PFI_12mmX178mm.

pdf + eps format

Journal of Child Psychology and Psychiatry **:* (2020), pp **–** doi:10.1111/jcpp.13261

Research review: A meta-analysis of the international

prevalence and comorbidity of mental disorders
in children between 1 and 7 years
Mira Vasileva,1,2 Ramona K. Graf,1 Tilman Reinelt,1 Ulrike Petermann,1 and
Franz Petermann1,*
1
Center for Clinical Psychology and Rehabilitation, University of Bremen, Bremen, Germany; 2Child and Community
Wellbeing Unit, Melbourne School of Population and Global Health, University of Melbourne, Melbourne, Vic.,
Australia

Background: Children younger than 7 years can develop mental disorders that might manifest differently than in
older children or adolescents. However, little is known about the prevalence of mental disorders at this early age.
Methods: We systematically searched the literature in the databases Web of Science, PsycINFO, PSYNDEX,
MEDLINE, and Embase to identify epidemiological studies of community samples published between 2006 and
2020. A series of meta-analyses was conducted to estimate the pooled worldwide prevalence of mental disorders in
general, specific mental disorders, and comorbidity in young children. Results: A total of ten epidemiological studies
reporting data on N = 18,282 children (12–83 months old) from eight countries met the inclusion criteria. The pooled
prevalence of mental disorders in general was 20.1%, 95% CI [15.7, 25.4]. Most common disorders were oppositional
defiant disorder (4.9%, 95% CI [2.5, 9.5]) and attention-deficit hyperactivity disorder (4.3%, 95% CI [2.5, 7.2]). The
prevalence of any anxiety disorders was 8.5%, 95% CI [5.2, 13.5], and of any depressive disorders was 1.1%, 95% CI
[0.8, 1.6]. Comorbidity was estimated at 6.4%, 95% CI [1.3, 54.0]. Conclusions: The literature search reveals that
the epidemiology of mental disorders in children younger than 7 years is still a neglected area of research. The
findings also indicate that there are a significant number of young children suffering from mental disorders who need
appropriate age-adapted treatment. Keywords: Mental disorder; comorbidity; preschool; prevalence; epidemiology.

‘the terrible twos’) is important for a child’s auton-

Introduction
omy and should not be classified as a mental
Children younger than 7 years can suffer from
disorder in the absence of functional impairment
mental health symptoms that impair their further
(Keenan & Wakschlag, 2000). Furthermore, young
development and mental health throughout their life
children regulate their emotions and behavior
span (Angold & Egger, 2007). However, the investi-
through their interactions with their caregivers
gation of mental disorder prevalence in very young
(Lincoln, Russell, Donohue, & Racine, 2017). It is,
children is still a neglected area of research
therefore, often unclear whether the emotional and
(Lyons-Ruth et al., 2017). Better knowledge of the
behavioral problems that a child manifests should
prevalence and comorbidity of mental disorders is
be interpreted as individual psychopathology or as
essential for effective service planning and the opti-
an expression of dysfunctional interpersonal care-
mization of treatments and assessment tools for this
giver-child relationship (von Klitzing, Dohnert,
age group (Egger & Angold, 2006; Polanczyk, Salum,
Kroll, & Grube, 2015).
Sugaya, Caye, & Rohde, 2015).
To address challenges in the classification and
There are several challenges facing the classifica-
diagnosis of mental disorders in young children, the
tion of mental disorders at this early age. Most arise
Task Force Zero to Three developed the Diagnostic
from recognition of the first seven years of life as an
Classification of Mental Health and Developmental
important developmental stage. Due to the rapid
Disorders of Infancy and Early Childhood (DC: 0–5;
physical, emotional, behavioral, and cognitive devel-
Zero to Three, 2016). The DC: 0–5 includes devel-
opment that young children experience, their emo-
opmentally sensitive classification criteria for disor-
tional and behavioral problems are often considered
ders in infants and young children through five
transient problems rather than mental disorders
years of age, considering age-specific manifesta-
(Egger & Angold, 2006). It is also difficult to
tions, important predictors of normal and maladap-
differentiate between interindividual variability in
tive development, and individual differences in
normal development and psychopathology. For
development (Zeanah et al., 2017). The DC: 0–5 is
instance, increasing oppositionality between the
designed to complement the diagnostic systems ICD
second and fourth years of age (also referred to as
(World Health Organization, 1992) and DSM (Amer-
ican Psychiatric Association, 2013). Despite adap-
tions to child development, the DC: 0–5 is still not
*Deceased.
widely used internationally (Lyons-Ruth et al.,
Conflict of interest statement: No conflicts declared.

© 2020 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for Child and
Adolescent Mental Health.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial License, which permits use, distribution and
reproduction in any medium, provided the original work is properly cited and is not used for commercial purposes.
2 Mira Vasileva et al.

2017). Instead, most diagnoses in young children

Methods
rely on the standard ICD and DSM diagnostic
Search strategy
systems. Moreover, they mainly include diagnoses
which were introduced to describe symptoms as We followed PRISMA guidelines for conducting and reporting
they become manifest in adults and do not account reviews and meta-analyses (Moher, Liberati, Tetzlaff, &
Altman, 2009) and the strategy used for the previous meta-
for developmental variations (Egger & Angold,
analysis of mental disorder prevalence rates in older children
2006). Only recently was the first preschool subtype and adolescents (Polanczyk et al., 2015). A systematic
of a disorder previously exclusively based on symp- literature search was conducted in the electronic databases
toms in adults – the post-traumatic stress disorder Web of Science, PsycINFO, PSYNDEX, MEDLINE, and
(PTSD), preschool subtype – included in the DSM-5, Embase. The search was limited to include records from
2006 onwards and was updated on February 21, 2020.
which might be an important step toward more
Search terms included ‘(mental health or mental disorder)
developmentally sensitive diagnostic criteria in the AND (preschool* or toddler*) AND (prevalence or epidemiol-
future (Vasileva, Haag, Landolt, & Petermann, ogy)’. This search yielded 3,027 potential studies (Figure 1).
2018). Additionally, the reference lists of all relevant empirical
Despite challenges in diagnostics with very young studies and reviews were scanned for potential publications
and a further 25 studies were identified. After extracting 655
children, few reviews have thus far estimated the
duplicates, titles and abstracts of 2,397 studies were
prevalence of mental disorders at this age. In a screened independently by two authors (R. G. and M.V.).
systematic literature search, McDonnell and Glod Subsequently, the authors (R. G. and M.V.) reviewed texts of
(2003) identified seven studies that estimated the potential epidemiological studies.
prevalence of specific mental disorders in children
ages 1–6 years as ranging from 0.1% to 26.4%. The Inclusion criteria
most common disorders observed were oppositional
defiant disorder (ODD) and anxiety disorders. These Studies had to satisfy the following criteria to be included:
estimates were based partly on samples from psy-
1. epidemiological study of community samples;
chiatric settings which might lead to overestimation
2. use of a standardized assessment procedure to diagnose at
of prevalence for the general population. In another least three disorders according to DSM-IV/DSM-IV-TR/
literature review of studies in community samples, DSM-5, ICD-9/ICD-10, or DC: 0–3/D-C: 0–5;
Egger and Angold (2006) reported prevalence rates 3. inclusion of children who are 12–83 months old;
for any DSM mental disorder ranging from 14.0% to 4. publication year between 2006 and 2020.
26.4%. These findings were based on four studies Language of the publication was not an exclusion criterion.
published between 1982 and 2005. Average preva-
lence rates were highest for serious emotional dis-
Coding of studies
turbance and anxiety disorders. Both reviews
indicated that approximately one quarter of children Each study was coded (by R. G.) according to study charac-
with a mental disorder showed one or more comorbid teristics, population characteristics, sample characteristics,
study methods, assessment methods, and effect size. Difficul-
disorders. Based upon a more recently published
ties with coding of studies were discussed with the first author
selective literature search, von Klitzing et al. (2015) (M.V.). Study characteristics included information about the
reported prevalence rates ranging from 16% to 18%. author(s), research affiliation, and year of publication. Char-
None of these reviews included a meta-analysis of acteristics of the population referred to country; the Human
the prevalence rates or investigated the sources of Development Index, according to the United Nations; and year
of data collection. To describe the sample, we coded mean age,
variability found between studies. For older children
age range, and percentage of male participants. We also coded,
and adolescents ages 4–18 years, there is a meta- if the sample included only children older than three years or
analysis estimating the worldwide pooled prevalence younger children (age group). Study methods were coded
of any mental disorder at 13.4%, 95% CI [11.3, 15.9] according to location (country; urban size: large city area with
(Polanczyk et al., 2015). 1,000,000 or more citizens, medium-sized city area with
500,000–999,999 citizens, or small area with less than
500,000 citizens), sampling strategy (kindergartens/schools,
The present study households, birth register, primary care including regular
health check-ups or other) and study design (one-stage design
The aims of the present study were to update the with one assessment or two-stage design including screening
literature review of epidemiological studies with and some further assessment of a subsample). Assessment
methods were coded referring to diagnostic criteria, assess-
children younger than 7 years since the review of
ment instrument (DISC = Diagnostic Interview Schedule,
Egger and Angold (2006), and to estimate the pooled Lucas, Fisher, & Luby, 1998; PAPA =Preschool Age Psychiatric
worldwide prevalence of mental disorders in this age Assessment, Egger & Angold, 2004; or other), and informant.
group. We focused on the prevalence of any mental We coded if functional impairment was required for the
disorder, as well as specific mental disorders in diagnosis (‘no’, ‘yes’, and ‘yes/no’ when it was required only
for some diagnoses). The effect size was reported as the
community samples. Furthermore, we estimated the
proportion of children with any mental disorder, a specific
comorbidity of mental disorders in very young chil- mental disorder, or one or more comorbid mental disorders
dren. Findings of this review and meta-analysis relative to the whole sample.
could guide future research, policy making, and In cases of longitudinal data, the first wave was considered
treatment planning. to avoid bias due to dropouts. If several studies were based on

© 2020 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for
Child and Adolescent Mental Health.
 Prevalence of mental disorders in young children 3

Records idenfied through Addional records idenfied

database searching through other sources

Idenficaon
(n = 3,027) (n = 25)

Potenal records Duplicates removed

(n = 3,052) (n = 655)

Records assessed for Records excluded

eligibility (n = 2,397) (n = 2,380)
( )
Eligibility

Studies sasfying the Records excluded because

inclusion criteria of idencal samples
(n = 17) (n = 7)

Studies included in the

Included

meta-analysis
(n = 10)

Figure 1 Flowchart of study selection in the systematic review (2006–2020)

the same data and reported the same outcomes, the study with mental disorder. We included moderators identified as a
the largest sample was included in the meta-analysis. In cases significant source of variation in the previous meta-analysis
of two-stage data, the second sample was used to estimate the by Polanczyk et al. (2015) as well as further predictors that
number of children with mental disorders and the initial might influence heterogeneity in the prevalence: location,
sample was considered as base. In studies that included sample frame, diagnostic instrument, study design, functional
prevalence rates with and without the requirement of func- impairment (no or partly required vs. required for all disor-
tional impairment, rates with a requirement of functional ders), number of diagnoses, and age group (older than three
impairment were extracted. To ensure better comparison years vs. samples of younger children). We conducted univari-
between studies, data were included for the meta-analysis ate metaregression to analyze the effect of each predictor on
regarding DSM-IV and the mother’s report, if there were heterogeneity.
optional diagnostic criteria or informants. Publication bias was tested with the two-tailed Egger’s
regression test that analyzes asymmetry in the funnel plot
(Egger, Smith, Schneider, & Minder, 1997).
Statistical analysis
The meta-analyses were performed using the metaphor pack-
age in R version 1.9-8 (Viechtbauer, 2010). Random-effects Results
meta-analyses were conducted to estimate the pooled preva- The systematic review identified 17 studies encom-
lence of any mental disorder, specific mental disorders, and passing k = 10 independent community samples.
comorbidity. Analyses on specific mental disorders were con- Studies of 18,282 children ages 12–83 months were
ducted for commonly reported diagnoses (k ≥ 4 study sam-
included in the meta-analysis (Table 1). Symptoms
ples). Extracted proportions were transformed into logits for
more precise estimation (Lipsey & Wilson, 2001). If studies were assessed using parent report referring to the
reported rates of 0%, we set these rates to 0.1% to define the present or recent past (up to 3 months). Studies
logarithm. We calculated overall pooled effect sizes and the often relied on birth registries (k = 4) or primary care
95% confidence intervals, first including all studies and, next, practices (k = 3). Studies were conducted in eight
excluding outliers. All values were then back-transformed
countries – most of them in Western Europe (k = 4)
using inverse logit transformation, to facilitate interpretation.
Effects were considered outliers when their studentized deleted or the USA (k = 3). Most studies were conducted in
residuals were greater than 1.96 (Viechtbauer & Cheung, large city areas (Al-Jawadi & Abdul-Rhman, 2007;
2010). Outliers were extracted from the meta-analysis only if Ezpeleta, de la Osa, & Domenech, 2014; Gleason
case deletion diagnostics showed that extraction of these et al., 2011; Lavigne, LeBailly, Hopkins, Gouze, &
effects would have an substantial effect on the fitted model
Binns, 2009; Rijlaarsdam et al., 2015; Skovgaard
(by analyzing Cook’s distances) or on the variance–covariance
matrix of the parameter estimates (by analyzing covariance et al., 2007). In two studies, data were collected in
ratios). medium-sized cities (Petresco et al., 2014; Wich-
Using the restricted maximum likelihood method, we esti- strom et al., 2012), and further two studies in small
mated heterogeneity with Cochrane’s Q-test, the actual stan- cities (Bufferd, Dougherty, Carlson, & Klein, 2011;
dard deviation on the logarithmic scale (^sÞ, and I2 statistics.
Carter et al., 2010).
Moderator analysis was conducted for the prevalence of any

© 2020 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for
Child and Adolescent Mental Health.
 4

Table 1 Description of studies identified by the systematic review

Mira Vasileva et al.

Age range in
Years data Study Response Sample months % Screening/diagnostic Diagnostic Number of
Study Country collection Frame design rate in % size (mean) male instrument criteria diagnoses

Al-Jawadi and Abdul- Iraq 2003/2004 Primary care One-stage 95.4 829 12–48 (30.0b) 55.1 Interview form including DSM-IV 24
Rhman (2007) DSM-IV criteria
Bufferd et al. (2011) USA 2004–2007 Commercial One-stage 66.4 541 36–50 (43.2) 54.1 PAPA DSM-IV 14
mailing lists
Carter et al. (2010) USA 2000–2004 Birth registers Two-stage 89.1 1,329a 36–72 (54.0b) 49.6 ASBR, DISC-IV DSM-IV 12
Ezpeleta et al. (2014) Spain 2009/2010 Census of Two-stage 58.7 1,341a 36–47 (45.6) 50.9 SDQ, DICA-PPC DSM-IV 38
infant schools
Gleason et al. (2011) Romania – Primary care Two-stage 69.5 1,003a 18–60 (41.2) 51.8 CBCL, PAPA DSM-IV 18
Lavigne et al. (2009) USA – Kindergarden, One-stage 47.5 796 47–60 (53.3) 49.1 DISC-YC DSM-IV 8
primary care
Petresco et al. (2014) Brazil 2004 Birth registers One-stage 84.7 3,585 72–83 (81.6) 51.3 DAWBA DSM-IV, 28
ICD-10
Rijlaarsdam et al. Netherlands – Birth registers Two-stage 67.4 6,172a 60–83 (72.4) 50.0 CBCL, DISC-YC DSM-IV 21
(2015)
Skovgaard et al. Denmark 2000 Birth registers One-stage 65.0 211 17–19 (18.0b) 52.0 Interview with items DSM-IV, 10
(2007) from DC:0-3
CBCL, CHAT, ITSCL,
MEI and clinical
judgment
Wichstrom et al. Norway 2007/2008 Primary care Two-stage 79.5 2,475a 46–63 (53.0) 49.1 SDQ, PAPA DSM-IV 17
(2012)

ASBR, Adaptive Social Behavior Ratings (Skovgaard et al., 2007); CBCL, Child Behavior Checklist (Achenbach & Rescorla, 2000); CHAT, Checklist for Autism in Toddlers (Baron-Cohen
et al., 2000); DAWBA, Development and Well-Being Assessment (Goodman, Ford, Richards, Gatward, & Meltzer, 2000); DICA-PPC, The Diagnostic Interview of Children and Adolescents
for Parents of Preschool Children (Ezpeleta et al., 2014); DISC-YC, Diagnostic Interview Schedule, Young Children (Lucas, Fisher, & Luby, 1998); ITSCL, The Infant–Toddler Symptom
Checklist (DeGangi et al., 1995); MEI, Mannheim Eltern Interview (Esser et al., 1989); PAPA, Preschool Age Psychiatric Assessment (Egger & Angold, 2004); SDQ, Strengths and
Difficulties Questionnaire (Goodman, 1997).
a
In cases of two-stage design, only part of the sample was interviewed after screening the full sample.
b
Estimated based on available data.

© 2020 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for
Child and Adolescent Mental Health.
 Prevalence of mental disorders in young children 5

Since most studies assessed prevalence rates for as an outlier, with high prevalence rates for ADHD
any depressive disorder or combined rates for dys- and sleep disorders. Even after excluding outliers,
thymia and major depression, which were the most the heterogeneity of studies was substantial for
common diagnoses in this diagnostic group, we almost all specific disorders, as indicated by the Q-
calculated the pooled prevalence for any depressive test (Q = 20.59–446.99, df = 3–8, p ≤ .001). How-
disorder and not for specific subcategories. We also ever, the Q-test showed no significant heterogeneity
calculated the prevalence of any anxiety disorder for reactive attachment disorder (Q = 7.53, df = 3,
and of specific anxiety disorders. p = .057) or selective mutism (Q = 1.78, df = 2,
p = .412). The percentage of variation across studies
that is due to heterogeneity rather than chance was
Prevalence of any mental disorder
between I2 = 60.5 and I2 = 98.98. Tau ranged
All 10 studies reported prevalence rates for any between ^s = 0.06 and ^s = 1.27.
mental disorder (Figure 2). The overall pooled preva-
lence was 20.13%, 95% CI [15.72, 25.41]. There was
Comorbidity
substantial heterogeneity between the studies
(Q = 438.66, df = 9, p < .001; ^s = 0.47, 95%, CI Seven studies assessed the prevalence of children
[0.32, 0.88]; I2 = 98.28%). However, no study was having two or more diagnoses. The pooled prevalence
identified as an outlier. of comorbidity was estimated at 5.51%, 95% CI
[3.86, 7.80]. The study of Petresco et al. (2014) was
identified as an outlier that exerted a substantial
Specific disorders
effect on the fitted model. Without this study, the
Table 2 provides an overview of the pooled preva- pooled prevalence was 6.44%, 95% CI [1.33, 54.02].
lence rates of specific disorders or groups of disor- Heterogeneity remained substantial (Q = 20.59,
ders (see Appendix S1 for forest plots of specific df = 6, p = .001; ^s = 0.25, 95%, CI [0.12, 0.74];
disorders). The highest pooled prevalence rates for a I2 = 84.19%).
specific disorder were identified for attention-deficit
hyperactivity disorder (ADHD: 2.73%–4.27%) and
Metaregression
ODD (3.90%–4.90%; Table 2). Other common diag-
noses were specific phobias (2.36%–3.23%), feeding Metaregression analysis revealed no significant pre-
disorders (1.36%–2.89%), and sleep disorders dictors of heterogeneity in the rates of any mental
(1.65%–2.89%). disorder (location: QM = 0.61, df = 1, p = .433; sam-
Most studies with effects that were detected as ple frame: QM = 0.19, df = 1, p = .664; diagnostic
outliers identified lower rates than the remaining instrument: QM = 1.91, df = 2, p = .384; study
studies (Table 2; Al-Jawadi & Abdul-Rhman, 2007; design: QM = 0.46, df = 1, p = .497; requirement of
Carter et al., 2010; Gleason et al., 2011; Petresco functional impairment: QM = 0.32, df = 1, p = .57,
et al., 2014; Rijlaarsdam et al., 2015; Wichstrom number of diagnoses: QM = 3.30, df = 2, p = .192,
et al., 2012). As regards to outliers with higher rates, age group: QM = 0.01, df = 1, p = .877).
the study by Al-Jawadi and Abdul-Rhman (2007)
was identified as an outlier for PTSD and reactive
Publication bias
attachment disorder. Bufferd et al. (2011) reported a
substantially higher prevalence of social phobias For most analyses, Eggers’ test was nonsignificant,
and selective mutism than the remaining studies. indicating no publication bias (|t| = 0.18–2.69,
The study by Ezpeleta et al. (2014) was also marked df = 3–8, p = .059 .866). There was significant

Figure 2 Pooled prevalence of any mental disorders

© 2020 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for
Child and Adolescent Mental Health.
 6

Table 2 Pooled prevalence of specific mental disordersa and comorbidity

Mira Vasileva et al.

Pooled prevalence
Study prevalence (%) (% [95% CI])

Al-Jawadi and Bufferd Carter Ezpeleta Gleason Lavigne Petresco Skovgaard

Abdul-Rhman et al. et al. et al. et al. et al. et al. Rijlaarsdam et al. Wichstrom Without
Diagnosis (2007) (2011) (2010) (2014) (2011) (2009) (2014) et al. (2015) (2007) et al. (2012) All studies outliers

ADHD 0.0b 2.0 8.7 3.7 0.4b 12.8 2.6 8.0 2.4 1.9 2.7 [1.3, 5.8] 4.3 [2.5, 7.2]
ODD 0.4b 9.4 8.4 6.9 0.9 13.4 2.0 10.1 – 1.8 3.9 [1.8, 8.3] 4.9 [2.5, 9.5]
CD 0.8 – 1.2 1.4 0.2 – 0.6 0.2b – 0.7 0.6 [0.4, 1.1] 0.8 [0.6, 1.2]
Depression 0.5 1.5 0.1b 3.4b 1.4 0.6 1.3 0.8 – 2.0 1.1 [0.7, 1.8] 1.1 [0.8, 1.6]
Any anxiety – 19.6 – 6.6 4.5 – 8.8 7.8 – 1.5b 6.4 [3.2, 12.4] 8.5 [5.2, 13.5]
disorders
Separation 0.8 5.4 2.2 2.2 1.3 – 3.2 0.9 – 0.3b 1.6 [0.8, 2.6] 1.9 [1.1, 3.2]
anxiety
GAD – 3.9 0.2 0.1 2.5 0.6 0.2 0.4 – – 0.6 [0.2, 1.6] –
Specific 0.0b 9.1 9.0 3.7 0.3 – 5.4 6.1 – 0.7 2.4 [0.8, 6.7] 3.2 [1.3, 08.0]
phobia
Social – 4.4b 0.3 1.9 0.6 – 0.1b 1.2 – 0.5 0.8 [0.3, 1.9] 0.8 [0.4, 1.5]
phobia
PTSD 11.0b – 0.3 0.0 0.2 – 0.8 0.1 – – 0.5 [0.1, 2.0] 0.2 [0.1, 0.6]
RAD 6.5b 0.9 – 0.0 2.0 – – – 0.9 – 0.7 [0.2, 3.1] 0.4 [0.1, 1.1]
Feeding 4.9 2.8 – 1.7 – – 0.0b – 2.8 – 1.4 [0.3, 6.3] 2.9 [1.7, 4.7]
disorders
Sleep – – – 13.3b 4.2 – – – 1.4 0.7 2.9 [0.8, 1.0] 1.7 [0.5, 4.5]
disorders
Selective 0.1 1.5b – 0.4 0.2 – – – – – 0.4 [0.1, 1.3] 0.3 [0.2, 0.6]
mutism
Tic disorders 1.2 1.7 2.1 0.4 0.0b 0.7 [0.2, 2.1] 0.1 [0.6, 2.4]
Comorbidity – 9.2 5.8 7.5 3.9 6.4 2.2b 6.7 – – 5.5 [3.9, 7.8] 6.4 [5.2, 7.9]

ADHD, attention-deficit hyperactivity disorder; CD, conduct disorder; GAD, generalized anxiety disorder; ODD, oppositional defiant disorder; PTSD, post-traumatic stress disorder; RAD,
reactive attachment disorder.
a
The table includes only mental disorders that were reported in at least four studies.
b
Outliers with studentized deleted residuals ≥1.96 and covariance <1.

© 2020 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for
Child and Adolescent Mental Health.
 Prevalence of mental disorders in young children 7

asymmetry in the funnel plot for ODD (t = 2.42, ongoing psychotherapy in older children and ado-
df = 7, p = .046) and reactive attachment disorder lescents had positive effects and reduced rates of
(t = 7.17, df = 3, p = .006). Especially for ODD, mental disorders (Weisz et al., 2017). Furthermore,
there was the tendency that large studies with there could be informant bias: symptoms were
small standard errors reported rather large effects reported by the parents in our meta-analysis, while
while smaller studies reported smaller effects. older children and adolescents rated their symptoms
Asymmetry was not more significant after exclud- on their own in the previous meta-analysis of
ing outliers (ODD: t = 2.07, df = 6, p = .084; Polanczyk et al. (2015).
reactive attachment disorder: t = 1.65, df = 2, The prevalence of specific disorders was similar for
p = .241). Overall, publication bias was not young and older children and adolescents for ADHD,
expected to be substantial, since the publication ODD, and any anxiety disorder (Polanczyk et al.,
of epidemiological studies usually does not depend 2015). One potential explanation for the similar
on significant results. prevalence rates could be a strong genetic risk for
these disorders (Demontis et al., 2019; Shimada-
Sugimoto, Otowa, & Hettema, 2015). On the other
Discussion hand, the findings might be interpreted in the
The current meta-analysis aimed to estimate the context of very early-onset environmental determi-
worldwide prevalence of mental disorders and nants of child psychopathology (Koss & Gunnar,
comorbidity in children younger than 7 years. Based 2018).
on ten studies from eight countries, we found a In the current meta-analysis, we found lower
20.13%, 95% CI [15.72, 25.41] pooled prevalence of prevalence rates for conduct disorder and depres-
any mental disorder. This means that every fifth sion than previously reported for older children and
child suffers from a mental health problem that adolescents (Polanczyk et al., 2015). For conduct
satisfies categorical diagnostic criteria. Except in one disorder, these discrepancies might be associated
study (Al-Jawadi & Abdul-Rhman, 2007), functional with increased rates of adolescence-limited antiso-
impairment was required for all or part of the cial behavior in the older sample (Moffitt, 1993). The
diagnoses. Furthermore, 6.44% of young children higher rates of depression in older children and
in community samples had two or more comorbid adolescents could be due to the progressing socio-
disorders. Hence, every third child with a mental emotional and cognitive development that is associ-
disorder fulfills the criteria for at least one further ated with internalizing symptoms (Kilford, Garrett, &
psychiatric diagnosis, which might lead to higher Blakemore, 2016).
functional impairment and more persistent symp- Although there was substantial heterogeneity
toms (Egger & Angold, 2006). between single study effects, we could not find
Our findings are consistent with previous reviews significant sources of variation. In contrast, Polanc-
estimating the prevalence of any mental disorder in zyk et al. (2015) identified study location, data
very young children between 14.0% and 26.4% frame, diagnostic instrument, and requirement of
(Egger & Angold, 2006; von Klitzing et al., 2015). functional impairment as accounting for a signifi-
We found a slightly higher prevalence of any mental cant proportion of the heterogeneity between stud-
disorder than a meta-analysis of older children and ies. Especially requirement of functional impairment
adolescents, which identified a prevalence of 13.4%, has been recognized important in the diagnostics of
95% CI [11.3, 15.9] (Polanczyk et al., 2015). Com- mental disorders in preschool children (Egger &
paring the results of the two meta-analyses of Angold, 2006). Our literature review showed that
younger and older children is difficult because there most epidemiological studies of young children have
were a different number of diagnoses investigated for also included functional impairment as critical
the different age groups and the diagnostic criteria aspect to differentiate normal variation in child
for young children have been criticized to be insen- behavior from mental disorders that might require
sitive to detect symptoms in this age group. The therapeutic help.
prevalence in the current meta-analysis usually We detected few studies as outliers showing pos-
referred to the past three months while studies of sible reasons for variation. For example, Al-Jawadi
older children and adolescents used different tame and Abdul-Rhman (2007) conducted an epidemio-
frames (current, 6-month, or 12-month prevalence). logical study in Iraq and found high prevalence rates
However, the slightly higher prevalence compared to for trauma- and stress-related disorders compared
older children and adolescents could be because to the remaining studies. These findings are most
some disorders typically manifest in younger chil- likely associated with children’s high exposure to
dren (e.g., sleep disorders and separation anxiety). violence and deprivation, due to the political and
From a developmental psychopathological point of humanitarian situation in that country (AlObaidi,
view, children who deviate from normal development Jeffrey, Scarth, & Albadawi, 2009). Another outlier
at an early stage can still achieve adequate adapta- effect that substantially changed the pooled preva-
tion and return to normal using their resources lence of sleep disorders was found in the study of
(Sroufe, 1997). It is also possible that previous or Ezpeleta et al. (2014) in Spain. The higher prevalence

© 2020 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for
Child and Adolescent Mental Health.
8 Mira Vasileva et al.

there might indicate potential cultural or environ- even hinders the successful treatment (Weisz et al.,
mental effects. Although the previous meta-analysis 2017).
of the prevalence of mental disorders in older chil- Our findings show that the epidemiology of mental
dren and adolescents did not detect any cultural health in young children is still a neglected area. While
influence (Polanczyk et al., 2015), we could not Polanczyk et al. (2015) found and synthesized preva-
preclude such an effect, because only eight countries lence rates from 198 studies of older children and
were represented. adolescents, we only found ten studies on indepen-
dent samples of children younger than 7 years. More
research is needed in this area to estimate prevalence
Limitations
rates more precisely and specifically to each develop-
There were certain limitations that should be mental stage (infants, toddlers, preschool children),
considered when interpreting our results. We and to detect potential sources of variability. Further-
selected only papers that reported multiple diag- more, future epidemiological studies should pay more
noses and excluded potential studies that esti- attention to developmental peculiarities of this age
mated the prevalence of a single diagnosis. Some such as heterogeneity in child development and
epidemiological studies did not report prevalence dependence on caregivers to regulate emotions.
rates for less-frequent disorders. Therefore, we Although the DC: 0–5 addresses developmental vari-
could not calculate pooled estimates for some ations in the classification of disorders in young
disorders like mixed disorders of conduct and children, only one study used these criteria, while
emotions. Furthermore, the small number of effects most were based on the DSM-IV. It should also be
that were included in some meta-analyses, espe- acknowledged that this review and meta-analysis is
cially of specific disorders, could lead to impreci- based on previously published studies and does not
sion in the prevalence estimates. In the answer questions such as to what extend mental
metaregression, including only a few studies could health problems in young children can be seen as
lead to insufficient statistical power to detect small disorders and if we have the reliable classification and
moderator effects on heterogeneity. We combined diagnostic tools to assess such disorders. However,
results for children younger and older than three since this is the first meta-analysis of the prevalence of
years. Even though in our analysis younger age mental disorders in children younger than 7 years,
could not explain heterogeneity between studies, the pooled prevalence rates we found could be used as
mental disorders are expected to manifest differ- benchmarks to compare against the future results of
ently at different developmental stages. It should epidemiological studies.
be acknowledged that, although our aim was to
estimate worldwide prevalence, we could not con-
clude that the pooled prevalence based on studies Supporting information
from eight countries is representative of worldwide Additional supporting information may be found online
prevalence. Finally, our results were mainly based in the Supporting Information section at the end of the
on DSM-IV and give no reference to DSM-5, ICD, article:
or DC: 0-5.
Appendix S1. Forest plots for the prevalence of specific
disorders.
Implications
The current results have important clinical and
research implications. The high prevalence of mental Acknowledgements
The authors would like to acknowledge the contribution
disorders emphasizes the necessity of appropriate
of Franz Petermann who passed away in 2019. We
treatment. There are some effective interventions for thank him for being our mentor and for his support in
young children that can be provided to children in this and many valuable projects in the field of child
need (von Klitzing et al., 2015). Most of the children clinical psychology.
with a mental disorder in our meta-analysis had a
single diagnosis. Hence, early interventions might
help to prevent mental problems from becoming Correspondence
more complex and developing comorbidity. On the Mira Vasileva, Child and Community Wellbeing Unit,
other hand, one third of young children with mental Centre for Health Equity, Melbourne School of Popula-
disorders fulfilled the criteria for more than one tion and Global Health, University of Melbourne, 207
diagnosis. These children might need more intensive Bouverie Street, Melbourne, VIC 3053, Australia;
help because comorbidity often complicates and Email: [email protected]

© 2020 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for
Child and Adolescent Mental Health.
 Prevalence of mental disorders in young children 9

Key points

 There are few previous reviews estimating the prevalence of mental disorders in children between 1 and 7
years. None has so far included a meta-analytical strategy.
 A meta-analysis based on ten epidemiological studies published between 2006 and 2020 estimated a 20.1%
pooled prevalence for any mental disorder in children ages 12–83 months. The pooled prevalence for a
specific disorder was between 0.1% and 4.9%, and 6.4% for comorbidity.
 More research with developmentally sensitive diagnostic criteria is needed to estimate prevalence rates more
precisely and to detect sources of variability.
 There is a substantial number of young children who need appropriate treatment.

Egger, M., Smith, G.D., Schneider, M., & Minder, C. (1997).

References Bias in meta-analysis detected by a simple, graphical test.
References marked with an asterisk indicate studies BMJ, 315, 629–634.
included in the meta-analysis. Esser, G., Blanz, B., Geisel, B., & Laucht, M. (1989).
Achenbach, T., & Rescorla, L. (2000). Manual for the ASEBA Mannheimer Elterninterview [Mannheim Parent Interview].
preschool form. Burlington: University of Vermont. Mannheim, Germany: Beltz.
*Al-Jawadi, A.A., & Abdul-Rhman, S. (2007). Prevalence of *Ezpeleta, L., de la Osa, N., & Domenech, J.M. (2014).
childhood and early adolescence mental disorders among Prevalence of DSM-IV disorders, comorbidity and impair-
children attending primary health care centers in Mosul, ment in 3-year-old Spanish preschoolers. Social Psychiatry
Iraq: A cross-sectional study. BMC Public Health, 7, 274. and Psychiatric Epidemiology, 49, 145–155.
AlObaidi, A.K., Jeffrey, L.R., Scarth, L., & Albadawi, G. (2009). *Gleason, M.M., Zamfirescu, A., Egger, H.L., Nelson, C.A., Fox,
Iraqi children’s rights: Building a system under fire. N.A., & Zeanah, C.H. (2011). Epidemiology of psychiatric
Medicine, conflict and survival, 25, 148–165. disorders in very young children in a Romanian pediatric
American Psychiatric Association (2013). Diagnostic and sta- setting. European Child & Adolescent Psychiatry, 20, 527–535.
tistical manual of mental disorders, 5th edn. Washington, Goodman, R. (1997). The Strengths and Difficulties Question-
DC: American Psychiatric Publishing. naire: A research note. Journal of Child Psychology and
Angold, A., & Egger, H.L. (2007). Preschool psychopathology: Psychiatry, 38, 581–586.
Lessons for the lifespan. Journal of Child Psychology and Goodman, R., Ford, T., Richards, H., Gatward, R., & Meltzer,
Psychiatry, 48, 961–966. H. (2000). The Development and Well-being Assessment:
Baron-Cohen, S., Wheelwright, S., Cox, A., Baird, G., Char- Description and initial validation of an integrated assess-
man, T., Swettenham, J., . . . & Doehring, P. (2000). Early ment of child and adolescent psychopathology. Journal of
identification of autism by the Checklist for Autism in Child Psychology and Psychiatry, 41, 645–655.
Toddlers (CHAT). Journal of the Royal Society of Medicine, Keenan, K., & Wakschlag, L.S. (2000). More than the terrible
93, 521–525. twos: The nature and severity of behavior problems in clinic-
*Bufferd, S.J., Dougherty, L.R., Carlson, G.A., & Klein, D.N. referred preschool children. Journal of Abnormal Child
(2011). Parent-reported mental health in preschoolers: Psychology, 28, 33–46.
Findings using a diagnostic interview. Comprehensive Psy- Kilford, E.J., Garrett, E., & Blakemore, S.-J. (2016). The
chiatry, 52, 359–369. development of social cognition in adolescence: An inte-
*Carter, A.S., Wagmiller, R.J., Gray, S.A., McCarthy, K.J., grated perspective. Neuroscience & Biobehavioral Reviews,
Horwitz, S.M., & Briggs-Gowan, M.J. (2010). Prevalence of 70, 106–120.
DSM-IV disorder in a representative, healthy birth cohort at Koss, K.J., & Gunnar, M.R. (2018). Annual research review:
school entry: Sociodemographic risks and social adaptation. Early adversity, the hypothalamic-pituitary-adrenocortical
Journal of the American Academy of Child & Adolescent axis, and child psychopathology. Journal of Child Psychology
Psychiatry, 49, 686–698. and Psychiatry, 59, 327–346.
DeGangi, G., Poisson, S., Sickel, R., & Wiener, A.S. (1995). *Lavigne, J., LeBailly, S.A., Hopkins, J., Gouze, K.R., & Binns,
Infant/Toddler Symptom Checklist: A screening tool for H.J. (2009). The prevalence of ADHD, ODD, depression, and
parents. Tucson, AZ: Therapy Skill Builders. anxiety in a community sample of 4-year-olds. Journal of
Demontis, D., Walters, R.K., Martin, J., Mattheisen, M., Als, Clinical Child and Adolescent Psychology, 38, 315–328.
T.D., Agerbo, E., . . . & Bækvad-Hansen, M. (2019). Discovery Lincoln, C.R., Russell, B.S., Donohue, E.B., & Racine, L.E.
of the first genome-wide significant risk loci for attention (2017). Mother-child interactions and preschoolers’ emotion
deficit/hyperactivity disorder. Nature Genetics, 51, 63–75. regulation outcomes: Nurturing autonomous emotion regu-
Egger, H.L., & Angold, A. (2004). The Preschool Age Psychiatric lation. Journal of Child and Family Studies, 26, 559–573.
Assessment (PAPA): A structured parent interview for diag- Lipsey, M.W., & Wilson, D.B. (2001). Practical meta-analysis.
nosing psychiatric disorders in preschool children. In D. Thousand Oaks, CA: Sage.
Delcarmen-Wiggens & A. Carter (Eds.), Handbook of infant, Lucas, C., Fisher, P., & Luby, J. (1998). Young-child DISC-IV
toddler, and preschool mental health assessment (pp. 223– research draft: Diagnostic interview schedule for children.
243). New York, NY: Oxford University Press. New York, NY: Columbia University.
Egger, H.L., & Angold, A. (2006). Common emotional and Lyons-Ruth, K., Manly, J.T., Von Klitzing, K., Tamminen, T.,
behavioral disorders in preschool children: Presentation, Emde, R., Fitzgerald, H., Paul, C., Keren, M., Berg, A., Foley,
nosology, and epidemiology. Journal of Child Psychology and M., & Watanabe, H. (2017). The worldwide burden of infant
Psychiatry, 47, 313–337. mental and emotional disorder: Report of the task force of

© 2020 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for
Child and Adolescent Mental Health.
10 Mira Vasileva et al.

the World Association for Infant Mental Health. Infant Sroufe, L.A. (1997). Psychopathology as an outcome of devel-
Mental Health Journal, 38, 695–705. opment. Development and Psychopathology, 9, 251–268.
McDonnell, M.A., & Glod, C. (2003). Prevalence of psy- Vasileva, M., Haag, A.C., Landolt, M.A., & Petermann, F.
chopathology in preschool-age children. Journal of Child (2018). Posttraumatic stress disorder in very young children:
and Adolescent Psychiatric Nursing, 16, 141–152. Diagnostic agreement between ICD-11 and DSM-5. Journal
Moffitt, T.E. (1993). Adolescence-limited and life-course-per- of Traumatic Stress, 31, 529–539.
sistent antisocial behavior: A developmental taxonomy. Viechtbauer, W. (2010). Conducting meta-analysis in R with the
Psychological review, 100, 674–701. metafor package. Journal of Statistical Software, 36, 1–48.
Moher, D., Liberati, A., Tetzlaff, J., & Altman, D.G. (2009). Viechtbauer, W., & Cheung, M. (2010). Outlier and influence
Preferred reporting items for systematic reviews and meta- diagnostics for meta-analysis. Research Synthesis Methods,
analyses: The PRISMA statement. Annals of internal medi- 1, 112–125.
cine, 151, 264–269. von Klitzing, K., Dohnert, M., Kroll, M., & Grube, M. (2015).
*Petresco, S., Anselmi, L., Santos, I.S., Barros, A.J., Fleitlich- Mental disorders in early childhood. Deutsches Arzteblatt
Bilyk, B., Barros, F.C., & Matijasevich, A. (2014). Prevalence International, 112, 375–386.
and comorbidity of psychiatric disorders among 6-year-old Weisz, J.R., Kuppens, S., Ng, M.Y., Eckshtain, D., Ugueto,
children: 2004 Pelotas Birth Cohort. Social Psychiatry & A.M., Vaughn-Coaxum, R., . . . & Chu, B.C. (2017). What five
Psychiatric Epidemiology, 49, 975–983. decades of research tells us about the effects of youth
Polanczyk, G.V., Salum, G.A., Sugaya, L.S., Caye, A., & Rohde, psychological therapy: a multilevel meta-analysis and impli-
L.A. (2015). Annual Research Review: A meta-analysis of the cations for science and practice. American Psychologist, 72,
worldwide prevalence of mental disorders in children and 79–117.
adolescents. Journal of Child Psychology and Psychiatry, 56, *Wichstrom, L., Berg-Nielsen, T.S., Angold, A., Egger, H.L.,
345–365. Solheim, E., & Sveen, T.H. (2012). Prevalence of psychiatric
*Rijlaarsdam, J., Stevens, G.W., van der Ende, J., Hofman, A., disorders in preschoolers. Journal of Child Psychology and
Jaddoe, V.W., Verhulst, F.C., & Tiemeier, H. (2015). Preva- Psychiatry, 53, 695–705.
lence of DSM-IV disorders in a population-based sample of World Health Organization (1992). The ICD-10 classification of
5- to 8-year-old children: The impact of impairment criteria. mental and behavioural disorders: clinical descriptions and
European Child & Adolescent Psychiatry, 24, 1339–1348. diagnostic guidelines. Geneva: Author.
Shimada-Sugimoto, M., Otowa, T., & Hettema, J.M. (2015). Zeanah, C.H., Carter, A.S., Cohen, J., Egger, H., Gleason,
Genetics of anxiety disorders: Genetic epidemiological and M.M., Keren, M., Lieberman, A., Mulrooney, K., & Oser, C.
molecular studies in humans. Psychiatry and Clinical Neu- (2017). Introducing a new classification of early childhood
rosciences, 69, 388–401. disorders: DC: 0–5TM. Zero to Three Journal, 37, 11–17.
*Skovgaard, A.M., Houmann, T., Christiansen, E., Landorph, S., Zero to Three (2016). DC:0–5TM: Diagnostic classification of
Jorgensen, T., Olsen, E.M., . . . & CCC 2000 Study Team mental health and developmental disorders of infancy and
(2007). The prevalence of mental health problems in children early childhood. Washington: Author.
1 1/2 years of age – the Copenhagen Child Cohort 2000.
Journal of Child Psychology and Psychiatry, 48, 62–70. Accepted for publication: 15 April 2020

© 2020 The Authors. Journal of Child Psychology and Psychiatry published by John Wiley & Sons Ltd on behalf of Association for
Child and Adolescent Mental Health.