicating that the greater the weight at birth and the
Father presence also correlated with lower scores of greater the child’s IQ, the fewer behaviour problems
behaviour problems. were shown by the child.
As far as family factors are concerned, Table 2 Given the diversity of factors correlated with the
shows that an indicator of maternal psychiatric child’s behaviour problems, a hierarchical multi-
disorder (SRQ score from 0 to 19) is most correlated variate linear regression was carried out aiming at
with behaviour problems, followed by home envir- examining the variance explained by these factors.
onment quality (HOME score from 15 to 53), the The conceptual model of the multivariate analysis
number of younger siblings (none to 3 siblings) and was based on the bio-ecological model of develop-
paternal care (score from 0 to 7). The greater the ment (Bronfenbrenner, 1994/1996), which allowed
prevalence of maternal psychiatric disorder and the for considering behaviour problems as a hierarchical
greater the number of younger siblings, the higher determination phenomenon from a psychological
the scores of behaviour problems. On the other perspective. It was also based on time considera-
hand, the better the home environment quality and tions, such as year of data collection and children’s
the more the father cared for the child, the lower the first or fourth year of life. The hierarchical relation-
prevalence of behaviour problems. The three sub- ship established between the factors follows the dif-
scales of the HOME show the same pattern of signi- ferent levels of organisation proposed by the author
ficant negative correlations with the CBCL. The (social, family and individual systems) as far as the
subscale affective interaction in particular was cor- most distant and most proximal determinants of
related with the CBCL, indicating that high scores in child development are concerned.
784 Luciana Anselmi et al.

Table 3 Summary of the hierarchical linear regression analysis of CBCL T scores (n ¼ 634)

Total R2
Step of regression Predictor factors B SE ß ß T sig adjusted**

1 Social demographic factors (birth) Maternal age ).15 .05 ).10 .00
Maternal education* )1.31 .51 ).10 .01 .04
2 Child factors (1st–4th year) No. of hospitalisations 1.64 .46 .13 .00 .07
3 Family factors (4th year) Maternal psychiatric 1.03 .08 .43 .00 .27
disorder – SRQ
No. of younger siblings 1.35 .58 .09 .02
4 Home environment quality (4th year) HOME ).14 .06 ).09 .04 .28

*Maternal education in years: no education ¼ 0; 1 to 4 ¼ 1; 5 to 8 ¼ 2; 9+ ¼ 3.

**F-test: F £ 001.

In the first step of the equation, social demo- variance changed to 7% with the inclusion of factors
graphic factors were introduced (maternal age and of the child him/herself, such as birth weight and
education and father presence). Paternal age did not the number of hospitalisations and of paternal care
integrate with the model because it correlated too in the first year of life. Of these factors, only the
highly with maternal age. These factors are consid- number of hospitalisations remained significant in
ered distant determinants as they usually have an the equation after controlling the effects on the CBCL
indirect effect on the child, acting through proximal score of the other factors included in the same level,
factors, and because, theoretically, they do exist as well as in the previous level of the equation. The
independent from the child. In addition, social entry into the equation of family factors regarding
demographic data were collected when the baby was the number of younger siblings and maternal psy-
born. chiatric disorder (SRQ) considerably increased the
As they were collected in the first year, some fac- variance explained by the model, reaching 27%. In
tors regarding the child him/herself, such as birth reality, this substantial increase occurred mainly
weight, number of hospitalisations and paternal due to the entry of the SRQ into the equation, which
care, were introduced in the second step of the by itself increased the variance by 19%, thus
equation. In a third step, information on maternal reaching 26%. The introduction of the remaining
psychiatric disorder and number of younger siblings factors, home environment quality (HOME) and
were introduced. These data were collected when the child’s IQ, increased the variance explained by only a
child was four and influence the child, but are less little, and only HOME remained significant in the
influenced by the child. Finally, in the fourth step of equation. Together, all the factors explained 28% of
the equation, information concerning the quality of the variance in the child’s behaviour problems.
family environment and the child’s IQ were intro-
duced, both collected when the child was four. These
factors are considered proximal determinants of
Discussion
development because the child plays an active role in
this relationship. Besides children’s characteristics, The prevalence of behaviour problems in the present
they refer to other people and objects in the close study was quite high (24%) and is similar to the one
environment. found in other Brazilian epidemiological studies,
Table 3 briefly illustrates the findings from the such as Almeida Filho (1985) and Miranda (1996).
hierarchical linear regression analysis. The first found a prevalence of 23.4% in children
The results indicate that for each additional year of aged 5 to 14 and the second found a prevalence of
maternal age and education, the mean scores on the 21% using CBCL in children starting schooling. The
CBCL decreased ().15 and )1.31, respectively). Each results of the present study also show that the rate of
hospitalisation led to an increase of 1.64 in the CBCL children classified in the clinical group (15%) is
mean scores. For each higher score on the SRQ and similar to the prevalence of psychiatric disorders
each additional sibling, the mean scores on the (16%) in preschool population-based samples
CBCL increased (1.03 and 1.35, respectively). On the (Gould, Wunsch-Hitzig, & Dohrenwend, 1981).
other hand, for each higher score on the HOME, Apart from the overall prevalence of behaviour
the mean on the CBCL decreased ().14). problems, the results showed high prevalence of
The social demographic factors included in the externalising (31.8%) and internalising behaviour
analysis (maternal age and education and father problems (15.2%), two categories of the CBCL. The
presence) explained 4% of the variance of the CBCL. greater prevalence of problems of externalisation in
Among these factors, maternal age and education the CBCL tends to decrease with increasing age,
remained significant in the equation after controlling which did not happen with the children from the
the effects on the CBCL of the other factors included present study, who were only four years old and
in the same level of the equation. The value of the were found to be in the minimum age limit of the
 Psychosocial determinants of behaviour problems in Brazilian preschool children 785

instrument. At this age, the rates tend to be high correlated with the behaviour problem scores (11 out
because the symptoms are usually confounded with of the 20 studied) and the effects found followed the
manifestations pertaining to the younger child’s expected direction, showing a trend consistent with
developmental level, such as crises of aggressiveness the literature on the psychosocial determinants of
and disobedience, which are usual in the young behaviour problems.
child who is searching for autonomy and needs to Many of the factors present in each of the three
affirm his/her position. This can make it difficult for levels studied, social demographic, family and indi-
mothers to discriminate common reactions to the vidual, also contributed to the prediction of the
child’s developmental phase from eventual psycho- child’s behaviour problems as seen in the regression
pathological symptoms. Moreover, children with analysis. The final regression equation showed that
externalising problems tend to present more cognit- behaviour problems were significantly associated
ive deficit, school and socialisation difficulties than with two social demographic factors (maternal age
children with internalising problems. With more and education), three family factors (maternal psy-
evident characteristics, it is easier for mothers to chiatric disorder, quality of the home environment
identify externalising behaviour problems. The high and number of younger siblings) and one individual
prevalence of externalising problems may also be factor (number of hospitalisations). These results
explained by the fact that a number of child patho- corroborate those of other studies reported in the
logies manifest themselves as conduct problems, literature which identified maternal age and educa-
using action to express conflicts and anxieties tion (Dubow & Luster, 1990), presence of maternal
(Campbell & Ewing, 1990). psychiatric disorder (Seifer et al., 1996), quality of
The present study did not reveal any association the home environment (Bastos & Almeida Filho,
between sex and ethnic group and the child’s beha- 1990), number of younger siblings (Teti et al., 1996)
viour problems. This is in contrast to the findings of and hospitalisations (Rutter, 1981) as factors asso-
Velez et al. (1989), which showed a greater preval- ciated with the child’s behaviour problems.
ence of behaviour problems among African American Among family factors, it was maternal psychiatric
and Hispanic children. The absence of ethnic differ- disorder, assessed in the child’s fourth year of life,
ences in the present study may be related to the that most explained the variance in behaviour
miscegenation found in Brazil, making black chil- problems. This was the factor most correlated with
dren less stigmatised and not more vulnerable than the child’s behaviour problems when analysed in
white children living under the same socioeconomic isolation, as well as when adjusted for the other
conditions. Another hypothesis is that since in this factors comprising the final equation of the multi-
age group children do not go to school and live most variate regression analysis. It is understood that
of the time in the home environment, they have not maternal psychiatric disorder may affect the child,
experienced the difference and consequent stigma- mainly through her interactional pattern. The
tisation which may provoke stress and make them greater the suspect of maternal psychiatric disorder,
vulnerable to developing emotional disorders. Fur- the lower the quality of the home environment as
thermore, there are methodological differences in the assessed by the HOME (r ¼ ).27), the less demon-
definition of ethnic groups. Some studies use the stration of positive affect on the mother’s part
observation of the interviewer, as in the present (r ¼ ).15), the less stimulation with play (r ¼ ).24)
study, whereas in others the mother informs the and the less variety of experiences (r ¼ ).26) offered
interviewer. This tends to produce different results, to the children. The results of the present study
making comparisons between studies difficult. As far suggest that maternal psychopathological symptoms
as the lack of association between sex and total score constitute an important factor to be investigated for
of behaviour problems is concerned, this was also determining the child’s behaviour problems. How-
found in other studies which have used the CBCL in ever, when the relationship between psychiatric
preschool (Samantiego, 1995; Rae-Grant, Thomaz, symptoms in the mother and in the child is dis-
Offord, & Boyle, 1989) and school children (Jensen, cussed, one ought to be cautious regarding the
Bloedau, Degroot, Ussery, & Davis, 1990). Overall, direction of the effect. It is important to note the
studies have been inconsistent, although the bulk of dynamic character of this association and the mu-
the evidence suggests that gender differences are not tual influence which both the mother and the child
marked in preschool children. This is so, despite the exert on each other (Sameroff, 1993). One can infer
converging evidence in school-age children that in- that not only may the parentsÕ behaviour have a role
dicates higher rates of externalising problems in in the determination of the child’s behaviour prob-
boys, as well as a shift towards more internalising lems but also the presence of an emotionally dis-
problems in girls by early adolescence. Un- turbed child may affect the family dynamics,
fortunately, studies of preschoolers do not appear to especially the parents. Even though it is the parents
clarify when these gender differences emerge who more extensively direct the interaction with
(Campbell, 1994). young children, the child’s characteristics interact
Several of the factors investigated from the infant’s with the quality of parenting and this may be at
birth to the fourth year of life were significantly the origin of behaviour problems. The relationship
786 Luciana Anselmi et al.

between low maternal age and child behaviour indicating the continuity of the adverse situation.
problems, found in the present study, suggests the Therefore, the children who lived with poor and low-
importance of adolescent pregnancy prevention education mothers in the first year of life carried on
programmes. living in this situation when they were four years old.
The number of younger siblings was also signific- Therefore it is difficult to determine whether the ef-
antly correlated with the behaviour problems of the fect of certain factors on the child’s behaviour prob-
children who took part in the present study. The lems is due to the previous adversity, the present
literature has pointed out that an increase in the adversity or, more probably, both. The results of the
number of children tends to lead to a decrease in present study suggest that the presence of behaviour
attention and individual responsiveness to the child problems is not determined solely by individual or
by mothers who are dealing with younger children social demographic factors present at the child’s
(Teti et al., 1996). This can be shown through the birth, but also by present conditions, especially the
results of the present study which revealed that experiences in the home environment. In reality,
the greater the number of children, the poorer the child development is also a product of the child’s
quality of the family environment (r ¼ ).21), affective developmental history, which includes not only
interaction (r ¼ ).23), stimulation with play genetic characteristics and past experiences but also
(r ¼ ).22) and variety of experiences (r ¼ ).22) pro- present circumstances (Bowlby, 1982).
vided to the child. The results of this study showed that several fac-
Finally, among the child’s individual factors tors present in each of the three domains suggested
examined, only the number of hospitalisations was a by Bronfenbrenner (1994/1996) constitute signific-
predictor of their behaviour problems. Hospitalisa- ant predictors of the child’s behaviour problems. The
tion has traumatic consequences for the child, rep- findings corroborate the perspective of multi-deter-
resenting a stressful event in his/her life which may mination of children’s behaviour problems, sug-
contribute to the onset of behaviour problems. Hos- gesting that they are the result of the interaction of
pitalisation may also be an indicator of an unfa- factors of different origins which operate concomit-
vourable physical condition in the child and the antly (Sameroff, Seifer, Baldwin, & Baldwin, 1993).
findings would therefore indicate an expected co- Therefore, being exposed to a great number of ad-
morbidity of physical and mental illnesses, already verse conditions, children of developing countries,
widely published in studies on infant mental health such as those of this sample, end up showing a high
(Eiser, 1990). prevalence of behaviour problems.
The fact that birth weight and gestational age did
not appear to be associated with the child’s beha-
viour problems may be related to the fact that they
Acknowledgements
do not have a long-standing effect, affecting the
child mainly in the first years of life. In the pre- The study received grant support from PRONEX/
school years, social and family factors gain more CNPq, Brazil, PPGE-UFPel, and was partially sub-
importance, outweighing the eventual effect of bio- mitted as a Master’s thesis by the first author to the
logical factors. Another hypothesis to be examined PPG Psychology/UFRGS/Brazil, under supervision
in future studies is that the biological factors of the second author.
operate as a vulnerability factor and, as a conse-
quence, only exert influence when other stressors
are present. Correspondence to
The results of the present study show that the
Luciana Anselmi, FURG, Rua Eng. Alfredo Huch
greater part of the variance of the behaviour prob-
475, DECC, CEP: 96201900. Rio Grande – RS –
lems was explained by factors associated with the
Brazil; Email: luciana.anselmi@furg.br
social and psychological context of the family mem-
bers and less by the characteristics intrinsic to the
child him/herself. The associations found between
References
the social demographic factors present at the child’s
birth and the behaviour problems in his/her fourth Achenbach, T.M. (1991). Manual for the Child Behaviour
year of life support the idea that certain early Checklist/4-18 and 1991 profile. Burlington, VT:
experiences increase the risk of psychiatric disorders University of Vermont,
in later developmental periods (Knorring, Bohman, & Achenbach, T.M., & Howell, C.T. (1993). Are American
children’s problems getting worse? A 13-year com-
Sivgardsson, 1982). Moreover, these associations
parison. Journal of the American Academy of Child
show the importance of the first year of life in child
and Adolescent Psychiatry, 32, 6.
development (Hay & Kumar, 1995). These results Almeida Filho, N. (1985). Epidemiologia das desordens
may also be interpreted considering the continuity in psiquiátricas da infância no Brasil. Salvador: Centro
the risk situation in which the child lives. For ex- Editorial e Didático da UFBA.
ample, family income and parental education were Barnard, K., Bee, H., & Hammond, M. (1984). Develop-
similar in the child’s first and fourth year of life, mental changes in maternal interactions with term
 Psychosocial determinants of behaviour problems in Brazilian preschool children 787

and preterm infants. Infant Behaviour and Develop- mento de crianças brasileiras a uma clı́nica-escola de
ment, 14, 203–215. Psicologia. Revista Interamericana de Psicologia, 28,
Baskett, L.M. (1984). Ordinal position differences in 61–72.
children’s family interactions. Developmental Psy- Goodyer, I.M. (1990). Family relationships, life events
chology, 20, 1026–1031. and childhood psychopathology. Journal of Child
Bastos, A.C.S., & Almeida Filho, N. (1990). Variables Psychology and Psychiatry, 31, 161–192.
económicosociales, ambiente familiar y salud mental Gould, M.S., Wunsch-Hitzig, R., & Dohrenwend, B.
infantil en un área urbana de Salvador (Bahia), (1981). Estimating the prevalence of childhood psy-
Brasil. Acta Psiquiatrica Psicologica da América chopathology: A critical review. Journal of the Amer-
Latina, 3, 147–154. ican Academy of Child and Adolescent Psychiatry, 20,
Bird, H.R., Canino, G., Rubio-Stipec, M., Gould, M.S., 462–476.
Ribera, J., Sesman, M., Woodbury, M., Huertas, S., Grizenko, N., Cvejic, H., Vida, S., & Sayegh, L. (1991).
Pagan, A., Sanchez-Lacay, A., & Moscoso, M. (1988). Behaviour problems of the mentally retarded. Cana-
Estimates of the prevalence of childhood maladjust- dian Journal of Psychiatry, 36, 712.
ment in a community survey in Puerto Rico. Archives Halpern, R. (1993). Poverty and infant development. In
of General Psychiatry, 45, 1120–1126. Charles H. Zeanah Jr. (Ed.), Handbook of infant
Bird, H.R. (1996). Epidemiology of childhood disorders mental health (pp. 73–86). New York: The Guilford
in a cross-cultural context. Journal of Child Psycho- Press.
logy and Psychiatry, 37, 35–49. Hay, D., & Kumar, R. (1995). Interpreting the effects of
Bordin, I., Mari, J., & Caeiro, M. (1995). Validação da mothersÕ posnatal depression on children’s intelli-
versão brasileira do ÔChild Behavioural ChecklistÕ gence: A critique and re-analysis. Child Psychiatry
(CBCL) (Inventário de Comportamentos da Infância and Human Development, 25, 165–181.
e Adolescência): Preliminary data. Revista ABP-APAL, Jensen, P.S., Bloedau, L., Degroot, J., Ussery, T., &
17, 55–66. Davis, H. (1990). Children at risk: I. Risk factors and
Bowlby, J. (1982). Attachment and loss: Vol. 1. Attach- child symptomatology. Journal American Academy of
ment (2nd edn). New York: Basic Books. Child and Adolescent Psychiatry, 29, 51–59.
Bradley, R.H., & Caldwell, B. (1984). Home Observa- Kaufman, S. (1972). A short form of the Wechsler
tion for the Measurement of Environment (rev. edn). Preschool and Primary Scale of Intelligence.
Mimeo, AR: University of Arkansas. Journal Consulting and Clinical Psychology, 39,
Bronfenbrenner, U. (1996). A ecologia do desenvolvi- 361–369.
mento humano: Experimentos naturais e planejados Knorring, A.L., Bohman, M., & Sigvardsson, S. (1982).
(M.A. Veronese, Trad.). Porto Alegre: Artes Médicas Early life experiences and psychiatric disorders: An
(Original paper published in 1994). adopted study. Acta Psychiatrica Scandinava, 65,
Campbell, S.B., & Ewing, L.J. (1990). Hard-to-manage 283–291.
preschoolers: Adjustment at age nine and predictors Lewis, M.D. (1993). Emotion–cognition interactions in
of continuum symptoms. Journal of Child Psychology early infant development. Cognition and Emotion, 7,
and Psychiatry, 31, 871–889. 145–170.
Campbell, S.B. (1994). Behaviour problems in pre- Lewis, M. (1995). Tratado de Psiquiatria da Infância e
school children: A review of recent research. Journal Adolescência. (I.S. Trad.). Porto Alegre: Artes Médi-
of Child Psychology and Psychiatry, 36, 113–149. cas. (Original paper published in 1991).
Caplan, H.L., Cogill, S.R., Alexandra, H., Robson, K.M., Mari, J., & Williams, P. (1986). A validity study of a
Katz, R., & Kumar, R. (1989). Maternal depression psychiatric screening questionnaire (SRQ-20) in prim-
and the emotional development of the child. British ary care in the city of São Paulo. British Journal of
Journal of Psychiatry, 154, 818–822. Psychiatry, 118, 23–26.
Crnic, K.A., Greenberg, M.T., Ragosin, A.S., Robinson, Minde, K. (1993). Prematurity and serious medical
N.M., & Basham, R.B. (1983). Effects of stress and illness in infancy: Implications for development and
social support on mothers and premature and full- intervention. In C.H. Zeanah Jr. (Ed.), Handbook of
term infants. Child Development, 54, 209–217. infant mental health (pp. 87–105). New York: The
Cunha, J.A. (1992). Manual do WPPSI: Administração e Guilford Press.
crédito dos testes. Unpublished manuscript. Miranda, M. (1996). Fatores psicossociais associados à
Dubow, E.F., & Luster, T. (1990). Adjustment of saúde mental de crianças no inı´cio da escolarização.
children born to teenage mothers: The contribution Master’s Thesis, Universidade Federal de São Paulo,
of risk and protective factors. Journal of Marriage and São Paulo.
the Family, 52, 393–404. Mrazek, D.A., Anderson, I.S., & Strunk, R.C. (1985).
Dubowitz, V., & Goldberg, C. (1970). Clinical assess- Disturbed emotional development of severely asth-
ment of gestational age in newborn infants. The matic preschool children. Journal of Child Psychology
Journal of Pediatrics, 1, 77. and Psychiatry, 26, 81–94.
Eiser, C. (1990). Psychological effects of chronic dis- Murray, L. (1992). The impact of posnatal depression
ease. Journal of Child Psychology and Psychiatry, 31, on infant development. Journal of Child Psychology
85–98. and Psychiatry, 33, 543–561.
Frankenburg, K.W., Dodds, I., Archer, P., & Bresnick, Osofsky, J.D., Hann, D.M., & Peebles, C. (1993).
B. (1990). Denver II: Technical manual and training Adolescent parenthood: Risks and opportunities for
manual. Denver: Denver Developmental Materials. mothers and infants. In C.H. Zeanah Jr. (Ed.), Hand-
Gimeniz, S.R., & Silvares, E.F.M. (1993). Relação entre book of infant mental health (pp. 106–119). New York:
ordem de nascimento e freqüência de encaminha- The Guilford Press.
788 Luciana Anselmi et al.

Rae-Grant, N., Thomaz, H., Offord, D.R., & Boyle, M.H. Parental psychopathology, multiple contextual risks,
(1989). Risk, protective factors and the prevalence of and one-year outcomes in children. Journal of Clinical
behavioral and emotional disorders in children and Child Psychology, 25, 423–435.
adolescents. Journal of the American Academy of Stern, M., & Karraker, K.H. (1990). The prematurity
Child and Adolescent Psychiatry, 28, 262–268. stereotype: Empirical evidence and implications for
Rohde, L.A., Busnello, E., Chachamovich, E., Vieira, G., practice. Infant Mental Health Journal, 11, 3–11.
Pinzon, V., & Ketzer, C. (1998). Transtorno de déficit Teti, D.M., Sakin, J.W., Kucera, E., Corns, E., & Eiden,
de atenção/hiperatividade: revisando conhecimen- R.D. (1996). And baby makes four: Predictors of
tos. Revista da Associação Brasileira de Psiquiatria- attachment security among preschool-age firstborns
APAL, 20, 166–178. during the transition to siblinghood. Child Develop-
Rutter, M. (1981). Stress, coping and development: ment, 67, 579–596.
Some issues and some questions. Journal of Child Velez, C.N., Johnson, J., & Cohen, P. (1989). A long-
Psychology and Psychiatry, 22, 323–336. itudinal analysis of selected risk factors for childhood
Rutter, M. (1996). Stress research: Accomplishments psychopathology. Journal of the American Academy of
and tasks ahead. In (R.J. Haggerty, L.R. Sherrod, Child and Adolescent Psychiatry, 28, 861–864.
N. Garmezy, & M. Rutter (Eds.), Stress, risk, and Victora, C.G., Barros, F.C., & Vaughan, J.P. (1988).
resilience in children and adolescents: Process, Epidemiologia da desigualdade. São Paulo: Hucitec.
mechanisms, and interventions (pp. 355–385). Victora, C.G., Barros, F.C., Halpern, R., Menezes, A.,
Cambridge: Cambridge University Press. Horta, B., Tomasi, E., Weiderpass, E., Cesar, J.,
Sameroff, A.J., Seifer, R., Zax, M., & Barocas, R. (1987). Olinto, M., Guimarães, P., Garcia, M., & Vaughan, J.
Early indicators of developmental risk: Rochester (1996). Estudo longitudinal da população materno-
Longitudinal Study. Schizophrenia Bulletin, 13, 383– infantil da região urbana do sul do Brasil, 1993:
394. Aspectos metodológicos e resultados preliminares.
Sameroff, A.J. (1993). Models of development and Cadernos de Saúde Pública, 30, 34–45.
developmental risk. In C.H. Zeanah Jr. (Ed.), Hand- Wechsler, D. (1991). Test de Inteligencia Para Preescol-
book of infant mental health (pp. 120–142). New York: ares (WPPSI) Manual. Buenos Aires: Paidos.(Original
The Guilford Press. paper published in 1963).
Sameroff, A.J., Seifer, R., Baldwin, A., & Baldwin, C. Weissman, M.M., Gammon, D.G., John, K., Merinkas,
(1993). Stability of intelligence from preschool to K.R., Warner, V., Prusoff, B.A., & Sholomskas, D.
adolescence: The influence of social and family risk (1987). Children of depressed parents: Increased
factors. Child Development, 64, 80–97. psychopathology and early onset of major depression.
Samantiengo, V.C. (1995). El Child Behaviour Checklist: Archives of General Psychiatry, 44, 847–853.
su estandartización y aplicación en un estudio epi- Winnicott, D.W. (1966). A criança e o seu mundo.
demiológico: Problemas comportamentales y sucessos (A. Cabral, Trad.). Rio de Janeiro: Zahar. (Original
de vida en niños de 6 a 11 años de edad. Master’s paper published in 1964).
Thesis, Universidad de Buenos Aires, Buenos Aires.
Seifer, R., Sameroff, A., Dickstein, S., Keitner, G., Manuscript accepted 24 June 2003
Miller, I., Rasmussen, S., & Hayden, L. (1996).