Journal of Child Psychology and Psychiatry 56:3 (2015), pp 370–393 doi:10.1111/jcpp.

12372

Annual Research Review: Secular trends in child and

adolescent mental health
Stephan Collishaw
Institute of Psychological Medicine and Clinical Neurosciences, MRC Centre for Neuropsychiatric Genetics and
Genomics, Cardiff University, Cardiff, UK

Background: Child and adolescent mental health problems are common, associated with wide-ranging functional
impairments, and show substantial continuities into adult life. It is therefore important to understand the extent to
which the prevalence of mental health problems has changed over time, and to identify reasons behind any trends in
mental health. Scope and Methodology: This review evaluates evidence on whether the population prevalence of
child and adolescent mental health problems has changed. The primary focus of the review is on epidemiological
cross-cohort comparisons identified by a systematic search of the literature (using the Web of Knowledge database).
Findings: Clinical diagnosis and treatment of child and adolescent psychiatric disorders increased over recent
decades. Epidemiological comparisons of unselected population cohorts using equivalent assessments of mental
health have found little evidence of an increased rate of ADHD, but cross-cohort comparisons of rates of ASD are
lacking at this time. Findings do suggest substantial secular change in emotional problems and antisocial behaviour
in high-income countries, including periods of increase and decrease in symptom prevalence. Evidence from low- and
middle-income countries is very limited. Possible explanations for trends in child and adolescent mental health are
discussed. The review also addresses how cross-cohort comparisons can provide valuable complementary
information on the aetiology of mental illness. Keywords: Time trends, secular change, depression, antisocial,
psychopathology.

mortality (Jokela, Ferrie, & Kivimaki, 2009; Mau-

Introduction
ghan, Stafford, Shah, & Kuh, 2014; Thapar et al.,
Epidemiological surveys indicate that psychiatric
2012). The prevention of child and adolescent mental
disorders in children and adolescents are common,
health problems is therefore an important health
affecting at least one in ten 5–16-year olds (Green,
priority (Institute of Medicine, 2009), and an accu-
McGinnity, Meltzer, Ford, & Goodman, 2005; Melt-
rate understanding of secular trends in child and
zer, Gatward, Goodman, & Ford, 2000). These
adolescent psychopathology is needed to gauge the
include neurodevelopmental disorders such as aut-
success of attempts to reduce the burden of child
ism spectrum disorder (ASD) and attention-deficit
psychiatric problems.
hyperactivity disorder (ADHD), mood and anxiety
Despite efforts to better understand child and
disorders, and conduct disorders. Emotional and
adolescent psychiatric conditions and how to pre-
behavioural problems more broadly defined are
vent them, there are concerns among parents,
considerably more common. Mental health problems
teachers and health professionals that today’s chil-
have wide-ranging detrimental impacts on young
dren are more prone to mental health problems than
people’s well-being, health and education. They are
were previous generations of children and young
also not transient phenomena (Maughan & Colli-
people (Barnard, Potter, Broach, & Prior, 2002;
shaw, 2015; Thapar, Collishaw, Pine, & Thapar,
Russell, Kelly, & Golding, 2010). Indeed, the past
2012). Long-term prospective studies highlight con-
20 years has seen considerable increases in the
tinuities between child and adult mental health, with
diagnosis and treatment of child and adolescent
over half of disorders in young adulthood preceded
psychiatric disorders (Fombonne, 2009; Getahun
by a psychiatric disorder in childhood (Kim-Cohen
et al., 2013; Ma, Lee, & Stafford, 2005; Smith,
et al., 2003), and conversely strong persistence and
Larkin, & Southwick, 2008). As discussed below,
recurrence of childhood psychiatric conditions into
there are a range of explanations for such trends,
adulthood (Birmaher et al., 2004). Psychiatric prob-
including increased help-seeking by parents and
lems account for a substantial and increasing world-
young people, improved screening and clinical rec-
wide burden of ill health (Murray et al., 2012).
ognition in schools and primary care, medicalization
Psychopathology is associated with risk for suicide,
of feelings and behaviours previously considered as
the second or third leading cause of death in
normal (Dowrick & Frances, 2013), and a broaden-
adolescents and young adults (Windfuhr et al.,
ing of diagnostic classifications of psychiatric disor-
2008); and with effects on chronic illness and early
ders. It is also possible that there have been real
changes in the prevalence of childhood psychiatric
Conflict of interest statement: No conflicts declared. conditions.

© 2014 Association for Child and Adolescent Mental Health.

Published by John Wiley & Sons Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main St, Malden, MA 02148, USA
doi:10.1111/jcpp.12372 ARR: Secular trends in child and adolescent mental health 371

The first aim of this review was to evaluate child psychopathology. A related issue is whether
evidence on whether the population prevalence of social inequalities in child mental health have
child and adolescent mental health problems has increased or decreased. Addressing health inequal-
changed. It is first necessary to consider the ities in morbidity and mortality is recognized as an
strengths and weaknesses of different methodologi- important policy priority nationally and internation-
cal designs that have been used to address this ally (Marmot, 2010). However, the extent of social
question. These include cross-sectional studies com- inequalities affecting child and adolescent mental
paring lifetime prevalence across different age health, and whether progress is being made in
cohorts, meta-analyses of epidemiological studies reducing such inequalities, is rarely considered.
using psychiatric interviews, cross-cohort compari- A full review of the myriad changes in children’s lives
sons in which symptom prevalence is directly com- that have occurred over recent decades is beyond the
pared, and time series tracking population suicide scope of this review. Instead, a number of illustrative
and offending rates. Using convergent evidence examples are discussed to highlight the methodolog-
across these different study methods, but with a ical and conceptual challenges involved.
primary focus on epidemiological cross-cohort com- A third aim was to highlight examples where cross-
parisons, this paper will review trends in the most cohort comparisons have provided novel insights on
common types of child and adolescent psychopa- the role of hypothesized risk factors. Sudden dis-
thology – neurodevelopmental problems, antisocial continuities in the social environment can approxi-
behaviour and emotional problems. In understand- mate features of a ‘natural experiment’ (Thapar &
ing these trends, it is important to consider possible Rutter, 2015). In addition, studies of secular change
variations between children and adolescents, boys for a range of phenotypes (e.g. height, age at pubertal
and girls, different socioeconomic groups and across maturation, cognitive ability) have pointed to some-
different countries. Social change affecting risks for what different conclusions regarding the role of
mental health problems is unlikely to have impacted environmental and heritable factors when compared
in the same way on these different groups. The to studies of within-population variation (Dickens &
review focuses on trends since the middle of the last Flynn, 2001; Silventoinen et al., 2003).
century, but recent fluctuations may be relatively Throughout, this review aims to identify important
minor when compared to very long-term historical knowledge gaps and to provide guidance on priorities
change in rates of mental ill health and violence for future research.
(Pinker, 2011). It is also important to note from the
outset that epidemiological ‘like-for-like’ compari-
sons of trends in population prevalence are at this Methodological considerations
stage only available for a small number of high- Studies have used various designs to consider
income countries, and almost completely lacking in whether the prevalence of child mental health prob-
low- and middle-income countries. lems has changed across time. Apart from research
Next, the review considers the reliability and tracking rates of diagnosis and treatment in clinical
validity of evidence about trends in child and practice, studies have also examined population
adolescent mental health, and in particular, the suicide and crime rates, generational differences in
possibility that changes in reporting confound lifetime rates of retrospectively reported mental ill
observed trends. A number of methods can help health, used meta-analytic methods to compare
point to more robust conclusions. These include the rates of interview-assessed disorder in epidemiolog-
consideration of convergent evidence across multiple ical surveys conducted at different time points, and
informants and study methods, data about predic- undertaken direct cross-cohort comparisons of epi-
tive validity where longitudinal data have been demiological studies using equivalent mental health
collected across multiple cohorts, and evidence for symptom screens.
specificity of trends in particular aspects of child Each of these approaches has pros and cons. Four
mental health. general issues are important to consider. First, to
Cross-cohort comparisons provide an opportunity what extent is a study able to compare ‘like with like’,
to address a number of other important questions both in terms of samples and measurement? Second,
beyond simply tracking trends in prevalence. to what extent does the coverage of a study extend to
A second aim was to consider possible explanations the whole population rather than focusing on
for any increases or decreases in the prevalence of selected subgroups? A third challenge relates to the
child mental health problems. The past 50 years has possibility that the way informants report mental
seen substantial societal changes which are likely to health problems has changed, even when equivalent
have had far-reaching impacts on children’s lives measures are used in different cohorts. Finally,
and well-being. Cross-cohort comparisons with linked data on hypothesized explanatory factors is
linked data on risk factors and mental health are required in order to test explanations for trends in
sparse, but where such data are available they mental health.
provide valuable opportunities for understanding More specifically, time trends studies can be
the possible impact of changes in risk factors on evaluated according to the extent to which they

© 2014 Association for Child and Adolescent Mental Health.

372 Stephan Collishaw J Child Psychol Psychiatry 2015; 56(3): 370–93

encompasses the following features: (a) comparison point. Studies have typically reported high response
of unselected representative epidemiological cohorts rates (>80%) or addressed the possibility of selective
using the same sampling frames; (b) equivalence in drop-out analytically, for example by using cohort-
response and methods for dealing with selective specific sample weights derived from prior more
nonresponse; (c) use of equivalent measures of complete data. All selected studies used validated
mental health symptoms or disorders; (d) the ability symptom screens with little or no variation across
to provide convergent data across multiple methods assessments. One longer term trends study, where
or informants; (e) data on impairment as well as similar but nonidentical assessments were com-
symptoms of psychopathology; (f) contemporaneous pared used calibration methods to ensure compara-
rather than retrospective assessment; (g) linked and bility of comparisons across time (Collishaw,
comparable data on hypothesized explanatory fac- Maughan, Goodman, & Pickles, 2004). Not included
tors; (h) longitudinal follow-up or data linkage are studies which have used bespoke items to
allowing a test of the predictive validity of mental examine trends in mental health-related phenom-
health trends with respect to independently assessed ena, including self-reported feelings of depression or
functional outcomes. anxiety, self-harm or suicidality, delinquency and
Finally, it is important to consider distinctions substance use. It was beyond the scope of this review
between age, period and cohort effects (Robertson & to fully consider these here, but they may also
Boyle, 1998). These are typically confounded in most provide valuable information on trends in child and
studies, but have important implications for under- adolescent mental health. One example is the cross-
standing reasons behind changes in prevalence. It is national World Health Organization Health Behav-
well known that patterns of disease vary by age. They iour among School-aged Children series of studies
might also vary for different periods of historical (see HBSC.org for details). Turning back to the
time, for example as a result of economic recession. studies included here, data from multiple infor-
Cohorts of individuals born at the same time may mants, information about functional impact of
also share risk for disease independent of their age symptoms, longitudinal validation of trends, infor-
and period of assessment. Comparisons of longitu- mation on whether trends differ by gender or soci-
dinal birth cohorts at multiple points in times are odemographic subgroup and linked data on
required to distinguish between period and cohort potential explanatory factors all provide important
effects (Keyes et al., 2014), but very few such studies strengths. Most studies included only one informant,
have been undertaken. In this paper, the term ‘cross- typically parent reports for younger children and
cohort comparison’ refers to comparisons of repeat youth self-reports in adolescence. One Finnish study
epidemiological samples assessed at different points of 8-year-old children also included child-based
in times. It is important to remember that any reports of symptoms (Sourander, Niemel€ a, Santa-
difference could reflect either period effects or cohort lahti, Helenius, & Piha, 2008). Impact of symptoms
effects or some combination of the two. on psychosocial functioning has rarely been
assessed, and only one study used independent data
on longitudinal outcomes to validate trends (Colli-
Trends in child and adolescent mental health shaw et al., 2004). Finally, while most studies
This section of the review focuses on trends in the examined whether trends differed by gender, only a
most commonly occurring types of child and adoles- minority of studies directly tested how far trends in
cent psychopathology. The main focus is on studies mental health might be accounted for by trends in
that meet the most important of the criteria outlined hypothesized explanatory factors.
above, namely those in which contemporaneous
representative population samples have been com-
Neurodevelopmental disorders
pared using equivalent assessments of mental
health. Perhaps, the most dramatic increases in child psy-
Tables 1–3 provide overviews of key studies of chiatric diagnoses relate to changes in ADHD and
children, adolescents and mixed age groups respec- autism spectrum disorders (ASD). In the 1960s
tively that were included in this section of the review around one in 2500 children were diagnosed with
(see online supporting information for description of autism (Lotter, 1966); the incidence of autism spec-
search strategy). In total, 21 studies were identified trum disorders has increased steadily to around 3–6
that met relevant inclusion and exclusion criteria. Of children per 1000 children at the beginning of the
these, the majority were undertaken in the United 21st Century (Baird et al., 2006; Rutter, 2005).
Kingdom (6), Finland (3), other Nordic countries (3) There is some evidence of a continuing increase in
or the Netherlands (3). Five used samples of chil- diagnoses of autism spectrum disorders in the 21st
dren, 11 of adolescents, and 5 included children and Century (Boyle et al., 2011; Russell, Rodgers, Ukou-
adolescents. Studies used various sampling munne, & Ford, 2014; but see also Taylor, Jick, &
approaches, including via schools, private house- MacLaughlin, 2013). A recent systematic review of
holds or follow-ups of birth cohorts. Most studies worldwide studies found a median prevalence of 62
used equivalent sampling approaches at each time per 10,000 (Elsabbagh et al., 2012), and in some

© 2014 Association for Child and Adolescent Mental Health.

 Table 1 Summary of cross-cohort comparison studies of children using mental health symptom screens

Study Age Response Trends by Explanatory

Lead author (year) Country years (years) Sample Ns rates (%) Informants Measures Outcomes subgroup factors Key findings
doi:10.1111/jcpp.12372

Sourander (2004, Finland 1989 8 HH/ 986 96 Parent Rutter A Means Gender SES Decreased parent-report
2008) 1999 SCH 831 86 Teacher Rutter B High Family type conduct and total problem
Santalahti (2005, 2005 HH/ 870 84 Child CDI score Life events high scores 1989–1999
2008) SCH (boys)
Ilola (2013) HH/ Increased child-reported
Luntamo (2012) SCH depression (girls)
Increased somatic
symptoms
Changes in bully/
victimization
Sellers (2014) UK 1999 7 HH 1,033 83 Parent SDQ Means Gender Demograph Mean problems, variance,
2004 HH 648 76 Teacher Variance SES and high problems all
2008 BC 13 857 72 High reduced (especially boys;

© 2014 Association for Child and Adolescent Mental Health.

score gender converged)
Impact Impact changed (esp boys)
Trends not explained by
demographic changes
Tick (van der Netherlands 1989 2–3 HH 394 90 Parent CBCL Means Gender No Small drop in mean
Ende, Koot, & 2003 279 80 High SES problems and high scores
Verhulst 2007) score
Matijasevich Brazil 1993 4 BC 634 87 Parent CBCL Means Gender Demograph, Substantial increases in
(2014) 2004 BC 3,750 91 High SES perinatal, means and abnormal
score maternal range scores, especially
psychiatric aggressive behaviour, but
attention problems
decreased; Similar trends
by gender; Increased
problems most marked for
low SES. Increases not
explained by explanatory
factors
McArdle (2003) UK 1973 7–9 SCH 515 Not Teacher Rutter B High No No Reduced rate of high scores
1994 SCH 1,044 reported score

HH, random sample of private households; BC, birth cohort; SCH, school-based sampling.
ARR: Secular trends in child and adolescent mental health
373
 Table 2 Summary of cross-cohort comparison studies of adolescents using mental health symptom screens
374
Lead author Study Age Response Trends by Explanatory
(year) Country years (yrs) Sample Ns rates (%) Informants Measures Outcomes subgroup factors Key findings

von Soest Norway 1992 16– SCH 2,994 97 Youth Depressive Means Gender Substance Increased mean
(2014) 2002 17 SCH 3,438 91 Mood Variance SES Conduct depression (boys) and
2010 SCH 2,813 84 Inventory High Family Eating high depression scores
score type Appearance (greater for boys) from
1992 to 2002
Increased variance
Stephan Collishaw

No interaction by SES,
fam type
Decreased conduct
problems (2002-10
boys)
Cannabis, eating
problems partly
mediate cohort effect
(92-02) on depression,
as well as appearance
in girls
Fleming New 2007 10– SCH 9,107 74 Youth RADS-SF High Gender Demograph Increased depression
(2014) Zealand 2012 17 SCH 8,500 68 SDQ score Age and SDQ emotional and
hyperactivity scores,
decreased conduct
scores
Torikka Finland 2000– 14– SCH 618 084 85–90 Youth R-BDI High Gender None Small increase severe
(2014) 2011 16 total score SES depression, more
(annual) marked increases in
low SES groups
Sourander Finland 1998 13– SCH 1,458 89 Youth SDQ Means Gender None No change in mean
(2012) 2008 17 SCH 1,569 92 High emotional or conduct
score problems, small
decrease in
hyperactivity (girls); no
change in high problem
scores
Collishaw UK 1986 16– BC 4,524–7,120a 32–50 Youth GHQ-12 Means Gender Parenting Increased depression
(2010, 2006 17 HH 747 53 Parent Malaise High Fam type Parent scores and high scorers
2012) Rutter score SES mental (girls), no differences in
Schepman health trends by SES or family
(2011) type; Increase
attenuated when
accounting for parental
distress; small increase
in parent-reported
conduct not accounted
by parenting; no
change in hyperactivity

© 2014 Association for Child and Adolescent Mental Health.

J Child Psychol Psychiatry 2015; 56(3): 370–93

(continued)
 Table 2 (continued)

Lead author Study Age Response Trends by Explanatory

(year) Country years (yrs) Sample Ns rates (%) Informants Measures Outcomes subgroup factors Key findings

Sweeting UK 1987 15 HH 505 65 Youth GHQ-12 Means Gender School Increased depression
(2009, 2010) 1999 SCH 2,195 79 High SES Family means and high scores
doi:10.1111/jcpp.12372

West (2003) 2006 SCH 3,194 81 score Worries for girls from higher
Lifestyle SES (1987–1999) and
for boys and girls (1999
–2006)
Increase attenuated
when controlling for
family arguments/
worries, and school
disengagement/
worries. Evidence for
increased exposure to
and vulnerability to
risk factors
Sigfusdottir Iceland 1997 14– SCH 3,913 78–90 Youth Depression Means Gender No Depression increased

© 2014 Association for Child and Adolescent Mental Health.

(2008) 2000 15 6,346 and (girls), anxiety
2003 3,556 anxiety increased (boys and
2006 7,430 dimension girls)
scales
Tick (2008) Netherlands 1993 11– SCH 1,095 82 Youth YSR Means Gender No Decreased problems
2003 18 810 83 High Age (boys), increased
score SES internalizing problems
(girls; especially
younger adolescents).
Wangby Sweden 1970 15 SCH 522 89 Youth The Means Girls No No change emotional
(2005) 1996 SCH 529 88 Symptom Variance only and antisocial mean
Quest High scores; increased
and variability antisocial;
Low more girls with high
scores and with low antisocial
behaviour
Fichter Greece 1980 11– SCH 2,631 NK Youth GHQ-28 Means Gender No Increased mean and
(2004, 2005) Greek 1998 26b SCH 2,920 90 ANIS High high emotional and
immigrants 1980 10– SCH 815 NK score eating disturbance
to Germany 1998 21 SCH 881 90 symptom scores in both
locations

(continued)
ARR: Secular trends in child and adolescent mental health
375
 376 Stephan Collishaw J Child Psychol Psychiatry 2015; 56(3): 370–93

countries including Sweden and the United States,

type or SES; Increased

problems (1974–1999)

income inequalities in
hyperactivity; Similar
validated using adult

Trends not explained

by changes in family
one in every 100 children now meets criteria for an

emotional problems

emotional problems
outcomes; Increase

systematic change

trends by gender,
autism spectrum disorder (Centres for Disease Con-
Key findings

Increased conduct

family type, SES;

(1986–1999); No
trol & Prevention, 2012; Idring et al., 2012). A recent
meta-analysis estimated that the most recent
changes to the diagnostic classification system,
DSM-5, will result in a decrease in the number of
children diagnosed (Kulage, Smaldone, & Cohn,
2014).
Explanatory

There have also been marked increases in clinical

structure

diagnoses of ADHD, and corresponding increases in

factors

Family

the prevalence of medication treatment for ADHD in

SES

many countries across the world, including in North

America (Getahun et al., 2013; Olfson, Gameroff,
Trends by

Marcus, & Jensen, 2003; Toh, 2006), Europe (Atla-

subgroup

Family
Gender

dottir et al., 2014; Hsia & Maclennan, 2009; McCar-

type
SES

thy et al., 2012; Wong, Murray, Novak-Camilleri, &

Stephens, 2004), Australia (Stephenson, Karanges,
School-based sample with some outliers-mean ages 14.8–15.8 years (SD = 1.3–1.6 years). Age included as covariate in analyses.
Outcomes

& McGregor, 2013) and Asia (Man et al., 2014;

Means

Scheffler, Hinshaw, Modrek, & Levine, 2007).

score
High

A recent meta-analysis of epidemiological studies,

however, found no evidence for any systematic
change in the prevalence of ADHD diagnoses
Rutter-A/
Measures

once methodological variability across studies is

SDQc

accounted for (Polanczyk, Willcutt, Salum, Kieling,

& Rohde, 2014). It seems likely that long-term
increases in incidence of clinical diagnoses of neu-
Informants

rodevelopmental disorders are largely explained by a

Parent

broadening of diagnostic definitions and better rec-

ognition by professionals (Elsabbagh et al., 2012).
Indeed, most studies which have directly com-
HH, random sample of private households; BC, birth cohort; SCH, school-based sampling.
Response

pared rates of parent- or teacher-endorsed reports of

rates (%)

hyperactivity/inattention using equivalent epidemi-

ological samples and symptom screens (see Tables
Calibration methods used to compare across similar but nonidentical measures.
62
45
83

1–3) have found no evidence for a systematic

increase in prevalence of ADHD-related difficulties
either in adolescent or child samples, with symptom
10 499

levels and the proportion of children with high scores

7,293

remaining stable or showing small reductions over

868
Ns

recent decades (e.g. Achenbach, Dumenci, & Resc-

orla, 2002a; McArdle, Prosser, Dickinson, & Kolvin,
Sample

2003; Collishaw et al., 2004; Tick, van der Ende,

HH
BC
BC

Koot, & Verhulst, 2007; Sourander, Niemela, Santa-

lahti et al., 2008; Collishaw, Maughan, Natarajan, &
(yrs)

16
15–
Age

Pickles, 2010; H€ olling, Schlack, Petermann, Ravens-

Sieberer, & Mauz, 2014; Sellers, Maughan, Pickles,
Thapar, & Collishaw, 2015; but see also Tick, van
Higher response for parent than youth.
Study

1986
1999
years

1974

der Ende, & Verhulst, 2007; Fleming et al., 2014).

Similar studies are lacking in relation to symptoms
of ASD, but constitute an urgent priority given public
concern about a possible real change in population
Country

prevalence.
UK
Table 2 (continued)

Affective disorders
Gore-Langton
(2004, 2007),

Service use for affective disorders – including help-

Lead author

seeking, diagnosis, and medication treatment – has

Collishaw

increased substantially in high-income countries

(2011)
(year)

over recent decades (Kosidou et al., 2010; Olfson,

Blanco, Wang, Laje, & Correll, 2014; Smith et al.,
b
a

© 2014 Association for Child and Adolescent Mental Health.

 Table 3 Summary of cross-cohort comparison studies of mixed age ranges using mental health symptom screens

Lead author Study Age Response Trends by Explanatory

(year) Country years (years) Sample Ns rates (%) Informants Measures Outcomes subgroup factors Key findings

H€olling Germany 2003–6 3–17 HH – 14 477 67 Parent SDQb Means Gender No Small increases emotional
(2014) 2009–12 quest 10 353a 39–73 High Age problems (adolescents),
HH- score SES conduct problems (children),
phone no change hyperactivity; Mean
trends only for low SES.
doi:10.1111/jcpp.12372

No change high scores

Maughan UK 1999 5–15 HH 10 438 83 Parent SDQ Means Gender SES Reduced total problems (parent-
(2008) 2004 HH 7,977 76 Teacher Age Family rated), conduct problems
Child SES type (parent, youth-rated),
(11+) Fam hyperactivity (parent-rated),
type emotional problems (youth-
rated, but reverse teacher-
rated). No variation in trends by
gender or most
sociodemographic variables.
Changes partly explained by
improving social conditions
Tick Netherlands 1983 6–16 HH 1,735 85 Parent CBCL Means Gender No Small increases in mean
(van der 1993 HH 1,715 82 Teacher TRF High Age problem and % high scores

© 2014 Association for Child and Adolescent Mental Health.

Ende, 2003 HH 1,417 75 score SES (parent reports). Similar trends
& Ethnic by gender; early increases only
Verhulst, for adolescents
2007)
Lin (2007) China 1993c 6–14 SCH 1,960 Not Parent Rutter A High Gender No Increased emotional problems
2003 SCH 3,711 known score (boys and girls); small increase
in externalizing behaviour
(girls)
Achenbach US 1976d 7–16/ HH 1,442 82 Parent CBCL Means Gender No Increase in parent problems
(2002a, 1989 18 HH 2,446 90 Teacher YSR High Age 1976–1989 and teacher
2002b, 1999 HH 1,641 93 Child TRF score SES problems 1981–1989 then
2003) decrease in problem scores but
remaining higher than 1976
(reverse pattern for competence
scores). No difference in trends
by gender, age or SES
Youth reports (11–18) also
show decline problems 1989–
1999

HH, random sample of private households; BC, birth cohort; SCH, school based sampling.
a
One third of sample newly recruited (39% response); rest resampled at older ages from baseline survey (73% response).
b
Different modes of administration – questionnaire at baseline; telephone at follow-up.
ARR: Secular trends in child and adolescent mental health

c
Prior time point not included because of noncomparability of measure.
d
Children referred for mental health service use omitted from 1976 sample and from some matched comparison analyses in later samples. Teacher reports from separate survey conducted
in 1981 (Achenbach et al., 2002a, 2002b).
377
378 Stephan Collishaw J Child Psychol Psychiatry 2015; 56(3): 370–93

2008; Stephenson et al., 2013). This increase has logical samples using prospective assessments is
been especially marked among adolescents (Olfson therefore essential.
et al., 2014; Potteg ard, Zo€
ega, Hallas, & Damkier, Evidence against a secular trend in depression
2014). For example, visits to doctors by adolescents comes from a meta-analysis of 26 epidemiological
in the United States during which depression was studies with data on rates of depressive disorder in
reported increased from 1.4 million to 3.2 million in children and adolescents between 1965 and 1999
less than a decade between 1995 and 2002 (Ma (Costello, Erkanli, & Angold, 2006). This analysis
et al., 2005). The debate about efficacy and safety of not only modelled effects of time, but also adjusted
certain antidepressant drugs in adolescents (FDA, for the substantial variations between studies in
2004; Thapar et al., 2012; Whittington et al., 2004) terms of diagnostic systems, assessment methods,
has not appeared to have impacted significantly on and sample characteristics such as age and gen-
these trends. Antidepressant use in young people der. The study found no evidence that rates of
has continued to increase substantially (Olfson depression had changed once methodological dif-
et al., 2014). Variations in clinical help-seeking ferences were accounted for. However, caution is
and treatment cannot by themselves be seen as also needed in interpreting these findings. The
accurate indicators of population prevalence, not wide variation in prevalence estimates from one
least because evidence shows that the majority of study to the next (prevalence estimates ranged
children and adolescents with psychiatric disorders, from <1% to >25%) may have made it difficult to
including anxiety and depression, do not access detect systematic secular trends in depression
medical or other public sector services for these prevalence.
illnesses (Ford, Hamilton, Meltzer, & Goodman, Cross-cohort comparisons in which epidemiolog-
2007; Potter et al., 2012). As with autism and ical studies at different time points have used
ADHD, trends in diagnosis and treatment of affec- identical screens for symptoms of anxiety and
tive disorders are likely in part to reflect differences depression are helpful as these studies allow a
in ascertainment, service provision, and variation in like-for-like comparison of unselected population
diagnostic thresholds for identifying and treating samples, albeit focusing on symptoms rather than
disorders. However, there is now accumulating disorder. Studies using this kind of design have
evidence that there has also been a real increase now been undertaken in several countries (see
in population symptom prevalence, though not all Tables 1–3). Many of these studies support the
evidence is consistent. view that adolescent emotional problems have
Initial indications that the population prevalence shown a long-term increase over the past 30 years,
of affective disorders might have increased came including in the United Kingdom, Greece, Iceland,
from several large epidemiological investigations of the Netherlands and Norway (Collishaw et al., 2004,
adults which compared lifetime rates of depression 2010; Fichter, Xepapadakos, Quadflieg, Georgop-
across different age groups. If there are no secular oulou, & Fthenakis, 2004; Sigfusdottir, Asgeirsdot-
changes in risk for depression then lifetime rates tir, Sigurdsson, & Gudjonsson, 2008; Sweeting,
would be expected to increase with age given Young, & West, 2009; Tick, van der Ende, &
individuals’ longer at-risk periods. However, the Verhulst, 2008; von Soest & Wichstrøm, 2014).
reverse has consistently been observed when differ- The majority of these studies also suggest that this
ent age cohorts are compared, with lifetime rates up increase has been greater for girls than for boys.
to four times higher in young adults compared to One exception is the study by W angby, Magnusson,
those aged 65 and over, and an increased preva- and Stattin (2005) finding no evidence of long-term
lence of (retrospectively reported) adolescent onset change in self-rated emotional problems in Sweden
depression (Burke, Burke, Rae, & Regier, 1991; between 1970 and 1996 respectively.
Kessler et al., 2005; Robins & Regier, 1991). Evi- In the United Kingdom, a number of independent
dence for this pattern has been observed in the cross-cohort comparisons of adolescent emotional
Epidemiological Catchment Area study, the problems have now been undertaken, allowing
National Comorbidity Survey and in other epidemi- clearer conclusions to be drawn. In one study,
ological studies of adults across the world (e.g. parent-reported symptoms of depressed mood, anx-
Cross-National Collaborative Group, 1992; Kessler iety and fearfulness among 15/16-year olds were
et al., 2005). However, there are a number of assessed across three national cohorts in the 1970s,
methodological artefacts that are difficult to 80s and 90s and showed a substantial increase
account for in retrospective cross-sectional compar- between 1986 and 1999 (Collishaw et al., 2004).
isons. For example, age is confounded with length Evidence from youth self-reports, including a com-
of recall period, older participants may be more parison of Scottish samples assessed in 1987, 1999
reluctant to report depression, and depression is and 2006 (Sweeting et al., 2009), and of English
associated with earlier mortality which may result teenagers in 1986 and 2006 (Collishaw et al., 2010)
in an underrepresentation of earlier-born individu- also points to a substantial increase in symptoms of
als suffering from depression (Cuijpers & Smit, anxiety/depression. Increases in emotional prob-
2002). A direct comparison of multiple epidemio- lems were not confined to those with low-level

© 2014 Association for Child and Adolescent Mental Health.

doi:10.1111/jcpp.12372 ARR: Secular trends in child and adolescent mental health 379

difficulties. In fact, secular increases were more resources to deal with mental health problems are
marked for those with more severe symptom levels. typically scarcer than in high-income countries.
These studies also undertook more detailed analyses Low- and middle-income countries also often expe-
of whether patterns of secular change varied by type rience more rapidly changing social conditions.
of affective symptoms (Collishaw et al., 2010; Sweet- Tracking trends in youth mental health in low-
ing et al., 2009). In one study, increases in problems income countries therefore constitutes an important
were evident across all symptoms assessed (Sweet- priority for future epidemiological research. Sugges-
ing et al., 2009). In the other, the increase was tive evidence for an increase in adolescent
specific to a subset of items (worry, irritability, emotional problems comes from two national
fatigue, sleep disturbance and feeling stressed; Col- population-based surveys of Vietnamese adoles-
lishaw et al., 2010). In terms of absolute effect sizes, cents which showed a high and increasing preva-
all three UK epidemiological studies point to non- lence of self-reported low mood over the past
trivial increases in prevalence of youth emotional decade, together with a marked increase in reports
problems. The proportion of parents reporting high of suicidal thoughts or self-harm (Le, Nguyen, Tran,
problem scores increased by 70% between 1986 and & Fisher, 2012).
1999 (Collishaw et al., 2004), the number of young In contrast to a sizeable body of evidence on
people meeting established GHQ ‘case criteria’ trends in adolescent depression, there is less evi-
almost doubled for Scottish boys and more than dence on long-term trends in emotional problems
doubled for Scottish girls between 1987 and 2006 among younger children. Parent-reported emotional
(Sweeting et al., 2009), and twice as many adoles- problems in preschool children in the Netherlands
cents endorsed five or more symptoms of anxiety or showed a reduction between 1989 and 2003 (Tick,
depression in 2006 compared to 1986 (15% vs. 7%; van der Ende, Koot, & Verhulst, 2007). A compar-
Collishaw et al., 2010). ison of 4-year olds in the 1993 and 2004 Pelotas
Epidemiological evidence from other countries birth cohorts in Brazil found no evidence of a
also indicates a rise in youth emotional problems. change in emotional problems, but an increase in
Cross-cohort comparisons from an increasing somatic complaints (Matijasevich et al., 2014). A
number of countries including Greece, Germany, German study which included both children and
Sweden, Iceland, Norway, China and New Zealand adolescents found evidence for recent increases in
have all shown increased self-reported adolescent adolescent, but not child emotional problems
symptoms of depression and anxiety from the 1980s (H€olling et al., 2014). In the United Kingdom, recent
onwards (Fichter et al., 2004; Fleming et al., 2014; evidence shows a reduction in younger children’s
H€olling et al., 2014; Lin & Wang, 2007; Kosidou emotional problems between 1999 and 2008 as
et al., 2010; Sigfusdottir et al., 2008; von Soest & rated by parents (Sellers et al., 2014). Very few
Wichstrøm, 2014), as well as increased sleep-onset studies of younger children have included children’s
difficulties (Pallesen et al., 2008). Tick, van der own rating of emotional symptoms. One exception is
Ende, & Verhulst (2007) also report long-term a Finnish study of 8-year olds comparing cohorts
increases in parent-reported emotional problems in assessed in 1989, 1999 and 2005 (Sourander et al.,
the Netherlands. Evidence of recent trends is rela- 2008). This study included children’s own ratings of
tively sparse. von Soest & Wichstrøm compared emotional difficulties, as well as parent and teacher
rates of youth-reported symptoms of depression in reports. Findings suggested an increase in girls’
Norway. There was a substantial increase in both self-reported depression between 1989 and 2005
male and female adolescent depression high symp- when assessed using the Children’s Depression
tom scores during the 1990s, but no significant Inventory, but no changes in parent- or teacher-
change between 2002 and 2010. Sourander et al. reported emotional difficulties (Sourander et al.,
(2012) found no change in adolescent emotional 2008). The same study also showed increases in
symptoms assessed using the SDQ in Finland child-rated sleep problems, fatigue, and somatic
between 1998 and 2008, although other Finnish complaints (Luntamo, Sourander, Santalahti, Aro-
evidence using the BDI suggests the possibility of an maa, & Helenius, 2012). These findings raise the
increase in the proportion of adolescents with mod- possibility that child-based assessments may in fact
erate or severe depressive symptoms between 2000 reveal trends over time in emotional difficulties even
and 2011, especially in more deprived subgroups when they are not apparent according to adult
(Torikka et al., 2014). informants.
Evidence from low- and middle-income countries Nevertheless, the possibility that trends in emo-
is largely lacking, but constitutes an important tional problems may be less marked in younger
research priority. There is considerable cross-cul- children relative to adolescents would perhaps not
tural variation in prevalence of child and adolescent be surprising given evidence for different aetiological
mental health problems (Crijnen, Achenbach, & features being associated with pre- and postpubertal
Verhulst, 1997), and the burden of psychiatric depression (Maughan, Collishaw, & Stringaris, 2013).
problems is often especially high in developing However, direct evidence awaits developmentally
countries (Kieling et al., 2011). Furthermore, informed approaches to cross-cohort comparisons

© 2014 Association for Child and Adolescent Mental Health.

380 Stephan Collishaw J Child Psychol Psychiatry 2015; 56(3): 370–93

in which not only cross-sectional symptom levels hospital (Hawton, Saunders, & O’Connor, 2012).
but also developmental trajectories in psychopa- The presentation of self-harm/attempted suicide in
thology are compared across cohorts. This remains hospitals in some high-income countries has
difficult given that it requires the comparison of increased over time (Hawton et al., 2003, 2012),
multiple longitudinal cohorts tracked from but epidemiological evidence about trends in self-
childhood onwards using equivalent assessment harm is lacking.
methods. Completed suicides are less common but repre-
An important question is whether longer term sent a leading cause of death globally in this age
change in adolescent symptomatology has been group, especially among young males (Hawton et al.,
sustained in the 21st Century. In the UK, compar- 2012). Annual data on suicide rates are available for
isons of the 1999 and 2004 British Child and many countries – though typically only for broad age
Adolescent Mental Health Surveys suggested that bands. Suicide rates among young males increased
emotional symptoms levelled off or began to reverse substantially between the 1960s and 1990s in many
(Maughan, Collishaw, Meltzer, & Goodman, 2008), countries across the world, but then showed a
and there was also no change in rates of anxiety or marked decline between the 1990s and mid-to-late
depressive disorders across these samples (Green 2000s (Hawton et al., 2012; McKeown, Cuffe, &
et al., 2005). As noted, a more recent study focusing Schulz, 2006; Mittendorfer-Rutz & Wasserman,
on younger children indicated a further reduction in 2004; V€ arnik et al., 2009). More recent data have
symptom levels up to 2008 in the United Kingdom shown a renewed upturn in young adult male suicide
(Sellers et al., 2014). An important issue concerns as the global economic crisis began in 2008 (Barr
the impact of the world-wide recession in 2008 on et al., 2012; Chang, Stuckler, Yip, & Gunnell, 2013),
trends in child and adolescent mental health. One and this was especially marked in Europe and North
recent study examined two very large Canadian and America (Chang et al., 2013). This is consistent with
US cohorts with data collected on multiple occa- similar upturns in suicide rates in countries affected
sions between 1997 and 2010 (Keyes et al., 2014). by previous recessions (Stuckler, Basu, Suhrcke,
This study is also one of very few to examine age, Coutts, & McKee, 2009).
period and cohort effects simultaneously. The study There are some difficulties in interpreting long-
focused primarily on symptoms of adult mental term trends in suicide. There is a possibility of
distress, but did also include adolescents in one of underrecording due to misclassification of some
the two cohorts. Findings provide evidence for suicides as accidental deaths. Misclassification rates
variations in distress by age, by period and by themselves may change over time, as they are
cohort. First, rates of psychological distress were affected by the quality of information available, and
highest in late adolescence and showed an addi- by factors such as social stigma that might influence
tional peak in symptoms in the late 40s. Second, decisions about recording a verdict of suicide (Haw-
the study indicated steadily increasing rates of ton et al., 2012). There is evidence that underre-
distress for cohorts born from the 1930s to the cording might have been greater in the past, and that
1990s. Finally, the study indicated a decline in the increase in suicides in the last Century might be
distress during the period 1997–2007, followed by over-estimated (Mohler & Earls, 2001). It is not clear,
an increase in distress between 2008 and 2010. however, how such biases would account for the
Similarly, annually collected Finnish data up to observed pattern which saw a rise and then a fall in
2011 also indicates a possible upturn in reports of recorded suicides.
depression in the final years of the series (Torikka Several explanations have been proposed for
et al., 2014). As discussed below, there is also now changing rates of self-harm and suicide. In addition
strong evidence that the global financial crisis to trends in mental ill health and socioeconomic
resulted in a significant up-turn in suicides among conditions, evidence also suggests that efforts to
the adult population (Barr, Taylor-Robinson, Scott- reduce access to and lethality of common methods of
Samuel, McKee, & Stuckler, 2012). In the light of suicide have been effective public health measures
these sets of findings, an important priority is to and have helped reduce the incidence of suicide, for
collect further comparable epidemiological data to example changes in pack sizes of drugs such as
track trends in child and adolescent mental health paracetamol, detoxification of gas supplies, or safer
postrecession. storage of guns and of agricultural pesticides (Gun-
nell, Middleton, & Frankel, 2000; Hawton et al.,
2004; Hawton et al., 2012).
Suicide and self-harm
Self-harm is common in adolescence. Community
Antisocial behaviour
studies suggest that around 10% report deliberate
self-injury or self-poisoning with or without suicidal Crime statistics, self-reports of offending and cross-
intent, and there is a higher prevalence of self- cohort comparisons of reports of conduct problems
harming behaviour in females than males. Of those in epidemiological surveys all point to substantial
who self-harm, around one in eight present to changes in rates of youth antisocial behaviour, with

© 2014 Association for Child and Adolescent Mental Health.

doi:10.1111/jcpp.12372 ARR: Secular trends in child and adolescent mental health 381

evidence both of substantial rises and falls over the Police-recorded crime and victim-reported crime
past 50–60 years. provide indexes of the total burden of criminal
activity rather than estimates for the numbers of
Offending. Crime rates since the middle of the last offenders in the population. It is well-established
century have followed broadly similar trends across than a small minority of offenders may account for a
many high-income nations. The most extensive large proportion of offences (Farrington & West,
information relates to trends in crime in the United 1993). Crime surveys also do not take account of
States. These show an overall 350% increase in co-offending (Reiss & Farrington, 1991), and provide
recorded crime between 1960 and 1990, with rising little if any information on the age and gender of
rates of violence, homicide and property crime over offenders. Even though there are well-documented
this period. The 1990s then saw a marked reduction peaks in offending in mid-to-late adolescence (Gott-
in all forms of crime (Federal Bureau of Investigation, fredson & Hirschi, 1990; Hirschi & Gottfredson,
2009; Van Dijk, van Kesteren, & Smit, 2008). In 1983), national crime rates do not therefore directly
most countries in North America, Europe, Asia and address whether and how the prevalence of youth
Oceania, crime rates including homicides peaked in offending has changed. Surveys assessing self-
the early/mid-1990s after prolonged periods of reported offending therefore in principle provide a
increase, and have since fallen substantially (Inter- valuable complement. Evidence is limited, but does
national Crime Victimization Survey, Van Dijk et al., suggest some important discrepancies which remain
2008; UN Economic and Social Council, 2014). unexplained. For example, data from the UK Offend-
However, these trends are not universal, and Central ing, Crime and Justice Survey indicated no evidence
and South America and some regions in Africa have of a fall in offending by adolescents and young adults
seen increasing rates of crime in the 21st Century during the late 1990s and early 2000s (Home Office,
(United Nations Economic & Social Council, 2014). 2008).
For example, a recent study estimated that homicide
rates in Brazil increased almost threefold between Conduct problems. Recent years have also seen
1980 and 2003, and remained substantially higher increased concern about youth antisocial behaviour,
in 2010 than in the 1990s (Murray, de Castro whether or not this is categorized as criminal
Cerqueira, & Kahn, 2013). Victim reports of nonle- offending. Epidemiological data indicate that adoles-
thal violent crime, such as physical aggression, cent conduct problems (e.g. lying, disobedience,
robbery and theft also showed substantial increases aggression, bullying) are far more prevalent than
in Brazil between 1988 and 2009 (Murray et al., criminal offending, but also reach a peak prevalence
2013). during adolescence (Moffitt, 1993).
Police registers, victim surveys and self-reports As shown in the accompanying Tables, a number
provide complementary information on trends in of cross-cohort comparisons have included mea-
offending. It is important to note that the level of sures of conduct problems, for example using the
credibility of register-based information may vary CBCL or SDQ. Collishaw et al. (2004) compared the
across countries, and police-recorded crime statis- prevalence of parent-reported conduct problems in
tics should be treated with caution, and not consid- three national UK cohorts of 15/16 year olds
ered in isolation. Many victims do not report a crime assessed in 1974, 1986 and 1999. There was a
to the police, and offences recorded by police are substantial increase in adolescent conduct prob-
typically only a subset of those reported. National lems, particularly nonaggressive behaviours (lying,
and local policy drivers can also influence the quality stealing, disobedience), over this 25-year period,
and quantity of crime data recorded by police (UK with similar increases for boys and girls, and for
Statistics Authority, 2014). Surveys of victim- young people from socially advantaged and disad-
reported crime and of self-reported offending are vantaged homes. Longitudinal analyses comparing
often seen as more accurate indicators of crime long-term outcomes for adolescents with conduct
trends. As expected, victim surveys suggest higher problems in 1974 and 1986 found similarly elevated
overall levels of crime than do police-recorded fig- odds of adult unemployment, homelessness, poor
ures, but there is close correspondence in the physical and mental health, problem drinking, and
direction of trends for both police-recorded and divorce in both cohorts. Pervasive adult psychosocial
victim-reported crime in countries where both dysfunction (four or more such problems) was
sources of information are available (Van Dijk, between three and four times more likely in both
Tseloni, & Farrell, 2012). For example, in the United cohorts for those with adolescent conduct problems
Kingdom, both sources indicate that rates of crime (Collishaw et al., 2004). The fact that conduct prob-
doubled between 1981 and 1995, and then showed a lems were as predictive of adult problems in both
fall up to 2013. Latest figures from the Crime Survey cohorts is important because it provides evidence
for England and Wales indicate a fall in total number that trends over time do not merely reflect changes in
of victim-reported offences to the lowest level since parental reporting. More recent evidence from the
the survey first began in 1981 (Office for National United Kingdom indicates a levelling off in rates of
Statistics, 2014). adolescent conduct problems since 1999 (Maughan

© 2014 Association for Child and Adolescent Mental Health.

382 Stephan Collishaw J Child Psychol Psychiatry 2015; 56(3): 370–93

et al., 2008), though rates in the middle of the last States than in European countries. Emerging evi-
decade remained considerably elevated compared dence from other countries, such as Brazil, indi-
with earlier cohorts. cates that rates of violent crime have not shown
Studies in the Netherlands have found evidence the same recent decline, and perhaps have contin-
for small increases in parent-rated but not self- ued to rise. This highlights how important it is to
rated conduct problems in adolescents in the bear in mind that there are marked international
Netherlands since the early 1990s (Tick, van der variations in rates and trends in crime, and other
Ende, & Verhulst, 2007, 2008). In the United forms of antisocial behaviour. As discussed, the
States, evidence has shown increases in parent-, evidence also highlights that there may be swift
teacher- and self-rated conduct problems between changes in rates of antisocial behaviour in partic-
the 1970s and 1980s, followed by a fall in the ular societies. More systematic tracking of antiso-
1990s (Achenbach, Dumenci, & Rescorla, 2002b, cial behaviour beyond police-recorded crime thus
2003; Achenbach et al., 2002a). In Sweden, there is remains a priority. This should include continued
evidence for an increase in self-reported offending evidence from crime victimization surveys, as well
and conduct problems in adolescent cohorts as epidemiological surveys of self-reported offend-
between 1970 and 1996 (W angby et al., 2005), ing or parent- and teacher-reported conduct prob-
followed by a decline in antisocial behaviour since lems.
the 1990s consistent with national trends in regis-
tered crime (Backman, Estrada, Nilsson, & Shan-
non, 2014; Estrada, Petterson, & Shannon, 2012). Changes in distribution of child and adolescent
Evidence for an increase in behaviour problems, at mental health problems
least for girls, has also been reported in China Beyond changes in overall population prevalence of
between 1993 and 2003 (Lin & Wang, 2007). mental health problems, it has been suggested that
However, other studies of adolescents in Finland the distribution of problems within the population
(Sourander et al., 2012), Germany (H€ olling et al., may have changed in important ways. Two ques-
2014), Norway (von Soest & Wichstrøm, 2014) and tions are relevant here. First, has there been a
New Zealand (Fleming et al., 2014) have found no convergence of gender differences? Some evidence is
evidence for an increase in antisocial behaviour in consistent with this possibility (e.g. Sellers et al.,
the 1990s and 2000s, with some evidence of a 2014; von Soest & Wichstrøm, 2014), but other
decline in conduct problems since 2000 (von Soest studies have found similar trends for males and
& Wichstrøm, 2014). females (e.g. Collishaw et al., 2004; Matijasevich
Relatively few studies have focused on younger et al., 2014), while others still point to a divergence
children, but available European evidence does not of gender differences, for example more marked
indicate any major long-term increase in child increases in adolescent emotional problems in girls
conduct problems (Sourander et al., 2008; Tick, than in boys (e.g. Sweeting et al., 2009). Second, is
van der Ende, Koot, & Verhulst, 2007). These there evidence for a polarization of mental health
studies have suggested stable or declining rates of problems in society? Most studies have not tested
behaviour problems in prepubertal children. Recent for changes in population variance, but studies that
evidence from the United Kingdom also suggests a have examined categorically defined high scores
more recent decline in behavioural problems alongside mean scale score have typically provided
between 1999 and 2008 that is more pronounced consistent results for both (see Tables 1–3). As
in boys than in girls (Sellers et al., 2014). discussed below, there is some evidence for increas-
A Brazilian study reminds us that trends in child ing social inequalities, particularly in relation to
behaviour problems may well differ across different emotional problems (Gore Langton, Collishaw,
world regions. In keeping with intercountry differ- Goodman, Pickles, & Maughan, 2011).
ences in crime trends, Matijasevich et al. (2014)
found evidence for a substantial increase in prob-
lem behaviours, especially aggression, in preschool Real changes in prevalence or changes in
children. reporting?
In summary, police-recorded and victim-reported Even when representative population samples are
crime, as well as self- and parent-reported conduct compared using identical symptom screens it is
problems in epidemiological studies highlight a necessary to consider whether observed trends
general pattern of rising rates of youth antisocial represent real changes in mental health, or instead
behaviour during the 1970s, 1980s and early a shift in how symptoms are perceived and reported
1990s. Rates of antisocial behaviour appear to by informants. For example, long-term increases in
have levelled off or fallen since the 1990s in many adolescent emotional problems and antisocial
but not all countries. Broadly speaking, these behaviour might reflect greater openness by infor-
trends have been similar across many high-income mants in more recent cohorts about reporting
countries, with the exception that a peak in rates problems, a shift in thresholds for rating symptoms
was observed somewhat earlier in the United and behaviours as problematic, or greater sensitiv-

© 2014 Association for Child and Adolescent Mental Health.

doi:10.1111/jcpp.12372 ARR: Secular trends in child and adolescent mental health 383

ity to children’s problems due perhaps to societal

Explaining trends in child and adolescent
medicalization of symptoms previously seen as
mental health
normal.
Recent decades have seen considerable changes
Several studies demonstrate that prevalence
affecting children’s environments as they grow up,
estimates can be sensitive to measurement and
and a variety of explanations for trends in mental
reporting effects. Studies of cross-cultural variations
health have been proposed (Eckersley, 2006; Fo-
in psychopathology suggest that intercountry com-
mbonne, 1998; Hagell, 2012; Rutter & Smith, 1995).
parisons of prevalence of psychopathology using
These can broadly be classified in terms of changes
symptom screens can diverge in important ways
in (a) individual vulnerability; (b) family life; (c)
from comparisons according to diagnostic interviews
extrafamilial influences on risk and (d) broader
(Goodman et al., 2012; Heiervang, Goodman, &
cultural change.
Goodman, 2008). Whether the same is true for
When considering each of these possibilities, it is
cross-cohort comparisons is unclear, but the possi-
important to recognize that testing reasons for
bility cannot be excluded. Studies also highlight that
secular changes in mental health is inherently
prevalence estimates and between-group compari-
difficult. Society has changed in many ways, and it
sons of psychopathology using questionnaire
is impossible to draw simple conclusions based on
screens can be very sensitive to minor variations in
two correlated trends at the population level. Con-
wording of items (Goodman, Iervolino, Collishaw,
siderable caution is needed in inferring explanations
Pickles, & Maughan, 2007) and to differences across
about differences between populations on the basis
informants (Collishaw, Goodman, Ford, Rabe-Hesk-
of correlations between individuals within each
eth, & Pickles, 2009), further highlighting the impor-
population (Robinson, 1950). Finally, very few
tance of being cautious about reporting effects.
cross-cohort comparisons have included like-for-like
Still, there are several reasons for thinking that
data on mental health and relevant hypothesized
time trends in symptom reports do in fact reflect
explanatory factors.
real changes in population prevalence of mental
health problems. The first is that there is a
convergence of evidence across multiple study Change in individual vulnerability
methods and multiple informants. For example,
One possibility is that changes in perinatal condi-
police-recorded and victim-reported crime, as well
tions have influenced patterns of individual vulner-
as parent-reported conduct problems, have fol-
abilities across successive generations of children.
lowed broadly similar trends. Second, it seems
For example, epidemiological evidence suggests risk
unlikely that there has been a general shift in
for neurodevelopmental disorders such as autism
openness about reporting symptoms, because there
and ADHD may be associated with increased survival
is considerable specificity in observed trends. For
of very premature or extreme low birth weight babies,
example, cross-cohort comparisons have demon-
in utero exposure to antidepressant medication, and
strated marked variations in rates of adolescent
advancing parental age (Frans et al., 2013; Lampi
emotional and conduct problems, but much less so
et al., 2012; Raj et al., 2013). However, several points
in symptoms of hyperactivity. This demonstrates
need to be considered. First, epidemiological evidence
that informants have not simply become more open
alone is typically insufficient to test the causal nature
about reporting difficult behaviour, or that parents
of hypothesized risk factors, and the causal role of
have become more sensitive to troublesome behav-
many early risk factors remains uncertain (Thapar &
iour in their children per se. The strongest evi-
Rutter, 2009). Second, as noted by Raj et al. (2013)
dence for real changes in mental health comes
even if one assumes causality, changes in relatively
from studies that provide independent validation of
rare risk factors such as antidepressant use in
informants’ reports and that directly compare the
pregnancy are unlikely to have any major effect on
validity of reports over time. Such studies are
the overall population prevalence of common mental
exceedingly rare, but are in principle possible. As
health problems. Third, there have also been notable
discussed above, parent-reported conduct prob-
improvements in early life conditions; for example,
lems increased substantially over time, but equally
decreased exposure to maternal smoking in preg-
strongly predicted a broad range of independently
nancy (Ekblad, Gissler, Korkeila, & Lehtonen, 2014).
assessed adult difficulties in successive longitudi-
Finally, as we have seen, population-level evidence
nal cohorts in the United Kingdom (Collishaw
for trends in childhood neurodevelopmental prob-
et al., 2004). Future research might also consider
lems remains weak. In contrast, there is much more
drawing on methodologies developed in the social
convincing evidence for trends in adolescent mental
sciences. One approach has been to develop direct
health. It is therefore also important to consider
measures of incomparability in reporting using
secular changes affecting individual vulnerabilities
‘anchoring vignettes’ and to take account of this
later in development.
in statistical comparisons (King, Murray, Salomon,
Secular trends in the timing of puberty may be
& Tandon, 2004; Salomon, Tandon, & Murray,
important here. Adolescence is associated with
2004).

© 2014 Association for Child and Adolescent Mental Health.

384 Stephan Collishaw J Child Psychol Psychiatry 2015; 56(3): 370–93

heightened risk for the onset of range of mental Although there are well-documented associations
health problems such as depression (Thapar et al., with child psychopathology, these in large part
2012). Evidence also suggests that early pubertal reflect the influence of more proximal family pro-
timing is associated with increased risks for mental cesses and living conditions that both precede and
health and psychosocial problems that persist follow parental separation (Rutter, 2013). It has
across adolescence and into adulthood (Golub et al., also been suggested that associations of divorce
2008). There was a marked decrease in age of with child mental health have reduced over time as
puberty up to the middle of the 20th Century, but parental divorce has become more common; possi-
recent evidence is more complex, suggesting a level- bly because of changes in the characteristics of
ling of trends in the timing of pubertal markers such parents who become separated, their reasons for
as age at menarche, followed by renewed decreases separation, improved awareness about potential
in timing over the past 25 years (Sorensen et al., impacts on children, or reduced social stigma
2012). However, the timing of hormonal changes associated with divorce (Amato & Keith, 1991; but
associated with puberty does not appear to have see also Amato, 2001). Evidence suggests that
changed over recent years (Sorensen et al., 2012). trends in adolescent conduct and emotional prob-
There is a lack of direct evidence testing the extent to lems (at least in the United Kingdom) are equally
which changes in pubertal maturation account for evident for single-parent as for two-parent families.
secular changes in mental health. Cross-cohort This suggests that changes in family composition
comparisons including measures of mental health are not a primary driver of mental health trends
and pubertal timing are needed. (Collishaw et al., 2004).
Finally, given evidence of co-occurrence between Understanding of how children’s proximal family
many forms of psychopathology, trends in adoles- environments have changed over time is relatively
cent psychopathology may be interlinked. Trends in limited, but studies are beginning to provide useful
affective symptoms might to some extent be attrib- information about changes in parental psychopa-
utable to changes in behavioural adjustment or thology, parent–child relationships and child mal-
vice versa (Collishaw et al., 2004). More generally, treatment experiences. One of the best established
it is possible that trends in related difficulties such risk factors for child and adolescent psychopathol-
as sleep problems or eating problems have contrib- ogy is having a parent who suffers with depression
uted to trends in adolescent depression (Pallesen (Thapar et al., 2012). Rates of adult depression have
et al., 2008; von Soest & Wichstrøm, 2014). Indeed, also risen considerably over time (Olfson et al.,
a recent meta-analytic review has highlighted a 2014; Ferri, Bynner, & Wadsworth, 2003), and this
substantial decline in sleep time among school- might have contributed to trends in adolescent
aged children and adolescents over the past emotional problems. A recent study in the United
100 years (Matricciani, Olds, & Petkov, 2012). Kingdom included data on maternal and adolescent
Meta-analyses showing substantial changes in per- symptoms of depression (Schepman et al., 2011).
sonality inventory scores across successive gener- Findings showed substantial increases in youth and
ations also indicate a possible contribution of maternal depression symptoms between 1986 and
personality traits to trends in mental health 2006. Cross-cohort differences in adolescent depres-
(Twenge, 2011). Further methodologically rigorous sion were attenuated, but remained significant when
studies are required to address each of these controlling for trends in maternal depression.
possibilities. The use of developmentally sensitive Studies have also examined possible changes in
data across multiple longitudinal cohorts would be parent–child relationships, with conflicting findings.
especially helpful. First, a cross-cohort comparison of epidemiological
samples of Scottish teenagers found that adoles-
cents reported increased worries about family life,
Changes in family life
and more frequent arguments with parents. Multi-
It has been suggested that substantial increases in variate models suggested a possible contribution to
nonmarital cohabitation and parental separation increased levels of adolescent depression (Sweeting,
might explain increases in youth psychosocial West, Young, & Der, 2010). However, another study
problems over the 20th Century (see Rutter & focusing on different youth-reported indicators of
Smith, 1995 for a discussion). However, drawing parenting and parent–child relationship quality
conclusions about the implications of correlated found no evidence for a ‘decline in parenting’. Levels
trends is particularly problematic when these are of youth-reported parental interest did not differ
distal risk markers whose ‘meaning’ has changed between 1986 and 2006, while reported quality time
over time. Nonmarital cohabitations are now much spent with parents, parental expectations of good
more common, but today often show little differ- behaviour and monitoring of out-of-home activities
ences in terms of child outcomes when compared to all increased over time. Furthermore, analyses sug-
marriages (Rutter & Solantaus, 2014). Similarly, gested that each of these parenting indicators
the relationships between parental separation was associated with lower levels of antisocial
and child mental health are not straightforward. behaviour in both cohorts, and that changes in

© 2014 Association for Child and Adolescent Mental Health.

doi:10.1111/jcpp.12372 ARR: Secular trends in child and adolescent mental health 385

parenting may have limited the observed increase in for young people (West & Sweeting, 2003). Worries
parent-reported youth conduct problems over this about school performance and school disengage-
time period (Collishaw, Gardner, Maughan, Scott, & ment increased over time in Scotland, and these
Pickles, 2012). factors may have contributed to cross-cohort differ-
Apart from normative changes in parenting, some ences in psychological distress, especially among
countries have also seen apparent changes in rates girls (Sweeting et al., 2010; West & Sweeting, 2003).
of child maltreatment, though accurate monitoring More detailed examination of temporal variation
of maltreatment rates is especially challenging. In within successive cohorts also showed that proxim-
the United States, both agency-recorded and ado- ity of data collection to school exams was associated
lescent self-reported rates of maltreatment have with greater differences in depressive symptoms
reduced since the early 1990s in the United States between earlier and later cohorts (West & Sweeting,
(Finkelhor, Turner, Ormrod, & Hamby, 2010), a 2003).
period which has also seen improvements in chil-
dren’s emotional and behavioural adjustment
Broader socioeconomic and cultural influences
(Achenbach et al., 2003). Maltreatment is an estab-
lished causal risk factor for child psychopathology Finally, broader socioeconomic and cultural factors
(Jaffee, Strait, & Odgers, 2012), and so a reduction are relevant for understanding mental health trends.
in maltreatment would seem likely to have contrib- There are vast global inequalities affecting children’s
uted to US trends in child psychopathology over this health, development and education (Ortiz, Daniels,
period. & Engilbertsd ottir, 2012), and the reduction in
Finally, many low- and middle-income countries health inequalities is also an important policy prior-
face different and profound challenges affecting ity within high-income countries (Marmot, 2010;
family life; for example, an estimated 17 million Wilkinson & Pickett, 2010). Many studies have
children have lost one or both parents to the AIDS established strong links between poverty and child
epidemic (UNICEF, 2013), and risks of this kind and adolescent mental health, and the mechanisms
are strongly associated with risk for psychopathol- that mediate effects of poverty on child mental health
ogy in children and young people (e.g. Cluver, are now better understood (Bradley & Corwyn, 2002;
Gardner, & Operario, 2007). Understanding the Costello, Compton, Keeler, & Angold, 2003; Lund
impact of changing health and social conditions on et al., 2010; Yoshikawa, Aber, & Beardslee, 2012).
families and on children’s mental health in low- Some cross-cohort comparisons suggest that mental
income and middle-income countries is an urgent health inequalities have become more marked over
priority. time, particularly with respect to adolescent emo-
tional problems (Gore Langton et al., 2011; Torikka
et al., 2014). For example, Gore Langton et al. (2011)
Changes in extrafamilial psychosocial influences on
found that the disparity in rates of emotional prob-
mental health
lems between low- and middle/high-income families
Two examples highlight that extrafamilial psychoso- increased fourfold over the last 25 years of the 20th
cial influences are also likely involved in explaining Century. Reasons behind emerging inequalities in
youth mental health trends. adolescent emotional problems are likely to include a
Bullying is a well-established risk factor for greater impact of social adversity on proximal family
mental health problems in children and adolescents factors including maternal distress, exposure to
(Arsenault, Bowes, & Shakoor, 2010). The majority stressful life events and family discord (Gore Langton
of studies that have examined long-term trends et al., 2011). In addition to social inequality, it has
point to a decline in peer victimization (Finklehor, been suggested that increasingly consumerist and
2013; Molcho et al., 2009; Rigby & Smith, 2011), individualistic values and attitudes have contributed
but caution is needed in drawing firm conclusions, to increasing levels of psychological distress among
given evidence of variation in trends by country young people (Eckersley, 2006), though these possi-
(Molcho et al., 2009), by informant, and by child bilities have been difficult to test empirically with the
gender (Ilola & Sourander, 2013). Most studies data currently available.
point to a reduction in rates of bullying, but there Finally, increased professional recognition of child
are exceptions. These include a Finnish study mental health problems, as well as developments in
which found a rise in teacher-rated bullying/vic- the areas of early intervention and prevention, may
timization among girls (but not boys) aged 8 years have resulted in better outcomes for children than
between 1989 and 2005 (Ilola & Sourander, 2013), would otherwise have been seen. However, while it
evidence of unchanged or increasing rates in the is true that more young people are in contact with
United Kingdom (Molcho et al., 2009), and a sub- mental health services than in the past, many
stantial rise in reported incidents of cyber-bullying young people with psychiatric problems still do not
(Jones, Mitchell, & Finklehor, 2012). receive professional help. Furthermore, recent aus-
Other research has examined the possibility that terity measures in many countries are placing
educational experiences have become more stressful considerable strains on social, health and mental

© 2014 Association for Child and Adolescent Mental Health.

386 Stephan Collishaw J Child Psychol Psychiatry 2015; 56(3): 370–93

health services for young people; it will take time to development that are related in important ways to
evaluate the impact of this on trends in mental children’s mental health. These include well-docu-
health. mented secular gains in intelligence (the ‘Flynn
effect’; Dickens & Flynn, 2001; Williams, 2013),
changes in pubertal timing (Sorensen et al., 2012),
Explaining mental health trends
and changes in physical health (Frederick, Snell-
It is clear that multifactorial models are needed to man, & Putnam, 2014).
account for mental health trends. Psychiatric dis- Third, it is now clear that different explanatory
eases represent complex phenotypes with multifac- frameworks are sometimes needed to account for
torial underpinnings. Trends in mental health are individual differences within populations and for
likely attributable to a variety of countervailing differences between populations. One example con-
social trends that have impacted on children’s risk cerns secular changes in height. Individual differ-
for mental health problems in diverse ways, some ences in height are highly heritable (Silventoinen et
for the better and others for the worse. There are a al., 2003), but the average height of the population
number of priorities for future research. First, there increased substantially over the 19th and 20th
is a need for high-quality data to address basic centuries, most probably due to better nutrition
questions about whether the prevalence of well- and improved health care early in life (Fredriks
established causal risk factors has changed. Sec- et al., 2000). Different types of explanation may
ond, linked data on hypothesized explanatory fac- also be needed to understand population trends in
tors and youth mental health across successive mental health. For example, progress has been
cohorts are required in order to assess changes in made in understanding biological and psychosocial
prevalence and ‘impact’ of risk factors, and to be mechanisms underlying cigarette addiction and
able to use statistical models to estimate the degree suicide. However, cultural norms and legislative
to which trends in mental health might be attribut- changes are probably more relevant for understand-
able to changes in risk factors. Most studies to date ing the substantial changes in smoking prevalence
have focused on broad-based sociodemographic risk seen over recent decades, while population-wide
markers (e.g. social class, family type), rather than efforts to reduce access to commonly used methods
on proximal mechanisms (e.g. quality of parent– of suicide have been an important factor in reduc-
child relationships). This is because few existing ing suicide rates. Cross-cohort comparisons have
cross-cohort comparisons of child mental health also been important in highlighting potential
also include comparable information on relevant broader cultural influences on mental health (Ec-
explanatory factors. Measurement harmonization kersley, 2006; Wilkinson & Pickett, 2010), and for
with a view to enabling cross-cohort comparisons developing theoretical models of secular change, for
is an urgent priority for future data collection in example social multiplier effects to account for
population cohorts, along with dedicated replica- increases in intelligence test scores (Dickens &
tions of earlier cohorts encompassing measures of Flynn, 2001).
mental health and explanatory factors of interest. Fourth, cross-cohort comparisons of populations
Finally, progress in understanding trends in psy- affected and unaffected by a universal change in
chopathology requires a better handle on testing risk can overcome the problem of selection effects,
causal explanations. This is a particular challenge and therefore sometimes provide conditions approx-
for time trends studies. imating a ‘natural experiment’ (Costello et al., 2003;
Shadish, Cook, & Campbell, 2002; Thapar & Rutter,
2015). Causal inferences are stronger where there
What can time trends studies tell us about the are multiple populations or subpopulations who
aetiology of child psychopathology? experienced a clearly defined change in an environ-
Studies of secular change provide an important mental condition at different times, or where there
complement to studies of within-population varia- otherwise equivalent affected and unaffected groups
tion, and can also provide valuable information (Shadish et al., 2002). Recent concerns about a link
about aetiological factors involved in child psycho- between the MMR vaccine and the secular increase
pathology. in autism diagnosis have now been dispelled using a
First, it is now clear that the prevalence, pattern- wide variety of scientific evidence (Rutter, 2005).
ing and ‘meaning’ of many risk factors has changed One study showed that when the MMR vaccine was
radically over time. An appreciation that the social discontinued in Japan in 1993 the rate of autistic
context of children’s lives is changing is also impor- spectrum disorder diagnoses did not reduce (Honda,
tant to identify new or emerging risks for child Simizu, & Rutter, 2005). In fact, the annual rate of
psychopathology, including for example develop- increase in the incidence of ASD diagnoses was as
ments in online and mobile technology (Livingstone great or greater than before the MMR vaccine was
& Smith, 2014). discontinued, and also greater than in many other
Second, cross-cohort comparisons are required to countries in which MMR continued to be used.
understand secular changes in other aspects of child A second example concerns possible links between

© 2014 Association for Child and Adolescent Mental Health.

doi:10.1111/jcpp.12372 ARR: Secular trends in child and adolescent mental health 387

economic factors such as unemployment and parisons of epidemiological samples undertaken in

suicide. These are typically hard to interpret high-income countries that underlying rates of men-
because individuals who commit suicide often have tal health problems have also changed. More such
a range of mental health problems which might also studies are now needed, particularly in low- and
impact on performance in the labour market. How- middle-income countries which often experience
ever, as highlighted above, there is now well-repli- particularly rapid social change. Where they have
cated evidence that suicide rates increase at times of been undertaken, cross-cohort comparisons have
economic recession, and one recent study showed not only helped address whether the prevalence of
that the impact of the 2008 recession on suicide mental health problems is changing, but have also
trends varied between geographical regions within provided novel insights on the role of environmental
the United Kingdom according to the extent to which influences on mental health. Reasons for population
unemployment rates increased in each region (Barr mental health trends are complex and multifaceted,
et al., 2012). This combination of temporal and but studies are beginning to evaluate potential
spatial data strengthens the interpretation of a explanations for trends in mental health. Evidence
causal effect of economic recession on suicide risk. of secular trends in mental health highlights that the
Finally, perhaps the clearest example of a cohort burden of child psychopathology may be amenable
effect on risk for mental illness comes from studies to change through public health interventions.
showing increased risk for schizophrenia in popu- Indeed, one important conclusion is that changes
lations exposed to famine in utero (St Clair et al., in policy or legislation can often lead to substantial
2005; Susser, Hoek, & Brown, 1998). Findings of health benefits in the population, even when indi-
these studies are convincing because exposure to vidual-level aetiological pathways remain poorly
famine was brief but severe, and so otherwise understood.
equivalent cohorts of individuals who were or were
not exposed during gestation could be compared.
Typically, however, population-level changes in Supporting information
environmental conditions do not occur suddenly, Additional Supporting Information may be found in the
and there may often be a long lag between the online version of this article:
environmental exposure and health outcome of Appendix S1. Search strategy.
interest. It is important to consider possible con-
founding factors that may differentiate those affected
and unaffected by population-level environmental Acknowledgements
change. It is also clear that individuals differ in how S.C. is supported by the Waterloo Foundation.
they respond to environmental stress. For all these Research on trends in child and adolescent mental
reasons, robust conclusions about causal risk fac- health is supported by a grant from the Medical
Research Council (MR/J01348X/1). The extremely
tors require convergent data using a variety of
helpful comments and suggestions by reviewers of an
different study designs (Thapar & Rutter, 2015).
earlier version of this manuscript are gratefully
acknowledged. This review article was invited by the
journal, for which the author has been offered a small
Conclusions honorarium towards expenses; the article has under-
gone full, external peer review. The author has declared
An accurate understanding of trends in child and
that he has no competing or potential conflicts of
adolescent psychopathology is essential for planning
interest in relation to this article.
population health priorities. It is also important for
gauging our success in addressing the burden of
childhood psychiatric illness. Findings on trends in
Correspondence
child and adolescent mental health are complex. It is Stephan Collishaw, Child & Adolescent Psychiatry
clear that marked increases in help-seeking and Section, Institute of Psychological Medicine and Clinical
treatment for many disorders likely reflect changes Neurosciences, MRC Centre for Neuropsychiatric
in diagnostic thresholds, and better public and Genetics and Genomics, Cardiff University, Hadyn Ellis
professional awareness. In addition, however, there Building, Maindy Road, Cardiff CF24 4HQ, UK; Email:
is now convincing evidence from cross-cohort com- [email protected]

© 2014 Association for Child and Adolescent Mental Health.

388 Stephan Collishaw J Child Psychol Psychiatry 2015; 56(3): 370–93

Key points
• Substantial increases in diagnosis and treatment of child psychiatric disorders in clinical practice do not
necessarily reflect changes in population prevalence. Like-for-like comparisons using unselected cohorts are
required to test mental health trends.
• Epidemiological evidence of this kind does not indicate any major change in prevalence of neurodevelop-
mental disorders, though evidence with respect to autism spectrum disorders remains very sparse.
• There are important long-term trends affecting adolescent emotional problems and antisocial behaviour, and
suicide in adolescents and young adults.
• Cross-cohort comparisons of child and adolescent mental health are of particular relevance in low- and middle-
income countries given the often rapid pace of social change, but few such studies have been undertaken.
• Testing explanations for trends in mental health is challenging, and mechanisms underlying secular change
remain poorly understood.
• Studies of secular change complement studies of individual differences within populations, but involve
additional methodological challenges.

in childhood differ from onset in adolescence? Journal of the

References American Academy of Child and Adolescent Psychiatry, 43,
Achenbach, T.M., Dumenci, L., & Rescorla, L.A. (2002a). Ten- 63–70.
year comparisons of problems and competencies for Boyle, C.A., Boulet, S., Schieve, L.A., Cohen, R.A., Blumberg,
national samples of youth: Self, parent and teacher S.J., Yeargin-Allsopp, M., . . . & Kogan, M.D. (2011). Trends
reports. Journal of Emotional and Behavioral Disorders, 10, in the prevalence of developmental disabilities in US
194–203. children, 1997–2008. Pediatrics, 127, 1034–1042.
Achenbach, T.M., Dumenci, L., & Rescorla, L.A. (2002b). Is Bradley, R.H., & Corwyn, R.F. (2002). Socioeconomic status
American student behaviour getting worse? Teacher-ratings and child development. Annual Review of Psychology, 53,
over an 18-year period. School Psychology Review, 31, 428– 371–399.
442. Burke, K.C., Burke, J.D., Rae, D.A., & Regier, D.A. (1991).
Achenbach, T.M., Dumenci, L., & Rescorla, L.A. (2003). Are Comparing age at onset of major depression in five US
American children’s problems still getting worse? A 23-year community populations. Archives of General Psychiatry, 48,
comparison. Journal of Abnormal Child Psychology, 31, 1– 789–795.
11. Centres for Disease Control and Prevention (2012). Prevalence
Amato, P.R. (2001). Children of divorce in the 1990s: An of autism spectrum disorders – autism and developmental
update of the Amato and Keith (1991) meta-analysis. disabilities monitoring network, 14 sites, United States,
Journal of Family Psychology, 15, 355–370. 2008. Morbidity and Mortality Weekly Report, 61, 1–19.
Amato, P.R., & Keith, B. (1991). Parental divorce and the well- Chang, S., Stuckler, D., Yip, P., & Gunnell, D. (2013). Impact of
being of children – a meta-analysis. Psychological Bulletin, 2008 global economic crisis on suicide: time trend study in
110, 26–46. 54 countries. British Medical Journal, 347, f5239.
Arsenault, L., Bowes, L., & Shakoor, S. (2010). Bullying Cluver, L., Gardner, F., & Operario, D. (2007). Psychological
victimization in youths and mental health problems: distress amongst AIDS-orphaned children in urban South
‘Much ado about nothing’? Psychological Medicine, 40, Africa. Journal of Child Psychology and Psychiatry, 48, 755–
717–729. 763.
Atladottir, H.O., Gyllenberg, D., Langridge, A., Sandin, S., Collishaw, S., Gardner, F., Maughan, B., Scott, J., & Pickles, A.
Hansen, S.N., Leonard, H., . . . & Leonard, H. (2014). The (2012). Do historical changes in parent-child relationships
increasing prevalence of reported diagnoses of childhood explain increases in youth conduct problems? Journal of
psychiatric disorders: A descriptive multinational Abnormal Child Psychology, 40, 119–132.
comparison. European Child and Adolescent Psychiatry, Collishaw, S., Goodman, R., Ford, T., Rabe-Hesketh, S., &
published online May 2014 [Epub ahead of print]. Pickles, A. (2009). How far are associations between child,
Backman, O., Estrada, F., Nilsson, A., & Shannon, D. (2014). family, and community factors and child psychopathology
The life course of male and female offenders. British Journal informant-specific and informant-general? Journal of Child
of Criminology, 54, 393–410. Psychology and Psychiatry, 50, 571–580.
Baird, G., Simonoff, E., Pickles, A., Chandler, S., Loucas, T., Collishaw, S., Goodman, R., Pickles, A., & Maughan, B. (2007).
Meldrum, D., & Charman, T. (2006). Prevalence of disorders Modelling the contribution of changes in family life to time
of the autism spectrum in a population cohort of children in trends in adolescent conduct problems. Social Science and
South Thames: The Special Needs and Autism Project Medicine, 65, 2576–2587.
(SNAP). Lancet, 368, 210–215. Collishaw, S., Maughan, B., Goodman, R., & Pickles, A. (2004).
Barnard, J., Potter, D., Broach, S., & Prior, A. (2002). Autism in Time trends in adolescent mental health. Journal of Child
schools: crisis or challenge? London: National Autistic Society. Psychology and Psychiatry, 45, 1350–1362.
Barr, B., Taylor-Robinson, D., Scott-Samuel, A., McKee, M., & Collishaw, S., Maughan, B., Natarajan, L., & Pickles, A.
Stuckler, D. (2012). Suicides associated with the 2008–10 (2010). Trends in adolescent emotional problems in
economic recession in England: time trend analysis. British England: A comparison of two national cohorts twenty
Medical Journal, 345, e5142. years apart. Journal of Child Psychology and Psychiatry,
Birmaher, B., Williamson, D.E., Dahl, R.E., Axelson, D.A., 51, 885–894.
Kaufman, J., Dorn, L.D., & Ryan, N.D. (2004). Clinical
presentation and course of depression in youth: Does onset

© 2014 Association for Child and Adolescent Mental Health.

doi:10.1111/jcpp.12372 ARR: Secular trends in child and adolescent mental health 389

Costello, E.J., Compton, S.N., Keeler, G., & Angold, A. (2003). Fleming, T.M., Clark, T., Denny, S., Bullen, P., Crengle, S.,
Relationships between poverty and psychopathology. Peiris-John, R., . . . & Lucassen, M. (2014). Stability and
A natural experiment. JAMA, 290, 2023–2029. change in the mental health of New Zealand secondary
Costello, E.J., Erkanli, A., & Angold, A. (2006). Is there an school students 2007–2012: Results from the national
epidemic of child or adolescent depression? Journal of Child adolescent health surveys. Australian and New Zealand
Psychology and Psychiatry, 47, 1263–1271. Journal of Psychiatry, 48, 472–480.
Crijnen, A.A.M., Achenbach, T.M., & Verhulst, F.C. (1997). Fombonne, E. (1998). Increased rates of psychosocial
Comparisons of problems reported by parents of children in disorders in youth. European Archives of Psychiatry and
12 cultures: Total problems, externalizing and internalizing. Clinical Neuroscience, 248, 14–21.
Journal of the American Academy of Child and Adolescent Fombonne, E. (2009). Epidemiology of pervasive developmental
Psychiatry, 36, 1269–1277. disorders. Pediatric Research, 65, 591–598.
Cross-National Collaborative Group (1992). The changing rate Ford, T., Hamilton, H., Meltzer, H., & Goodman, R. (2007).
of major depression. Journal of the American Medical Child mental health is everybody’s business: The prevalence
Association, 268, 3098–3105. of contact with public sector services by type of disorder
Cuijpers, P., & Smit, F. (2002). Excess mortality in depression: among British school children in a three-year period. Child
A meta-analysis of community studies. Journal of Affective and Adolescent Mental Health, 12, 13–20.
Disorders, 72, 227–236. Frans, E.M., Sandin, S., Reichenberg, A., Langstrom, N.,
Dickens, W.T., & Flynn, J.R. (2001). Heritability estimates Lichtenstein, P., McGrath, J.J., & Hultman, C.M. (2013).
versus large environmental effects: The IQ paradox resolved. Autism risk across generations. A population-based
Psychological Review, 108, 346–369. study of advancing grandpaternal and paternal age.
Dowrick, C., & Frances, A. (2013). Medicalising unhappiness: JAMA Psychiatry, 70, 516–521.
New classification of depression risks more patients being Frederick, C.B., Snellman, K., & Putnam, R.D. (2014).
put on a drug treatment from which they will not benefit. Increasing socioeconomic disparities in adolescent obesity.
British Medical Journal, 347, f7140. Proceedings of the National Academy of Sciences of the
Eckersley, R. (2006). Is modern Western culture a health United States of America, 111, 1338–1342.
hazard? International Journal of Epidemiology, 35, 252– Fredriks, A.M., Van Buuren, S., Burgmeijer, R.J.F.,
258. Meulmeester, J., Beuker, R.J., Brugman, E., . . . & Wit,
Ekblad, M., Gissler, M., Korkeila, J., & Lehtonen, L. (2014). J.M. (2000). Continuing secular growth change in the
Trends and risk groups for smoking during pregnancy in Netherlands 1955–1997. Pediatric Research, 47, 316–323.
Finland and other Nordic countries. European Journal of Getahun, D., Jacobsen, S.J., Fassett, M.J., Chen, W.,
Public Health, 24, 544–551. Demissie, K., & Rhoads, G.G. (2013). Recent trends in
Elsabbagh, M., Divan, G., Koh, Y.J., Kim, Y.S., Kauchali, S., childhoood attention-deficit/hyperactivity disorder. Journal
Marcın, C., . . . & Fombonne, E. (2012). Global prevalence of of the American Medical Association Pediatrics, 167, 282–
autism and other pervasive developmental disorders. Autism 288.
Research, 5, 160–179. Golub, M.S., Collman, G.W., Foster, P.M.D., Kimmel, C.A.,
Estrada, F., Petterson, T., & Shannon, D. (2012). Crime and Rajpert-De Meyts, E., Reiter, E.O., . . . & Toppari, J. (2008).
criminology in Sweden. European Journal of Criminology, 9, Public health implications of altered pubertal timing.
668–688. Pediatrics, 2008 , S218.
Farrington, D.P., & West, D.J. (1993). Criminal, penal and life Goodman, A., Heiervang, E., Fleitlich-Bilyk, B., Alyari, A.,
histories of chronic offenders: Risk and protective factors Patel, V., Mullick, M.S., . . . & Goodman, R. (2012). Cross-
and early identification. Criminal Behaviour and Mental national differences in questionnaires do not necessarily
Health, 3, 492–523. reflect comparable differences in disorder prevalence.
FDA Center for Drug Evaluation and Research (2004). Social Psychiatry and Psychiatric Epidemiology, 47, 1321–
Background on suicidality associated with antide- 1331.
pressant drug treatment (memorandum). www.fda.gov/ Goodman, R., Iervolino, A.C., Collishaw, S., Pickles, A., &
ohrms/dockets/ac/04/briefing/4006B1_03_Background% Maughan, B. (2007). Seemingly minor changes to a
20Memo%2001-05-04.htm. questionnaire can make a big difference to mean scores: A
Federal Bureau of Investigation. Crime in the United States cautionary tale. Social Psychiatry and Psychiatric
(2009). Available from: http://www2.fbi.gov/ucr/cius2009/ Epidemiology, 42, 322–327.
data/table_01.html [last accessed 29 November 2014]. Gore Langton, E., Collishaw, S., Goodman, R., Pickles, A., &
Ferri, E., Bynner, J., & Wadsworth, M. (2003). Changing Maughan, B. (2011). An emerging income differential for
Britain, changing lives. London: Institute of Education. adolescent emotional problems. Journal of Child Psychology
Fichter, M.M., Quadflieg, N., Georgopoulou, E., Xepapadakos, and Psychiatry, 52, 1081–1088.
F., & Fthenakis, E.W. (2005). Time trends in eating Gottfredson, M., & Hirschi, T. (1990). A general theory of crime.
disturbances in young Greek migrants. International Stanford, CA: Stanford University Press.
Journal of Eating Disorders, 38, 310–322. Green, H., McGinnity, A., Meltzer, H., Ford, T., & Goodman, R.
Fichter, M.M., Xepapadakos, F., Quadflieg, N., Georgopoulou, (2005). Mental health of children and young people in Great
E., & Fthenakis, W.E. (2004). A comparative study of Britain, 2004. London: Palgrave Macmillan.
psychopathology in Greek adolescents in Germany and in Gunnell, D., Middleton, N., & Frankel, S. (2000). Method
Greece in 1980 and 1998–18 years apart. European availability and the prevention of suicide: A re-analysis of
Archives of Psychiatry and Clinical Neuroscience, 254, 27– secular trends in England and Wales 1950–1975. Social
35. Psychiatry and Psychiatric Epidemiology, 35, 437–443.
Finkelhor, D. (2013). Trends in bullying and peer victimization. Hagell, A. (2012). Changing adolescence: Social trends and
Crimes against Children Research Center bulletin. Available mental health. Bristol: Policy Press.
from: http://cola.unh.edu/sites/cola.unh.edu/files/CV280_ Hawton, K., Hall, S., Simkin, S., Bale, L., Bond, A., Codd, S., &
Bullying_Peer_Victimization_Bulletin_1-23-13_with_toby_edits. Stewart, A. (2003). Deliberate self-harm in adolescents:
pdf [last accessed October 2014]. A study of characteristics and trends in Oxford, 1990–2000.
Finkelhor, D., Turner, H., Ormrod, R., & Hamby, S.L. (2010). Journal of Child Psychology and Psychiatry, 44, 1191–1198.
Trends in childhood violence and abuse exposure: Evidence Hawton, K., Saunders, K.E.A., & O’Connor, R.O. (2012). Self-
from 2 national surveys. Archives of Paediatric and harm and suicide in adolescents. The Lancet, 379, 2373–
Adolescent Medicine, 164, 238–242. 2382.

© 2014 Association for Child and Adolescent Mental Health.

390 Stephan Collishaw J Child Psychol Psychiatry 2015; 56(3): 370–93

Hawton, K., Simkin, S., Deeks, J., Cooper, J., Johnston, A., with mental disorder: Developmental follow-back of a
Waters, K., . . . & Simpson, K. (2004). UK legislation on prospective-longitudinal cohort. Archives of General
analgesic packs: Before and after study of long term effect on Psychiatry, 60, 709–717.
poisonings. British Medical Journal, 329, 1076. King, G., Murray, C.J.L., Salomon, J.A., & Tandon, A. (2004).
Heiervang, E., Goodman, A., & Goodman, R. (2008). The Enhancing the validity and cross-cultural comparability of
Nordic advantage in child mental health: Separating health measurement in survey research. American Political Science
differences from reporting style in a crosscultural Review, 98, 191–207.
comparison of psychopathology. Journal of Child Kosidou, K., Magnusson, C., Mittendorfer-Rutz, E., Hallqvist,
Psychology and Psychiatry, 49, 678–685. J., Hellner Gumpert, C., Idrizbegovic, S., . . . & Dalman, C.
Hirschi, T., & Gottfredson, M. (1983). Age and the (2010). Recent trends in levels of self-reported anxiety,
explanation of crime. The American Journal of Sociology, mental health service use and suicidal behaviour in
89, 552–584. Stockholm. Acta Psychiatrica Scandinavica, 22, 47–55.
H€olling, H., Schlack, R., Petermann, F., Ravens-Sieberer, U., Kulage, K.M., Smaldone, A.M., & Cohn, E.G.. (2014). How will
Mauz, E., & KiGGS Study Group. (2014). Psychopathological DSM-5 affect autism diagnosis? A systematic review and
problems and psychosocial impairment in children and meta-analysis. Journal of Autism and Developmental
adolescents aged 3–17 years in the German population: Disorders, 44, 1918–1932.
prevalence and time trends at two measurement points Lampi, K.M., Lehtonen, L., Lien Tran, P., Suominen, A.,
(2003–2006 and 2009–2012). Results of the KiGGS study: Lehti, V., Banerjee, P.N., . . . & Sourander, A. (2012). Risk
First follow-up (KiGGS Wave 1). Bundesgesundheitsblatt, of autism spectrum disorders in low birth weight and
57, 807–819. small for gestational age infants. Journal of Pediatrics,
Home Office (2008). Young people and crime: Findings from the 161, 830–836.
2006 Offending, Crime and Justice Survey. Colchester: Le, M.T.H., Nguyen, H.T., Tran, T.D., & Fisher, J.R. (2012).
Home Office Statistical Bulletin. Experience of low mood and suicidal behaviors among
Honda, H., Simizu, Y., & Rutter, M. (2005). No effect of MMR adolescents in Vietnam: findings from two national
withdrawal on the incidence of autism: A total population population-based surveys. Journal of Adolescent Health,
study. Journal of Child Psychology and Psychiatry, 46, 572– 51 , 339–348.
579. Lin, H., & Wang, Y. (2007). Child behavioural problems:
Hsia, Y., & Maclennan, K. (2009). Rise in psychotropic drug Comparative follow-up study two decades – socio-cultural
prescribing in children and adolescents during 1992–2001: comments. World Cultural Psychiatry Research Review, Oct
A population-based study in the UK. European Journal of 2007, 2, 128–132.
Epidemiology, 24, 211–216. Livingstone, S., & Smith, P.K. (2014). Annual research review:
Idring, S., Raj, D., Dal, H., Dalman, C., Sturm, H., Zander, E., Harms experienced by child users of online and mobile
. . . & Magnusson, C. (2012). Autism spectrum disorders in technologies: the nature, prevalence and management of
the Stockholm Youth Cohort: design, prevalence, and sexual and aggressive risks in the digital age. Journal of
validity. PLoS ONE, 7, e41280. Child Psychology and Psychiatry, 55, 635–654.
Ilola, A.M., & Sourander, A. (2013). Bullying and victimization Lotter, V. (1966). Epidemiology of autistic conditions in young
among 8-year old children: A 16-year population-based children. Social Psychiatry, 1, 124–135.
time-trend study. Nordic Journal of Psychiatry, 67, 171– Lund, C., Breen, A., Flisher, A.J., Kakuma, R., Corrigal, J.,
176. Joska, J.A., . . . & Patel, V. (2010). Poverty and common
Institute of Medicine (2009). Preventing mental, emotional and mental disorders in low income countries: A systematic
behavioral disorders among young people: progress and review. Social Science and Medicine, 71, 517–528.
possibilities. Washington DC: National Academies Press. Luntamo, T., Sourander, A., Santalahti, P., Aromaa, M., &
Jaffee, S.R., Strait, L.B., & Odgers, C.L. (2012). From Helenius, H. (2012). Prevalence changes of pain, sleep
correlates to causes: Can quasi-experimental studies and problems and fatigue among 8-year old children: Years
statistical innovations bring us closer to identifying the 1989, 1999, and 2005. Journal of Paediatric Psychology, 37,
causes of antisocial behaviour? Psychological Bulletin, 138, 307–318.
272–295. Ma, J., Lee, K.V., & Stafford, R.S. (2005). Depression treatment
Jokela, M., Ferrie, J., & Kivimaki, M. (2009). Childhood during outpatient visits by US children and adolescents.
problem behaviors and death by midlife: The British Journal of Adolescent Health, 37, 434–442.
National Child Development Study. Journal of the Man, K.K.C., Ip, P., Hsia, Y., Chan, E.W., Chui, C.S.L., Lam,
American Academy of Child and Adolescent Psychiatry, 48, M.P., . . . & Wong, I.C. (2014). ADHD drug prescribing trend
19–24. is increasing among children and adolescents in Hong Kong.
Jones, L.M., Mitchell, K.J., & Finkelhor, D. (2012). Trends in Journal of Attention Disorders. Advance online publication.
youth internet victimization: Findings from three youth doi: 10.1177/1087054714536047.
internet safety surveys 2000–2010. Journal of Adolescent Marmot, M. (2010). Fair society, healthy lives: The marmot
Health, 50, 179–186. review. Strategic review of health inequalities in England
Kessler, R.C., Berglund, P., Demler, O., Jin, R., Merikangas, post-2010. London: The Marmot Review.
K.R., & Walters, E.E. (2005). Lifetime prevalence and age-at- Matijasevich, A., Murray, E., Stein, A., Anselmi, L., Menezes,
onset distributions of DSM-IV disorders in the National A.M., Santos, I.S., . . . & Victora, C.G. (2014). Increase in
Comorbidity Survey Replication. Archives of General child behaviour problems among urban Brazilian four-year-
Psychiatry, 62, 593–602. olds: 1993 and 2004 Pelotas Birth Cohorts. Journal of Child
Keyes, K.M., Nicholson, R., Kinley, J., Raposo, S, Stein, M.B., Psychology and Psychiatry, 55, 1125–1134.
Goldner, E.M., & Sareen, J. (2014). Age, period, and cohort Matricciani, L., Olds, T., & Petkov, J. (2012). In search of lost
effects in psychological distress in the United States and sleep: secular trends in the sleep time of school-aged
Canada. American Journal of Epidemiology, 179, 1216–1227. children and adolescents. Sleep Medicine Reviews, 16,
Kieling, C., Baker-Henningham, H., Belfer, M., Conti, G., 203–211.
Ertem, I., Omigbodun, O., . . . & Rahman, A. (2011). Child Maughan, B., & Collishaw, S. (2015). Development and
and adolescent mental health worldwide: Evidence for psychopathology: A lifecourse perspective. In A. Thapar, D.
action. Lancet, 378, 1515–1525. Pine, J.F. Leckman, S. Scott, M.J. Snowling & E. Taylor
Kim-Cohen, J., Caspi, A., Moffitt, T.E., Harrington, H., Milne, (Eds.), Rutter’s child and adolescent psychiatry (6th edn),
B.J., & Poulton, R. (2003). Prior juvenile diagnoses in adults Chichester: John Wiley and Sons.

© 2014 Association for Child and Adolescent Mental Health.

doi:10.1111/jcpp.12372 ARR: Secular trends in child and adolescent mental health 391

Maughan, B., Collishaw, S., Meltzer, H., & Goodman, R. Pinker, S. (2011). The better angels of our nature: why violence
(2008). Recent trends in UK child and adolescent mental has declined. London: Penguin Books.
health. Social Psychiatry and Psychiatric Epidemiology, 43, Polanczyk, G.V., Willcutt, E.G., Salum, G.A., Kieling, C, &
305–310. Rohde, L.A. (2014). ADHD prevalence estimates across three
Maughan, B., Collishaw, S., & Stringaris, A. (2013). decades: An updated systematic review and meta-regression
Depression in childhood and adolescence. Journal of the analysis. International Journal of Epidemiology, 43, 434–442.
Canadian Academy of Child and Adolescent Psychiatry, 22, Potteg
ard, A., Zo€ega, H., Hallas, J., & Damkier, P. (2014). Use
35–40. of SSRIs among Danish children: A nationwide study.
Maughan, B., Stafford, M., Shah, I., & Kuh, D. (2014). European Child and Adolescent Psychiatry, 23, 1211–
Adolescent conduct problems and premature mortality: 1218.
Follow-up to age 65 years in a national birth cohort. Potter, R., Mars, B., Eyre, O., Legge, S., Ford, T., Sellers, R., . . .
Psychological Medicine, 44, 1077–1086. & Thapar, A.K. (2012). Missed opportunities: Mental
McArdle, P., Prosser, J., Dickinson, H., & Kolvin, I. (2003). disorder in children of parents with depression. British
Secular trends in the mental health of primary school Journal of General Practice, 62, e487–e493.
children. Irish Journal of Psychological Medicine, 20, 56– Raj, D., Lee, B.K., Dalman, C., Golding, J., Lewis, G., &
58. Magnusson, C. (2013). Parental depression, maternal
McCarthy, S., Wilton, L., Murray, M.L., Hodgkins, P., antidepressant use during pregnancy, and risk of autism
Asherson, P., & Wong, I.C.K. (2012). The epidemiology of spectrum disorders: Population case-control study. British
pharmacologically treated attention deficit hyperactivity Medical Journal, 346, f2059.
disorder (ADHD) in children, adolescents and adults in UK Reiss, A.J., & Farrington, D.P. (1991). Advancing knowledge
primary care. BMC Pediatric, 12, 1–11. about co-offending: Results from a prospective longitudinal
McKeown, R.E., Cuffe, S.P., & Schulz, R.M. (2006). US survey of London males. Journal of Criminal Law and
suicide rates by age group, 1970–2002: An examination of Criminology, 82, 360–395.
recent trends. American Journal of Public Health, 96, Rigby, K., & Smith, P.K. (2011). Is school bullying really on the
1744–1751. rise? Social Psychology of Education, 14, 441–455.
Meltzer, H., Gatward, R., Goodman, R., & Ford, T. (2000). The Robertson, C., & Boyle, P. (1998). Age-period-cohort analysis
mental health of children and adolescents in Great Britain: of chronic disease rates. I: modelling approach. Statistics in
summary report. London: The Stationary Office. Medicine, 17, 1305–1323.
Mittendorfer-Rutz, E., & Wasserman, D. (2004). Trends in Robins, L.N., & Regier, D.A. (1991). Psychiatric disorders in
adolescent suicide mortality in the WHO European Region America: The epidemiologic catchment area study. New York:
2004. European Child and Adolescent Psychiatry, 13, 321– John Wiley and Sons.
331. Robinson, W.S. (1950). Ecological correlations and the
Moffitt, T.E. (1993). Adolescence-limited and life-course- behavior of individuals. American Sociological Review, 15,
persistent antisocial behavior: A developmental taxonomy. 351–357.
Psychological Review, 100, 674–701. Russell, G., Kelly, S., & Golding, J. (2010). A qualitative
Mohler, B., & Earls, F. (2001). Trends in adolescent suicide: analysis of lay beliefs about the aetiology and prevalence of
Misclassification bias? American Journal of Public Health, autistic spectrum disorders. Child Care Health and
91, 150–153. Development, 36, 431–436.
Molcho, M., Craig, W., Due, P., Pickett, W., Harel-Fisch, Y., Russell, G., Rodgers, L.R., Ukoumunne, O.C., & Ford, T.
Overpeck, M., & HBSC Bullying Writing Group. (2009). (2014). Prevalence of parent-reported ASD and ADHD in the
Cross-national time trends in bullying behaviour 1994– UK: findings from the Millennium Cohort Study. Journal of
2006: Findings from Europe and North America. Autism and Developmental Disorders, 44, 1–10.
International Journal of Public Health, 54, S225–S234. Rutter, M. (2005). Incidence of autism spectrum disorders:
Murray, J., de Castro Cerqueira, DR., & Kahn, T. (2013). Crime Changes over time and their meaning. Acta Paediatrica, 94,
and violence in Brazil: Systematic review of time trends, 2–15.
prevalence rates and risk factors. Aggression and Violent Rutter, M. (2013). Annual research review: Resilience – clinical
Behavior, 18, 471–483. implications. Journal of Child Psychology and Psychiatry,
Murray, C.J., Vos, T., Lozano, R., Naghavi, M., Flaxman, A.D., 54, 474–487.
Michaud, C., . . . & Memish, Z.A. (2012). Disability-adjusted Rutter, M., & Smith, D. (1995). Psychosocial disorders in young
life years (DALYs) for 291 diseases and injuries in 21 people: time trends and their causes. Chichester: John Wiley
regions, 1990–2010: A systematic analysis for the Global and Sons.
Burden of Diseases Study 2010. Lancet, 380, 2197–2223. Rutter, M., & Solantaus, T. (2014). Translation gone awry:
Office for National Statistics (2014). Crime in England and Differences between common sense and science. European
Wales, Year ending September 2013. Statistical Bulletin. Child and Adolescent Psychiatry, 23, 247–255.
Available from: http://www.ons.gov.uk/ons/dcp171778_ Salomon, J.A., Tandon, A., & Murray, C.J.L. (2004).
349849.pdf [last accessed 29 November 2014]. Comparability of self rated health: Cross sectional multi-
Olfson, M., Blanco, C., Wang, S., Laje, G., & Correll, C.U. country survey using anchoring vignettes. British Medical
(2014). National trends in the mental health care of children, Journal, 328, 258.
adolescents, and adults by office-based physicians. JAMA Santalahti, P., Aromaa, M., Sourander, A., Helenius, H., &
Psychiatry, 71, 81–90. Piha, J. (2005). Have there been changes in children’s
Olfson, M., Gameroff, M.J., Marcus, S.C., & Jensen, P.S. psychosomatic symptoms? A 10-year comparison from
(2003). National trends in the treatment of attention deficit Finland. Pediatrics, 115, e434–e442.
hyperactivity disorder. American Journal of Psychiatry, 160, Santalahti, P., Sourander, A., Aromaa, M., Helenius, H.,
1071–1077. Ik€
aheimo, K., & Piha, J. (2008). Victimization and bullying
Ortiz, I., Daniels, L.M., & Engilbertsd ottir, S. (2012). Child among 8-year-old children. A 10-year comparison of rates.
poverty and inequality. New perspectives and challenges. European Child and Adolescent Psychiatry, 17, 463–472.
New York: United Nations Children’s Fund. Scheffler, R.M., Hinshaw, S.P., Modrek, S., & Levine, P. (2007).
Pallesen, S., Hetland, J., Sivertsen, B., Samdal, O., Torsheim, The global market for ADHD medications. Health Affairs, 26,
T., & Nordhus, I.H. (2008). Time trends in sleep-onset 450–457.
difficulties among Norwegian adolescents: 1983–2005. Schepman, K., Collishaw, S., Gardner, F., Maughan, B., Scott,
Scandinavian Journal of Public Health, 36, 889–895. J., & Pickles, A. (2011). Do changes in parent mental health

© 2014 Association for Child and Adolescent Mental Health.

392 Stephan Collishaw J Child Psychol Psychiatry 2015; 56(3): 370–93

explain trends in youth emotional problems? Social Science Thapar, A., & Rutter, M. (2009). Do prenatal risk factors cause
and Medicine, 73, 293–300. psychiatric disorder? Be wary of causal claims. British
Sellers, R., Maughan, B., Pickles, A., Thapar, A., & Collishaw, Journal of Psychiatry, 195, 100–101.
S. (2015). Trends in parent- and teacher-rated emotional, Thapar, A., & Rutter, M. (2015). Using natural experiments
conduct and ADHD problems and their impact in pre- and animal models to study causal hypotheses in relation to
pubertal children in Great Britain: 1999–2008. Journal of child mental health problems. In A. Thapar, D. Pine, J.F.
Child Psychology and Psychiatry, 56, 49–57. Leckman, S. Scott, M.J. Snowling & E. Taylor (Eds.), Rutter’s
Shadish, W.R., Cook, T.D., & Campbell, D.T. (2002). child and adolescent psychiatry (6th edn), Chichester: John
Experimental and quasi-experimental designs for Wiley and Sons.
generalized causal inference. Boston: Houghton Mifflin. Tick, N.T., van der Ende, M.S., Koot, H.S., & Verhulst, F.C.
Sigfusdottir, I.D., Asgeirsdottir, B.B., Sigurdsson, J.F., & (2007). 14-year changes in emotional and behavioral
Gudjonsson, G.H. (2008). Trends in depressive symptoms, problems of very young Dutch children. Journal of the
anxiety symptoms and visits to healthcare specialists: A American Academy of Child and Adolescent Psychiatry, 46,
national study among Icelandic adolescents. Scandinavian 1333–1340.
Journal of Public Health, 36, 361–368. Tick, N.T., van der Ende, J., & Verhulst, F.C. (2007). Twenty-
Silventoinen, K., Sammalisto, S., Perola, M., Boomsma, D.I., year trends in emotional and behavioural problems in Dutch
Cornes, B.K., Davis, C., . . . & Kaprio, J. (2003). Heritability children in a changing society. Acta Psychiatrica
of adult body height: A comparative study of twin studies in Scandinavica, 116, 473–482.
eight countries. Twin Research, 6, 399–408. Tick, N.T., van der Ende, J., & Verhulst, F.C. (2008). Ten-year
Smith, R.P., Larkin, G.L., & Southwick, S.M. (2008). Trends in trends in self-reported emotional and behavioural problems
US emergency department visits for anxiety-related mental of Dutch adolescents. Social Psychiatry and Psychiatric
health conditions, 1992–2001. Journal of Clinical Epidemiology, 43, 349–355.
Psychiatry, 69, 286–294. Toh, S. (2006). Trends in ADHD and stimulant use among
Sorensen, K., Mouritsen, A., Aksglaede, L., Hagen, C.P., children, 1993–2003. Psychiatric Services, 57, 1091.
Mogensen, S.S., & Juul, A. (2012). Recent secular trends Torikka, A., Kaltiala-Heino, R., Rimpel€ a, A., Marttunen, M.,
in pubertal timing: Implications for evaluation and diagnosis Luukkaala, T., & Rimpel€ a, M. (2014). Self-reported depre-
of precocious puberty. Hormone Research in Paediatrics, 77, ssion is increasing among socio-economically disadvantaged
137–145. adolescents – repeated cross-sectional surveys from Finland
Sourander, A., Koskelainen, M., Niemel€ a, S., Rihko, M., from 2000 to 2011. BMC Public Health, 14, 408.
Ristkari, T., & Lindroos, J. (2012). Changes in adolescents Twenge, J.M. (2011). Generational differences in mental
mental health and use of alcohol and tobacco: A 10-year health: Are children and adolescents suffering more, or
time-trend study of Finnish adolescents. European Child less? American Journal of Orthopsychiatry, 81, 469–472.
and Adolescent Psychiatry, 21, 665–671. UK Statistics Authority. (2014). Assessment of compliance
Sourander, A., Niemel€ a, S., Santalahti, P., Helenius, H., & with the Code of Practice for Official Statistics. Statistics on
Piha, J. (2008). Changes in psychiatric problems and service crime in England and Wales (produced by the Office for
use among 8-year old children: A 16-year population-based National Statistics). Assessment Report 268.
time-trend study. Journal of the American Academy of Child United Nations Children’s Fund (2013). Towards an AIDS-free
and Adolescent Psychiatry, 47, 317–327. generation: Children and AIDS sixth stock-taking report.
Sourander, A., Santalahti, P., Haavisto, A., Piha, J., Ik€aheimo, UNICEF.
K., & Helenius, H. (2004). Have there been changes in United Nations Economic and Social Council. (2014). World
children’s psychiatric symptoms and mental health service Crime Trends and emerging issues and responses in the field of
use? A ten-year comparison from Finland. Journal of the crime prevention and criminal justice. Available from: http://
American Academy of Child and Adolescent Psychiatry, 43, www.unodc.org/documents/data-and-analysis/statistics/
1134–1145. crime/ECN.1520145_EN.pdf [last accessed October 2014].
St Clair, D., Xu, M., Wang, P., Yu, Y., Fang, Y., Zhang, F., . . . & Van Dijk, J., Tseloni, A., & Farrell, G. (2012). The international
He, L. (2005). Rates of adult schizophrenia following crime drop: New directions in research. Basingstoke:
prenatal exposure to the Chinese famine of 1959–1961. Palgrave Macmillan.
Journal of the American Medical Association, 294, 557– Van Dijk, J.J.M., van Kesteren, J.N., & Smit, P. (2008).
562. Criminal victimisation in International Perspective, key
Stephenson, C.P., Karanges, E., & McGregor, I.S. (2013). findings from the 2004–2005 ICVS and EU ICS. The Hague:
Trends in the utilisation of psychotropic medications in Boom Legal Publishers.
Australia from 2000 to 2011. Australia and New Zealand V€
arnik, A., K~olves, K., Allik, J., Arensman, E., Aromaa, E., van
Journal of Psychiatry, 47, 74–87. Audenhove, C., . . . & Hegerl, U. (2009). Gender issues in
Stuckler, D., Basu, S., Suhrcke, M., Coutts, A., & McKee, M. suicide rates, trends and methods among youths aged 15–
(2009). The public health effect of economic crises and 24 in 15 European countries. Journal of Affective Disorders,
alternative policy responses in Europe: An empirical 113, 216–226.
analysis. Lancet, 374, 315–323. von Soest, T., & Wichstrøm, L. (2014). Secular trends in
Susser, E., Hoek, H.W., & Brown, A. (1998). Neurodevelopmental depressive symptoms among Norwegian adolescents from
disorders after prenatal famine. The story of the Dutch Famine 1992 to 2010. Journal of Abnormal Child Psychology, 42,
Study. American Journal of Epidemiology, 147, 213–216. 403–415.
Sweeting, H., West, P., Young, R., & Der, G. (2010). Can we Wangby, M., Magnusson, D., & Stattin, H. (2005). Time trends
explain increases in young people’s psychological distress in the adjustment of Swedish Teenage Girls: A 26-year
over time? Social Science and Medicine, 71, 1819–1830. comparison of 15-year-olds. Scandinavian Journal of
Sweeting, H., Young, R., & West, P. (2009). GHQ increases Psychology, 46, 145–156.
among Scottish 15 year olds 1987–2006. Social Psychiatry West, P., & Sweeting, H. (2003). Fifteen, female and
and Psychiatric Epidemiology, 44, 579–586. stressed: Changing patterns of psychological distress over
Taylor, B., Jick, H., & MacLaughlin, D. (2013). Prevalence and time. Journal of Child Psychology and Psychiatry, 44, 399–
incidence rates of autism in the UK: Time trend from 2004- 411.
2010 in children aged 8 years. BMJ Open, 3, e003219. Whittington, C.J., Kendall, T., Fonagy, P., Cottrell, D.,
Thapar, A., Collishaw, S., Pine, D.S., & Thapar, A.K. (2012). Cotgrove, A., & Boddington, E. (2004). Selective serotonin
Depression in adolescence. The Lancet, 379, 1056–1067. reuptake inhibitors in childhood depression: Systematic

© 2014 Association for Child and Adolescent Mental Health.

doi:10.1111/jcpp.12372 ARR: Secular trends in child and adolescent mental health 393

review of published versus unpublished data. Lancet, 363, psychotropic medications. Archives of Disease in
1341–1345. Childhood, 89, 1131–1132.
Wilkinson, R., & Pickett, K. (2010). The spirit level. London: Yoshikawa, H., Aber, J.L., & Beardslee, W. (2012). The effects
Penguin. of poverty on the mental, emotional, and behavioral health of
Williams, R.L. (2013). Overview of the Flynn effect. Intelligence, children and youth: implications for prevention. American
41, 753–764. Psychologist, 67, 272–284.
Windfuhr, K., While, D., Hunt, I., Turnbull, P., Lowe, R.,
Burns, J., . . . & Kapur, N. (2008). Suicide in juveniles and Accepted for publication: 4 November 2014
adolescents in the United Kingdom. Journal of Child Published online: 12 December 2014
Psychology and Psychiatry, 49, 1155–1165.
Wong, I.C.K., Murray, M.L., Novak-Camilleri, D., & Stephens,
P. (2004). Increased prescribing trends of paediatric

© 2014 Association for Child and Adolescent Mental Health.