The South Australia
Early Childhood Data Project (SA ECDP)
John Lynch
Professor of Epidemiology and Public Health
University of Adelaide
Adult social and health
disadvantage
Labour market and income
disadvantage
Child health disadvantage
birth
Developmentaldisadvantage
0-2
Early learning disadvantage
3-5
Educationaldisadvantage
6-17
Adult social and health
disadvantage
Inter-generational social and health inequality
Early Childhood Systems
Health Education
- 9 months 0 1 2 3 4 5 6 …
Healthy BirthHealthy ChildImmunization
Maternal Health
HealthyPregnancy
Feeding, GrowthSleeping, Settling, Attachment
Child careLanguage
Reading
PreschoolSelf regulation
Play
LiteracyNumeracy
Cognitive abilitySocioemotional development
HearingVision
Child Development
The Early Childhood Government ‘System’
Child Protection
Child Protection
5
EMPOWER CREHealth systems, disadvantage and child well-being
CIA Lynch Epidemiology Public HealthCIB Sawyer Child psychiatry Robinson Research InstCIC Mol OBGYN Robinson Research InstCID Roberts Placental physiology Robinson Research Inst CIE Dekker OBGYN Robinson Research Inst CIF Stocks GP MedicineCIG Schurer Economics U SydneyCIH Gurrin Biostatstics U MelbourneCII Dwyer CE, SA Women’s and Children’s Health NetworkCIJ Director, SA Child and Family Health Service
6
Interventional Epidemiology
Platform 2: Follow-up & Scale-upWhole-of-population linked data for
children in SA and NTN~ 300,000
Platform 1: Pragmatic RCTs and Natural Experimentsin pregnancy, postnatal and community-based care
Platform 4: Methodology CoreRisk Stratification
Simulation of RCTs in observational dataCausal mediation analysis
Human capability production functionCost:Benefit analysis
Platform 3: Cohort Simulation of RCTs & Risk StratificationCohorts include - SCOPE; LSAC; LSIC; HILDA (Aus)
ALSPAC, MCS (UK)
7
Platform 2
Data Linkage
8
SA Early Childhood Data Project
• Every complete birth cohort
• Born 1999-2014
• N = ~ 300,000 children
• N = ~ 12,000 Aboriginal children
• Over 7 million records
• ~ 20 different government sources
• Public good resource - ~ 20 approved
researchers including within government
SA PoliceDASSACAMHS
9
Perinatal Epidemiology
Potentially Preventable
HospitalisationsService Change
One Platform Multiple purposes
Less More “translational”
10
Standard Epidemiological Research Outputs
11
Anemia in Pregnancy and AEDI Vulnerability (N = 13,654 imputed)
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
UnadjustedAdjusted*
* Adjusted for twin, maternal age, smoking in pregnancy, no. antenatal visits, parity, inter-pregnancy interval, maternal occupation, paternal occupation, Aboriginal or Torres Strait Islander status, remoteness, SEIFA-IRSD.
Smithers et al. Pediatric and Perinatal Epidemiology (2014)
12
Gestational Age (weeks) and AEDI vulnerability at age 5 (N = 12,601)
0.8
1
1.2
1.4
1.6
1.8
2
2.2
37 38 39 40 41 42-45
CommunicationEmotionalLanguagePhysicalSocialOne or more
Weeks of gestation
Adju
sted
ORs
37 weeks: 7%38 weeks: 20%39 weeks: 23%
40 weeks: 37%41 weeks: 13%42-45 weeks: 0.7%
Smithers et al. BJOG (2015)
13
14
More “Translational” Research Outputs
15
Infographic that accompanied release of the Nyland Commission Report
BetterStart Research Report March 2017
AGE
0.0%
5.0%
10.0%
15.0%
20.0%
25.0%
30.0%
age 0 1 2 3 4 5 6 7 8 9 10
Cumulative Incidence of Child Protection Contact Types (1999 – 2015)Denominator = Total Population of SA Children (~128,000)
1 in 4 childrenby age 10
Notification
Screened in
Investigation
Substantiation
Out-of-Home-Care
2% or 1 in 50 children by
age 10
What does this look like for the rest of Australia?
2010
“…slightly more than 1 in 4 (26.7%) of all NSW children and young people under the age of 18 were ever reported to Community Services by the end of 2008/09.”
“…two in 100 NSW children and young people currently under the age of 18 had ever lived in OOHC by the end of 30 June 2009.”
2012
“The likelihood that a child born in 2011 will notified by the time they are aged 18, is 23.5% - slightly less than 1 in 4…”
2013
The number of children known to Child Safety Services… 1 in 4.2 of all Queensland Children and 1 in 1.6 Aboriginal and Torres Strait Islander children”
2016
“By 11 years of age, 7.8% of Aboriginal children had entered out-of-home-care compared to 0.96% of non-Aboriginal children”
1.4% of all children will have experienced OOHC by age 10
2008
By age 16, 22.5% of children born in 1991 were subject to at least one child protection notification
BetterStart Child Health and Development Research Group | http://health.adelaide.edu.au/betterstart/
UK1.8% of all children born 2001-2003 experienced OOHC by age 10 (Mc Grath-Lone, 2016)
France1.9% of all children experienced OOHC by 2011( Seraphin, 2017)
19
The cumulative incidence of contact with the child protection system in South Australia by age 10
What’s happened over time?
20
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Cum
ulat
ive
inci
denc
e (%
)
Age (years)
Cumulative incidence of notifications to the child protection system by age and year of birth, 1999-2005
1999
2000
2001
21
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Cum
ulat
ive
inci
denc
e (%
)
Age (years)
Cumulative incidence of notifications to the child protection system by age and year of birth, 1999-2005
1999
2000
2001
2002
2003
2004
22
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Cum
ulat
ive
inci
denc
e (%
)
Age (years)
Cumulative incidence of notifications to the child protection system by age and year of birth, 1999-2005
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
23
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Cum
ulat
ive
inci
denc
e (%
)
Age (years)
Cumulative incidence of notifications to the child protection system by age and year of birth, 1999-2005
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
1999: 3% <1
2013: 8% <1
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
age 0 1 2 3 4 5 6 7 8 9 10
Notification Screened in notification Investigation Substantiation Out of home care
Cumulative Incidence of Child Protection Contact Types (1999 – 2015)Denominator = Number of Children Notified to Child Protection (n=33,235)
First 1400 days 44%
46%
56%
51%
62%
16% < 1
No contact with Child Protection
1+ contact, not screened-in
1+ screened-in contact, no
investigation
1+ investigation,
no substantiation
1+ substantiation,
no out of home care
1+ out of home care
episodeTotal
N (column %) N (column %) N (column %) N (column %) N (column %) N (column %) N (column %)Physical health and wellbeing
Not vulnerable 12,578 (92.8) 782 (82.7) 846 (80.6) 349 (75.7) 254 (67.9) 121 (68.0) 14,930 (90.2)
Vulnerable 975 (7.2) 164 (17.3) 203 (19.4) 112 (24.3) 120 (32.1) 57 (32.0) 1,631 (9.8)
Social competence
Not vulnerable 12,524 (92.4) 774 (81.9) 840 (80.1) 340 (73.7) 284 (75.9) 116 (64.8) 14,878 (89.9)
Vulnerable 1026 (7.6) 171 (18.1) 209 (19.9) 121 (26.3) 90 (24.1) 63 (35.2) 1,680 (10.1)
Emotional maturity
Not vulnerable 12,413 (92.0) 774 (82.3) 845 (81.0) 346 (75.2) 290 (78.4) 116 (65.9) 14,784 (89.7)
Vulnerable 1076 (8.0) 166 (17.7) 198 (19.0) 114 (24.8) 80 (21.6) 60 (34.1) 1,694 (10.3)
Communication and general knowledge
Not vulnerable 12,698 (93.7) 821 (86.8) 898 (85.6) 385 (83.7) 295 (78.7) 144 (80.9) 15,241 (92.0)
Vulnerable 853 (6.3) 125 (13.2) 151 (14.4) 75 (16.3) 80 (21.3) 34 (19.1) 1,318 (8.0)
Language and cognitive skills
Not vulnerable 12,931 (95.6) 841 (89.3) 918 (87.8) 376 (81.7) 281 (75.7) 142 (79.3) 15,489 (93.7)
Vulnerable 601 (4.4) 101 (10.7) 128 (12.2) 84 (18.3) 90 (24.3) 37 (20.7) 1,041 (6.3)
Total* 13,532 (100.0) 942 (100.0) 1,046 (100.0) 460 (100.0) 371 (100.0) 179 (100.0) 16,530 (100.0)
Developmental vulnerability at age 5 on the Australian Early Development Census- All children
*Total numbers vary across domains due to differing response rates
~ 3 times higher vulnerability
~ 2 times higher vulnerability
26
0.0%
10.0%
20.0%
30.0%
40.0%
50.0%
60.0%
70.0%
80.0%
90.0%
100.0%
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15Other Govt Anonymous Child Care FSAFamily/Friend/Neighbour Family Law system Health Non-GovtOther Police School/preschool Self report/Unknown
Pattern of reporters to child protection system by age of the child
40%
27
A Public Health Approach
To Early Intervention
28
1,364 childrennot screened in
1,649 childrennot investigated
829 childrennot substantiated
4,700 children in every birth cohort~25% of children
At least 1Notification
SA Population by age 10
(20,000 births per year)
3,842 children per birth cohort (~80%)no further action by
child protection agency
3,336 children in every birth cohort~19% of children
Screened inNotification
1,687 children in every birth cohort~10% of children
Investigated
858 children in every birth cohort~5% of children
Substantiated
Out-of-home Care
355 children in every birth cohort~1.8% of children
29
Whole-of-Population Data Linkage – South Australia(Health, Welfare, Justice, Child Protection, Education…)
Service Delivery – Evaluationsincluding services offered by Government and NGOs
Federal Government Data(AEDC, Centrelink, PBS, MBS … Census)
Pragmatic RCTs - Innovations
SA Early Childhood Data Project – Information, Evaluation, and Innovation
Integrated research information system + government ‘business intelligence’ system –
combines linked administrative data, cohorts, service delivery evaluations and RCTs
Enriched phenotypes from cohorts, special collections and
genetic data
Info
rmat
ion
Syst
ems
Eval
uatio
n an
d In
nova
tion
SCOPE GuthrieCardsPrems
data
30
Use linked administrative data to routinely and cost effectively evaluate programs
and service delivery innovations
• using RCTs for medium term follow-up in school
• propensity score matched comparison groups
• regression discontinuity designs
• instrumental variables
Data Linkage and Routine Evaluation Studies
31
• IF we knew who was in and not in the systems – coverage in the population
• IF we knew which services children and parents accessed
• and IF we were able to regularly, routinely collect relevant outcomes like child development, mental health, physical health, …
• and we were able to create, document and measure changes in service provision,e.g., “integration” such as Children’s Centres
• Then a population-wide data linkage information system could be used to helpquantitatively evaluate service innovations on a routine, cost-effective andsustainable basis
• And potentially in close to real time to provide clinical and practice feedback to improve service quality
Information Systems and Improving Effectiveness of Services
32
The science of the “bleeding obvious” …?