Soc Psychiatry Psychiatr Epidemiol (2011) 46:787–796
DOI 10.1007/s00127-010-0248-5
ORIGINAL PAPER
Behavioral problems among children living in orphanage facilities
of Karachi, Pakistan: comparison of children in an SOS Village
with those in conventional orphanages
Zohra S. Lassi • Sadia Mahmud • Ehsan U. Syed •
Naveed Z. Janjua
Received: 5 August 2009 / Accepted: 7 June 2010 / Published online: 24 June 2010
Ó Springer-Verlag 2010
Abstract
Purpose This study compared the behavioral problems of
children living in an SOS Village, which attempts to pro-
vide a family setup for its children, with those living in
conventional orphanages.
Methods We conducted a cross-sectional survey of 330
children, aged 4–16 years, living either in an SOS or other
conventional orphanages of Karachi, and assessed their
behavioral problems using strengths and difficulty ques-
tionnaire (SDQ). Behavioral problems on composite SDQ
and subscales, rated by foster mothers, were compared
between children in the two groups using v2 test of inde-
pendence. Multivariable models were built, using gen-
eralized estimating equations (GEE) regression approach,
to identify factors independently associated with behav-
ioral problems.
Results The overall prevalence of behavioral problems
was 33%. On univariate comparison, we found that groups
did not differ in their overall behavioral problems, while
they were significantly different on the peer problem scale
(P = 0.026). The model for composite SDQ behavioral
problems identified five factors: wasting, \5 years length
of stay in the facility and foster mother’s depression, while
facility type and sex of the child were part of a significant
interaction. Foster mother depression, child’s malnourish-
ment and fewer years of stay at the facility were associated
with conduct problems, while the child’s sex, facility type
and child’s parental living status were part of two
interactions.
Conclusion We found a high burden of behavioral
problems among children living in orphanages of Karachi,
Z. S. Lassi (&)
S. Mahmud
E. U. Syed
N. Z. Janjua
Aga Khan University, Karachi, Pakistan
e-mail: zohra.lassi@aku.edu
Pakistan. Foster mothers’ depression and child’s nutritional
status, which are associated with behavioral problems, can
be target of interventions to reduce behavioral problems of
children living in orphanages.
Keywords Mental health
Orphaned children

Orphanages in Karachi
Behavioral problem

Conduct problem
Introduction
Globally, one in every five children and adolescents suffers
from a mental disorder, and two out of five who require
mental health services do not receive them. It is expected
that by 2020, childhood neuropsychiatry disorder will rise
to over 50% and will become one of the five most common
reasons of morbidity, mortality and disability among chil-
dren [1]. Despite these alarming figures, mental disorders
are often unidentified or diagnosed too late.
Behavioral problems can occur in children of all ages
and very often start early in life. Children with these
problems can be rude, and have tantrums and a higher
tendency of developing an oppositional defiant disorder/
conduct disorder, with difficult temperament, learning or
reading difficulties, depression and hyperactivity [2]. In
addition, among these children, hitting and kicking other
people are common [2]. The importance of early detection
of emotional and behavioral problems is recognized
worldwide and a number of researches have been con-
ducted in developed countries. However, there has been
little systematic research into childhood psychiatric disor-
ders in developing countries [3]. A cross-sectional survey
from Karachi on 5- to 11-year-old children attending
mainstream private and public schools estimated a
123
788
prevalence of 34% for behavioral problems [4], whereas
the burden reported from other developing countries such
as Bangladesh and India was 15% [5] and 12% [6],
respectively. Many risk factors have been identified for the
occurrence of mental disorders, among which social factors
are clearly implicated in the genesis and maintenance of
these disorders and their extension into adulthood [7].
Other factors include quality of parental relationship, par-
ent’s mental health, education, occupation and broader
environmental circumstances such as adequacy of housing
[8–10]. Studies have also investigated specific emotional,
behavioral, conduct, hyperactivity and peer problems in
children. A study from Pakistan has identified a prevalence
of 9.3% for emotional and behavioral problems among
school children [11]. However, research on specific mental
health problems is yet to gain a significant role in devel-
oping countries, and researchers are struggling to identify
the overall mental health status of children and adolescents
[4–6, 10, 12–20].
Children living in orphanages are one of the most vul-
nerable groups of children in a society: many of them live
in a state of repeated neglect, abuse or fear. Hence, a safe
new home that they can trust is not by itself sufficient to
repair the damage imposed by abnormal early stress on the
developing nervous system [21]. Thus, children in foster
care have a disproportionately high prevalence of mental
health disorders [22], and studies have indicated that
between 50 and 80% of children in foster care suffer from
moderate to severe mental health problems [23, 24]. One
study on Eritrean orphanages found children with mood
disturbances, language delays and disturbed social inter-
action with peers, and concluded that orphanages are
necessarily the breeding grounds of psychopathological
problems [25].
The situation clearly signifies that the mental health of
orphaned children is vulnerable, and in a country like
Pakistan they further face the problem of inadequacy in the
number of orphanages and the quality of care provided.
Notwithstanding that Karachi is the largest and the most
populated city of Pakistan (with a population of 14.5 mil-
lion), it has only five registered orphanages (the list of
facilities was provided by the City Police Liaison Com-
mittee). These include one operated by the government and
four by non-governmental organizations. Among these,
SOS and Edhi homes are the only ones that have outreach
activities and children of both sexes from multiple catch-
ments areas. SOS is an international movement that aims to
provide housing and care to orphaned children. It was
started in Imst, Austria, in 1949 after World War II by
Herman Gmeiner [26]. Gmeiner believed that all children
needed a mother, a home, siblings and a family. It then
became the mission of the SOS organization to meet these
needs and to ensure that every child raised in an SOS
123
Soc Psychiatry Psychiatr Epidemiol (2011) 46:787–796
Village was equipped for transition into a successful and
productive adulthood.
This survey was aimed not only to obtain credible
baseline data for estimating the prevalence of behavioral
problems among children living in orphanage facilities in
Karachi, but also intended to compare the behavioral
problems of children in an SOS facility with that in con-
ventional orphanages. The latter objective is important
because unlike an SOS facility, the conventional orphan-
ages have segregated facilities for male and female chil-
dren and the concept of family is not really built in. Our
study hypothesizes that children living in the SOS facility
have a better mental health status, as SOS provides a
family setup for its children. If this hypothesis is validated,
SOS can be recommended as a role model for the care of
orphaned children.
Methods
Setting and design
We contacted all the five registered orphanages in Karachi;
the authorities of only three gave us permission to conduct
the survey, while two refused. The facilities that gave
permission included SOS, a government facility (Al-Banat
Darul-Atfal that housed girls only) and Edhi home. Edhi
home maintains separate facilities for boys and girls;
however, they gave us permission to conduct the survey
only in their boy’s facility. Hence, the final study popula-
tion included children from SOS, Edhi boys arm and the
government’s (girls) facility; the last two were grouped
together as the conventional orphanage category. A cross-
sectional survey was conducted in these orphanages from
July to September 2007 on children aged 4–16 years, who
gave assent and whose foster mothers gave informed
consent.
Sample size was calculated to compare the prevalence
of behavioral problems between SOS and conventional
orphanage children. For the sample size calculation, we
assumed a prevalence of 34% of behavioral problems [4]
among SOS children. To detect an odds ratio of 2
(assuming that children in conventional orphanages are at
higher risk) with 5% significance level and 80% power,
147 children from each group, and a total of 294 children
were required to meet the study objectives [27]. During the
actual sampling process, we found 126 eligible children
from the SOS Village, 195 from Edhi boys arm and 22
children from the government (girls) arm, respectively.
From the SOS and Edhi Village, all the sampled children
participated in the study. However, from the government’s
arm, we could collect data from only nine girls because the
remaining children did not return from their summer break,
Soc Psychiatry Psychiatr Epidemiol (2011) 46:787–796
when they visited their relatives, due to heavy rains. Hence,
the overall participation rate was 96.2%. The total sample
size achieved was 330 with 126 children in the SOS arm
and 204 in conventional orphanages.
Data collection and instruments
A structured interviewer-administered questionnaire was
developed to collect information related to foster mothers
and children. Foster mothers were asked about their edu-
cational status and number of children under their care and
for children information sought included age, sex, educa-
tion, length of stay in the facility, parent’s living status,
prior living arrangements and anthropometric parameters.
Pre-testing of the questionnaire was done on 30 children
(approximately 10% of the total sample size). Some of the
questions that required changes due to phrasing were
amended after pre-testing. Foster mother’s anxiety and
depression were assessed with a validated tool, the Aga
Khan University anxiety and depression scale (AKUADS)
that has a specificity of 79% and sensitivity of 66% [28].
Behavioral problems of children were assessed using
strengths and difficulty questionnaire (SDQ). SDQ is the
most widely used children’s mental health assessment tool
that was developed by Goodman in 1997 [29]. SDQ is a
brief (25 items) multi-informant behavioral screening
questionnaire. The 25 items are divided among five sub-
scales of five items each, generating scores for conduct
problems, inattention hyperactivity, emotional symptoms,
peer problems and pro-social behavior. In a British survey,
multi-informant SDQ identified 5- to 15-year-old children
having a psychiatric diagnosis with a specificity of 94.6%
and a sensitivity of 63.3%, and identified over 70% of
children with conduct, hyperactivity, depressive and some
anxiety disorders [30].
SDQ is available in more than 40 languages including
Urdu (national language of Pakistan). The parent version of
SDQ was validated in Urdu in 2005 that suggested a cutoff
of 18 for overall behavioral problems, and has an overall
sensitivity and specificity of 69 and 71%, respectively [31].
The sensitivity and specificity of cutoffs for the subscales
in this validation study were above 55% [31]. In this study,
we used the validated SDQ Urdu version that categorizes
children as normal and abnormal with regard to their
behavioral problems. Conduct problem scale (CPS) was
not validated in Urdu; hence, for this subscale we adopted
the cutoff suggested by Goodman, as most of the Urdu
validated cutoffs on subscales and composite scale corre-
spond to original Goodman’s cutoffs [30]. As SDQ is a
multi-informant behavioral tool, SDQ ratings for children
were obtained from both foster mothers and teachers. The
Urdu SDQ was validated for the parent version only;
hence, the same version was used for the assessment of
789
children’s behavioral problems by teachers. We computed
the prevalence estimates of overall behavioral problems
and those on subscales for both foster mother and teacher
ratings, and compared these ratings between the two
facility types, respectively. However, the multi-variable
analysis was done using foster mother ratings only. This is
because children spend more time with the foster mothers
as compared to teachers, and it seems plausible to assume
that the former know and can assess children’s behavior
better. Moreover, as pointed above the SDQ teacher ver-
sion was not validated in Urdu.
To explore if children’s behavioral problems were
associated with their nutritional health, we assessed chil-
dren’s nutritional health with their height, weight, age and
sex, according to National Center for Health Statistics
(NCHS) criteria of 1978. Three indicators were computed
to measure children’s nutritional status: stunting, i.e.,
children \-2 standard deviation (SD) from median height
for age; underweight, i.e., children \-2 SD from median
weight for age; and wasting, i.e., children \-2 SD from
median weight for height [32].
Ethics
The Aga Khan University Ethical Review Committee
reviewed and approved the study.
Statistics
Data were analyzed on SAS 9.1.3 [33]. Descriptive anal-
ysis of socio-demographic variables was done to describe
the characteristics of our study population. Mean and
standard deviation were computed for the continuous and
proportions for the categorical variables. Prevalence of
children’s behavioral problems was estimated on SDQ
composite rating scores, as well as on subscale ratings. A
univariate comparison of the prevalence of behavioral
problems on the composite and subscales of SDQ between
the two orphanage facility types was done using Chi-square
test of independence or Fisher’s exact test. We also com-
pared the mean composite scores of overall behavioral
problems between the two facility types using T test for
two independent samples. Kappa statistics was applied to
assess the agreement between foster mother’s and teacher’s
ratings on composite SDQ and subscale. Multivariable
analysis was conducted using the generalized estimating
equations (GEE) [34], a regression approach that takes
clustering structure of the data into account. In our data,
children were clustered within the foster or house mother.
As a mother was taking care of and raising a number of
children, it was expected that children’s behavior in one
cluster was possibly correlated. Multivariable GEE analy-
sis was performed using PROC GENMOD [34], and
123
790
models were built for overall behavioral problems and
conduct problems. Variables for inclusion in the final
model were selected based on biological importance and
statistical significance (P value \ 0.25 in univariate anal-
ysis). Scale examination for continuous variables was done
using quartile analysis and multicolinearity was checked
for each independent variable [35]. Confounding was
assessed by change in beta coefficient (corresponding to an
independent variable) by greater than 10% [36]. Biologi-
cally meaningful interactions were also assessed for
inclusion in the final model. Confidence intervals (CI) and
Wald P values were used to examine the significance of
each independent variable in the multivariable analysis.
Results
Children in conventional orphanages were significantly
younger (9.7 ± 2.1 years) as compared to those in SOS
(11.5 ± 3.1 years), and their length of stay in the orphan-
age (2.6 ± 2.4 years) was significantly shorter than that of
SOS children (3.9 ± 2.7 years). It was also found that 44%
of the SOS children were in age-appropriate classes as
compared to only 20% of conventional orphanage children.
With respect to nutritional health, children from conven-
tional orphanages performed better; 18% of the children in
the SOS facility were stunted and 22% were malnourished,
whereas in conventional orphanages these proportions were
only 6 and 8%, respectively. The proportion of wasting in
children was not significantly different between the two
facility types, and overall 7% of children in the study
sample were wasted (Table 1).
Foster mothers in the two facility types were not sig-
nificantly different with regard to their education and
depression status; overall, 52% of the foster mothers had
[10 years of schooling and 3 out of 23 were found to have
anxiety and depression. The mean number of children
under the care of a conventional orphanage mother was
much larger (24.7 ± 8.9) as compared to that under the
care of an SOS mother (8.7 ± 1.6) (Table 1).
There was no significant agreement between foster
mother’s and teacher’s rating on overall behavioral prob-
lems (kappa = 0.073, P = 0.184), hyperactivity subscale
(kappa = 0.104, P = 0.07), peer problem subscale
(kappa = 0.073, P = 0.351) and pro-social symptom sub-
scale (kappa = 0.083, P = 0.128). However, significant
but poor agreement was found on emotional symptoms
subscale (kappa = 0.199, P \ 0.001) and conduct problem
subscale (kappa = 0.149, P = 0.011).
On the behavioral subscales of SDQ, about 9% of
children in the study sample were rated as abnormal on the
emotional scale, 50% on conduct problems, 13% on
hyperactivity, 84% on peer problems and 47% on pro-
123
Soc Psychiatry Psychiatr Epidemiol (2011) 46:787–796
social behavior subscale by the foster mother. On the
composite SDQ, 33% of the children were rated as
abnormal by foster mothers. Univariate facility-wise
comparison of mother’s behavioral rating shows that a
significantly higher proportion of SOS children (89.7%)
were rated as abnormal on peer problem scale as compared
to children from conventional orphanages (80.4%). How-
ever, there was no significant difference in foster mother’s
rating on the overall SDQ and the other subscales between
the two facility types (Table 2).
On the other hand, as per teacher rating of SDQ, about
9% children were rated abnormal on the emotional scale,
49% on conduct problems, 17% on hyperactivity, 88% on
peer problems and 42% on pro-social behavior subscale.
On the composite SDQ, 39% of the children were rated as
abnormal by teachers. There was no significant difference
in teacher’s rating on the overall SDQ and subscale
between two facility types (Table 2).
The multivariable regression model for overall behav-
ioral problems identified five factors: wasting (OR-
adj = 1.77, 95% CI: 0.92–3.41), B5 years of stay in the
facility (OR-adj = 1.83, 95% CI: 0.86–3.90) and foster
mother’s depression (OR-adj = 1.93, 95% CI: 0.76–4.89),
with facility type and sex of the child being part of a sig-
nificant interaction. Analysis of this interaction indicated
that there was no difference in the behavioral problems of
male children from SOS and conventional orphanages
(OR-adj = 0.87, 95% CI: 0.29–2.60); however, female
children living in conventional orphanages were at a much
higher risk relative to those living in SOS (OR-adj = 9.86,
95% CI: 3.22–30.16) (Table 3).
The multivariable regression model for conduct prob-
lems identified six factors: foster mother depression (OR-
adj = 4.48, 95% CI: 2.22–9.03) and child’s malnourish-
ment (OR-adj = 1.94, 95% CI: 1.25–2.92), while every
1 year increase in length of stay at the facility had a pro-
tective effect (OR-adj = 0.87, 95% CI: 0.77–0.98), and
with child’s sex, facility type and child’s parental living
status being part of two interactions (facility 9 sex and
parental living status 9 sex). Analysis of interaction
between facility and sex indicates that there was no dif-
ference in the conduct problems of male children between
SOS and the conventional orphanage (OR-adj = 0.43, 95%
CI: 0.15–1.24); however, female children in the conven-
tional orphanage were at a much higher risk relative to
those in SOS (OR-adj = 6.98, 95% CI: 2.44–19.93). The
interaction between child’s sex and his/her parental living
status indicated that there was no difference in the conduct
problems of male children with regard to their parental
living status (OR-adj = 1.01, 95% CI: 0.69–1.45), whereas
female children had a protective effect against conduct
problems when they had no parent alive (OR-adj = 0.25,
95% CI: 0.09–0.67) (Table 3).
Soc Psychiatry Psychiatr Epidemiol (2011) 46:787–796 791

Table 1 Socio-demographic
Characteristics children (n = 330) Mean ± SD or n (%) P value
characteristics of children and
foster mothers in an SOS SOS Village Conventional
Village and in conventional n = 126 n = 204
orphanages, Karachi, Pakistan
Age (years)
\6 12 (9.5) 12 (5.9) \0.001
6–12 55 (43.7) 151 (74.0)
[12 59 (46.8) 41 (20.1)
Sex
Male 71 (56.3) 195 (95.6) \0.001
Female 55 (43.7) 9 (4.4)
Educational class of children
Pre-Primary 21 (16.7) 118 (57.8) \0.001
Primary 74 (58.7) 78 (38.2)
Secondary 31 (24.6) 8 (3.9)
Length of stay in facility 3.9 ± 2.7 2.6 ± 2.4 \0.001
Academic class in accordance with age
Yes 55 (43.7) 41 (20.1) \0.001
No 71 (56.3) 163 (79.9)
Living status of parent
No parent 28 (22.2) 93 (45.6) \0.001
At least one parent alive 98 (77.8) 111 (54.4)
Previous stay
With parents 93 (73.8) 112 (54.9) \0.001
Other than parents 33 (26.2) 92 (45.1)
Stunting \ -2SD
Stunting 22 (17.5) 12 (5.9) 0.001
Otherwise 104 (82.5) 192 (94.1)
Wasting \ -2SD
Wasting 7 (5.6) 17 (8.3) 0.345
Otherwise 119 (94.4) 187 (91.7)
Malnourishment \ -2SD
Malnourished 28 (22.2) 17 (8.3) \0.001
Otherwise 98 (77.8) 187 (91.7)
Foster mother (n = 23) n = 15 n=8
Educational Level of Foster mother
Less than or equal to 10 years of schooling 8 (53.3) 4 (44.4) 00.673
P value of categorical variables
Greater than 10 years of schooling 7 (46.7) 5 (55.6)
is obtained from Fisher exact
test or v2 test of independence Number of children under care 8.7 ± 1.6 24.7 ± 8.9 0.001
where appropriate Depression status
P value for continuous variables Depressed 2 (13.3) 1 (11.1) 1.0
is obtained from t test for two Normal 13 (86.7) 8 (88.9)
independent samples

Discussion and 39% based on teacher ratings for children living in

To the best of our knowledge, this is the first epidemiologic
study using validated instrument in local language on
behavioral problems of children living in orphanage facil-
ities in this region. Our study estimated a prevalence of
33% of behavioral problems based on foster mother ratings
orphanage facilities of Karachi.
Most of the studies from a general population of chil-
dren reported a prevalence of 10–20% [6, 7, 37]; however,
the estimates of the prevalence largely depend on the
choice of instruments used, study design employed and the
source of the sample. Our data were from orphanages, and

123
792 Soc Psychiatry Psychiatr Epidemiol (2011) 46:787–796

Table 2 Number and (percentage) of children rated abnormal by foster mothers and teachers in SDQ subscales and composite ratings: by
orphanage home
Categories Foster mother rated Teacher rated
n (%) P Value Total n (%) P Value Total
SOS Conventional SOS Conventional
n = 126 orphanage n = 126 orphanage
n = 204 n = 204

Emotional symptom scale (ESS) 8 (6.3) 23 (11.3) 0.136 31 (9.4) 11 (8.7) 17 (8.4) 0.921 28 (8.5)
Conduct problem scale (CPS) 65 (51.6) 100 (49.0) 0.650 165 (50.0) 66 (52.4) 96 (47.5) 0.392 162 (49.4)
Hyperactivity scale (HS) 14 (11.1) 28 (13.7) 0.489 42 (12.7) 22 (17.5) 32 (15.8) 0.701 54 (16.5)
Peer problem scale (PPS) 113 (89.7) 164 (80.4) 0.026* 277 (83.9) 109 (86.5) 181 (89.6) 0.394 290 (88.4)
Prosocial symptom scale (PSS) 53 (42.1) 103 (50.5) 0.136 156 (47.3) 53 (42.1) 85 (42.1) 0.998 138 (42.1)
Composite scale, overall 35 (27.8) 74 (36.3) 0.111 109 (33.0) 50 (39.7) 79 (39.1) 0.918 129 (39.3)
behavioral problems
Composite scale comprised ESS, CPS, HS and PPS
P value of categorical variables is obtained from Fisher exact test and v2 test of independence where appropriate
* Significant at 5% significance level

comparable data from this population were not available.
Our prevalence estimates were high, as we used screening
tools that yielded higher estimates than diagnostic tools.
Yet, our results were comparable to a recent study that
found 34% of school children rated as abnormal by parents
on overall behavioral problems, using SDQ, in mainstream
public and private schools in Karachi.
We used multi-informant ratings by foster mothers and
teachers, respectively. We found that there was poor or no
significant agreement in their ratings, which is usually
reported in studies that utilize multi-informant tools [38,
39]. It has been found that routine mental health screening,
which uses only parent reports, is likely to under-identify
symptoms and functional problems that would be uncov-
ered if reports were to be solicited from teachers [38].
Hence, it suggests the development of an algorithm to help
clinicians or researchers determine when to solicit teacher
input and how to interpret parent and teacher contrasts.
We found that the prevalence of abnormal behavior on
foster mother-rated overall SDQ was somewhat higher in
children in conventional orphanages (36%) as compared to
that (28%) in SOS children, though the difference was not
statistically significant. We also compared the mean com-
posite score of overall behavioral problems between the
two facility types; mean score from SOS mothers
(15.6 ± 4.1) was significantly lower than that from con-
ventional orphanage mothers (16.6 ± 3.8). This provides
some evidence that provision of a home environment and
the role of family have a positive impact on child’s mental
and social development. However, we note that the prev-
alence of foster mother-rated peer problems among chil-
dren living in SOS (90%) was significantly higher relative
to that for conventional orphanage children (84%).
We found an association of wasting with overall
behavioral problems, and malnourishment with conduct
problems. Studies have shown that wasting negatively
affects brain growth and development, which predisposes
antisocial and violent behavior by affecting cognitive
functions [40]. Children with malnutrition have been found
to have poor attention span and memories, more isolation
and fewer positive peer relationships. A dose–response-
relationship has also been reported between degrees of
malnutrition and externalizing behavior at ages 8–17 years
[41]. A similar association has been found in our study,
where the risk for conduct problems was about twice in
malnourished relative to well-nourished children.
We also found that risk for behavioral and conduct
problems in children were twofold and fourfold, respec-
tively, when the mother had anxiety and depression.
Mothers’ depression was always found to be highly asso-
ciated with children’s behavioral problems in previous
studies [42]. Studies have found that children of depressed
or anxious parents are themselves at a substantially
increased risk (two to fivefold) for psychiatric disorders
[43–45], the reason being that the parents direct a higher
number of commands and criticisms toward their children,
who in turn respond with increased noncompliance and
deviant child behavior.
An interesting relation was seen between children’s
behavioral problems and length of stay at a facility, sug-
gesting that orphanages were protective against children’s
behavioral and conducts problems. Orphanages, which
were always considered as a breeding ground for behav-
ioral problems, were found to be protective! A possible
reason for this finding may be the children’s previous
environment: mostly, they were brought to the orphanage

123
Soc Psychiatry Psychiatr Epidemiol (2011) 46:787–796 793

Table 3 Univariate and multivariable GEE models: unadjusted and adjusted odds ratios for overall behavioral and conduct problems
Variables Overall behavioral problems Conduct problem
Unadjusted OR (CI) Adjusted OR (CI) Unadjusted OR (CI) Adjusted OR (CI)
Independent variables Exchangeable working correlation 0.23a Exchangeable working correlation 0.17a

Facility
SOS 1 VPI 1 VPI
Conventional orphanage 1.51 (0.54–4.26) 0.87 (0.36–2.06)
Age categories
Adolescent 1 1 1 VEM
School age 0.98 (0.46–2.09) 0.80 (0.36–1.77) 1.03 (0.57–1.86)
Pre-school age 1.55 (0.63–3.85) 1.32 (0.51–3.33) 1.29 (0.57–2.94)
Sex
Male 1 VPI 1 VPI
Female 0.59 (0.29–1.20) 0.58 (0.25–1.33)
Length of stay
Continuous 0.89 (0.81–0.99) 0.87 (0.77–0.98)
[5 years of stay 1 1
B5 years of stay 1.96 (0.87–4.45) 1.83 (0.86–3.90)
Class according to age
With accordance 1 VEM 1 VEM
Not in accordance 1.20 (0.75–1.91) 0.99 (0.68–1.43)
Education of foster mother
B10 years 1 VEM 1 VEM
[10 years 1.38 (0.44–4.35) 1.25 (0.52–3.02)
Depression of foster mother
Normal 1 1 1 1
Depressed 1.39 (0.56–3.46) 1.93 (0.76–4.89) 3.18 (1.71–5.94) 4.48 (2.22–9.03)
Parental living status
At least one parent 1 VEM 1 VPI
No parent 0.81 (0.49–1.35) 0.68 (0.53–0.89)
Prior living arrangement
Without parents 1 VEM 1 VEM
With parents 1.29 (0.67–2.47) 1.36 (0.94–1.95)
Children under care of the mother 1.01 (0.97–1.05) VEM 0.99 (0.95–1.04) VEM
Malnourishment \ -2SD
Otherwise 1 VEM 1 1
Malnourished 1.25 (0.75–2.10) 1.93 (1.29–2.87) 1.94 (1.25–2.92)
Wasting \ -2SD
Otherwise 1 1 1 VEM
Wasting 1.78 (0.95–3.32) 1.77 (0.92–3.41) 0.62 (0.32–1.20)
Stunting \ -2SD
Otherwise 1 VEM 1 VEM
Stunted 0.77 (0.38–1.54) 1.07 (0.58–1.95)
Interactions
Effect of facility on male and female children
Male child from SOS Village 1 1
Male child from conventional orphanage 0.87 (0.29–2.60) 0.43 (0.15–1.24)
Female child from SOS Village 1 1
Female child from Conventional orphanage 9.86 (3.22–30.16) 6.98 (2.44–19.93)

123
794
Table 3 continued
Variables Overall behavioral problems
Unadjusted OR (CI)
Independent variables Exchangeable working correlation 0.23a
Effect of parental living status on male and female children
Male child having at least one parent alive
Male child having no parent alive
Female child having at least one parent alive
Female child having no parent alive
a
Exchangeable working correlations for multivariable models
OR odds ratio, CI confidence level, VEM variable excluded from the model, VPI variable is part of an interaction
because of the inability of caretakers to provide them with
food and basic necessities, or because the parent(s) died or
remarried and the children were living with stepparents.
Therefore, orphanages may provide a better environment in
comparison to the previous disturbing living situations.
However, comparable studies on orphanages are few and
none has identified this phenomenon; this evidence needs
to be explored in future studies.
Multivariable analysis indicated that the effect of facil-
ity type on the behavioral and conduct problems was dif-
ferent for male and female children. The overall behavioral
and conduct problems of male children were similar in the
two settings, suggesting that male children at Edhi (males
from the conventional arm belonged to Edhi home) per-
formed as well as male children at SOS. However, female
children at the government facility (females from the
conventional arm belonged to the government facility)
were at a much higher risk for overall behavioral and
conduct problems relative to SOS girls.
We also found that the effect of parental living status on
conduct problems was different for male and female chil-
dren. There was no difference in the conduct problems of
male children with regard to their parental living status;
however, females had a protective effect against conduct
problems when they had no parent alive. The reason for this
effect needs to be further explored in future studies, yet the
possible reason could be the control of female children by
foster mothers when the child does not have a parent alive.
This study has provided important baseline data
regarding mental health of children living in orphanage
facilities in Karachi, Pakistan. This is particularly relevant
as data pertaining to the prevalence of child psychiatric
disorders among those living in orphanages are sparse in
this region. Secondly, the use of SDQ that has strong
psychometric properties, as well as research and clinical
utility, has increased the reliability of prevalence that the
tool has captured.
The study had some limitations: first, data was cross-
sectional in nature, so it cannot provide direct support for
123
Soc Psychiatry Psychiatr Epidemiol (2011) 46:787–796
Conduct problem
Adjusted OR (CI) Unadjusted OR (CI) Adjusted OR (CI)
Exchangeable working correlation 0.17a
1
1.01 (0.69–1.45)
1
0.25 (0.09–0.67)
causal inferences. However, the importance of such study
lies in suggesting potential factors associated with the
outcome of interest; second, we used screening tools rather
than diagnostic tool for identifying behavioral problems,
which could have resulted in higher prevalence rates; third,
in this study we used parent SDQ version for teachers,
since the teacher version was not validated in Urdu; fourth,
conduct problem scale (CPS) was not validated in Urdu,
hence for this subscale Goodman’s cutoff was adopted;
however, this does not seem to be a serious limitation as
most of the Urdu-validated cutoffs on subscales and com-
posite scale correspond to the original Goodman’s cutoffs;
fifth, the population of females in the conventional arm was
quite small, which was mainly because the authorities of
the Edhi girl’s facility refused to participate in the study;
sixth, GEE analysis requires large number of clusters
(about 30 or more), and in our sample we had 23 clusters.
However, the number of clusters in our sample was not
much smaller than 30 and the GEE analysis offered a better
analysis than ordinary logistic regression that ignores the
clustering structure in the data.
Conclusion and recommendations
Our study provided some evidence that the SOS Village
can be used as a model facility for orphaned children. Their
philosophy of providing children with a mother, siblings
and home can be applied to other orphanages as it has
shown some positive impact on the child’s mental and
social development. However, data from Karachi SOS
implied that the nutritional health of its children required
improvement.
Opportunities to identify children showing early signs of
serious behavioral problems are limited in orphanages.
Children entering foster care should be routinely screened
and referred to child mental health experts. As far as pre-
vention is concerned, this study has identified certain
amendable risk factors, which can be alleviated by proper
Soc Psychiatry Psychiatr Epidemiol (2011) 46:787–796
and timely measures. Findings suggest that improving
maternal mental health is not only self-evidently valuable
for foster mothers, but it is also likely to have wider ben-
efits for children’s behavioral problems. Hence screening
and treating house mothers for anxiety and depression will
have a beneficial impact on the children under their care.
Further research can enhance the understanding of the
patterns of co-morbidity, perceived treatment needs and
psychosocial correlates. Longitudinal studies should be the
next step in understanding the natural history of children’s
mental disorders and to further establish temporality and
causal associations.
Acknowledgments We thank all the study participants, profes-
sionals and employees of the orphanages from where the data were
collected: SOS Village, Karachi; Edhi home, Karachi; and Al-Banat
Darul-Atfal, Karachi, as well as all the research team members who
have worked and contributed to this research. Our special thanks are
due to Professor Kausar S Khan, who was kind enough to give us
guidance related to the subject.
References
1. National Institute for Health Care Management (2005) Children’s
mental health: an overview and key considerations for health
system stakeholders. A report by National Institute for Health
Care Management
2. Wattie B (2003) The importance of mental health of children,
Canadian Mental Health Association
3. Nikapota AD (1991) Child psychiatry in developing countries. Br
J Psychiatry 158:743–751
4. Syed EU, Hussein SA, Mahmud S (2007) Screening for emo-
tional and behavioral problems amongst 5–11 years old school
children in Karachi, Pakistan. Soc Psychiatry Psychiatr Epi-
demiol 42(5):421–427
5. Mullick M, Goodman R (2005) The prevalence of psychiatric
disorders among 5–10 years old in rural, urban and slums areas of
Bangladesh: an exploratory study. Soc Psychiatry Psychiatr Ep-
idemiol 40:663–671
6. Srinath S, Sirimaji S, Ruraj G, Seshadri S, Subbakrishna D, Bhola
P (2005) Epidemiological study of child and adolescent psychi-
atric disorders in urban and rural areas of Bangalore, India. Indian
J Med Res 162:67–79
7. Rutter M (1989) Pathways from childhood to adult life. J Child
Psychol Psychiatry 30:23–51
8. Fleitlich B, Goodman R (2001) Social factors associated with
child mental health problems in Brazil: a cross-sectional survey.
BMJ 323:599–600
9. Goodman R, Slobodskaya H, Knyazev G (2005) Russian child
mental health: a cross-sectional study of prevalence and risk
factors. Eur Child Adolesc Psychiatry 14:28–33
10. Hackett R, Hackett L, Bakta P, Gowers S (1999) The prevalence
and association of psychiatric disorders in children in Kerala,
South India. J Child Psychol Psychiatry 40:801–807
11. Javed MA, Kundi MZM, Khan PA (1992) Emotional and
behavioral problems among school children in Pakistan. J Pak
Med Assoc 42(2):181–183
12. Durkin MS, Hasan ZM, Hasan KZ (1998) Prevalence and cor-
relates of mental retardation among children in Karachi, Pakistan.
Am J Epidemiol 147:281–288
795
13. Mirza I (2001) Common mental disorders in urban vs. rural
Pakistan. Br J Psychiatry 178:475–476
14. Chisholm D, Sekar K, Kumar K, Saeed K, James S, Mubbahar M
et al (2000) Integration of mental health care into primary care.
Br J Psychiatry 174:581–588
15. Mubbashar MH (2001) Mental health literacy in developing
countries. Br J Psychiatry 179:75
16. Naqvi HA, Khan MM (2006) Pathway to psychiatric care in
Karachi. JCPSP 16(6):438–439
17. Goodman R, Renfrew D, Mullick M (2000) Predicting type of
psychiatric disorder from strengths and difficulties questionnaire
(SDQ) scores in child mental health clinics in London and Dhaka.
Eur Child Adolesc Psychiatry 9(2):125–134
18. Durkin MS, Khan NZ, Davidson LL, Huq S, Munir S, Rasul E
et al (2000) Prenatal and postnatal risk factors for mental retar-
dation among children in Bangladesh. Am J Epidemiol 152:
1024–1033
19. Rahman A, Mubbashar MH, Gater R, Goldberg D (1998) Ran-
domized trial of impact of school mental health programme in
rural Rawalpindi, Pakistan. Lancet 352:1022–1025
20. Mumford DB, Mihhas FA, Akhtar I, Akhter S, Mubbashar MH
(2000) Stress and psychiatric disorder in urban Rawalpindi. Br J
Psychiatry 177:557–562
21. Hughes DA (1999) Building the bonds of attachment: awakening
love in deeply troubled children. Jason Aronson, London
22. Halfron N, Zepeda A, Inkelas M (2002) Mental health services
for children in foster care. UCLA Center for Healthier Children,
Families and Communities, Los Angeles
23. DosReis S, Zito JM, Safer DJ, Soeken KL (2001) Mental health
services for youths in foster care and disables youth. Am J Public
Health 91(7):1094–1099
24. Simms MD, Dubowitz H, Szilagyi MA (2000) Health care
needs for children in the foster care system. Pediatrics 106(4):
909–918
25. Wolff PH, Fesseha G (1998) The orphans of eritrea: are
orphanages part of the problem or part of the solution. Am J of
Psychiatry 155(10):1319–1324
26. SOS children village Pakistan. http://www.sos.org.pk/121int.html
27. Epi Info 6 (2001) A word processing, database and statistics
program for public health. In: 6.04d ed. Center for Disease
Control, World Health Organization, Geneva, Switzerland
28. Badar S, Hashim R, Murad M, Imtiaz J (1998) Development of
an indigenous screening instrument in Pakistan: the Aga Khan
University Anxiety and Depression Scale. J Pak Med Assoc
48:261–265
29. Goodman R (1997) The strengths and difficulties questionnaire: a
research note. J Child Psychol Psychiatry 38(5):581–586
30. Goodman R, Ford T, Simmons H, Gatward R, Meltzer H (2000)
Using the strengths and difficulties questionnaire (SDQ) to screen
for child psychiatric disorders in a community sample. Br J
Psychiatry 177:534–539
31. Samad L, Hollis C, Prince M, Goodman R (2005) Child and
adolescent psychopathology in a developing country: testing the
validity of the strengths and difficulties questionnaire (Urdu
version). Int J Methods Psychiatr Res 14(3):158–166
32. Cogill B (2003) Anthropometric indicators measurement guide.
Food and Nutrition Technical Assistance Project, Academy for
Educational Development,Washington, DC
33. Inc S. SAS for windows. In. SAS 9.1.3 ed
34. Allison PD (1999) Logistic regression using the SAS system:
theory and application. SAS Institute Inc, North Carolina
35. Hosmer DW, Lameshow S (2000) Applied logistic regression,
2nd edn edn. Wiley, New York
36. Rothman KJ, Greenland S (1998) Modern epidemiology, 2nd edn
edn. Lippincott-Raven, Philadelphia
123
796
37. Roberts RE, Atkinson CC, Rosenblatt A (1998) Prevalence of
psychopathology among children and adolescents. Am J Psy-
chiatry 155:715–725
38. Brown JD, Wissow LS, Gadomski A et al (2006) Parent and
teacher mental health ratings of children using primary-care
services: inter-rater agreement and implications for mental health
screening. Amb Pediatr 6:347–351
39. Dix KL, Askell-William H, Lawson MJ (2008) Different mea-
sures, different informants, same outcomes? Investigating mul-
tiple perspectives of primary school students’ mental health.
Paper presented at Australian Association for Research in Edu-
cation conference, Brisbane 2008
40. Ferris M, Holm-Hansen C (2007) Mahnomen county rural mental
health outreach project: evaluation report. Wilder Research,
Minnesota
123
Soc Psychiatry Psychiatr Epidemiol (2011) 46:787–796
41. Liu J, Raine A, Venable PH, Mednick SA (2004) Malnutrition at
age 3 and externalizing behavioral problems at ages 8, 11 and
17 years. Am J Psychiatry 161(11):2005–2013
42. Williams A (2005) Depression research in nursing: global per-
spective. Springer, New York
43. Last CG, Hersen M, Kazdin A (1991) Anxiety disorders in
children and their families. Arch Gen Psychiatry 48:928–934
44. Lieb R, Isensee B, Hofler M (2002) Parental major depression
and other mental disorders in offspring. Arch Gen Psychiatry
59:365–374
45. Hammen C, Brennan PA (2003) Severity, chronicity and timing
of maternal depression and risk for adolescent offspring diagnosis
in a community sample. Arch Gen Psychiatry 60:253–258