Transcript Secular trends in adolescent mental health

Methods for testing trends in mental health – is it
really possible to compare ‘like with like’?
Dr Stephan Collishaw
Cardiff University
[email protected]
NCRM Research Methods Festival, Oxford, July 2014
Outline
• Prevalence and burden
• Trends in diagnosis – need for epidemiological data
• Cross-cohort comparisons using symptom screens
• Issues to think about in comparing ‘like with like’
• Results – adolescent mental health 1970s-2000s
• Replication and validation
• Trends in child mental health 1999-2008
Child psychiatric disorders:
Burden and prognosis

1 in 10 has a clinically significant psychiatric disorder

Impact on family life, friendships, learning

Suicide and self harm (3rd leading cause of death)

Long-term prognosis

Most child/adolescent disorders persist to adulthood

>50% of adult mental disorders have onset <18 years

Parenting, employment, social exclusion, illness, mortality

Economic burden
Green et al., 2005; Kim-Cohen et al, 2003; Thapar et al., 2012; Windfuhr et al., 2008; Maughan et al., 2014
Diagnosis and treatment:
autism, ADHD, depression, anxiety
•
Increased help seeking, diagnosis and treatment
•
Substantial increases in psychotropic medication
•
Rates increased 3-5 fold per decade 1990s and 2000s
•
Similar trends in many countries
Getahun et al., 2013; Kosidou et al, 2010; Olfson et al., 2014; Stephenson et al, 2013
Trends in diagnosis and treatment
•
Important for planning service provision
But:
Increases in referrals and diagnoses may be due to
•
Increased public awareness & clinical recognition
•
Changing diagnostic criteria and practice
•
Treatment availability and perceived efficacy
•
‘Medicalisation’ of normal behaviour?
Also:
Majority still don’t access services
Ford et al., 2007; Potter et al., 2012
Epidemiological evidence

Two major meta-analyses of depression and ADHD

Epidemiological studies using structured diagnostic interviews

Meta-analyses: no evidence of increase in depression or ADHD
But

Variability in methods (samples, measures, diagnostic system)

Rates of depression range from <1% to >25%

‘Noise’ and variability likely to make trends difficult to detect
Costello et al., 2006; Polanczyk et al., 2014
‘Like-for-like’ cross-cohort comparisons

Comparable representative cohorts with equivalent measures


e.g. UK cohorts since 1960s have included Rutter/SDQ
Threats to comparability

Selective attrition

Minor changes to questionnaire make a big difference


Disobedience: “applies somewhat” (33%) vs “sometimes” (75%)

Calibration can be effective for aligning non-identical instruments
Change in reporting
Goodman et al., 2007
UK cross-cohort comparisons: 1974-1999
• Large nationally representative surveys (NCDS, BCS70, BCAMHS)
assessed in 1974, 1986, 1999
• Age 15-16
• Parent rated Rutter or SDQ
• Emotional problems
• Conduct problems
• Hyperactivity/inattention
• Calibration data used to align SDQ and Rutter questionnaires
• Study-specific weights using prior predictors of non-response
Collishaw et al, 2004
Emotional problems: high scores
Cohort 3 vs. cohort 2
OR = 1.72
25
% high scores
20
15
Boys
Girls
10
5
0
1974
N = 10,499
1986
1999
N = 7,293
N = 868
Collishaw et al, 2004
% high scores
Conduct problems: high scores
18
16
14
12
10
8
6
4
2
0
Boys
Girls
1974
1986
1999
Total OR = 1.56 per cohort
Collishaw et al, 2004
Hyperactivity: mean scores
mean hyperactive score
1.5
1.25
1
Boys
Girls
0.75
0.5
0.25
1974
1986
1999
Collishaw et al, 2004
Limitations
• Only parent reports
• Imperfection of Rutter/SDQ calibration?
• Crude measures
• Are population shifts also occurring at extremes?
• What about ‘change in reporting’?
Need for replication and validation
Replication:
The Youth Trends study (1986 & 2006)
Two nationally representative surveys of English youth

1986: BCS70 age 16 (N = 9,766)

2006: HSE follow-up ages 16-17 (N = 747)

Identical self rated symptom screens (GHQ/Malaise)
Questions

Increase in youth-reported symptoms

Variation in trends by severity?
Collishaw et al, 2010
Adolescent emotional symptoms
(youth reports)
youth rated symptoms
2.5
2
Girls
Boys
ES = 0.36; p < .001
1.5
ES = 0.13; p = .06
1
0.5
0
1986
2006
Collishaw et al., 2010
Trends by severity
OR (2006 vs 1986)
3
2.5
2
1.5
1
0.5
0
1+
2+
3+
4+
5+
6+
7+
8+
Youth emotional problem score
cohort differences significant at all thresholds, p<.01; Interaction p < .05
Collishaw et al, 2010
General shift in reporting?
No change in hyperactivity
0.5
0.4
hyperactivity
(z-score)
0.3
0.2
0.1
1986
0
2006
-0.1
-0.2
-0.3
-0.4
-0.5
Boys
Girls
Collishaw et al, 2010
Do trends reflect a change in reporting?
•
•
Shift in informant ‘thresholds’?
(e.g. different expectations about normal behaviour)
Greater willingness to report problems than in the past?
But
•
•
Specificity of findings (no increase in hyperactivity)
Validation using external criteria desirable…
Conduct problems:
Age 30 outcomes NCDS & BCS70 cohorts
4
NCDS
BCS70
Odds ratio
3
2
1
0
Unemployed
Sacked
Benefits
Homeless
Teen parent Poor health
Collishaw et al, 2004
Adolescent conduct problems and risk of
pervasive adult dysfunction:
4+ adverse outcomes age 30
5
Odds ratio
4
3
2
1
0
NCDS
BCS70
Collishaw et al, 2004
Child mental health trends: 1999-2008

1999: BCAMHS 7-year olds (n = 1034)

2004: BCAMHS 7-year olds (n = 648)

2008: MCS 7-year collection (n = 13,489)

Parent & teacher SDQ symptoms & impact

Weights used to adjust for attrition and stratified design
Sellers et al, in press
Child mental health trends: 1999-2008
SDQ total and subscale mean scores all declined

Boys: total score effect size = -0.27

Girls: total score effect size = -0.12

Bigger drop in problem scores for boys than girls (p = 0.027)

Similar conclusions based on parent and teacher reports

Drop in children scoring in abnormal range (11%, 10%, 8%)

But: increase in impact of problems, e.g. classroom learning
Sellers et al, in press
Conclusions
•
Comparing ‘like-with-like’ essential for testing trends
•
Replication and validation important
•
Long-term change in adolescent mental health
•
Recent data: improvements in child mental health
•
Latest data 2008, what has happened since?
Acknowledgements
Barbara Maughan (KCL)
Andrew Pickles (KCL)
Robert Goodman (KCL)
Anita Thapar (Cardiff)
Ruth Sellers (Cardiff)
Frances Gardner (Oxford)
Jacqueline Scott (Cambridge)
Ginny Russell (Exeter)
National Centre for Social Research; Department of Health
Medical Research Council; Nuffield Foundation; Waterloo Foundation
References
Collishaw et al (2004). Time trends in adolescent mental health. J Child Psychol Psych, 45, 1350-1362.
Collishaw et al (2010). Trends in adolescent emotional problems in England. J Child Psychol Psych, 51, 885-94.
Costello et al (2006). Is there an epidemic of child and adolescent depression? J Child Psychol Psych, 47, 1263-71
Ford et al (2007). Child mental health is everybody’s business. Child Adolescent Mental Health, 12, 13-20.
Getahun et al (2013). Recent trends in childhood ADHD. JAMA Pediatrics, 167, 282-8.
Goodman et al (2007). Seemingly minor changes to a questionnaire. Soc Psych Psych Epi, 42, 322-327.
Green et al (2005). Mental health of children and young people in GB, 2004. Palgrave Macmillan
Kim-Cohen et al (2003). Prior juvenile diagnoses in adults with mental disorders. Archives General Psychiatry, 60. 709-17
Kosidou et al (2010). Recent trends. Acta Psychiatrica Scandinavica, 22, 47-55.
Maughan et al (2014). Adolescent conduct problems and premature mortality. Psych Med, 44, 1077-86.
Olfson et al (2014). National trends in the mental health care of children, adolescents and adults. JAMA Psych, 71, 81-90
Polanczyk et al (2014). ADHD prevalence estimates across three decades. Int J Epidemiology, online first
Potter et al (2012). Missed opportunities mental disorder in children of parents with depression. BJGP, 62, e487
Sellers et al (in press). Trends in parent- and teacher-rated emotional, conduct. J Child Psychol Psych, in press.
Stephenson et al (2013). Trends in the utilisation of psychotropic medication. Austr New Zealand J Psychiatry, 47, 74-87.
Thapar et al (2012). Depression in adolescence. Lancet, 379, 1056-67.
Windfuhr et al (2008). Suicide in juveniles and adolescents in the United Kingdom. J Child Psychol Psych, 49, 1155-65