Content uploaded by Michael Parsonage
Author content
All content in this area was uploaded by Michael Parsonage on Sep 14, 2016
Content may be subject to copyright.
REPORT
Building a better future
Michael Parsonage, Lorraine Khan & Anna Saunders
The lifetime costs of childhood behavioural
problems and the benefits of early intervention
Centre for Mental Health REPORT Building a better future
22
Contents
Executive summary 3
1. Introduction 8
2. Early behavioural problems - background 9
Case study: Ollie 11
3. Outcomes over the life course 12
Case study: Tommy 20
4. The costs of behavioural problems 22
Case study: Alyssa 26
5. Evidence on the effectiveness of parenting programmes 27
Case study: Liah 32
6. Thecostsandbenetsofintervention 33
References 41
Key
EDUCATION
CRIMINAL
JUSTICE
NHS BULLYING
LOCAL AUTHORITY
HOUSING
SOCIAL SERVICES
3
Centre for Mental Health REPORT Building a better future
This report is part of a wider programme of
work on early intervention undertaken by
Centre for Mental Health with funding from the
Esmée Fairbairn Foundation. It uses economic
analysis to explore the long-term consequences
of severe behavioural problems that start in
childhoodandthebenetsofeffectiveearly
intervention, including possible savings in
public expenditure.
Early behavioural problems –
background
About 5% of children aged 5-10 display
behaviouralproblemswhicharesufciently
severe, frequent and persistent that they justify
diagnosis as a mental health condition
(‘conduct disorder’). A further 15-20% have
problems which fall below this threshold but are
still serious enough to merit concern.
Depending on age, problem behaviours may
include: persistent disobedience, angry
outbursts and tantrums, physical aggression,
ghting,destructionofproperty,stealing,lying
and bullying. For about half of the children
affected, serious problems will persist into
adolescence and beyond.
A wide range of risk factors, both genetic and
environmental, may be implicated in the early
development of severe behavioural problems.
Particular importance attaches to adverse
inuenceswithinthefamilyenvironment,such
as maltreatment and harsh, inconsistent or
neglectful parenting.
Outcomes over the life course
Severe and persistent behavioural problems in
young children are associated with a wide range
of adverse outcomes, not only in childhood but
throughout the life course and even extending
into succeeding generations. Many different
domains of life may be affected.
Executive Summary
Mental health: continuing mental health
difcultiesinadolescenceandadulthood,
including increased rates of depression and
anxiety, alcohol and drug abuse, personality
disorder, self-harm and suicide.
Physical health: increased rates of morbidity,
disability and premature mortality, often
associated with risky behaviours such as
smoking and alcohol or drug misuse; high rates
of teenage pregnancy.
Child protection: high risk of being placed on
child protection registers and being taken into
care.
Education: high rates of truancy and school
exclusion; frequent involvement in bullying as
both perpetrator and victim; poor educational
attainment.
Employment: high rates of unemployment;
increased likelihood of employment in low-paid
jobs and in jobs held for short periods of time;
increaseddependencyonwelfarebenets.
Crime: high rates of involvement in all types of
criminal activity including violent crime, often
starting at an early age.
Homelessness: substantially increased risk of
experiencing homelessness.
Social networks: few if any friends, whether in
childhood or in later life; limited involvement in
social activities.
Relationships: high rates of involvement in
personal relationships which are short-lived and
characterised by abuse and violence, including
mutual violence.
Parenthood: increased rates of child abuse
and maltreatment; children at increased risk
of being taken into care and of developing
behavioural problems.
Children with severe behavioural problems are
generally around four to ten times more likely to
experience these adverse outcomes in later life
than those with no such problems.
Centre for Mental Health REPORT Building a better future
4
The costs of behavioural problems
Various attempts have been made to estimate
the long-term costs of severe behavioural
problems,usuallywithinspecicageranges
rather than across the full life course, but all
of these are likely to under-estimate the true
costs by a substantial margin. Reasons for
this include the very broad range of adverse
outcomestobeincludedandthedifcultyof
expressing some of these in monetary terms,
e.g. the impact of severe behavioural problems
on wellbeing and quality of life.
Most studies focus on costs falling on the
public sector, particularly during childhood
and adolescence. A broad average of these
estimates suggests an annual cost to the
Exchequer of around £5,000 per child with
severe behavioural problems, taking into
account the extra costs falling on health, social
care, education and, from age 10 onwards, the
criminal justice system.
One attempt to measure costs from a societal
rather than public sector perspective suggests
that the overall lifetime costs of severe
behavioural problems amount to around
£260,000 per case. (The lifetime costs of
moderate problems are put at around £85,000
per case.) Costs relating to crime are the biggest
single component, accounting for more than
two-thirds of the total.
This estimate still omits many costs. For
example, bullying is a common behaviour of
people with severe behavioural problems and
one study estimates that, after taking into
accountotherinuences,thelifetimeearnings
of a victim of serious bullying are reduced by
around £50,000 on average. Moreover, this
gurerelatestolostearningsforasinglevictim
of bullying and there are many more victims
than there are perpetrators.
The effectiveness of parenting
programmes
Because quality of parenting is a critical
determinant of child outcomes, a number of
behavioural training programmes have been
developed which aim to improve the quality
of parent-child relationships and the skill of
parents in managing child behaviour.
A large body of evidence shows that, if well
implemented, these programmes can be
very effective in improving child behaviour,
particularly by encouraging positive parenting.
They also improve the behaviour of siblings and
the mental health and wellbeing of participating
parents.
In broad terms, the effectiveness of parenting
programmes is much the same across a wide
range of family types and ethnic groups. The
programmes are also at least as effective for
children with the most severe behavioural
problems as for those with more moderate
difculties.
The main gap in the evidence concerns the
extenttowhichthebenetsofparenting
programmes in terms of improved child
behaviour are maintained over time. This mainly
reectsalackofstudiescollectinglong-term
follow-up data rather than any clear evidence
that initial improvements are not sustained.
The effectiveness of parenting programmes
is reduced by half or more if they are poorly
implemented, e.g. by employing staff who are
not properly trained or supervised.
The costs and benets of intervention
Few studies of parenting programmes have
collected detailed economic data, particularly
in relation to outcomes. Assessment of costs
andbenetsmeasuredinmonetarytermsthus
requires a modelling approach which combines
quantitative data from effectiveness trials with
economic information from other sources.
5
Centre for Mental Health REPORT Building a better future
Taking a broad average of the results of four
economic modelling studies, it is estimated that
every £1 invested in parenting programmes
yieldsmeasurablebenetstosocietyofatleast
£3. In addition, the costs of intervention are
more than covered by subsequent savings in
public spending.
These are almost certainly conservative
conclusions,mainlybecauseofomittedbenets
but also because all four studies use a range
of conservative assumptions and modelling
procedures which in combination do not
necessarily yield the most likely result.
Detailed year-by-year analysis of costs and
benetsisavailableinoneofthefourstudies
and this suggests that up to 60% of the costs of
parenting programmes are recovered within two
years through savings in public expenditure and
allcostswithinaroundveyears.Thesesavings
largely accrue to health and education budgets.
In later years savings start to build up in the
criminal justice system and in the long term it
is this part of the public sector that secures the
biggestnancialreturns.
Finally, economic analysis highlights the
importance of implementing programmes
effectively, as failure to do this carries a very
heavypenaltyintermsofbenetsforgone.
Relatively low-cost measures which reduce
the likelihood that participating parents will
drop out mid-way through a programme, e.g.
provision of free transport and crèche facilities,
are likely to have a particularly high return. And,
because the funds for early intervention are
always likely to be constrained, it is important
that parenting programmes are targeted at
those families and children who are likely to
benetmost.
A number of studies have pursued this
approach, providing information on two main
questions: are parenting programmes good
value for money for society as a whole and, on
a narrower view, do these programmes pay for
themselves through future savings in public
spending?
All studies under-estimate the aggregate returns
from early intervention because of omitted
benets.Inparticular,noattemptsaremade
to include an imputed monetary valuation for
thebenetsofimprovedmentalhealthamong
children with behavioural problems in terms of
its impact on their wellbeing and quality of life.
Suchbenetsarethefundamentaljustication
forserviceprovisionandyettheyndnoplace
in the economic literature on early intervention.
Also omitted in all studies are a range of third-
party effects such as the impact of parenting
programmes on the mental health and quality of
life of parents and siblings and on the wellbeing
of others such as the victims of bullying and
children in the next generation.
Even allowing for these limitations, the available
evidence indicates that parenting programmes
are very good value for money, both for society
as a whole and from the narrower perspective of
the public sector. This is not surprising, as the
costs of intervention are relatively low while the
potentialbenetsareextremelyhigh,reecting
the many costly consequences of severe
behavioural problems that may be mitigated by
intervention.
Studies suggest that the average cost of
bringing a child with conduct disorder below
a clinical threshold as a result of a parenting
programme is around £1,750 per case. Set
against this, the lifetime costs of conduct
disorder, measured against a baseline of
moderate behavioural problems, have been put
at around £175,000 per case. Lifetime costs
thus need to be reduced by just 1% to cover the
costs of the intervention – a strikingly small
proportion. In practice, only a fraction of long-
term costs are likely to be saved, but the general
point that the costs of early intervention are very
lowrelativetothepotentialbenetsremains
valid.
Centre for Mental Health REPORT Building a better future
6
WHAT ARE SEVERE BEHAVIOURAL PROBLEMS?
HOW COMMON ARE THEY?
WHAT IS THE COST?
£260,000
ESTIMATED LIFETIME COST
OF SEVERE BEHAVIOURAL
PROBLEMS
CONDUCT DISORDER IS
TWICE AS HIGH AMONG
BOYS AS GIRLS
£1,300
ESTIMATED OF COST OF A
PARENTING PROGRAMME
All children behave badly from time to time,
but some display behavioural problems
which are so severe, frequent and persistent
that they justify diagnosis as a mental health
condition: conduct disorder.
For about half of the children concerned,
these problems will persist into adolescence
and beyond and are associated with a wide
range of damaging and costly outcomes
throughout the life course and even
extending into succeeding generations.
Depending on age, problem behaviours may
include: persistent disobedience, angry
outbursts and tantrums, physical aggression,
ghting,destructionofproperty,stealing,
lying and bullying.
RATES OF CONDUCT DISORDER ARE
HIGHER AMONG CHILDREN FROM
DISADVANTAGED BACKGROUNDS
5%OF CHILDREN AGED 5-10 HAVE CONDUCT DISORDER
ABOUT
A FURTHER 15-20% DISPLAY BEHAVIOURAL PROBLEMS
PER CHILD
PER CHILD
WHICH FALL BELOW THIS THRESHOLD BUT ARE STILL SERIOUS ENOUGH TO MERIT
CONCERN BECAUSE OF THE INCREASED RISK OF ADVERSE OUTCOMES IN LATER LIFE
7
Centre for Mental Health REPORT Building a better future
WHAT CAN WE DO?
WHY INVEST IN THIS INTERVENTION?
Provided that they are well implemented, parenting programmes are very good value for money.
Thepotentialbenetsofearlyinterventionaresohighrelativetoitscostthatonlyamodest
improvement in outcomes is needed to support a strong economic case.
These programmes more than pay for themselves through future savings in public spending,
spread across a range of budgets including education, health, social care and criminal justice.
Andtherearealsosubstantialbenetstowidersocietyandtoindividualsandtheirfamilies,not
all of which can easily be measured in monetary terms.
NEGATIVE OUTCOMES
MORE LIKELY TO LEAVE
SCHOOL WITH NO
QUALIFICATIONS
MORE LIKELY TO
BE DEPENDENT ON
DRUGS
MORE LIKELY TO
BE ON THE CHILD
PROTECTION REGISTER
MORE LIKELY TO
BECOME A TEENAGE
PARENT
x
4
8
2
3
x
xx
MORE LIKELY TO
END UP IN PRISON
20 x
MORE LIKELY
TO DIE BEFORE
AGE 30
6x
Parenting is a critical determinant of child outcomes. In particular, positive parenting protects
children from developing severe behavioural problems. Proven parenting programmes improve the
quality of parent-child relationships and the skill of parents in managing challenging behaviour in
their children.
This report looks in detail at the large and compelling body of evidence which demonstrates the
effectiveness and cost-effectiveness of this type of early intervention.
Centre for Mental Health REPORT Building a better future
8
1. Introduction
This report is one of a number of outputs that
have resulted from a 30-month programme
of work on early intervention for children with
behavioural problems undertaken by Centre
for Mental Health, with funding from the
Esmée Fairbairn Foundation. The main focus
of this programme is on how to improve the
implementation of evidence-based parenting
programmes in support of children aged up to
11withbehaviouraldifculties.
Therstmajoroutputoftheprogrammewas
A chance to change, a report published in late
2012 which analysed in detail the key factors
that determine the successful delivery of
evidence-based parenting interventions and the
main barriers that currently hamper such efforts
(Brown, Khan & Parsonage, 2012). Prominent
amongthebarrierswereidentiedarangeof
issues relating to funding. Most obviously these
includethedifcultyofsecuringresourcesfor
any new or expanded form of service provision
at a time when public sector budgets are
under severe restraint. But over and above the
perennial challenge of competing with other
programmes for limited funds, it is clear that
early intervention programmes face additional
hurdles.Inparticular,thenancialbenetsof
these programmes accrue over long periods
of time and they are also distributed across a
wide range of different agencies in the public
sector and beyond. Under current budgetary
arrangements in central and local government,
the funding of early intervention programmes is
always likely to fall victim to some combination
of short-termism and free-riding (i.e. attempts
bydifferentagenciestoreapthenancial
rewards of early intervention without incurring
any of the costs).
There is no easy solution to these problems
and possible reforms such as pooled or
shared budgets are likely to take many
years to implement on a national scale. In
the meantime, this report seeks to inform
and improve decision making under existing
arrangements, by mapping out in detail the
fullcostsofchildhoodbehaviouraldifculties
andhencethepotentialbenetsofeffective
early intervention, year by year and budget by
budget. Most existing studies in this area tend
to understate both the overall scale of costs
associated with severe behavioural problems in
childhood and the possible scope for relatively
quick pay-offs from early intervention. More
detailed and comprehensive information on
costsandbenetsmayhelptobringhometo
decision makers the huge impact of behavioural
problems among children and why these
problems matter to a wide range of different
agencies. The analysis also highlights the
importance of maximising the returns from
early intervention, for example by targeting
programmesatthosewhowillbenetmost
and by avoiding false economies in programme
delivery which may greatly reduce effectiveness.
9
Centre for Mental Health REPORT Building a better future
emerging as early as two or three. Early onset
is associated with a high degree of persistence
into later life; indeed, problematic behaviour
in the early years has been shown to have
the highest continuity into adulthood of all
measured human traits except intelligence
(Scott, 2004). According to NICE, about half of
all children with early-onset conduct disorder
have serious problems that persist into later
life (NICE, 2013). In the second group, conduct
disorder begins in adolescence and continues
beyond this phase of development in only a
small minority of cases.
In line with the focus of our work programme
on early intervention, the rest of this report
concentrates on children with early-onset
problems. The terms ‘conduct disorder’ and
‘severe behavioural problems’ are used
interchangeably to describe problems which
aresufcientlyseveretomeetthecriteria
for diagnosis of a mental health condition,
and similarly the terms ‘conduct problems’
and ‘behavioural problems’ are used
interchangeably to describe problems which fall
short of a diagnostic threshold.
Prevalence
Accordingtothemostrecentofcialsurveyof
mental health in children and young people
(Green et al., 2005), the prevalence of conduct
disorder among children aged 5-10 is 4.9%,
equivalent to around 30,000 children in
each one-year cohort in this age range in
England. More than twice as many boys are
affected as girls: 6.9% of all boys in the 5-10
age range against 2.8% of girls. Conduct
disorder also has a strong gradient by socio-
economic class, being nearly three times as
common among children from unskilled and
workless households as among those from the
professional and managerial groups.
2: Early behavioural problems – background
Behavioural problems
Serious behavioural or conduct problems among
children and young people may take a variety
of forms and these are to some extent age-
specic.Forexample,amongchildrenaged3-7
they typically include persistent disobedience,
angry outbursts and tantrums, provocation,
physical aggression towards other children,
destruction of property and blaming others. At
ages 8-11 the list may also include swearing,
lying,stealing,rule-breaking,physicalghts,
bullying, cruelty to animals and other children
andre-setting.Andatages12-17theproblem
behaviours can include not just many of the
foregoing but also violence, robbery, vandalism,
substance misuse, persistent truanting, running
away from home, early sexual activity and
teenage pregnancy.
In some cases these problems become so
severe, frequent and persistent that they justify
a diagnosis of ‘conduct disorder’, a mental
health condition recognised in all major illness
classicationsystems,inwhichthescaleof
conduct problems is such as to impair a child’s
ownfunctioningaswellascausingsignicant
distress to others.
Conduct disorder is the most common mental
health condition found among children and
young people. At the same time, much larger
numbers display behavioural problems which,
whiledistressing,areinsufcientlysevere
to merit a clinical diagnosis. It is, however,
important not to overlook children with such
sub-threshold problems, as the evidence
shows very clearly that these can still signal an
elevated risk of adverse outcomes in later life.
The population of children and young people
with conduct disorder divides into two sub-
groups,distinguishedbyageofonset(Moftt,
1993).Intherstgroup,thedisorderbecomes
apparent at an early age, i.e. before ten, with
evidence of serious behavioural problems often
Centre for Mental Health REPORT Building a better future
10
The different risk factors are not all of equal
importance. In particular, the evidence suggests
that parenting is the single most consistently
powerfulinuenceontheemotionaland
behavioural development of children. For
example, one study has suggested that
parenting with poor supervision and lack of
warmth is responsible for 30-40% of problem
behaviour in children (Patterson et al., 1989).
There is also evidence that the association
between severe behavioural problems and
variables such as large family size and
parenthood may be largely mediated by
parenting practices, rather than these variables
actingasindependentinuencesintheirown
right (Scott, 2004).
The contributory factors leading to adolescent-
onset conduct disorder are rather different
and the condition has been described as
“the product less of individual risks than of
frustrations attendant on the ‘maturity gap’
[individuals reaching physical maturity some
years before achieving economic and social
independence] and social mimicry of deviant
peers” (Rutter et al., 2006). Social roles are
therefore much more important than in the case
of early-onset conduct disorder.
Less detailed and reliable information is
available on the numbers of children displaying
behavioural problems which fall short of
a diagnosable mental health condition,
essentially because there is no universal
agreementontheappropriatedenitionand
classicationofsuchcases.However,most
studies suggest that, in addition to those with
diagnosable conduct disorder, a further 15-20%
ofchildrencanbeidentiedashavingproblems
of moderate severity that may carry an elevated
risk of adverse long-term consequences.
Risk factors
Research on the causes of early-onset
conductdisorderhasidentiedawide
range of environmental risk factors as well
as an important genetic component. The
environmental risk factors include:
• socio-economic variables, such as large
family size, single parenthood, family
poverty and deprived neighbourhoods;
• parental characteristics, such as mental
illness, substance misuse, involvement
in criminal activity and low educational
attainment; and
• family relationships, such as harsh,
inconsistent or neglectful parenting,
physical or sexual abuse and family discord.
These risk factors tend to have a cumulative
effect. The likelihood of conduct disorder rises
progressively as an individual is subject to an
increasingnumberofadverseinuencesinearly
life (Murray et al., 2010).
Risk factors may also interact, with gene-
environment interactions being particularly
important. This means that genetically
determined differences between individuals
maycontroltheirsusceptibilitytoaspecic
environmental risk. For example, one study
found that early maltreatment was associated
with an increase of 24% in the probability of
conduct disorder among children at high genetic
risk, but only 2% among children at low genetic
risk (Jaffée et al., 2003).
11
Centre for Mental Health REPORT Building a better future
Ollie
Theirrstsonwasverylaidback,evenasa
baby. In contrast, Ollie was a poor sleeper
right from the start and felt much trickier
to manage. From the age of two and a half
he had frequent and severe tantrums. John
and Anna disagreed on how to deal with
his behaviour, frequently undermining
each other’s decisions. Ollie’s tantrums
worsened as he grew older, affecting his
performance at school and the whole family.
Anna and John were increasingly at a
loss about how they should manage him.
For his part, Ollie felt sad and singled
out; he felt his brother got preferential
treatment and was loved much more than
him. This made him all the angrier and
he became increasingly aggressive. On
three occasions, John was forced to leave
work early to help Anna contain Ollie’s
behaviour at home. On many occasions
family outings and holidays were cancelled.
Ollie often took out his anger on his more
placid brother and dominated him despite
his younger age. At school, teachers raised
concerns about his attention levels in the
classroom and his controlling and bullying
behaviour with other children. He often had
to be taken out of class or managed through
teaching assistant support.
Both parents were highly stressed. John
coped by spending more and more time
at work or away from the family; Anna
felt increasingly powerless, isolated and
trapped and began drinking heavily on her
own every evening to manage her anxiety
and distress.
As Ollie approached his mid-teens
he became increasingly angry at his
parents’ coldness – both with each
other and with him. He hated his
mother’s drinking and the arguments
which followed when she was drunk.
He hated himself even more and
regularly secretly self-harmed. His
school performance dramatically
deteriorated. He began staying away
from home as often as he could.
Initially, he stayed with family friends
but later lied and stayed with a series
of friends drinking and taking drugs.
He was expelled from school before
his exams for bringing drugs on to the
school campus. His parents refused to
have him home and after sleeping in
a young people’s crisis centre he was
then placed in local bed and breakfast
accommodation. He has been unable
to secure employment and is on
medication for depression (which he
frequently mixes with alcohol). He has
been arrested twice for drink-fuelled
assaults in public houses.
John and Anna’s
marriage became
strained after seven
years following the
birth of their second
child Ollie.
PUBLIC SECTOR BUDGETS THESE COSTS
FALL ON:
PLUS SOCIETAL COSTS:
including the
effects of bullying
Centre for Mental Health REPORT Building a better future
12
schoolwithnoqualications’.Itdoesnot
necessarily follow that all of the difference in
educational attainment can be attributed to
conduct disorder. For example, children with
this condition tend to come from more deprived
backgrounds than other children and also to
have below-average cognitive ability. Both of
these factors are known to result in poorer
educational outcomes, irrespective of whether
or not a child also has conduct disorder.
Depending on data availability, statistical
analysis can be used to take into account the
impact of these and other so-called confounding
variables and thus isolate a ‘pure’ conduct
disorder effect. Wherever possible, we use
adjustedndingsofthistype.
Outcomes in the early years
A distinguishing feature of early-onset conduct
disorder is not only its strong tendency to
persist over the life course but also the very
wide and diverse range of life domains that
are adversely affected by the condition. The
number of these domains is necessarily limited
in early childhood, but pervasiveness of impact
is still apparent, as shown by the evidence set
out below in six areas: mental health; physical
health; child protection; education; family
relationships; and peer relationships.
Mental health
The 2004 national survey of mental health
among children and young people shows that
7.7% of all children aged 5-10 suffer from some
kind of diagnosable mental health condition
(Green et al., 2005). A prevalence of 4.9% for
conduct disorder means that this is the single
most important mental health condition in
childhood and also the one which leads to the
most referrals to specialist child and adolescent
mental health services (NICE, 2013).
Clinical diagnosis of conduct disorder depends
not only on the scale and severity of behavioural
problemsbutalsoonevidenceofsignicant
distress and social impairment of the child.
In contrast to ‘internalising’ problems such as
3: Outcomes over the life course
Introduction
A very substantial body of evidence
demonstrates that early-onset conduct disorder
is associated with a wide range of adverse
outcomes, not only in childhood but throughout
the life course. This chapter provides a selective
review, with information presented separately
for outcomes in the early years (up to age 11), in
adolescence and in adulthood.
The most persuasive evidence on the enduring
consequences of childhood conduct disorder
comes from longitudinal data, particularly birth
cohort studies which track the experiences over
the life course of samples of individuals born in
the same year. This country has been a pioneer
in the development and use of such studies,
the longest-running relating to a nationally
representative cohort of children born in 1946.
Subsequent national studies cover samples
of children born in 1958, 1970 and 2000 and
these are supplemented by other, more local
studies such as the Avon Longitudinal Study of
Parents and Children (ALSPAC), which is tracking
a sample of over 14,000 children born in the
Avon area in the early 1990s. Similar studies in
other countries, particularly New Zealand, also
provide important data of particular relevance
for this report.
Data from birth cohort studies can be analysed
from two perspectives: looking forward from
childhood to adulthood, to answer a question
such as, ‘what proportion of children who
had early-onset conduct disorder went on to
become persistent offenders in later life?’; and
looking back from adulthood to childhood, to
answer the related but different question, ‘of
all those who became persistent offenders
in later life, what proportion suffered from
conduct disorder when they were children?’.
Both these approaches provide valid and useful
information, depending on the context in which
they are used.
Surveydatamayindicateandingsuchas,
‘children with early-onset conduct disorder
are twice as likely as other children to leave
13
Centre for Mental Health REPORT Building a better future
(Romeo et al., 2006). Even more strikingly, as
many as 40% had been admitted to hospital,
for an average stay of 8 days. These children
were hospitalised because their impulsive or
disobedient behaviour resulted in a range of
outcomes such as concussions, head injuries,
scalds and burns. A further quarter of the
children were taken to A&E departments for
similar reasons, attending on average twice
during the year.
The physical maltreatment of children by
parents is an important contributory cause
of conduct disorder, but it may also be a
consequence, as parents resort to physical
punishment as a means of dealing with bad
behaviour. Such a response not only risks
physical injury to the child but is also likely to
reinforce behavioural problems, by conveying
the message that aggression is a normal part
of family relationships and an effective way of
controlling others.
Child protection
Children with conduct disorder are at high risk
of being placed on child protection registers
and of being taken into care. For example, a
detailed study of child protection registers in
West Sussex found that children with conduct
disorder were 7.6 times more likely than other
children to be registered, even after allowing
for potential confounding factors such as family
socio-economic status (Spencer et al., 2005).
This was a much higher risk than for children
with any other mental or physical health
condition,includinglearningdifcultiesand
speech and language disorders. A national
study carried out in 2002 found that the
prevalence of conduct disorder in looked-after
children aged 5-10 was 36.5%, which is 7.5
times as high as the prevalence of this condition
among children generally (Meltzer et al.,
2003a).
Child protection is most commonly needed
because of parental abuse or neglect. As already
noted in the case of physical maltreatment, the
causal connection with conduct disorder may
runinbothdirectionsandtheguresgiven
above should be interpreted in this light. In
some cases parental abuse or neglect may be
the cause of conduct disorder while in others it
may be precipitated by it. The outcome in terms
anxiety, conduct disorder may be described as
an ‘externalising’ condition in which the child’s
distress takes the form of behaviours such
asaggressionordeancewhichdirectlyand
adversely affect others. It is all too easy in this
context to focus on the antisocial consequences
of conduct disorder and to ignore the distress
of the child which leads to these outcomes. Few
would deny that a child’s wellbeing and quality
of life may be severely compromised by anxiety
or depression; the same is also true of conduct
disorder and this is a major but often forgotten
cost of the condition.
The 2004 survey of childhood mental health
shows that more than a third of all children
with conduct disorder have another psychiatric
disorder as well (Green et al., 2005). Their
numbers divide roughly equally between those
who have conduct disorder combined with an
emotional disorder (most commonly anxiety)
and those who have conduct disorder along
withattentiondecithyperactivitydisorder
(ADHD). According to NICE, these co-existing
conditions are often missed when a diagnosis of
conduct disorder is being made (NICE, 2013).
Physical health
A high proportion of children with conduct
disorder suffer from physical and developmental
problems and are also very prone to injury.
It was found in the 2004 national survey that
the parents of children with conduct disorder
were more than three times as likely as other
parents to assess their child’s health as ‘fair’ or
‘bad’(17%comparedto5%)andsignicantly
less likely to say that it was ‘very good’ (50%
compared to 70%) (Green et al., 2005). Physical
or developmental problems were reported by
parents in about two-thirds of all children with
conduct disorder, with particularly high rates of
prevalence, in comparison with other children,
for speech and language problems, coordination
difcultiesandbed-wetting.
A detailed study of 80 children aged 3-8 who
were referred to specialist mental health
services for severe behavioural problems
found that during the previous 12 months a
clear majority (71%) had been taken to the
GP for reasons connected to their behaviour
Centre for Mental Health REPORT Building a better future
14
of child protection is the same but the pathways
to it may be different.
Another reason for child protection relating to
conduct disorder is parental inability to cope.
Someparentssimplyndtheirchild’sbehaviour
overwhelming and give up the child to be cared
for by the local authority (NICE, 2013).
Education
Children with conduct disorder make additional
demands on educational services from the
outset. For example, the study of 3-8 year-olds
by Romeo et al. cited above found that two-
thirds of parents made extra use of nursery
services because of their child’s behaviour and
a third of the children had been seen by an
educational psychologist and were receiving
special educational provision.
More generally, the 2004 national survey
of child mental health found that 52% of all
children with conduct disorder were reported
by teachers as having special educational
needs, compared with 15% of children with no
disorder (Green et al., 2005). The same source
also found that well over half of children with
conduct disorder were rated as having ‘some’
or‘marked’difcultywithreading,mathsand
spelling, which is more than twice the number
among those with no disorder. And 59% of
children with conduct disorder were assessed
as being behind by at least a year in their overall
intellectual development, against 24% of other
children.
Children with conduct disorder are often
disruptive in class, which hampers the learning
of other children and is a major cause of stress
for teachers. Bullying is covered in the section
below on peer relationships.
Family relationships
Children with conduct disorder place great
strain on their families, as their challenging
behaviour requires constant vigilance and
makes it harder for parents to carry out everyday
tasks. The study by Romeo et al. found that
this added eight hours a week to household
tasks, not including the time spent on repairing
damage caused by the child. It was also noted
that parents’ employment was sometimes
disrupted by having to take time off work as a
result of their child’s behaviour, particularly
when the child was sent home from school.
Siblings may also be affected, as a child’s
disruptive behaviour can be expected to have a
negative effect on the whole family. The 2004
national survey found that family functioning
was rated as ‘unhealthy’ in 42% of families
containing a child with conduct disorder,
compared with 17% of other families (Green et
al., 2005). The same survey also found, on the
basis of responses to a validated mental health
questionnaire (GHQ-12), that no fewer than 48%
of parents of children with conduct disorder
were assessed as having a severe emotional
problem, against 23% of other parents. Again
thisndingneedstobeinterpretedwithcaution
because of possible reverse causation, i.e.
depression or anxiety in parents may develop as
aresponsetoachild’sdifcultandaggressive
behaviour but may also be a prior contributory
cause.
Peer relationships
Children with severe behavioural problems
tend to have poor peer group relationships.
For example, the 2004 national survey found
that these children are around three times
morelikelythanaveragetohavedifcultyin
making friends and around eight times more
likelytohavedifcultyinkeepingthem.About
20% of children with conduct disorder have no
friends or only one, compared with 5% of other
children.
There are also strong links with bullying,
although the relationship here is quite complex,
because although many bullies do suffer from
seriousbehaviouraldifcultiesthereisalso
evidence that the highest rates of behavioural
problems are found in so-called bully/
victims, i.e. children who are simultaneously
perpetrators and victims of bullying. This was
oneofthendingsofastudybyWolkeet al.
(2000) which collected detailed information on
bullying in 31 primary schools in Hertfordshire
and North London. Overall, the study found
that4.3%ofallchildrencouldbeclassiedas
bullies, 39.8% as victims and 10.2% as bully/
15
Centre for Mental Health REPORT Building a better future
victims. Other research has demonstrated
that being bullied can be highly stressful and
psychologically damaging, with important long-
term consequences for educational and labour
market performance as well as mental health
(Copeland et al., 2013).
Outcomes in adolescence
As noted in Chapter 2, early-onset conduct
disorder persists into adolescence and beyond
in about half of all cases. At the same time the
overall prevalence of conduct disorder rises
from 4.9% among children aged 5-10 to 6.6%
among those aged 11-16 (Green et al., 2005).
(There is also a change in the gender balance: in
the younger age group, for every ten boys with
conduct disorder there are four girls, whereas in
the older age group the ratio is ten to six.) The
explanation for this apparent contradiction is
that the fall in numbers associated with non-
persistence among children with early-onset
conduct disorder is more than offset by the
increase associated with the development of
conduct disorder which begins in adolescence.
Many of the behaviours and associated
outcomes of conduct disorder are common
across the early-onset and adolescent-onset
sub-types, but the causal antecedents are
different, disorder of adolescent onset is
much less likely to persist into adulthood and
there is also some evidence that early onset is
associated with greater general severity of the
condition. Cross-section surveys showing the
outcomes of conduct disorder in adolescence
rarely distinguish between cases of early
and adolescent onset and, while some use is
made of such data below, preference is given
wherever possible to longitudinal studies
which provide a direct link between early-onset
conduct disorder and outcomes in adolescence
for the same individuals.
Mental health
A study based on birth cohort data has
reported that 36% of all those in the sample
who had severe behavioural problems at age
8 met diagnostic criteria for at least one major
depressive episode between the ages of 16
and 18 (Fergusson & Lynskey, 1998). This is
after adjustment for a wide range of potentially
confounding variables and compares with a
prevalence rate for depression of 17% among
those with no conduct problems in childhood.
Similarly, 26% of those with conduct disorder
atage8wereclassiedashavingananxiety
disorder between the ages of 16 and 18,
compared with 14% among those with no
problems in childhood. And 11% of those with
early-onset conduct disorder attempted suicide
during the period from ages 16 to 18, against
2.5% among those with no early problems.
All of these comparisons suggest that severe
conduct problems in childhood are associated
with substantially increased risks of emotional
as well as behavioural problems in adolescence.
The same study reports even more pronounced
risks in relation to alcohol and drug use. Thus
it was found that 31% of those with early-onset
conduct disorder met criteria for alcohol abuse
or dependence at ages 16-18 compared with
13% among those with no childhood problems,
whilethecorrespondingguresfordrugmisuse
or dependence (mainly cannabis) were 27% and
9%.
Physical health
Early-onset conduct disorder is associated with
a wide range of risky behaviours in adolescence
that may compromise physical health. Alcohol
and drug misuse fall into this category and
so too does smoking, with the 2004 national
survey of mental health in children and young
peoplendingthat30%ofallthosewith
conduct disorder at ages 11-16 were regular
smokers, compared with 5% of those with no
disorder (Green et al.,2005).Theseguresdo
not, however, distinguish between cases of
early onset and adolescent onset, nor do they
adjust for confounding variables. Adjusted birth
cohort data suggest a less dramatic but still
signicantdifference,withonestudynding
that 22% of those had conduct disorder at age 8
were nicotine dependent at age 18, against 11%
of those with no early behavioural problems
(Fergusson & Lynskey, 1998).
Conduct disorder is also strongly implicated in
sexual health risks and teenage parenthood.
One study, based on birth cohort data, divided
the sample into four equal-sized groups
according to the severity of behavioural
problems at ages 5-11 and found that those
with the greatest problems in childhood were
Centre for Mental Health REPORT Building a better future
16
2.2 times as likely as those with no problems to
engage in under-age sex and 1.9 times as likely
toengageinriskysexualbehaviour,denedas
having three or more partners before age 21
and “never or only sometimes” using a condom
(Ramrakha et al., 2007). Another study found
that children with conduct disorder at ages 7-9
were subsequently nearly three times more
likely become teenage parents than those with
no early behavioural problems (Fergusson et al.,
2005).
Education
By the time they reach secondary school the
behaviour of children with conduct disorder
has already put them at an educational
disadvantage. Academic failure then appears
to contribute to worsening problems as
the children reject regular activities and
schoolmates and associate more often with
like-minded peers (Masten et al., 2005). As this
example illustrates, the adverse consequences
of conduct disorder may often interact with each
other, with failure in one domain aggravating
failure in another, leading to a cascade or
downwardspiralofdifculties.
School attendance among children with
conduct disorder is often disrupted because
of exclusions and unauthorised absences
including truancy. One study found that having
persistent conduct disorder increased the
likelihood of being excluded from school by
almost 25 times compared with no disorder,
the most common reasons for exclusion being
aggressive or violent behaviour, being rude or
disrespectful to teachers, stealing or vandalism
and general bad behaviour (Meltzer et al.,
2003b). Other evidence suggests that children
with conduct disorder are more than three times
as likely as other children to have unauthorised
absences (28% against 8%), with over seven
timesasmanybeingidentiedbyteachersas
having played truant (Green et al., 2005).
All of these factors contribute to poorer
educationalattainment,withonestudynding
that nearly a third (31%) of all children who
had conduct disorder at age 8 leave secondary
schoolwithoutanyqualications(Fergusson
& Lynskey, 1998). This compares with 17%
among those with no behavioural problems in
childhood and is estimated after taking into
account confounding variables such as cognitive
ability and socio-economic background.
Criminality
Criminal activity is strongly age-related, rising
rapidly during adolescence to a peak at age 17,
when the number of offenders as a proportion
of the population reaches 6% among males and
2% among females, and then falling steadily
back (ONS, 2009). However, more than half of
those who offend do so only once and the great
bulk of crime is concentrated in the hands of
a relatively small minority of persistent and
prolicoffenders.
Thebestpredictorofprolicoffendingisthe
ageatwhicharstoffenceiscommitted.For
example, evidence from the Cambridge Study
in Delinquent Development, which has been
tracking a sample of 411 boys born in inner
London in 1953, shows that among all those in
thesamplewhocommittedtheirrstoffenceat
ages 10-13, no fewer than 91% became repeat
offenders, compared with only 37% of those
whorstoffendedatages21-30(Farringtonet
al., 2006). This group of very young offenders,
representing 8% of the overall sample,
accounted for 39% of all crimes recorded in
the study. Such a pattern of offending strongly
suggests a link with early behavioural problems
andthisisconrmedbyevidenceinthe
Cambridgestudythat90%ofprolicadolescent
offenders had conduct disorder at age 8
(Farrington, 1995).
Another longitudinal study has shown that,
looking forward from childhood, 18% of all
those in the survey who had conduct disorder at
age 8 received a court conviction for any offence
during the 12-month period from age 17 to 18
(Fergusson & Lynskey, 1998). This proportion is
nearly seven times as high as among those who
had no behavioural problems in childhood.
17
Centre for Mental Health REPORT Building a better future
medication for a mental health problem and
19 times as likely to have received inpatient
treatment in a psychiatric hospital (Odgers et
al., 2007).
Physical health
Alcohol and drug misuse along with a range
of other risky behaviours are associated with
impaired physical health and a commensurate
increase in the use of physical health services.
Again using data from the study by Odgers
et al., the odds of various physical health
outcomes at ages 26-32 for someone with
early-onset conduct disorder compared with no
early problems are increased as follows: risk
of cardiovascular disease 1.6 times, chronic
bronchitis symptoms 3.1 times, nicotine
dependency 8.7 times, gum disease 3.5 times,
serious injury 2.0 times, non-sports-related
injury 3.6 times and hospitalisation 3.4 times.
There are also increased risks of premature
mortality. For example, a Swedish study found
that children with severe behavioural problems
were subsequently 5.5 times more likely to die
before age 30 than those with no problems,
from a range of causes including suicide,
homicide, drug overdoses and accidental
poisoning as well as illness (Kratzer & Hodgins,
1997). Another study, based on the 1958
British birth cohort survey, divided the sample
into four equal-sized groups according to the
extent of behavioural problems at ages 7 and
11 and found that the probability of death by
age 46 was 2.3 times greater in the highest
quartile (i.e. the 25% of the sample with the
worst problems in childhood) than in the lowest
quartile (Jokela et al.,2009).Similarndings
relating to the risk of death or disability by
age 48 are reported in a study based on the
Cambridge Study in Delinquent Development
(Shepherd et al., 2009).
Labour market outcomes
Lackofeducationalqualications,continuing
behavioural problems and in some cases a
criminal record mean that people who had
conduct disorder as children are generally at
increased risk of doing badly in the labour
market. For example, a number of studies show
that those with childhood conduct disorder
Outcomes in adulthood
Many of the adverse behaviours and their
associated outcomes described above persist
into adult life. As before, in illustrating the scale
of these effects, use is made wherever possible
ofndingsderivedfromlongitudinalstudies
which provide a direct link between conduct
disorder in childhood and its consequences in
adulthood for the same sample of individuals.
Mental health
Uniquely among childhood mental health
conditions, early-onset conduct disorder is a
risk factor for all major psychiatric disorders
in adults and one study has estimated that
if conduct disorder in childhood could be
prevented, the prevalence of mental illness
among adults would be reduced by 25-50%
(Kim-Cohen et al., 2003).
Antisocial personality disorder (ASPD) is the
condition in adults most strongly linked with
conduct disorder in children. Indeed, the
former is best seen as a direct continuation of
the latter and in the Diagnostic and Statistical
Manual of Mental Disorders produced by the
American Psychiatric Association evidence of
symptoms of conduct disorder in childhood is
actually a requirement for diagnosis of ASPD.
The conversion rate from childhood conduct
disorder to adult ASPD varies from 40% to 70%
depending on the study (NICE, 2013).
Risks of other mental illnesses in adulthood
are less pronounced but still substantial. For
example, birth cohort data suggest that males
with early-onset conduct disorder are about
three times as likely as those with no early
problems to suffer from depression or anxiety in
their late 20s and early 30s, eight times as likely
to have post-traumatic stress disorder, twice
aslikelytobealcoholdependent,vetimesas
likely to be drug dependent and 25 times as
likely to have attempted suicide (Odgers et al.,
2007).
These higher rates of psychiatric morbidity
carry through into greater use of mental
health services. For example, in the study just
described, members of the sample with early-
onset conduct were four times more likely than
those with no early problems to have received
Centre for Mental Health REPORT Building a better future
18
spend more time out of work than others
when in their 20s and 30s, are more likely to
bedependentonwelfarebenetsand,when
employed, are more likely to be in less-skilled
and lower-paid jobs (Fergusson & Horwood,
1998).Lackofeducationalqualicationsis
probably the critical mediating variable in
explainingthesendings.Forexample,atthe
end of 2009, only 56% of people with no or
lowqualicationswereinpaidworkcompared
with77%ofpeoplewithallotherqualications
(Barrett, 2010).
One contrary result is reported in a study based
on the 1970 British birth cohort study which
ndsthatseverebehaviouralproblemsat
age 10 are associated for males, though not
females, with higher than average earnings
at age 30 (Knapp et al., 2011). As the authors
note,thisisanunexpectedndingandone
which has not been reported elsewhere in the
research literature. A possible explanation is
that elements of antisocial behaviour such as
risk-taking and aggression may be adaptive
in some workplace contexts. Further research
is needed to test this hypothesis and for the
moment the consensus in the literature is that
the labour market consequences of conduct
disorder in childhood are very largely negative.
Homelessness
Various studies have shown that people with
a history of severe behavioural problems are
at much higher risk of homelessness than
the general population. One found that 20%
of all men with early-onset conduct disorder
had some experience of homelessness
(including being taken in by others) between
the ages of 26 and 32, which represented a
ten-fold increase in the odds of homelessness
compared with men who had no early history of
behavioural problems (Odgers et al., 2007).
Crime
The strong links between conduct disorder and
criminality noted in adolescence are largely
maintained when individuals are in their 20s
and 30s. Thus it has been found in one study
that people who had conduct order at ages 7-9
were 3.2 times more likely than those with no
early problems to engage in property-related
offending at ages 21-25, 4.1 times more likely
to engage in violent offending and no less than
19 times more likely to have served a prison
sentence (Fergusson et al., 2005). Another
study based on birth cohort data found that the
10% of the males in the sample who had severe
behavioural problems in childhood accounted
for 72% of the time spent in prison up to age 32
by all members of the sample combined (Odgers
et al., 2007).
Basedonsuchndings,ithasbeenestimated
that 30.0% of all crime in this country is
committed by people who had conduct disorder
as children (Centre for Mental Health, 2009).
Taking into account the independent effects on
offending of differences in cognitive ability and
socio-economic background, the proportion
of crime attributable to conduct disorder is
estimated at 21.7%.
Personal relationships
The psychosocial functioning of adults who
had conduct disorder as children is generally
poor and this carries though into the sphere of
personal relationships. In particular, intimate
relationships are much more likely than average
to be short-lived and characterised by abuse
and violence. One study found that adults who
had conduct disorder at ages 7-9 were twice as
likely as those with no early problems to have
multiple (10+) sexual partners at ages 21-25
and more than three times as likely to have
been involved in inter-partner violence in the 12
months up to age 25 (Fergusson et al., 2005).
19
Centre for Mental Health REPORT Building a better future
Another study found that at age 32 men with
conduct disorder since childhood were ten times
more likely than those with no early problems
tobeinictingcontrollingabuseontheir
partners (humiliating, restricting, intimidating
or stalking) and, relative to their numbers in
the sample, these men accounted for six times
their share of convictions for rape and violence
againstwomen(Mofttet al., 2002).
According to another study, girls with conduct
disorder were, at 21 years of age, three times
more likely to have been victims of partner
violence and four times more likely to have been
in a mutually violent relationship in the past
year than their peers (Bardone et al., 1996).
Inter-generational effects
There is a good deal of evidence to suggest that
children with conduct disorder are likely in their
turn to become the parents of children who
themselves display the same severe behavioural
problems (see for example Jaffée et al., 2006,
and Pajer, 1998). This continuity across the
generationsreectsanumberofinteracting
inuences.
First, as already seen, there is an important
genetic component in the development of
early-onset conduct disorder. Second, there is
evidence of assortative mating between people
with behavioural problems, which compounds
the genetic risk and also exposes children to
adverseinuencesinthehomeenvironment
such as relationship violence (Krueger et al.,
1998). Third, there is evidence that parents
with a history of behavioural problems are more
hostile and harsh in their parenting styles than
other parents (Bosquet & Egeland, 2000) and
also more likely to abuse their own children
(Verona & Sachs-Ericsson, 2005). Both of these
are major risk factors for the early development
of conduct disorder. Finally, parents with
behavioural problems are likely to expose their
children to a range of wider environmental
factors that are also implicated in the
development of early-onset conduct disorder
such as low socio-economic position and living
in deprived neighbourhoods.
It was noted in Chapter 2 that the risk factors
for conduct disorder tend to have a cumulative
effect, with the likelihood of disorder rising
progressively as an individual is subject to an
increasingnumberofadverseinuencesinearly
life. The children of parents who themselves
have a history of severe behavioural problems
are among those most likely to face such
multiple risks.
Centre for Mental Health REPORT Building a better future
20
Tommy
Tommy was always challenging to manage
from an early age. He was aggressive with
his siblings, often biting them and making
them cry. He often repeatedly banged his
head against the wall of his room when in a
tantrum.
At pre-school, his nursery worker regularly
called Lisa in at the end of the school
daytohighlightdifcultieswithTommy’s
behaviour. He was aggressive, wouldn’t
share, spat at other children and found
itdifculttosettle.Athome,Lisawas
becoming increasingly stressed. She had
given up part-time work and suffered
regular panic attacks. She became harsher
in her disciplining of Tommy although this
justmadehimmoredeant.Tommywas
constantly on the go; he ended up in A&E
after hurling himself downstairs during
a tantrum. On this occasion, all children
were placed on the at-risk register due to
evidence of bites and bruising and because
of general concern over the children’s
welfare.
Duringprimaryschool,hewasdifcult
to teach, needing regular input from
a teaching assistant to help focus his
attention. His understanding of tasks was
poor. He had few friends and was often
both bullied and removed from class.
Eventually, he received a statement of
educational needs although later this was
not transferred to his secondary school.
During secondary school, he found that he
was able to gain friends and feel good about
himself through being the ‘class clown’
and through dominating others. He and
a small group of friends systematically
targeted other pupils and used physical
and psychological force to extract money
which was spent on cannabis and other
drugs.
By the age of 15 he had been temporarily
excluded twice for violence against
teachers and other pupils. He attended
the local Pupil Referral Unit but eventually
stopped attending just before he was
due to sit his exams. He gained no
qualications.Hewasveryaggressive
with his mother, sleeping away from
home and mixing with older friends from
his estate. The group regularly caused
damage to local shops, sabotaged lifts
andleftgraftionthestairwellsoflocal
ats.Hehadfrequentbrusheswithlocal
housingofcersandthepolice.Someof
his friends were known to be involved in
local drug dealing. His mother repeatedly
asked for help from social services but
the children had long since been removed
from child protection registers and
Tommywasconsideredinsufcientlyat
risk to be in need of support from social
services.
By the age of 16, he had cautions for
offences of criminal damage and theft.
By 18, he had children with two separate
partners. These relationships ended
following violence. His sleeping patterns
were out of synch and he was routinely
using cocaine and alcohol. He had been
unable to hold down a job, leaving after
only a few weeks. He was regularly in
ghtsinlocalpubsendingupeitherin
A&E departments or in police custody. He
condedtoonehealthworkerthathe‘felt
better’ when he felt physical pain.
Lisa had three
children. Her second
child, Tommy, was
born during a violent
break-up from her
partner.
21
Centre for Mental Health REPORT Building a better future
Eventually he was arrested for an
unprovoked and violent attack on someone
in a pub with a broken glass. He was
sentenced to 18 months in youth custody
transferring to adult prison at 18. Health
screening picked up signs of depression.
After release from prison, he lived in a
shabby bedsit and was unable to get work.
He became increasingly reliant on alcohol
and drugs and his pattern of violence
continued with three further returns
to prison. A mugging of a woman at a
cashpoint to get money for drugs resulted in
a further lengthy period of imprisonment.
PUBLIC SECTOR BUDGETS THESE COSTS
FALL ON:
PLUS SOCIETAL COSTS:
including the
effects of bullying
SOCIAL SERVICES
Centre for Mental Health REPORT Building a better future
22
Introduction
The evidence summarised above demonstrates
beyond doubt that the consequences of early-
onset conduct disorder are both wide-ranging
and long-lasting. One way of highlighting
the overall scale of these adverse outcomes
is to express as many of them as possible in
monetary terms and then combine these into
a single estimate of lifetime costs. It should
already be apparent that any such estimate is
likely to be extremely high, whether measured
from the perspective of the individuals who
suffer from conduct disorder or from wider
perspectives such as the impact on the
Exchequer or on society as a whole. Various
attempts have been made to estimate the long-
term costs of conduct disorder, usually within
specicagerangesratherthanacrossthefull
life course, but all of these are likely to under-
estimate the true costs by a substantial margin.
This is for a variety of reasons.
First, many of the adverse outcomes of conduct
disorderaredifculttoquantify,letalonevalue
inmonetaryterms.Aspecicexamplemight
be classroom disruption in schools and its
impact on teachers and other pupils, but this
is merely one instance among many of a wide
range of antisocial behaviours which in one
way or another cause distress or unhappiness
to others. And, as already noted, individuals
with conduct disorder may themselves
suffersignicantdistressbecauseofthe
consequences of their behaviour, such as poor
personal relationships and social isolation.
These wide-ranging impacts on the quality of
life are inherently hard to measure and value.
Second, there are some effects of antisocial
behaviour on third parties which can be
quantiedbutwhicharerarelyifeverincluded
in the estimated costs of conduct disorder.
For example, bullying is a common behaviour
by people with severe conduct problems,
particularly – but not only – in childhood.
Evidence from longitudinal studies shows
that being bullied can have serious long-term
economic consequences, with one study
estimating that, after taking into account a wide
rangeofotherinuences,thelifetimeearnings
of a victim of serious bullying are reduced by
around £50,000 on average (Hummel et al.,
2009, drawing on data in Brown & Taylor, 2008).
This is clearly an attributable cost of conduct
disorder but not one which features in any
published estimate of overall costs. Moreover,
thegureof£50,000relatestolostearningsfor
a single victim of bullying and, as the evidence
shows, there are many more victims of bullying
than there are perpetrators.
Third, most estimates of the costs of conduct
disorder are based on surveys in which the
numbers of people with this condition are
relatively small. This raises the possibility
that outcomes of low probability but very
high unit cost may be altogether missed. For
example, children with conduct disorder are
at considerably greater risk of being taken
into care than other children, but in absolute
terms the risk is still relatively small. On the
other hand, each single instance is extremely
expensive, with the cost of a child looked after
in a local authority care home being estimated
at over £150,000 a year (Curtis, 2012).
Finally, most estimates of the costs of conduct
disorder are based on ‘bottom-up’ methods
of calculation which use individual-level data
collected in surveys to derive an overall national
average or total. For example, a recent study of
the economic impact of childhood psychiatric
disorders on public services has estimated that
the overall national cost of children’s mental
health services amounts to around £64 million
a year in 2008 prices (Snell et al., 2013). This is
calculated by combining survey data on service
use (e.g. numbers of psychiatric inpatients
stays) with independently estimated unit costs
for each type of service and then grossing up to
anationaltotal.However,a‘top-down’gure
based on budgetary and accounting data is
published by the Department of Health and this
shows that aggregate spending on children’s
mental health services in 2008/09 was £680
million (DH, 2013). In other words, the ‘bottom-
up’ approach produces an estimate of NHS
4: The costs of behavioural problems
23
Centre for Mental Health REPORT Building a better future
2012/13 prices. There is, however, much less
agreement on public service costs, as two of the
studies put these at around £2,000 a year while
the other gives an estimate of over £11,000
a year. The latter is based on a very small
sample and the cases it includes may also be of
above-average severity and therefore cost. On
the other hand, the other two studies include
the one by Snell et al. which, as noted above,
appears to under-estimate NHS costs by a very
large margin. A simple average of the three
studies suggests that the annual cost of conduct
disorder which falls on public services is around
£5,000 per child.
All three studies agree that the largest share of
public service costs for children with conduct
disorder is borne by the education sector. Again
taking a simple average, this comes out at
around £3,000 a year per child.
This is over and above the normal cost of
schooling and covers both the costs associated
with special educational needs and extra costs
falling on frontline education services (e.g. extra
help provided in the school by teachers and
teaching assistants). Averaged over the three
studies, costs falling on the NHS are around
£1,400 a year and on social services around
£600 a year.
£3000
THE ANNUAL
COST OF
CONDUCT
DISORDER TO
EDUCATION *
*ESTIMATE OF THREE STUDIES
spending on children’s mental services which is
less than a tenth of a known national total.
Various reasons may explain such
discrepancies. For example, there may be
differences in the coverage of services in the
two estimates. Information on service use
collected in surveys often depends on the
ability of respondents to recall past events,
which tends to result in under-estimation. And
the estimates of unit costs used in ‘bottom-up’
calculations may not always capture all relevant
categories of expenditure. Whatever the reasons
in any particular case, wider evidence suggests
that ‘bottom-up’ calculations are always likely to
result in the under-estimation of costs.
Costs in childhood
Three British studies provide estimates of the
costs of conduct disorder in childhood (Knapp
et al., 1999; Romeo et al., 2006; and Snell
et al.,2013).Allofthesegiveguresforthe
costs borne by public services (mainly health,
socialservicesandeducation),whiletherst
twoalsoincludeestimatesofthequantiable
costs falling on families such as time off work.
For the purposes of comparison, costs in all
three studies have been converted to a common
2012/13 price base, with costs from earlier
years being assumed to rise in line with general
ination.(Thisisaconservativeassumption,
because there is a well-established tendency for
health and other public service costs to increase
at a somewhat faster rate over time than prices
generally.)
Thetwostudiesgivingguresforcostsfalling
on families come up with broadly similar
estimates, each suggesting that on average
any family which includes a child with conduct
disorder faces extra costs, including lost
earnings, amounting to around £6,500 a year in
£5000
THE ANNUAL
COST OF
CONDUCT
DISORDER
TO PUBLIC
SERVICES *
*ESTIMATE OF THREE STUDIES
PER
CHILD
£1400
THE ANNUAL
COST OF
CONDUCT
DISORDER TO
THE NHS *
*ESTIMATE OF THREE STUDIES
PER
CHILD
PER
CHILD
Centre for Mental Health REPORT Building a better future
24
Costs of crime
The study by Scott et al. suggests that the
criminal justice system in this country incurs
costs of over £3,000 a year as a result of
crimes committed during adolescence and
early adulthood by each individual who had
conduct disorder at age 10. The high cost of
adolescent crime is also highlighted in a recent
reportbytheNationalAuditOfce,onthecost
to the criminal justice system of a cohort of
young offenders (NAO, 2011). This examined
83,000 young offenders who committed their
rstprovenoffencein2000andanalysedthe
subsequent offending behaviour of this cohort
for the period 2000-2009. It found that on
average each young offender cost £8,000 a year
to the criminal justice system and that each of
the most costly 10% cost £29,000 a year. As
noted earlier, there is evidence that around 90%
ofallprolicadolescentoffenders,i.e.thosein
the most costly 10%, had conduct disorder as
children.
Highastheseguresare,theyrepresentonly
a fraction of the overall costs of crime, as they
make no allowance for costs falling outside
the criminal justice system. Comprehensive
estimates of the costs of crime, both in total
andbytypeofoffence,wererstpublishedby
theHomeOfcein2000(Brand&Price,2000)
andpartiallyupdatedveyearslater(Dubourg
et al., 2005). These show, for example, that
the total cost of crime in England and Wales
in 1999/2000 was around £60 billion. Only
about 20% of the aggregate cost was incurred
by the criminal justice system, with most of
the remainder falling on the victims of crime,
including the value of stolen or damaged
property, losses in earnings resulting from
crime-related injuries and an imputed value
of the emotional and physical impact of crime
on victims. The estimate of over £3,000 a year
in Scott et al. for costs falling on the criminal
justice system should therefore be grossed up
to over £15,000 a year for the societal cost of
crime committed up to age 28 by an average
individual who had conduct disorder at age 10.
Costs in adolescence and early
adulthood
The most detailed estimates of the costs of
conduct disorder in adolescence and early
adulthood are those given in a study by Scott
et al.(2001),whichprovidesguresforpublic
sector costs incurred by a sample of children
followed up between ages 10 and 28. The
sample is divided into three groups, covering
those with conduct disorder at age 10, those
with sub-threshold conduct problems at the
same age and those with no problems.
Measured in 2012/13 prices, the average total
cost of public services used by individuals over
the period from ages 10 to 28 was as follows:
• individuals with conduct disorder at age 10:
£95,926 per head
• those with sub-threshold conduct problems
at age 10: £33,324 per head
• those with no problems at age 10:
£10,170 per head.
Measured on a net basis, i.e. over and above
the costs incurred by someone with no conduct
problems at age 10, the aggregate public sector
cost of conduct disorder up to age 28 works
out at £85,756 per individual, or £4,764 a
year, while the equivalent cost of sub-threshold
conduct problems is £23,514, or £1,286 a year.
Disaggregation of the total for conduct disorder
shows that 67% of the extra costs fall on the
criminal justice system, 18% on the education
sector and 13% on health and social services.
A similar study in the US has estimated the
costs of conduct disorder over a shorter time
period, from ages 12 to 18 (Foster & Jones,
2005). Again using a ‘no problems’ baseline,
thisndsthatthepublicsectorcostsofconduct
disorder in adolescence work out at £3,369 a
year, measured in 2012/13 UK prices. This is
notdissimilartotheequivalentguregivenin
Scott et al., but the breakdown of the total is
different, with the US study showing 24% of
the extra costs falling on the criminal justice
system, 22% on the education sector and as
much as 54% on health services.
25
Centre for Mental Health REPORT Building a better future
25
Taking this approach a stage further, a US study
has sought to measure the lifetime costs of
crimeassociatedwithprolicoffending(Cohen
& Piquero, 2009, updating Cohen, 1998).
Prolicoffendersinthisstudyaredenedas
those who commit six or more offences over
the course of a criminal career. Longitudinal
evidence in the US suggests that this group
represents about 15% of all offenders and
is responsible for half of all recorded crime.
Crime costs include criminal justice service
costs, costs to victims and lost productivity of
offenders who are imprisoned.
Measured on this basis, it is estimated that total
crime-relatedcostsforasingleprolicoffender
are in the range $2.1-$3.7 million (2007
dollars) when discounted back to birth. This is
equivalent to about 45-80 times annual GDP per
head in the US. Applying the same multiples
to UK GDP per head, it may be calculated
that in this country the lifetime costs of crime
committedbyasingleprolicoffenderareinthe
range £1.1-£1.9 million.
The average lifetime cost imposed on society
byaprolicoffenderisnotofcoursethesame
as the average lifetime cost of crime committed
by every individual with early-onset conduct
disorder. Not all children with serious conduct
problems go on to offend and, among those who
do,onlyaminoritybecomeprolicoffenders.
For example, using imprisonment as a proxy
forprolicoffending,birthcohortdatashow
that among all people with conduct disorder
at ages 7-9 only 14% were imprisoned at any
time between ages 17 and 25, the peak time for
criminal activity (Fergusson et al., 2005). On the
other hand, looking back rather than forward,
itremainsthecasethatalmostallprolic
offenders had conduct disorder in early life.
Lifetime costs
A broad-based estimate of the lifetime costs
of conduct disorder, covering not only crime
but also adverse outcomes in adult life relating
to mental illness, drug misuse, smoking,
suicide and unemployment, is given in Friedli &
Parsonage (2007). This combines information
on adult outcomes derived from birth cohort
data with costings based on relevant sources
suchastheHomeOfceguresforthecosts
of crime. The study estimates that the overall
lifetimecostoftheidentiedadverseoutcomes
among people who had early-onset conduct
disorder is around £260,000 per case in
2012/13 prices, while the lifetime cost among
those who had sub-threshold behavioural
problems is put at around £85,000 per case. In
each case the point of comparison is given by
the outcomes in adult life experienced by people
who had no conduct problems in childhood. Of
the overall lifetime costs of conduct disorder,
71% are attributable to costs relating to crime,
13% to costs associated with adult mental
illness and 7% to the costs of adverse labour
market outcomes.
Each one-year cohort of children in England
between the ages of 5 and 10 includes about
600,000 individuals, of whom around 30,000
meet the criteria for a diagnosis of conduct
disorder. If the lifetime cost of conduct disorder
is £260,000 per case, then the aggregate cost
of this condition for a single one-year cohort is
nearly£8billion.Sizeableasthisgureis,itis
almost certainly an under-estimate.
£15,000
THE ANNUAL
SOCIETAL
COST OF CRIME
COMMITTED UP
TO AGE 28*
*BY AN INDIVIDUAL WHO HAD
CONDUCT DISORDER AGE 10
Centre for Mental Health REPORT Building a better future
26
Alyssa
This depression was not picked up until
after her third pregnancy. She had also
been in a violent relationship which
resulted in her children being placed on
the child protection register for three
years. Alyssa had always been the stronger
more aggressive of the twin girls. She was
considered ‘a handful’ by her mother from
an early age; her mother struggled to cope
on her own. During primary school, Alyssa
was sparky and quick to learn but was
often in trouble for her behaviour and for
bullying other girls.
By the age of 12, Alyssa’s mother was
unable to cope with her daughter’s
behaviour at home or at school. She
decided her daughter needed a fresh start
and arranged for Alyssa to live with her
natural father. His working patterns meant
that she was poorly supervised after
school. She began spending more time
with older friends and also began self-
harming. Within months the placement
with her father broke down, with Alyssa
calling the police after coming to blows
with her father and new stepmother. She
returned to her mother and sister but
struggled to settle back into school and
continued to have frequent rows at home.
Her school performance was poor with
ongoing complaints from the school to
Alyssa’s mother both about her behaviour
and about bullying.
Alyssa’s mother was told that she was
regularly mixing with older men in the town.
By the age of 14 she was staying away from
home; her mother felt unable to control her
behaviour and her mother’s mental health
also spiralled out of control – she feared
that Alyssa was being given money, drugs
and alcohol for sex and social services
subsequently took her into voluntary care
placing her in a children’s home. Within
a month she had run away, sleeping on
friends’ sofas and then eventually returning
home. She continued to misuse substances.
She had regular STD infections with frequent
appointments with the local GUM clinic.
She continued self-harming, ending up in
A&E on two occasions after overdosing on
paracetemol and alcohol.
By 15, she was pregnant but would not
disclose the father’s name. She moved
to local authority supported housing for
teenage parents, continuing her schooling
but passing only two of her GCSEs. By the
age of 19 she had moved in with a partner.
This was a violent relationship and she now
had three children under the age of 5 – all of
whom had themselves been placed on the
child protection register. She now regularly
had cycles of anxiety and felt unable to leave
the house. She continued to experience
regular cycles of depression requiring
intermittent treatment and interfering with
her ability to sustain employment. Her
own children struggled in school and two
were statemented due to attentional and
behaviouraldifculties.
Alyssa’s mother
was a teenage
parent and
experienced
severe postnatal
depression after
giving birth to
Alyssa and her
twin sister.
PLUS SOCIETAL COSTS:
including the
effects of bullying
PUBLIC SECTOR BUDGETS THESE COSTS
FALL ON:
SOCIAL SERVICES
27
Centre for Mental Health REPORT Building a better future
Introduction
This chapter provides a brief review of published
evidence on the effectiveness of parenting
interventions aimed at preventing or treating
early-onset conduct disorder. The main focus
is on the impact of parent training programmes
which support families with children aged
3-11 who are showing early signs of severe
behavioural problems. However, there is also
reference to Family Nurse Partnerships, an
intervention increasingly available in this
country which provides earlier and longer-
term support to vulnerable teenage mothers,
both during pregnancy and for up to two years
afterwards.
Parent training programmes
As noted in Chapter 2, quality of parenting
is a critical determinant of child outcomes.
Starting from this premise, a number of
behavioural training programmes have been
developed which aim to improve the quality
of the parent-child relationship and the skills
of parents in managing child behaviour.
Well-known examples include the Incredible
Years Programme and the Positive Parenting
Programme (Triple P). Although details vary,
effective programmes tend to share the
same general approach. For example, rather
than simply prescribing techniques, they
emphasise principles of good parenting, such
as the importance of parent-child interaction,
consistency and positive reinforcement; they
encourage parents to use active problem-
solving to apply these principles to their own
situation; and they focus on behaviour change.
Programmes are generally delivered in group
settings, involve eight to eighteen sessions
of around two hours each and are delivered
by trained facilitators, with some potential to
provide programmes on a one-to-one basis for
hard-to-reach parents.
Behavioural parent training has been described
as “the most extensively studied treatment for
children’s conduct problems” (Scott, 2008).
The evidence base includes well over a hundred
randomised controlled trials (RCTs), with the
ndingssummarisedandassessedinanumber
of systematic reviews and meta-analyses,
including a Cochrane review (Furlong et al.,
2012) and a review by NICE (2013). All of these
reviews agree that parenting programmes are
an effective intervention for childhood conduct
problems.Keyndingsfromtheresearch
literature are summarised below.
Research ndings
First, there is good evidence that parenting
programmessignicantlyimprovethequality
of parenting. For example, the Cochrane review
referenced above found the following effect
sizes for the impact of parenting programmes
on positive and negative parenting practices
(the effect size is a widely used method of
quantifying the overall effectiveness of an
intervention and as a rough rule of thumb
an effect size of around 0.2 indicates a small
impact, one of around 0.5 a moderate impact
and one of around 0.8 a large impact):
• positive parenting practices (parent repor ts)
0.53
• positive parenting practices (independent
reports) 0.47
• negative parenting practices (parent reports)
0.77
• negative parenting practices (independent
reports) 0.42.
There is also some evidence that parenting
interventions reduce child maltreatment
(Lundahl et al., 2006).
Second, by improving the quality of parenting,
parent training programmes are also effective
in reducing child problem behaviour. Overall,
around two-thirds of children with conduct
disorder show some improvement in their
5: The effectiveness of parenting programmes
Centre for Mental Health REPORT Building a better future
28
Issues
The evidence summarised above indicates
that parenting programmes generate multiple
positive outcomes. An overall assessment of
the effectiveness of these programmes should,
however, also take into account the following
issues which have been raised in the research
literature.
First, many studies of the effectiveness of
parenting interventions have been carried
out under research conditions and this raises
the question of whether the programmes are
equally successful in ‘real world’ settings. On
the whole the evidence suggests a positive
conclusion. For example, the Cochrane review
ndsnostatisticallysignicantdifferencesin
outcomes according to the setting in which
trials take place. Perhaps the best example in
this country of a study undertaken in a ‘real
world’ service setting is the North Wales Sure
Start trial mentioned above, which resulted in
particularly large improvements over a range of
outcomes, including effect sizes of 0.9 for the
impact of parenting training on child conduct
problems, 0.7 for the impact on parenting stress
and 0.5 for the impact on parental depression
(Hutchings et al., 2007).
A second important question is whether the
effectiveness of parenting interventions is
broadly the same across different types of
families and children, or is it the case that
some groups do better than others. Early
studiesappearedtondsomeevidenceof
group-related differences in outcomes, and in
particular that more disadvantaged families
(e.g. single parents, parents on very low
incomes or those with depression) tended to
do worse. Some trials also found that severity
of child problem behaviour was associated with
poorer outcomes. More recent evidence and
analysishaslargelyoverturnedthesendings,
leading to the more optimistic conclusion that
parenting programmes work equally well across
behaviour following a parenting programme
and the majority of these move below the
clinical threshold for a mental health diagnosis
(NICE, 2013). Effect sizes for the overall impact
on child problem behaviour are estimated in
the Cochrane review at 0.53 based on parent
reports and 0.44 based on independent
reports (Furlong et al., 2012). Similar results
are reported in other studies. For example, a
systematic review of 57 RCTs carried out by
Dretzke et al. (2009) shows an effect size for
child behaviour change of 0.67 based on parent
reports. The evidence also suggests that the key
factor in promoting better child behaviour is an
improvement in positive parenting rather than a
reduction in negative parenting (Gardner et al.,
2010).
Third, there is evidence that parenting
programmes improve behaviour among the
siblings of children with conduct disorder. For
example, an RCT to evaluate the effectiveness of
the Incredible Years programme for parents with
pre-school children with serious behavioural
problems who were attending Sure Start centres
in North Wales found that the impact on the
behaviour of siblings, while smaller than among
the children who were the main focus of the
intervention, was still big enough to result in
an effect size in the moderate to large range:
0.39 on one measure of behaviour and 0.69 on
another (Hutchings et al., 2007).
And fourth, there is also good evidence to show
that participation in a parenting programme
improves the mental health and wellbeing
of the parents themselves as well as of their
children. For example, a meta-analysis of
the effectiveness of parenting programmes
in improving maternal psychosocial health
combined the results from 15 RCTs which
included relevant data and found statistically
signicantimprovementsfromtheintervention
in the following areas: depression (effect size
0.3); anxiety or stress (effect size 0.5); self-
esteem (effect size 0.4); and relationship with
partner (effect size 0.4) (Barlow et al., 2002).
The Cochrane review found an effect size of 0.36
for the overall impact of parenting programmes
on parental mental health.
29
Centre for Mental Health REPORT Building a better future
the parent to maintain the improved parenting
style, further reinforcing the child’s response
and so on in a positive feedback loop. Improved
child behaviour may also set up positive
interactions in other settings, for example at
school or in peer relationships. Plausible as
these arguments may be, further research is
needed for empirical validation of their scale
and impact on the maintenance of treatment
gains.
A fourth question is whether parenting
interventions can be taken to scale, so that
they become embedded in service systems
and thus able to support a large proportion
ofallthefamilieswhomightbenet.Thisisa
major challenge, but some positive evidence
comes from the US Triple P System Population
Trial, which sought to reduce maltreatment
in a whole population of children aged 0-8
in South Carolina (Prinz et al., 2009). (The
TriplePsystemhasvelevelsofintensity,
with level four corresponding to group parent
training programmes; level one consists of a
media campaign, levels two and three cover
interventions in primary care for children
withmildbehaviouralproblemsandlevelve
corresponds to intensive individual parent
training for families of children with the most
severe problems.)
In this trial nine counties in the state were
randomly assigned to dissemination of the
Triple P parenting system while nine matched
comparison counties received services as usual.
Outcomes after two years measured by county-
level indicators showed a large impact of the
intervention, as shown by the following effect
sizes:
• substantiated cases of child maltreatment
1.09
• out-of-home placements 1.22
• child maltreatment injuries requiring
hospital treatment 1.14.
a wide range of family and child risk variables
(Gardner et al., 2010). The Cochrane review
ndsnostatisticallysignicantdifferencesin
outcomes in relation to either the severity of
childhood conduct problems or family socio-
economic status (Furlong et al., 2012). There is
also evidence that parenting programmes work
equally well for different ethnic groups (Stewart-
Brown & Schrader-Mcmillan, 2011).
A third question concerns the extent to which
thebenetsofparentinginterventions,
particularly improved child behaviour, persist
over time. This is an under-researched area, as
few studies have collected follow-up data on
outcomes for periods longer than three or six
months. Where longer-term information has
been collected, this provides some evidence
that treatment gains are maintained at 12 and
18 months (Bywater et al., 2009), at 4 years
(Muntz et al., 2004) and at 8-12 years (Webster-
Stratton et al., 2011), but as noted in the
Cochrane review other studies have found poor
maintenance of gains at 12-month follow-up
(Furlong et al., 2012).
A general argument supporting the persistence
of treatment gains is that effective intervention
may set up a self-reinforcing cycle of change
that helps to promote and sustain improved
behaviour over time. This is in effect a reversal
of the argument noted in Chapter 3, that in
the absence of intervention conduct disorder
tends to persist over time because of negative
interactions, with failure in one domain
aggravating failure in another, leading to a
cascadeordownwardspiralofdifculties.
Change in the parent-child relationship as a
result of parent training provides the possibility
of setting up a cycle going in the opposite
direction, starting from the fact that this
relationship is always bi-directional; in other
words,parentingstylenotonlyinuenceschild
behaviourbutmayalsobeinuencedbyit.Thus
an improvement in parenting style evokes better
behaviour by the child, which in turn encourages
Centre for Mental Health REPORT Building a better future
30
In a community with 100,000 children aged
0-8, these effects would translate into 688
fewer cases of child maltreatment, 240 fewer
out-of-home placements and 60 fewer children
requiring hospital treatment. In addition to
these gains relating to maltreatment, there
wouldalsobewiderbenetsfromreduced
incidence and severity of childhood conduct
disorder,notquantiedinthisstudy.
Finally, the evidence on parenting interventions
highlights the major importance of effective
implementation as a determinant of programme
success, including such factors as therapist
adherence to treatment protocols, quality
of therapist training, practical delivery (e.g.
providing transport and crèches for parents
attending training programmes), ongoing
supervision and organisational support. The
Cochrane review shows the following effect
sizes for interventions with high and low
programmedelity,inthetablebelow.Asthe
table shows, poor implementation reduces
the impact of parenting programmes by half or
more.
highdelity lowdelity
child conduct problems (parent reports) 0.58 0.28
child conduct problems (independent reports) 0.53 0.22
positive parenting practices (parent reports) 0.61 0.37
negative parenting practices (independent reports) 0.50 0.04
Effect sizes for interventions
31
Centre for Mental Health REPORT Building a better future
Family Nurse Partnerships
The Family Nurse Partnership is a preventive
programmeforvulnerablerst-timeyoung
mothers, based on home visiting by trained
nurses from early pregnancy until the child is
two. The support provided by nurses covers
wide range of issues, including parenting and
child behaviour. Initially developed in the
US over 30 years ago, the programme was
introduced in this country in 2007 on a pilot
basis and it is now planned to increase the
overall number of families in the programme at
any one time to 13,000 by 2015.
Three large randomised controlled trials have
tested the programme in the US and these
havedemonstratedarangeofbenetsfor
both mothers and children. Importantly, these
studies have tracked – and indeed continue to
track – outcomes continuously over time and
published results are available for the impact
of the programme on a wide range of measures
throughout childhood and adolescence for the
children included in the trials.
Benetsforthechildrenincludethefollowing
(DH, 2011):
• 48%reductioninveriedcasesofchild
abuse and neglect by age 15;
• 50% reduction in language delay at 21
months;
• 67% reduction in behavioural and emotional
problems at age 6;
• 28% reduction in anxiety and depression at
age 12;
• 67% reduction in use of cigarettes, alcohol
and marijuana at age 12;
• 59% reduction in arrests by age 15.
The programme also generates a number
ofbenetsformothersintheprogramme,
including increased employment, reduced
welfare dependency and reduced offending
(61% fewer arrests and 72% fewer convictions
among mothers by the time children were aged
15).
Centre for Mental Health REPORT Building a better future
32
Liah
Liahhadadifcultchildhood.Shehada
distant relationship with her own mother
who was strict and emotionally unreliable.
She lived on the streets at 16, had a history
of substance misuse during this time and
was put on medication for anxiety during
her teens.
By the age of 21, things had stabilised a
little. She lived with her partner, had her
own tenancy and had a 4 year old son. She
was also pregnant with her second child.
Here relationship was volatile,
characterised by anger and regular violent
outbursts. The police had regularly attended
heratfollowingconcernsaboutdomestic
violence. Her son had been placed on a
child protection plan as a result of concerns
about family violence and his safety. His
behaviour was also getting more and
moredifcultforLiahtomanage.Hehad
frequent tantrums and staff regularly called
her in to complain about his behaviour at
nursery. Social workers told Liah that her
new-born son would also go straight onto a
child protection plan at birth.
Liah was persuaded by her social worker
to attend a local parenting support
group. Although initially sceptical about
attendance, she very quickly settled in and
was exceptionally positive about what she
learnt both from the group facilitators and
from other parents. She described picking
up very practical skills which quickly made
a difference to her son’s behaviour. The
groupalsomadeherreectonother
aspects of her life including her own
parenting and her volatile relationship.
Liah’s mental health and stress levels
also improved greatly to the extent that
she approached her GP to plan coming off
her medication.
Liah learnt quickly that she needed to
be consistent with the techniques she
had learnt, otherwise she would almost
immediately witness a deterioration in
her son’s behaviour. But this was getting
easier and the strategies were becoming
second nature to her now. After three
months, both children were removed
from child protection plans thanks to
the progress made by the family. Liah
describedfeelingsomuchmorecondent
and was considering returning to college
with a view to accessing employment.
PUBLIC SECTOR BUDGETS THESE COSTS
FALL ON:
SOCIAL SERVICES
33
Centre for Mental Health REPORT Building a better future
Introduction
This chapter seeks to address two main
questions:rst,areparentinginterventionsa
good use of scarce resources; and second, do
they pay for themselves through future savings
in public spending? Both these questions
involveweighingupcostsandbenets,inthe
rstcasefromtheperspectiveofsocietyas
a whole and in the second from the narrower
perspective of the public sector.
Apositiveanswertotherstquestionis–or
ought to be – a necessary condition for going
ahead with any programme in the public sector,
butitisnotalwaysasufcientcondition.
Particularly at a time when public funds are
under severe restraint, decision makers will also
be very concerned with affordability. This raises
particulardifcultieswiththefundingofearly
intervention for children with severe behavioural
problems, partly because any subsequent
savings in public spending may accrue over
long periods of time and partly because these
savings are also distributed across a wide range
of different agencies in the public sector. Who
pays,whobenetsandwhen?Alloftheseare
crucial questions for decision makers.
Evidence with which to answer these questions
is in relatively short supply, as few of the
studies of effectiveness have collected much
economic data, particularly in relation to
outcomes. The lack of long-term follow-up
data in the effectiveness literature is also a
major constraint. To get round these problems,
various studies have made use of an economic
modelling approach in which quantitative data
from effectiveness trials are incorporated into
intervention-specicmodelsandcombined
with economic information from other sources
toproduceestimatesofcostsandbenets
measured in monetary terms.
In some cases this is very straightforward. For
example, effectiveness studies often include
some data on the use of health services before
and after an intervention, e.g. numbers of GP
consultations or outpatient attendances. Such
information can readily be translated into
monetary equivalents using published national
data on unit costs.
In other cases the modelling requires additional
steps. For example, children with conduct
disorder are more likely than others to leave
schoolwithnoeducationalqualications.Inthe
absence of detailed data on the individuals in an
effectiveness study, the impact on their future
earnings can be simulated using information
taken from wider studies of the long-term
relationshipbetweeneducationalqualications,
employment and pay. As this example
shows, the modelling approach necessarily
involves the use of some assumptions in the
estimation process, in this case that a historical
relationship between education and earnings
for the general population will continue into the
future and can be applied without adjustment
to a particular sub-group. Results should be
therefore treated with a degree of caution.
A more important limitation in the economic
literature on early intervention is the almost
invariableomissionofmajortypesofbenet
in value-for-money assessments. Ideally, cost-
benetanalysesshouldseektoincorporateall
the effects of an intervention, where necessary
imputing monetary values to items which are not
conventionally traded or included in GDP. Most
obviously, in the area of mental health, this
means including an imputed monetary value for
thebenetsofimprovedmentalhealthinterms
of its impact on wellbeing and quality of life.
Suchbenetsarethefundamentaljustication
forserviceprovisionandyettheyndnoplace
in the economic literature on early intervention.
6: The costs and benefits of intervention
Centre for Mental Health REPORT Building a better future
34
To illustrate, appropriate information in
effectiveness studies combined with data on
programme costs can be used to calculate the
average cost of bringing a child with conduct
disorder below a clinical cut-off or threshold
as a result of a parenting intervention. Two
economic studies included in the Cochrane
review have been analysed in this way and
come up with virtually identical results, with
the cost in question being estimated at around
£1,750 per case in 2012/13 prices (Furlong
et al., 2012). A similar analysis in a study by
Dretzske et al. (2005), covering a larger number
oftrials,yieldsbroadlycomparablegures
(£1,875 per child brought below the clinical
threshold by community-based group parenting
programmes and £1,315 for clinic-based group
programmes).
It was noted in Chapter 4 that on one estimate
the lifetime cost of a child with conduct disorder
is around £260,000 while the lifetime cost of
a child with moderate (sub-threshold) conduct
problems is around £85,000, in each case
measured against the baseline of a child with no
problems. Effectiveness trials show that among
those children whose behaviour improves
following a parenting intervention, some move
into the ‘no problems’ category while others
make the smaller move into the ‘moderate
problems’ group. For example, work by NICE
indicates that among 3-year-old children with
conduct disorder who move below the clinical
cut-off following a parenting intervention, 46%
move into the ‘moderate problems’ group and
54% into the ‘no problems’ group (NICE, 2013).
Taking a conservative approach, suppose that
all those who improve remain with moderate
problems.Onthisbasis,thepotentialbenets
of intervention may be represented by the
difference in lifetime costs between a child
with conduct disorder and one with moderate
problems, i.e. £175,000 (£260,000 minus
£85,000).
Also omitted in all studies are a range of third-
party effects such as the impact of parenting
interventions on the mental health and
quality of life of parents and siblings and on
the wellbeing of others such as the victims of
bullying and children in the next generation.
In some cases, even when relevant economic
information is available, such as the impact of
bullying on the future earnings of victims, this
too is omitted.
For just the same reasons as the costs
of behavioural problems are invariably
understated,sotooarethebenetsofeffective
early intervention. None of this is to deny the
difcultiesofmonetaryvaluationintheareas
justdescribed.Itis,however,toagupa
warningthatthendingsofcost-benetstudies
reported below need very careful interpretation.
Inparticular,andingthatcostsexceed
measuredbenetsinanyparticularcasedoes
not necessarily mean that the intervention in
question is poor value for money. Unless the
full scale of any improvements in mental health
and quality of life are also taken into account,
no such conclusion can be drawn. In general,
society spends money on health care, not
because this might save money elsewhere or
later on, but because better health is a good
in its own right - and one which is valued very
highly. This applies as much to mental health as
to physical health.
Threshold analysis
Thendingsofhealtheconomicstudiescan
be presented in a number of ways. The most
usualistocomparethetotalbenetsofan
intervention, discounting those accruing in the
future back to a present value, with programme
costs,togiveameasureofnetbenets(B-C)
orabenet/costratio(B/C).Anyintervention
showingpositivenetbenetsorabenet/cost
ratio greater than one may be judged good value
for money.
An alternative approach is threshold analysis,
which explores such issues as the minimum
level of effectiveness that is needed for a
programme to pass a value for money test. This
approach has its limitations but can be helpful
when the available data on outcomes are limited
or uncertain.
35
Centre for Mental Health REPORT Building a better future
Evidence is presented using four main studies,
all of which combine economic modelling
with data on the effectiveness of programmes
derived from meta-analyses, using pooled
results from large numbers of individual trials.
These sources are:
1. Washington State Institute for Public
Policy (WSIPPP, 2013): for some years the
WSIPP have been producing estimates of
economic returns for a range of public sector
programmes,usingacommoncost-benet
framework and methodology. Initially the
work focused mainly on programmes in
the criminal justice area but coverage has
been steadily extended and now includes
a number of interventions relating to
child mental health, including parenting
programmes.
2. Social Research Unit (SRU, 2013): SRU
in Dartington has adopted the WSIPP
framework of analysis and sought to re-
populate it using UK data. While there is
a good deal of overlap between these two
sources,thereremainsufcientdifferences
to justify including results from both.
3. NICE (2013): NICE’s most recent guideline
on conduct disorder includes evidence
on the cost-effectiveness of a range of
interventions for this condition including
parenting programmes, based on economic
modelling.
4. Bonin et al. (2011a; see also 2011b): this
is an economic modelling study undertaken
for the Department of Health as a part of
a wider project on the economic case for
mental health promotion and mental illness
prevention.
Comparingcostsandbenets,theaveragecost
of bringing a child with conduct disorder below
the clinical threshold is estimated at £1,750
andthepotentialbenetintermsofreduced
lifetime costs at £175,000. Lifetime costs thus
need to be reduced by just 1% to cover the costs
of the intervention. This is a strikingly small
number.
As a more demanding test, the Cochrane review
includes estimates not only of the cost of
bringing an average child with conduct disorder
below the clinical threshold but also the cost
of achieving this for a child with the highest
level of conduct problems. The latter cost is
put at around £6,650 in 2012/13 prices. Even
usingthishighergure,lifetimecostsneedto
be reduced by just 4% to cover outlays on the
intervention.
Theseguresshowthatthecostsofeffective
intervention for conduct disorder are very small
relativetothepotentialbenets.Inpractice
many other factors also need to be considered.
For example, the calculations implicitly
assumethatthebenetsofintervention
persist throughout the life course, but long-
term follow-up data are not available to
substantiate this. Also, around half of children
with conduct disorder recover naturally, without
any need for intervention. On the other hand,
the costs of conduct disorder and hence the
potentialbenetsofeffectiveinterventionare
under-estimated in all studies. Even with the
qualicationsjustnoted,itremainsthecase
that only a modest degree of success is needed
for parenting programmes to be good value for
money.
Benet/cost ratios
This section reviews the evidence on the returns
fromparentingprogrammesusingthebenet/
cost ratio as an overall measure of performance.
A ratio of say 4:1 indicates that every £1
investedinaprogrammeyieldsbenetsvalued
at £4. Where information is available, estimates
are given separately for the returns measured
from a societal perspective and from the
perspective of the public sector.
Centre for Mental Health REPORT Building a better future
36
to subsequent savings in public spending of
over £2. Against this, there is some evidence
that the cost of providing parenting programmes
is higher in this country than in the US, which
would obviously reduce the net returns (SRU,
2013). The reasons for this difference in costs
are not entirely clear, although it may in part be
because these programmes are provided on a
larger scale in the US, which may result in some
economies.
Comparing the four sources of evidence used in
thetable,thedifferencesinestimatedbenet/
costratiosreectanumberoffactors.These
include the following.
Time period: theNICEstudymodelsbenets
only up to age 11 following an intervention at
age 3, whereas Bonin et al.followbenetsup
to age 30 after an intervention at age 5. WSIPP
and SRU also model an intervention at age 5
andseektomeasurelifetimebenets,although
in practice those accruing beyond childhood are
extremely small.
Effectiveness of the intervention: effect sizes
used in the modelling vary partly because of
differences in the coverage of trials in the meta-
analyses, but also because WSIPP and SRU
incorporate large downward adjustments to
Forreasonsdiscussedbelow,thesegures
show quite a wide spread. Taking a simple
unweighted average of the various estimates,
thebenet/costratioforparentingprogrammes
is 3.4:1 when measured from a societal
perspective, indicating that these programmes
are very good value for money when all
benetsaretakenintoaccount,and1.4:1
when measured from a purely public sector
perspective, showing that programme costs are
more than covered by subsequent savings in
public spending.
The returns to the public sector are in fact
higher than this if measured in a UK context.
This is because the WSIPP estimates relate to
the US, where a higher proportion of spending
on programmes such as health comes from
non-public sources. For example, the WSIPP
studyofTriplePshowsthat78%ofallbenets
come from savings in healthcare costs, yet only
33%oftotalbenetsaccruetothetaxpayer.If
all savings in health costs in this study and the
oneforIncredibleYearsinWSIPParere-dened
as savings in public spending, the public sector
benet/costratioaveragedoverallseven
estimates given in the table above increases
from 1.4:1 to at least 2.1:1. In other words,
every £1 spent on parenting programmes leads
Summary results for group parenting programmes from these studies are as follows:
Benet/costratio
society public sector only
WSIPP
• Triple P 5.6 : 1 1.9 : 1
• Incredible Years 1.2 : 1 0.4 : 1
SRU
• Generic programme 1.7 : 1 1.2 : 1
• Triple P 0.9 : 1 0.7 : 1
• Incredible Years 1.4 : 1 0.9 : 1
NICE
• Generic programme n/a 1.6 : 1
Bonin et al.
• Generic programme 7.9 : 1 2.9 : 1
37
Centre for Mental Health REPORT Building a better future
not also those falling on frontline education.
The paper by Snell et al. suggests that the latter
account for the bulk of extra education costs.
Costs associated with crime are important only
in Bonin et al. In the case of NICE, this is very
largely because the time period for analysis
is truncated at age 11, while for WSIPP and
SRU the explanation appears to be that the
impact of parenting interventions on antisocial
behaviour is very short-lived. SRU also base
their modelling on the assumption that children
with conduct disorder are no more likely than
average to become involved in criminal activity,
despite the very large body of evidence to the
contrary. The WSIPP and SRU studies of Triple P
includesomebenetsintheformofincreased
earnings and reduced health costs resulting
from a favourable impact of this programme
on parental depression; otherwise all impacts
on parents and also on siblings are omitted in
all studies. WSIPP and SRU allow for a small
increase in the subsequent earnings of children
whobenetfromaparentingintervention,but
this is not covered in NICE or Bonin et al.
The extent of these and other differences in
methodologyclearlymakesitdifculttodraw
direct comparisons between the four studies.
A further complication is that even relatively
small changes in modelling procedures can
have a sizeable impact on the overall results.
For example, the NICE study models a parenting
intervention at age 3 for the child, with
benetsbeingtrackedforthefollowingseven
years, a combination of assumptions which
means that any impact on juvenile offending
is almost completely ruled out. To show the
possible effect of a change in the child starting
age for the analysis, it may be noted that
the NICE report also includes an economic
analysis of another type of intervention for
conduct disorder, one which is given at age
7 and is child- rather than parent-focused.
This has a broadly similar cost to a parenting
programmebutasignicantlysmallerimpact
on behaviour. Despite this, the estimated
publicsectorbenet/costratioisalmosttwice
as large, at 3.1:1 compared with 1.6:1. This
is mainly because the follow-up period (up
to age 14) now includes a time when some
children with conduct disorder start becoming
allow for perceived methodological weaknesses
in some of the studies used. There is some
justicationfortheseadjustments.Ontheother
hand,theyresultinsignicantlysmallereffect
sizes than those used in NICE and Bonin et al.,
even though the latter are broadly in line with
those given in the Cochrane review, which only
includes studies meeting very high standards of
methodological quality.
Drop-out: Bonin et al. assume that as many
as 44% of all parents drop out of a training
programme before completion, with the further
assumptionthatchildrenderivenobenet
if parental attendance is incomplete. No
allowance appears to be made for drop-out in
NICE.TheestimatesofbenetsinWSIPPand
SRU are based on intention-to-treat analysis,
which means that outcomes are measured for all
participants in the original sample of a trial and
not simply those who complete the programme.
Drop-out is therefore taken into account, but the
scale of this is not explicitly stated.
Persistence of treatment gains: Bonin et al. and
NICE both assume a 50% rate of relapse among
children whose behaviour initially improves as
a result of a parenting intervention; in other
words, the behaviour of these children rapidly
reverts to its original level. It is also assumed
that some of those who do improve would have
done so anyway, in line with the natural rate of
recovery from conduct disorder. The modelling
procedures jointly used by WSIPP and SRU are
different from those in Bonin et al. and NICE but
have the effect that treatment gains generally
are smaller and more transient.
Coverage of benets: as noted in the table, the
NICEstudyincludesonlythosebenetswhich
take the form of savings in public expenditure.
Estimated savings in child healthcare costs are
broadly similar in WSIPP, NICE and Bonin et al.
but much lower in SRU. This is probably because
the savings in SRU are based on the costs of
child health care given in the paper by Snell et
al. (2013), which as noted earlier substantially
under-estimates these costs, perhaps by a
factor of ten. No savings in education costs
areidentiedinWSIPP,whilethosegiven
in NICE and Bonin et al. cover only the costs
associated with special education needs and
Centre for Mental Health REPORT Building a better future
38
Theseguresshowthatinterventioncostsare
fully recovered in terms of public expenditure
savings after seven years, with about a third of
the savings accruing in the short term (years 1
and 2). The returns are distributed over a range
ofagencies,withsignicantsavingstotheNHS
and education in both the short and medium
termandsomebenetsaccruingtothecriminal
justice system towards the end of the 7-year
period. Going beyond the 7-year cut-off, savings
in the criminal justice system build up steadily
and in the long term the costs of the intervention
are fully covered both by savings in this system
and, separately, by savings in the NHS.
Estimated savings in the education system are
signicantlylowerthanthoseintheNHS,butas
noted above the baseline costs against which
education savings are measured cover only the
costs of special education needs and not also
the extra costs falling on frontline education
services resulting from children’s severe
behavioural problems. The costing analysis by
Snell et al., which was not available at the time
of the Bonin et al. study, suggests that extra
frontline costs account for around 62% of all
additional educational costs.
regularoffenders.Iftheestimatedbenets
of this intervention in terms of reduced costs
in the criminal justice system are included
inthebenetsoftheparentingintervention,
thebenet/costratioofthelatterwouldrise
by more than half, to 2.5:1. This is clearly a
mechanistic calculation, but does illustrate
the potential sensitivity of modelling results to
specicassumptionsandprocedures.
Short-term budgetary impacts
Particularly when allowance is made for omitted
benets,itisclearthatwell-designedgroup
parenting programme are very good value
for money from a societal perspective. The
evidence also indicates that, over time, these
programmes more than pay for themselves
in public expenditure terms. This still leaves
open some questions relating to affordability.
In particular, what is the short-term impact
on public sector budgets and which agencies
withinthepublicsectorbenetmost?
To address these questions use is made mainly
of the study by Bonin et al., for which detailed
year-by-year information has been made
available. Measured in 2012/13 prices, the
average cost of a parenting programme is put
at £1,282 per child in this study. Public sector
savings in the short term (combined total for
years 1 and 2 following implementation of the
programme) and in the medium term (combined
total for years 3-7) are estimated as follows,
measured in £s per child:
years 1-2 years 3-7 total over 7 years
NHS 231 320 551
Personal social services 33 21 54
Education 181 165 346
Criminal justice system 0 322 322
Total public sector 445 828 1273
PUBLIC SECTOR SAVINGS FOLLOWING PROGRAMME IMPLEMENTATION (£ PER CHILD)
39
Centre for Mental Health REPORT Building a better future
The WSIPP analysis shows that both these
programmes are very good value for money.
This particularly applies to the Triple P system,
astotalbenetsareestimatedtoexceedthe
costs of the programme by a factor of six. Most
ofthebenetsrelatetoreductionsinchild
abuse and neglect and savings in expenditure
onout-of-homeplacements.Benetsaccruing
to taxpayers account for nearly 40% of the
total, implying that every $1 invested in the
programme yields savings to the taxpayer of
around $2.30.
The Family Nurse Partnership programme is a
relatively intensive intervention and its costs
are estimated at $9,600 per family, spread over
twoyears.Totalbenetsareestimatedatnearly
$23,000perfamily,implyingabenet/cost
ratio of 2.4:1 from the perspective of society
asawhole.Nearlytwo-thirdsofthebenets
are linked to better outcomes for the mothers
participating in the programme, mainly in the
form of higher earnings, and the remainder
are mainly accounted for by reduced crime and
higherearningsamongthechildren.Benets
accruing to taxpayers are estimated at around
$6,200 per family, or around two-thirds of the
cost of the programme.
Implementation and targeting of
programmes
Finally, economic analysis may be used to
underline the importance of implementing
programmes effectively and also ensuring that
they are targeted at the families and children
whoarelikelytobenetmost.
When budgets are tight, there is a natural
tendency to cut corners in the implementation
of programmes, for example by employing
staff who are not fully trained, but in all
likelihood such attempts will turn out to be
false economies, with any immediate savings
being greatly outweighed by the loss of future
benets.
If the estimates in Bonin et al. are increased
inlinewiththisgure,educationsavingsrise
to £478 in years 1-2 and £436 in years 3-7.
For the public sector as a whole, total savings
increase to £742 in years 1-2, corresponding
to nearly 60% of the cost of the intervention,
and to £1,099 in years 3-7. Full cost recovery is
achievedwithinveyearsandoverthe7-year
period as a whole the public sector saves £1.44
for every £1 invested.
The use of a 7-year follow-up period for analysis
also allows a direct comparison with the
modelling study by NICE, which uses the same
time span. Measured in 2012/13 prices, the
cost of the intervention in the NICE study is put
at £1,230 per child and subsequent savings in
total public spending at £2,012. Intervention
costs are fully recovered after about four
years. Of the savings over seven years, 57%
(£1,157) accrue to health and personal social
services, 40% (£804) to education and the small
remainder (£51) to criminal justice.
As in the study by Bonin et al.,theguresfor
education relate only to the costs of special
education needs. If allowance is also made for
extra frontline education costs, total savings in
public spending in the NICE analysis increase
to £3,331 over 7 years, implying a public sector
benet/costratioof2.7:1.Savingsineducation
costs alone rise to £2,123. Set against an
intervention cost of £1,230, this implies that for
the education sector investment in parenting
programmesismorethanjustiedonpurely
nancialgrounds.
Other programmes
All of the analysis presented above relates to
group parenting programmes, but results may
also be presented for two other interventions
which are covered in the economic modelling
work undertaken by WSIPP: the Triple P multi-
level parenting system which has been trialled
on a population-wide basis in South Carolina,
and Family Nurse Partnerships.
Centre for Mental Health REPORT Building a better future
40
refreshments amounted to £2,700, or 17% of
the total. Spread over say ten parents, a cost
of £270 per head is clearly very small when set
againstthebenetofupto£18,200percase
from reducing drop-out.
As a second illustration of the importance of
effective implementation, reference may be
madetothendingintheCochranereview
mentioned in Chapter 5, that the effectiveness
ofparentinginterventionsissignicantly
reducediftheyhavelowprogrammedelity.
For example, the effect size for impact on child
conduct problems (independent reports) is
0.44 when averaged over all programmes in the
Cochrane review, but only half this (0.22) for
thosewithlowdelityand0.53forthosewith
highdelity(Furlonget al.,2012).Lowdelity
thusmeanssacricingatleasthalfthebenets
of an average intervention. Again using the
study by Bonin et al. as a reference point, this
amounts to a loss of £5,100 per child in terms of
totalbenetsand£1,850intermsofsavingsin
public spending.
In the case of targeting, it was noted in Chapter
5 that parenting programmes are at least as
effective for children with severe behavioural
problems as they are for children with more
moderatedifculties.Ontheotherhand,the
long-term costs of behavioural problems, and
hencethepotentialbenetsofintervention,are
around three times as high among the 5% of
childrenwhoseproblemsaresufcientlysevere
to merit a diagnosis of conduct disorder as they
are among the 15-20% with more moderate,
sub-threshold problems. Effective targeting
of programmes at those with the most severe
problems will therefore yield much higher
returns. Budgetary constraints in the public
sector are always likely to impose some limits
on the availability of parenting programmes
and it is an important responsibility of budget-
holders to ensure that limited resources are
used to best overall effect.
Two examples may be used to illustrate this.
First, it was noted above that in the study by
Bonin et al. it is assumed that 44% of parents
drop out before completing a behavioural
training programme and that in consequence
the children of these parents make no
improvementintheirbehaviour.Thegureof
44% is at the upper end of a range suggested by
evidence reviews and was adopted as a means
of avoiding optimism in the overall results.
Whatevertheprecisegure,itisclearthatdrop-
out has a high cost.
Atrstsightitmightappearthatthescaleof
this cost is given by the loss of the average
valueofbenetsassociatedwithaparenting
programme. Thus the study by Bonin et al.
estimates that a parenting intervention yields
totallong-termbenetsperchildofaround
£10,200 from the perspective of society as a
whole and around £3,700 in the form of reduced
public spending. However, these are averages
whichincludebenetsofzerofor44outof
every100children,meaningthatbenetsfor
the remaining 56 must be considerably higher
thantheoverallaverage.Benetsperfamily
completing the intervention work at around
£18,200fortotalbenets(i.e.£10,600÷0.56)
and around £6,600 for public expenditure
savings(£3,700÷0.56)andthesearethe
guresthatrepresentthecostofonedrop-out.
Parents drop out of programmes for a variety
of reasons, but the available evidence
suggests that it is often practical problems
suchasdifcultieswithtransportorchildcare
that constitute the main barrier to initial
and continuing engagement. Conversely,
engagement can often be improved by holding
groups in convenient locations and providing
refreshments. Money spent on addressing
these practical issues is likely to yield a high
return, as their cost is very modest relative to
thebenetsofreducingdrop-out.Datafromthe
successful Sure Start trial in north Wales show
that in 2012/13 prices the total cost of running
a parenting group for 8-12 participants over
12 sessions amounted to £15,895 (Edwards
et al., 2007). Within this, the combined cost of
transport, crèche facilities, venue rental and
41
Centre for Mental Health REPORT Building a better future
Cohen, M. & Piquero, A. (2009) New evidence on the
monetary value of saving a high risk youth. Journal of
Quantitative Criminology, 25, 25-49.
Copeland, W., Wolke, D., Angold, A. & Costello, J.
(2013) Adult psychiatric outcomes of bullying and
being bullied by peers in childhood and adolescence.
JAMA Psychiatry, 70(4), 419-426.
Curtis, L. (2012) Unit Costs of Health and Social Care
2012. PSSRU, University of Kent.
Department of Health (2011) FNP Evidence Summary
Leaet. Available at www.gov.uk/government/
uploads/system/uploads/attachment_data/
le/215542/dh_128008.pdf.
Department of Health (2013) Programme Budgeting
Aggregate PCT Expenditure for Financial Years
2003/04 to 2011/12. Available at www.networks.
nhs.uk/nhs-networks/health-investment-network/
news/2011-12-programme-budgeting-data-now-
available.
Dretzke, J., Frew, E. et al. (2005) The effectiveness
and cost-effectiveness of parent training/education
programmes for the treatment of conduct disorder,
includingoppositionaldeantdisorder,inchildren.
Health Technology Assessment, vol. 9, no. 50.
Dretzke, J., Davenport, C. et al. (2009) The clinical
effectiveness of different parenting programmes for
children with conduct problems: a systematic review
of randomised controlled trials. Child and Adolescent
Psychiatry and Mental Health, 3,7.
Dubourg, R., Hamed, J. & Thorns, J. (2005) The
Economic and Social Costs of Crime against
Individuals and Households 2003/04. Home
Ofce Online Report 30/05. Available at www.
gov.uk/government/uploads/system/uploads/
attachment_data/le/191498/Green_Book_
supplementary_guidance_economic_social_costs_
crime_individuals_households.pdf.
Edwards, R., Coéilleachair, A. et al. (2007) Parenting
programmes for parents of children at risk of
developing conduct disorder: cost effectiveness
analysis. BMJ, 334:682.
Farrington, D. (1995) The development of offending
and antisocial behaviour from childhood: key
ndingsfromtheCambridgestudyindelinquent
development. Journal of Child Psychology and
Psychiatry, 360(6), 929-964.
Bardone,A.,Moftt,T.,Caspi.A.,Dickson,N.&Silva,
P. (1996) Adult mental health and social outcomes
of adolescent girls with depression and conduct
disorder. Development and Psychopathology, 8, 811-
829.
Barlow, J., Coren, E. & Stewart-Brown, S. (2002)
Meta-analysis of the effectiveness of parenting
programmes in improving maternal psychosocial
health. British Journal of General Practice, March
2002, 223-233.
Barrett, R. (2010) Disadvantaged groups in the
labour market. Economic and Labour Market Review,
4(6), 18-24.
Bonin, E., Stevens, M., Beecham, J., Byford, S. &
Parsonage, M. (2011a) Parenting interventions for
the prevention of persistent conduct disorders. In
Knapp, M., McDaid, D. & Parsonage, M. (eds.) Mental
Health Promotion and Mental Illness Prevention: the
economic case. London: Department of Health.
Bonin, E., Stevens, M., Beecham, J., Byford, S.
& Parsonage, M. (2011b) Costs and longer-term
savings of parenting programmes for the prevention
of persistent conduct disorder: a modelling study.
BMC Public Health 2011, 11:803.
Bosquet, M. & Egeland, B. (2000) Predicting
parenting behaviours from antisocial practices
content scale scores of the MMPI-2 administered
during pregnancy. Journal of Personality Assessment,
74(1), 146-162.
Brand, S. & Price, R. (2000) The Economic and Social
Costs of Crime. Home Ofce Research Study 217.
London:HomeOfce.
Brown, E., Khan, L. & Parsonage, M. (2012) A
Chance to Change: Delivering effective parenting
programmes to transform lives. London: Centre for
Mental Health.
Brown, S. & Taylor, K. (2008) Bullying, education
and earnings: evidence from the National Child
Development Survey. Economics of Education Review,
27, 387-401.
Bywater, T., Hutchings, J. et al. (2009) Long-term
effectiveness of a parenting intervention for children
at risk of developing conduct disorder. British Journal
of Psychiatry, 195, 318-324.
Centre for Mental Health (2009) The Chance of a
Lifetime: Preventing early conduct problems and
reducing crime. London: Centre for Mental Health.
Cohen (1998) The monetary value of saving a high
risk youth. Journal of Quantitative Criminology, 14,
5-33.
References
Centre for Mental Health REPORT Building a better future
42
Jaffée,S.,Moftt,T.,Caspi,A.&Taylor,A.(2003)
Lifewith(orwithout)father:thebenetsofliving
with two biological parents depend on the father’s
antisocial behaviour. Child Development, 74, 109-
126.
Jokela, M., Ferrie, J. & Kivimaki, M. (2009) Childhood
problems behaviours and death by midlife. Journal
of the American Academy of Child and Adolescent
Psychiatry, 48(1), 19-24.
Kim-Cohen, J., Caspi, A. et al. (2003) Prior juvenile
diagnoses in adults with mental disorders. Archives
of General Psychiatry, 60, 709-717.
Knapp, M., King, D., Healey, A. & Thomas, C.
(2011) Economic outcomes in adulthood and their
associationswithantisocialconduct,attentiondecit
and anxiety problems in childhood. Journal of Mental
Health Policy and Economics, 14(3), 137-147.
Knapp, M., Scott, S. & Davies, J. (1999) The cost of
antisocial behaviour in younger children. Clinical
Child Psychology and Psychiatry, 4(4), 457-473.
Kratzer, L. & Hodgins, S. (1997) Adult outcomes of
child conduct problems: a cohort study. Journal of
Abnormal Child Psychology, 25(1), 65-81.
Krueger,R.,Moftt,T.et al. (1998) Assortative
mating for antisocial behaviour: developmental and
methodological implications. Behaviour Genetics,
28(3), 173-186.
Lundahl, B., Nimer, J. & Parsons, B. (2006)
Preventing child abuse: a meta-analysis of parent
training programs. Research on Social Work Practice,
16(3), 251-262.
Masten, A., Roisman, G. et al., (2005) Developmental
cascades: linking academic achievement and
externalizing and internalizing symptoms over 20
years. Developmental Psychology, 41(5), 733-746.
Meltzer, H., Gatward, R., Corbin, T., Goodman, R. &
Ford, T. (2003a) The Mental Health of Young People
Looked After by Local Authorities in England. London:
TSO.
Meltzer, H., Gatward, R., Corbin, T., Goodman, R. &
Ford, T. (2003b) Persistence, onset, risk factors and
outcomes of childhood mental disorders. London:
TSO.
Moftt,T.(1993)Adolescence-limitedversus
life-course-persistent antisocial behaviour: a
developmental taxonomy. Psychological Review,
100, 674-701.
Moftt,T.,Caspi,A.,Harrington,H.&Milne,B.
(2002) Males on the life-course-persistent and
adolescent-limited pathways: follow-up at age 26
years. Development and Psychopathology, 14, 179-
207.
Farrington, D., Coid, J. et al. (2006) Criminal Careers
up to Age 50 and Life Success up to Age 48: New
ndings from the Cambridge Study in Delinquent
Development. Home Ofce Research Study 299.
London:HomeOfce.
Fergusson, D. & Horwood, J. (1998) Early conduct
problems and later life opportunities. Journal of Child
Psychology and Psychiatry, 39(8), 1097-1108.
Fergusson, D., Horwood, J. & Ridder, E. (2005) Show
me the child at seven: the consequences of conduct
problems in childhood for psychosocial functioning
in adulthood. Journal of Child Psychology and
Psychiatry, 46(8), 837-849.
Fergusson, D. & Lynskey, M. (1998) Conduct
problems in childhood and psychosocial outcomes
in young adulthood. Journal of Emotional and
Behavioral Disorders, 6(1), 2-18.
Foster, M. & Jones, D. (2005) The high costs of
aggression: public expenditures resulting from
conduct disorder. American Journal of Public Health,
95(10), 1767-1772.
Friedli, L. & Parsonage, M. (2007) Mental Health
Promotion: Building an economic case. Belfast:
Northern Ireland Association for Mental Health.
Furlong, M., McGilloway, S. et al. (2102) Behavioural
and cognitive-behavioural group-based parenting
programmes for early-onset conduct problems in
children aged 3 to 12 years. The Cochrane Library,
Issue 2.
Gardner, F., Hutchings, J., Bywater, T. & Whitaker,
C.(2010)Whobenetsandhowdoesitwork?
Moderators and mediators of outcomes in an
effectiveness trial of a parenting intervention. Journal
of Clinical Child and Adolescent Psychology, 39(4),
568-580.
Green, H., McGinnity, A., Meltzer, H., Ford, T. &
Goodman, R. (2005) Mental Health of Children
and Young People in Great Britain, 2004. Crown
Copyright. Basingstoke: Palgrave Macmillan.
Hummel, S., Naylor, P. et al. (2009) Cost-
effectiveness of Universal Interventions which
Aim to Promote Emotional and Social Wellbeing
in Secondary Schools. ScHARR, University of
Shefeld.Availableatwww.nice.org.uk/nicemedia/
live/11991/45517/45517.pdf.
Hutchings, J., Bywater, T. et al. (2007) Parenting
intervention in Sure Start services for children at
risk of developing conduct disorder: pragmatic
randomised controlled trial. BMJ, 29 March 2007,
334.
Jaffée, S., Belsky, J. et al. (2006) When parents have
a history of conduct disorder: how is the caregiving
environment affected? Journal of Abnormal
Psychology, 115(2), 309-319.
43
Centre for Mental Health REPORT Building a better future
Scott, S. (2004) Childhood antecedents of juvenile
delinquency. In Bailey, S. and Dolan, M. (eds.)
Adolescent Forensic Psychiatry. London: Arnold.
Scott, S. (2008) An update on interventions for
conduct disorder. Advances in Psychiatric Treatment,
14, 61-70.
Scott, S., Knapp, M., Henderson, J. & Maughan, B.
(2001) Financial cost of social exclusion: follow-up
study of antisocial children into adulthood. BMJ, 323,
28 July 2001.
Shepherd, J., Shepherd, I., Newcombe, R. &
Farrington, D. (2009) Impact of antisocial lifestyle on
health: chronic disability and death by middle age.
Journal of Public Health, 31(4), 506-511.
Snell, T., Knapp, M. et al. (2013) Economic impact
of childhood psychiatric disorder on public sector
services in Britain: estimates from national survey
data. Journal of Child Psychology and Psychiatry,
54(8), 977-985.
Social Research Unit (2013) Investing in Children.
Available at http://investinginchildren.eu/.
Spencer, N., Devereux, E. et al. (2005) Disabling
conditions and registration for child abuse and
neglect: a population-based study. Pediatrics,
116(3), 609-613.
Stewart-Brown, S. & Schrader-Mcmillan, A. (2011)
Parenting for mental health: what does the evidence
say we should do? Health Promotion International,
26(S1), i10-i28.
Verona, E. & Sachs-Ericsson, N. (2005) The
intergenerational transmission of externalizing
behaviors in adult particpants: the mediating role
of child abuse. Journal of Consulting and Clinical
Psychology, 73(6), 1135-1145.
Washington State Institute for Public Policy (2013)
Benet-Cost Results. Available at
www.wsipp.wa.gov/BenetCost.
Webster-Stratton, C., Rinaldi, J. & Reid, J. (2011)
Long-term outcomes of Incredible Years parenting
program: predictors of adolescent adjustment. Child
and Adolescent Mental Health, 16(1), 38-46.
Wolke,D.,Woods,S.,Bloomeld,L.&Karstadt,L.
(2000) The association between direct and relational
bullying and behaviour problems among primary
school children. Journal of Child Psychology and
Psychiatry, 41(8), 989-1002.
Muntz, R., Hutchings, J. et al. (2004) Economic
evaluation of treatments for children with severe
behavioural problems. Journal of Mental Health
Policy and Economics, 7, 99-106.
Murray, J., Irving, B., Farrington, D., Colman, I. &
Bloxsom, A. (2010) Very early predictors of conduct
problems and crime: results from a national cohort
study. Journal of Child Psychology and Psychiatry,
5(11), 1198-1207.
NationalAuditOfce(2011)The Cost of a Cohort
of Young Offenders to the Criminal Justice System.
Available at www.nao.org.uk/wp-content/
uploads/2010/12/1011663_technical_paper.pdf.
National Institute for Health and Clinical Excellence
(2013) Antisocial Behaviour and Conduct Disorders in
Children and Young People: Recognition, intervention
and management. National Clinical Guideline
Number 158. Available at publications.nice.org.
uk/antisocial-behaviour-and-conduct-disorders-in-
children-and-young-people-recognition-intervention-
cg158.
Odgers, C., Caspi. A. et al. (2007) Prediction of
differential adult health burden by conduct problem
subtypes in males. Archives of General Psychiatry,
64, 476-484.
OfceforNationalStatistics(2009)Social Trends
2009, chapter on crime and justice. Available at
www.ons.gov.uk/ons/rel/social-trends-rd/social-
trends/social-trends-39/index.html.
Pajer, K. (1998) What happens to “bad” girls? A
review of the adult outcomes of antisocial adolescent
girls. American Journal of Psychiatry, 155, 862-870.
Patterson, G, DeBaryshe, B & Ramsey, E. (1989) A
developmental perspective on antisocial behaviour.
American Journal of Psychology, 44, 329-335.
Prinz, R., Sanders, M. et al. (2009) Population-based
prevention of child maltreatment: the US Triple P
System population trial. Prevention Science, 10,
1-12.
Ramrakha, S., Bell, M. et al. (2007) Childhood
behaviour problems linked to sexual risk taking
in young adulthood: a birth cohort study. Journal
of the American Academy of Child and Adolescent
Psychiatry, 46(10), 1272-1279.
Romeo, R., Knapp, M. & Scott, S. (2006) Economic
cost of severe antisocial behaviour in children – and
who pays it. British Journal of Psychiatry, 188, 547-
553.
Rutter, M., Kim-Cohen, J., & Maughan, B. (2006)
Continuities and discontinuities in psychopathology
between childhood and adult life. Journal of Child
Psychology and Psychiatry, 47(3), 276-295.
Published January 2014
Photograph: istockphoto.com/ayala_studio
£10 where sold
© Centre for Mental Health, 2014
Recipients (journals excepted) are free to copy or
use the material from this paper, provided that the
source is appropriately acknowledged.
Register for our monthly email bulletins and get copies of our publications at
www.centreformentalhealth.org.uk
Centre for Mental Health
134-138 Borough High Street, London SE1 1LB
Tel 020 7827 8300
Fax 020 7827 8369
www.centreformentalhealth.org.uk
Charity registration no. 1091156. A company
limited by guarantee registered in England and
Wales no. 4373019.
REPORT
Building a better future
Michael Parsonage, Lorraine Khan & Anna Saunders
The lifetime costs of childhood behavioural
problems and the benefits of early intervention
Building a better future