To quantify developmental vulnerability at age 5 by child protection contacts before school in two Australian states.

All children with birth, child protection and/or 2009, 2012, 2015 and 2018 Australian Early Development Census (AEDC) data in New South Wales (NSW) and South Australia (SA) were grouped according to child protection contact before school: no contact, child protection reports, screened-in reports, investigations, substantiations and out-of-home care (OOHC). The outcome was developmental vulnerability on ≥1 AEDC domains or medically diagnosed conditions with support needs at school entry.

56 650 (14.2%) NSW children and 12 617 (15.6%) SA children had ≥1 child protection contact before school. Developmental vulnerability on ≥1 domains or medically diagnosed conditions was lowest in the no child protection group (NSW, 21–22%; SA, 24–25%), with progressively higher risk in the child protection report (NSW, 35%; SA, 41–46%) through to the OOHC (NSW, 50–54%; SA, 59–66%) groups in all AEDC years. Developmental risk was higher among children aged <2 years at first contact and those with more reports. Children with only one child protection report before school had approximately 65% higher developmental risk than the no child protection group in both states.

A single child protection report before school was an early indicator of higher developmental risk at age 5, with higher developmental risks among children with earlier, more serious and frequent child protection contacts. Beyond child safety screening, child protection reports represent an opportunity to mobilise early health and social support for children with developmental support needs.

Data may be obtained from a third party and are not publicly available. No data are available because Australian privacy laws prohibit us from making the individual-level de-identified data publicly available. The data used for this study were provided by several Australian State and Commonwealth government agencies under agreements with the researchers led by KF (in NSW) and JWL (in SA), the NSW Centre for Health Record Linkage (CHeReL) and SANT Datalink, following approval from multiple ethics committees and data custodians. Data are only able to be accessed by researchers who are approved users by the relevant ethics committees and data custodians. Data can be accessed through an application and approval process administered by the independent data linkage authorities, NSW CHeReL or SANT Datalink.

https://creativecommons.org/licenses/by/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/.

Children who experience abuse and neglect have worse health, developmental and long-term life outcomes than other children.1–3 Numerous reports have highlighted the crisis-oriented child protection system and the urgent need to invest in the prevention of child maltreatment and its consequences.4–9 The planning and implementation of primary and secondary prevention measures such as targeted supportive interventions relies on population-level data infrastructure and intelligence on early indicators of a child’s future risk of poor outcomes that services, if properly resourced, can respond to. If early contact with the child protection system reliably indicates the population burden of future developmental risk for children, such descriptive epidemiological evidence can inform resourcing and mobilisation of health system-led screening, assessment and early, non-judgemental support focused on child health and development in at-risk families.

There are prevention opportunities at every stage of the child protection system. The first indicator of risk is an initial report of safety concerns from health professionals and others to the child protection helpline. These reports may then be screened-in, investigated, substantiated and possibly lead to a court order to remove the child to out-of-home care (OOHC). Five previous publications have used whole population data on child protection and teacher-reported developmental vulnerability at school entry; all excluded children with medically diagnosed conditions with substantial support needs.2 10–13 One Canadian report showed developmental vulnerability among more than half of children in OOHC, 44% of children with other child protection contacts and one-quarter of children with no child protection contact.2 Another Canadian study used similar data from Manitoba, comparing developmental vulnerability among children in OOHC versus all other children, thus combining children with other child protection contacts and those who had no contact.13 Three Australian studies10–12 all estimated associations between different categorisations of child protection contact on developmental vulnerability at school entry. The modelling process involved adjustment for various covariates, such as socioeconomic and perinatal health characteristics. Covariate adjustment creates an ‘imaginary world’14 where the estimated quantity (eg, an odds ratio) represents the association between an exposure and outcome as if all individuals had the same level of the covariate. Of course, in the real world they do not. Such estimates are not quantities that are actionable by policymakers. Paraphrasing Rose et al,—relative risk is for researchers, decision-making requires absolutes.15

The purpose of the current study was to generate actionable targets for public health prevention, using best practice descriptive epidemiology,16 17 by quantifying the burden of teacher-reported developmental vulnerability on ≥ 1 domains and medically diagnosed conditions at school entry across all levels of child protection contact in whole child populations. We do this in two Australian states with different child protection legislation and practices and across four whole population censuses of child development at age 5 in 2009, 2012, 2015 and 2018.

A key finding from this two-jurisdiction study is that even a single child protection report in the first 2000 days of children’s lives was a robust indicator of developmental risk at age 5. It is perhaps not surprising that children with higher levels of child protection contact, such as substantiations or OOHC, have higher developmental risks, but we also showed that children with only a single report before age 5 without any more serious child protection contact had 1.6–1.8 times higher relative risk of developmental vulnerability on ≥ 1 domains or medically diagnosed conditions. While these reports to child protection were not considered ‘child protection matters’, they nevertheless carry information on the risk of poor developmental outcomes by age 5. While this may, in part, reflect the higher proportion of socioeconomic and health disadvantage among children in contact with child protection, it nevertheless represents the true population burden. An accurate description of the population burden linked to child protection contact requires crude risk estimates.16 27 So beyond its role in ensuring the immediate safety of children, information reported to child protection is also an important source of public health insight that could be better used to inform supportive targeted prevention responses.

Despite different child protection systems and procedures, we found that the scale and patterns of developmental risk among child protection populations were strikingly similar in two Australian jurisdictions and over almost a decade. In both states, one in seven children had a child protection report or more serious contact before school. There was progressively higher developmental risk among whole populations of children with more serious child protection contact, an earlier age at first child protection contact and more child protection reports before starting school. A previous Canadian study also found that children in OOHC had the highest risk of teacher-reported developmental vulnerability; however, children with previous medically diagnosed conditions were excluded from the study population and outcomes were not examined across other levels of child protection system contact.2 Even within each of the most serious child protection contact groups in our study, children with more reports had a higher developmental risk. This highlights the strength of the developmental risk signal that child protection reports carry at the population level, beyond the system’s own triaging processes rightly focused on immediate safety.

For the first time, we show that children with medically diagnosed conditions are disproportionately represented in early childhood child protection populations, in addition to children identified as developmentally vulnerable by teachers. Although children with previous medically diagnosed conditions have been assessed by the health system prior to school entry, it is important that their substantial health and development support needs are counted, alongside those with previously unidentified developmental vulnerability, in any data and intelligence that may be used to plan the supportive interventions from the health, child protection, early education and school systems.

This study highlights the importance of ensuring the developmental support needs of children reported to child protection, including First Nations children, are a focus in holistic and culturally responsive prevention approaches to promote the best outcomes for children before starting school. Similar to countries such as Canada, USA and New Zealand,28–31 First Nations children were over-represented in child protection systems, before starting school, in both Australian states. Systemic racism is a driver of the ‘causes’ of child protection contact, such as poverty, family violence and substance use, and past and continuing child welfare policies and practice resulting in the over-surveillance, reporting and potential pathologising of First Nations children and removal of children from their families.32 33 Recent reviews of child protection systems4 5 34 and broader policies35–37 emphasise the need for urgent reform, both jurisdictionally and nationally.

A limitation of this study is that administrative data does not capture all maltreatment experiences or risks experienced by children.38 However, it is also the case that adult recall of child maltreatment in interviews and surveys under-ascertains safety concerns reported to child protection.38 Given it is not feasible or ethical to collect self-reported maltreatment risk/experience from very young children, child protection administrative data offers the earliest opportunity to inform prevention in whole child populations.

Child protection reports represent an underused asset to plan and deliver—at the earliest possible opportunity—universal and targeted support services for children at higher developmental risk during early childhood. The health system has near-universal reach during pregnancy, birth and the early postnatal period, when it is common to screen for at-risk infants and families and make referrals for assessment and support. Child protection services do not have a statutory responsibility to respond to the unmet needs of children who are not screened in at risk of ‘serious harm’ (safety) or further investigated. However, there is potential to re-envisage the use of child protection reports to inform responses from adjacent agencies, such as the health system, as part of a ‘child health and well-being system’ response to support child development. While progress towards multiagency, whole-population data assets varies across jurisdictions internationally,39 40 this study demonstrates the potential to generate timely, contemporary, actionable targets for public health prevention in whole child populations using data on system contacts and outcomes that are routinely collected by different government agencies.

This two-jurisdiction study has attempted to reframe how data on child protection contacts—experienced by one in seven children by school age—could be used to inform supportive services led by agencies in health, human services and early education and care for children. While the child protection system’s primary function is rightly to deliver statutory responses to safety risks and actual harm, health, human services and early education and care are well placed to deliver non-stigmatising, supportive interventions to children and families with early indicators of developmental risks.

Data may be obtained from a third party and are not publicly available. No data are available because Australian privacy laws prohibit us from making the individual-level de-identified data publicly available. The data used for this study were provided by several Australian State and Commonwealth government agencies under agreements with the researchers led by KF (in NSW) and JWL (in SA), the NSW Centre for Health Record Linkage (CHeReL) and SANT Datalink, following approval from multiple ethics committees and data custodians. Data are only able to be accessed by researchers who are approved users by the relevant ethics committees and data custodians. Data can be accessed through an application and approval process administered by the independent data linkage authorities, NSW CHeReL or SANT Datalink.

We acknowledge the children and families of New South Wales (NSW) and South Australia (SA) whose data are included in this study, including those with experiences of child protection systems. We thank the NSW Centre for Health Record Linkage (CHeReL) and SANT Datalink for managing and conducting the data linkage for the NSW Child E-Cohort Project (led by KF) and the BEBOLD (Better Evidence, Better Outcomes, Linked Data) data platforms (led by JWL). This study uses data from the Australian Early Development Census (AEDC). The AEDC is funded by the Australian Government Department of Education. We also thank the NSW Ministry of Health, NSW Register of Births, Deaths and Marriages, NSW Department of Education, NSW Department of Communities and Justice, SA Health, SA Births, Deaths and Marriages Registry and SA Department for Child Protection for use of the other data sources in this study. We thank Merran Butler and Caroline Anderson at the Families and Community Services Insights, Analysis and Research (FACSIAR), Department of Communities and Justice, for their review and feedback on the manuscript. We thank staff at the Aboriginal Health and Medical Research Council of New South Wales and the NSW Child, Family and Community Peak Aboriginal Corporation (AbSec) for their input on the study and manuscript, including Sally Cowling, Annaliese Gielingh and Shiny Varghese. The findings and views reported in this study are those of the authors and should not be attributed to any agency or government department.