"No capacity to allocate" — The Department of Communities and Justice's practice of closing ROSH (Risk of Significant Harm) reports
No capacity to allocate - DCJs practice of closing ROSH reports
PDF - 2MB
No capacity to allocate - DCJs practice of closing ROSH reports
PDF - 2MB
A newly released NSW Ombudsman investigation has denounced DCJ’s longstanding practice of closing ROSH reports on the stated basis that it has inadequate resourcing to undertake the necessary investigation and assessment of risk. The Ombudsman has found that this practice is contrary to DCJ’s statutory responsibilities, but has become ingrained as ‘business as usual’. As a result of these findings, the NSW Ombudsman has recommended DCJ conduct a comprehensive review and reform of its ROSH (and related) policies, practices and tools.
It is generally accepted, and is recognised by the UN Convention on the Rights of the Child, that the family is the fundamental group of society and the natural environment for the growth and well-being of all its members and particularly children. Parents or legal guardians have the primary responsibilities for safeguarding and promoting the welfare and wellbeing of their children.
However, and again as the UN convention makes clear, the government has a duty to ensure all children receive such protection and care as is necessary for their well-being, taking into account the rights and duties of their parents or legal guardians. To this end, it is required to take all appropriate legislative and administrative measures to do so.
This includes, where a child is at risk of significant harm – including serious physical, sexual or emotional abuse or neglect – the government has a responsibility to take such steps as are necessary to protect them. This does not necessarily, or even usually, mean removing them from the family unit, and can include a range of interventions and services to support families, address risk and promote safety and wellbeing.
Critical to the NSW government’s performance of its child protection responsibilities is the legally-prescribed system of receiving, investigating and assessing ‘ROSH reports’.
A ROSH report is a report made to the Department of Communities and Justice (DCJ) when a person reasonably suspects that a child is ‘at risk of significant harm’ (ROSH). Many of these reports come from ‘mandatory reporters’, those who have a legal duty to report to DCJ if they suspect a child is at ROSH – including teachers, police, health workers and childcare workers.
DCJ receives over 240,000 ROSH reports each year. ‘Re-reports’ are common – on average, each child reported will be the subject of 2 reports.
DCJ has a statutory duty to consider each ROSH report it receives, and to undertake such investigation and assessment as it considers necessary to determine whether or not the child is at ROSH. Assessing a child as being at ROSH does not dictate what happens next in terms of the action DCJ may take. However, it is the critical first step in the statutory regime toward identifying those children most in need of the State’s care and protection, and then what form that should take.
For decades, oversight bodies – including the Ombudsman, the Audit Office, the Coroner’s Court and Parliamentary Committees – have been raising concerns at the large number of ROSH reports that DCJ ‘closes’ citing a lack of resources to investigate and assess.
More than two thirds of the ROSH reports received by DCJ in any year are closed with a label of ‘no capacity to allocate [to a caseworker]’ (NCTA) and others are closed with a similar label of ‘competing priorities’. This performance has not been improving, including after the threshold for reporting and responding was lifted from ‘risk of harm’ to ‘risk of significant harm’ following the special commission of inquiry into child protection almost 20 years ago.
Previous public reports and commentary have tended to focus on this top-level concern – the high percentage of ROSH reports closed without apparent investigation – to rightly call out a system that is evidently not working as it should. Others have reviewed or highlighted individual cases when – following a death or other adverse event – it becomes apparent that a particular child had been ‘known to DCJ’ through previous ROSH reports.
However, the high-level numbers are not improving, and we still see individual cases where ROSH reports were closed citing a lack of capacity and there are later, sometimes tragic, outcomes.
In this investigation we have sought to contribute something new. While acknowledging the resource-constrained environment under which DCJ of course operates, we focus on a closer examination of what DCJ’s statutory responsibilities actually require, and examine the policies and day-to-day practices currently used by DCJ to perform them. While this has included a sampling review of some cases (and some of these are included as illustrative case studies in this report), we have not investigated or made findings about the conduct of DCJ in respect of any particular case, and nor have we investigated the decision-making or conduct of any individual officers of DCJ.
Instead, we hope that by undertaking a close analysis in this area, this report will contribute to specific and meaningful lessons, and resultant reform. Reform should consider all of those things together – law, policy and practice – to bring coherence and greater simplicity to the ROSH system, and to address the current misalignment, complexity and confusion that is prevalent, including within DCJ itself.
Our investigation acknowledges that DCJ’s child protection work is, and always will be, difficult. DCJ is also beset by significant workforce and resourcing challenges. Caseworkers have some of the most difficult, yet also important and impactful jobs, in society. They should be supported by laws, policies and practices that are clear and consistent, and that support them to make the best decisions to identify and respond to those children most at risk.
Paul Miller
NSW Ombudsman
The Department of Communities and Justice (DCJ) is the lead agency for the NSW child protection system. Its functions are primarily governed by the Children and Young Persons (Care and Protection) Act 1998 (Care Act).
Under the Care Act, anyone who suspects a child is at risk of significant harm (ROSH) may make a report to DCJ’s Child Protection Helpline (Helpline).
A child is at ROSH when certain circumstances are present, to a significant extent, that mean that there are current concerns for that child’s safety, welfare or wellbeing. Those circumstances include physical and sexual abuse, neglect (including physical, psychological, medical and educational neglect), exposure to domestic violence, and serious psychological harm.[1]
People such as teachers, police, health workers and childcare workers are mandatory reporters – they must report when they have reasonable grounds to suspect, in the course of their work, that a child is at ROSH. Most of the approximately 247,000 ROSH reports that DCJ receives annually are made by mandatory reporters.
DCJ is required by the Care Act to assess any ROSH report it receives, including by making any investigation and assessment considered necessary to determine whether the child in question is, in fact, at ROSH. DCJ may decide to take no further action in response to a report if it considers there is insufficient reason to believe the child is at ROSH.
Assessing whether a child or a young person[2] is at ROSH is one of DCJ’s core responsibilities and is typically the first step toward considering any child protection interventions in respect of that child.
DCJ’s current practice is that any reports concerning children that are received at the Helpline are ‘screened’ as either a ROSH report or a non-ROSH report. A ROSH report is one which, on its face, appears to meet the statutory ROSH definition. ROSH reports are ’screened-in’ and then transferred, typically to a Community Service Centre (CSC),[3] for allocation, investigation and assessment by a DCJ caseworker.
What happens next has been the subject of serious and repeated concerns expressed, over many decades, by oversight agencies and past inquiries, including in the context of coronial inquests and child death reviews conducted by the Ombudsman.
In particular, alarm has been raised about DCJ’s practice of closing a large proportion of the screened-in ROSH reports without a field assessment or face-to-face assessment having been undertaken by a DCJ caseworker, and without any determination being made as to whether the child is or is not at ROSH. Instead, the majority of such reports are closed on the stated basis that DCJ has ‘no capacity to allocate’ the report to a caseworker or, in a smaller number of cases, because the caseworker to which the report was allocated has other ‘competing priorities’.
In 2024, the NSW Ombudsman’s Protecting Children at Risk (PCAR) report included unpublished data for the 2022-23 financial year showing that as many as two thirds or 64% of the 112,592 children involved in ROSH reports that year were ‘screened-in’ as ROSH by the Helpline at the initial triage stage but were not allocated to a caseworker for further investigation because of ‘no capacity to allocate’. In that report, we announced that we would be undertaking this investigation.
The figures we have since obtained in this investigation show no improvement:
This investigation set out to establish whether the conduct of DCJ when assessing, responding to and closing ROSH reports may amount to or involve administrative wrong conduct (that is, maladministration) of a kind described in s 26 of the Ombudsman Act 1974.
In doing so we have sought to move beyond the obvious and well-recognised conclusion that a child protection system in which more than two thirds of ROSH reports are closed on the stated basis of inadequate resourcing is one that is not working as it should.
We have accordingly sought to examine in more detail DCJ’s practices when triaging and assessing ROSH reports – practices which it has itself described as a ‘black hole’.
Our methodology included a desktop review of earlier system reviews, coronial inquests, legislation and decisions of courts and statutory bodies, published and unpublished DCJ data and our complaints holdings. We conducted compulsory witness interviews of certain officials, and surveyed DCJ staff in relevant roles.
We also conducted a desktop review of the first 380 triage records where a ROSH report was closed with the label ‘no capacity to allocate’ during June 2023 (Chapter 1 and Appendix A-4).
Some of the cases we reviewed are summarised in de-identified case studies where they illustrate issues raised by the investigation. However, we have not investigated or made findings about the conduct of DCJ in respect of any particular case. Similarly, we have not investigated the decision-making or conduct of any individual officers of DCJ, who perform critical social work under challenging circumstances.
While we have proceeded without making assumptions about the impact of resources on the large volume of ROSH reports that are ‘closed’, we also acknowledge the practical reality that DCJ operates within a context of finite resources, with limited financial funding and workforce resourcing challenges.
This report starts (as appropriate) with an examination of the Care Act’s requirements and DCJ’s statutory responsibilities (Chapter 2), before considering DCJ’s policies and practices (Chapter 3).
The key issues raised by the investigation are discussed in Chapters 4 to 9, and these are summarised below.
The closure problem is significant
The statutory expectation of the Care Act is that DCJ will respond by investigating and assessing each and every ROSH report it receives, where there is reason to believe a child is at ROSH. Clearly that is not happening.
However, while only a quarter of reports screened-in as ROSH are receiving a face-to-face assessment[4] by a caseworker, this observation alone does not demonstrate that DCJ has failed its statutory duty in every one of the three quarters of cases where no such assessment was conducted. This is because the Care Act does not expressly require a face-to-face assessment in all cases.
In some of the 75% of cases that are being closed, including those that are closed on the stated basis of ‘no capacity to allocate’, it is possible that a DCJ caseworker had in fact, and consistently with the Act, made a decision (based on the report itself and/or any desktop investigations) that there were no concerns that the child is at ROSH and that face-to-face investigation was not necessary.
However, our review of sample cases and our survey of staff indicates that this does not apply to the majority of closures. Rather, in most cases, reports are being closed on resourcing grounds and without field assessment despite DCJ continuing to have concerns, and in some cases serious concerns, that the child is at ROSH.
In the sample of cases we reviewed (where reports had been closed due to ‘no capacity to allocate’), in more than two thirds of them there was no evidence on the record to suggest that any view had been formed that the child was not at risk, or that a face-to-face assessment was not necessary. This conclusion was borne out by our survey of relevant DCJ staff, most of whom confirmed they ‘regularly’ close ROSH reports despite not being satisfied that the child was not at ROSH.
Whenever DCJ closes a ROSH report because of inadequate resources and without undertaking such investigations as it considers necessary to determine whether a child is at ROSH, it is failing its statutory responsibility.
The 28-day closure rule does not align with legislative intent
DCJ’s Triage Policy generally provides, with some exceptions, that reports screened-in as ROSH but not allocated for a face-to-face assessment within 28 days of receipt by the Helpline are to be closed.
Although data weaknesses mean the number of ROSH reports closed based on the 28-day rule is unknown, it is likely to be significant. In our sample, 28% of the cases closed due to ‘no capacity to allocate’ had recorded the reasons as being ‘exceeded or likely exceeded timeframes’.
The 28-day rule is being applied to cases where DCJ has reason to believe, including in some cases ‘serious concerns’, that the child is at ROSH, and where it has not undertaken the necessary investigations to dismiss those concerns.
The application of such a rule does not align with the Care Act. Although the Care Act does not fix any specific deadline for decisions, it does require that all reports will be the subject of a decision under section 30 – that is, either a decision to take such action as will enable a ROSH determination to be made, or a decision that no further action is required because there is insufficient reason to believe a child is at ROSH.
The question of whether face-to-face or field assessments are required in all cases
There appears to be a widespread, though we understand contested, view within DCJ that once a report has been screened-in as a ROSH report, a determination as to whether the respective child is (or is not) in fact at ROSH can never be made without a face-to-face assessment.
The Care Act requires DCJ to conduct such investigations as it considers necessary, without prescribing what form those investigations must take. There will, of course, be good reason in many cases why DCJ would and should take the view, when considering a particular report, that face-to-face investigation is necessary. This is particularly so having regard to the nature of the risks that are reported in ROSH reports, which typically concern harms that occur within families and behind closed doors.[5]
However, we raise this point because it appeared to us, from our review of sample cases, that there are (a relatively small) sub-set of cases where DCJ has in fact closed the case following desktop investigation, in substance because those investigations had allayed concerns. Even so, the closure was labelled as being on the basis of ‘no capacity to allocate’ because there had been no ‘face-to-face’ assessment as such.
This practice – and the inability of caseworkers to record full and accurate reasons for why they have closed a report beyond citing resource constraints – obscures the situation. It makes it difficult to be certain as to when resourcing constraints actually resulted in DCJ failing to fulfil its responsibility under the Act – that is, when it has closed a ROSH report while still having reason to believe the child or young person is at ROSH.
Allocation does not always result in timely and adequate support
Even when reports are allocated for a face-to-face assessment, DCJ has struggled to meet the response timeframes set by the Helpline during the initial triage. DCJ’s data for 2023-24 presented at Budget estimates in March 2025 showed the following:
Closing cases on the stated basis of resource limitations has become standard practice for DCJ. It is ‘business as usual’.
There is an apparent lack of any routine post-closure scrutiny of closure decisions, which reinforces a cultural norm that closing ROSH reports with labels such as ‘no capacity to allocate’ whenever a report does not result in a face-to-face assessment is an acceptable and even a preferred basis for closing reports.
Default use of ‘no capacity to allocate’ and equivalent labels
There has been no attempt to ensure that, at the very least, the closure labels are applied only as a last resort and where the closure is genuinely and unavoidably because of resource constraints.
Instead, (as discussed in Chapter 4 and above) it appears that the labels of ‘no capacity to allocate’ and ‘competing priorities’ are used as default labels whenever a report has not had a face-to-face assessment – even if that may not have been the primary rationale for deciding to close the report.
DCJ has failed to provide guidance or tools to enable staff to accurately and comprehensively record any substantive assessment they may have made in relation to risk, their reasons for that assessment, or how the assessment has affected their decision to close the report.
Tolerating such an approach obscures the decisions that are actually being made, impeding proper transparency and accountability, and limiting opportunities for learning. It also means that DCJ does not know the true extent to which finite resourcing prevents it from responding to those children for whom there are seriously held concerns about ROSH.
The tolerance of confusion, and delays in clarifying legislative responsibilities
It was not until an internal 2023 triage review that DCJ’s existing processes were mapped to its legislative framework. DCJ has laboured under confusion about various aspects of the nature and extent of its statutory functions and yet did not seek definitive legal advice concerning the complexities of its statutory duties and powers in respect of ROSH reports.
In an environment where the workforce is primarily focused on applying critical skills to provide trauma-informed and effective services to families but is simultaneously making administrative decisions that require an understanding of the statutory framework, the importance of clear guidance on administering those duties and powers cannot be understated. Several specific examples of confusion in the interpretation and the application of the Care Act are discussed in section 5.3.
In circumstances where DCJ has for decades said that its limited resources mean that it cannot investigate and respond to ROSH reports in all, or even most cases, it is even more imperative that its processes and practices ensure effective and efficient decisions are made to allocate and prioritise resources to the highest-risk cases.
The outcome of face-to-face assessments
In 2022-23, actual harm or risk of harm was ‘substantiated’ in 48% of the children (13,444 of 27,782) who had been assessed face-to-face after a ROSH report. The proportion was similar in 2023-24.
This means that most children who receive a field assessment are not assessed as being at ROSH.
No direct conclusion can be drawn from this as to whether DCJ is, through its triage process, prioritising the right reports for field assessment – that is whether, if it is the case that DCJ can only ‘see’ around a quarter of the children reported at ROSH, it is prioritising those reports with the highest risk. However, it does reinforce the importance of DCJ making optimal risk-based triaging decisions, especially since there are other cases (including some highlighted following the deaths of children) where field assessment was not undertaken despite the existence of serious concerns.
The impact of performance incentives on prioritisation decisions
DCJ has an appropriate goal of increasing the number of children ‘seen’ in response to ROSH reports. On a per capita basis, the target equates to face-to-face assessments being conducted on average for approximately 20 children per caseworker per year.
However, our staff survey results confirm anecdotal evidence suggesting that the performance targets create a perverse incentive for some staff to preference the allocation of certain ROSH reports, or avoid allocation, in such a way as to increase the count of children seen. We asked those staff who had a direct role in allocation about their views of the impact of the targets.
Nearly half said the targets influenced some staff to allocate in a way that would increase the children seen statistics, instead of allocating to those children and young people assessed to be most at risk. This includes, for example, preferencing reports about large families where more children can be seen simultaneously.
DCJ informed us during the investigation that it has adjusted its children seen measure so that it now also counts second and subsequent visits to children during a 12-month period, to address concerns that the performance target had previously only counted ‘first’ visits and therefore discouraged prioritisation of reports that would involve a subsequent visit to the same child.
De-prioritisation of some serious ROSH reports
DCJ’s current approach to allocation decisions at times results in ROSH reports being closed due to ‘no capacity to allocate’ where the threat, though less immediate, is more severe and/or likely than other reports. Such decisions operate on an apparent assumption that ‘urgency’ can be a proxy for ‘importance’.
However, this approach has the potential to skew triaging prioritisation away from some cases that may be serious even if the timeframe for responding may be less urgent. This potential bias is most visible in serious cases of chronic neglect, including educational and medical neglect.
Requirement for first priority response to be given to mandated cases
Children alleged to have experienced sexual abuse, serious physical abuse and serious neglect are to be provided with a comprehensive and co-ordinated safety, criminal justice and health response through the Joint Child Protection Response Program (JCPRP). Certain DCJ policies provide that reports managed by JCPRP that need to remain open at the conclusion of JCPRP’s assessment and transferred to a CSC must be allocated to a caseworker and/or prioritised.
We found evidence of JCPRP transferred cases where no action was taken by the receiving CSC. Approximately a quarter of cases that are ‘substantiated’ as at ROSH by the JCPRP and referred to CSCs for further casework are then closed on the stated basis of ‘current competing priorities’.
Similarly, ROSH reports that concern a child who is already in care must be allocated, unless approved otherwise by senior staff. However, not all districts have dedicated teams available to respond to such reports and some may be closed either due to ‘no capacity to allocate’ or ‘competing priorities’.
Poor prioritisations regarding reports involving disability
A standout issue in the cases we reviewed concerned either children with disability or children whose parents or carers had a disability. We observed inadequate information gathering during triage, over-reliance on the involvement of other services when deciding to close the report and failing to give sufficient weight to vulnerability when assessing risk to the child with disability.
Under-utilisation of desktop investigation
A 2021 DCJ time and motion study found that more than half (64%) of caseworker time during the triage process was spent on information gathering. Despite this, in our sample of cases closed due to ‘no capacity to allocate’, desktop investigation options were routinely underutilised.
In circumstances where most ROSH reports are closed on the basis of ‘no capacity to allocate’, an early, efficient and thorough desktop investigation is essential to ensuring that only the lowest-risk reports are the ones that will be closed.
Reasonably available information not always collected
Staff reported to us that, during triage, attempts to contact the child or family occur at least ‘sometimes’ and that other desktop-based activities occur ‘often’ and in some cases ‘always’.
However, 35% of survey respondents also reported that they had closed cases where reasonably available information that was necessary for deciding if a child was at ROSH was not collected.
We also observed a practice of not making reasonably straightforward enquiries before closing reports, sometimes months after the report was received.
Requests for information sent to third parties, but not followed up and closed
DCJ has information-gathering powers that permit it to obtain information from other agencies, for example NSW Health, Department of Education and Police. The timeliness of such information exchange is clearly critical.
We have seen cases of ROSH reports closed before information requested from third parties had been received. In other cases, attempts to contact third parties were made, but abandoned if a connection could not be quickly established.
The 28-day rule in particular discourages staff from waiting for key third-party information before closing matters. This risks bias towards reports where information can be confirmed quickly and may deter complex, multi-agency matters that take longer. The approach may also be inconsistent with the broader intent of the legislation providing for agency exchange of information.
Contextual information to enable holistic assessment of cumulative risk/multiple vulnerability
When assessing ROSH reports, staff are required to conduct database checks to ‘form a holistic picture of risk’. They must review the child’s history and that of their siblings or other children in the household, their parents and caregivers. They are also required to check for previous reports (particularly recent ones) closed without a response.
History checks are a key source of information about escalating risk. Our survey of staff indicates that checks are routinely completed during triage, with 67% (55 out of 82) of respondents saying that this always happens.
Nevertheless, we observed cases where multiple (and escalating) reports were repeatedly closed on the basis that there was ‘no capacity to allocate’, even when earlier decisions had been made with closure commentary that explicitly stated that, if a further report is received, the next report should be allocated to a caseworker.
We also observed cases where multiple risks or vulnerabilities had been reported (often in separate reports) but the actions taken did not reflect an adequate assessment of all the important risk information on the file.
Under-utilisation of Interagency Case Discussion prior to closure
Interagency Case Discussions are important consultation processes for managing complex risk and service coordination. However, they are not often used before a report is closed due to ‘no capacity to allocate’. Only 4% of reports in our sample (15 of 380) indicated in the triage record that the case had been taken to an Interagency Case Discussion prior to closure. This was confirmed by our staff survey responses.
Over-reliance on referrals/other agency involvement as a basis for closing cases
DCJ may refer families involved in ROSH reports to other agencies and service providers. This could be done as a way to prevent or mitigate risk, in which case the referral may be relevant to DCJ’s subsequent decision as to whether the child in question is (post-referral) at ROSH.
However, it is not possible to know which referrals are made by DCJ because it was considered the most appropriate response under the Care Act to that case and because the referral genuinely and fully addresses any risk concerns, and when instead the referral is being done as a better-than-nothing alternative in circumstances where the more appropriate response (resources permitting) would have been to allocate the matter to a DCJ caseworker for investigation and any further response.
Despite this obscurity, DCJ frequently relies on its referrals and the involvement of other agencies to decide that a report should be de-prioritised at triage for closure. That happens despite there being no assessment about whether the referral has or will adequately address risk. We saw cases where DCJ has made decisions to close a report citing the fact that another agency or service provider is involved with the child or family, even when that other agency or provider was itself the source of the ROSH report, and where the concerns raised included the family’s refusal to engage with them.
Most closures due to ‘no capacity to allocate’ are made without any service referrals
Around 80% of the children who are reported at ROSH and not seen by a caseworker also do not receive a referral. However, data limitations mean the true numbers are unknown.
In our sample of closures due to ‘no capacity to allocate’, the need for a referral was considered by DCJ in only 11% of cases (40 of 380). Of these, a referral was only made in 21 cases.
Another challenge arising from referrals being driven by lack of capacity to allocate the case to a caseworker, is that there can be little, if any, substantive assessment of the family’s service needs, resulting in the potential for inappropriate or ineffective referrals with a mismatch between service and needs.
Re-reporting and no-reporting loops
Children and their families can get ‘stuck’ in closure and re-report loops for long periods. Re-reporting arises frequently in cases where reports are closed after service referral. In 2022-23, 56% of all children who exited a family preservation service were re-reported at ROSH within 9 months, regardless of whether they completed the program.
These loops can also produce the opposite effect. Reporting fatigue can occur where reporters stop reporting where they see no action or receive little feedback. The coroner has also previously heard evidence that service providers were effectively being ‘educated’ not to make further reports to DCJ, again because such reports would likely not be acted upon, and because providers themselves were expected to address risks with their client families.[6]
Helpline closure of ‘duplicates’
The Helpline does not generally have a role in closing reports that are ‘screened-in’ as ROSH reports. However, the Helpline can close such a report if it is assessed to be a ‘duplicate’.
Helpline staff told us that duplicate reports, once closed by the Helpline, are only forwarded to the CSC if the triage record (for the original report) is still open at the CSC or there is an open case. This means that, if the earlier report was screened-in as ROSH and closed due to lack of capacity, the duplicate would also be immediately closed, without even being referred to the CSC for its information or consideration.
We saw several cases where reports were closed as duplicates in circumstances where new information had been provided.
We also saw reports closed as duplicates where, although reporting the same or similar kinds of information as previously reported, a significant period had elapsed since the original report was made. This included in cases where the concerns related to potential ongoing neglect.
Incorrectly deciding that a ROSH report is merely a duplicate means that evidence that may help inform a view about whether there is an ongoing or escalating risk may be missed.
The adoption and use of SDM tools has not been the primary focus of this investigation, and we have not conducted an in-depth review of the suite of SDM comprised of tools created and licensed by a Californian company, now known as Evident Change. However, in examining DCJ’s practices in relation to ROSH reports, it has been necessary to give some consideration to the use of these tools as they are applied to make decisions about ROSH reports.
In Appendix A-3 we outline some of the limitations in design (and related flaws in use) that are known to DCJ or have become apparent to us during this investigation. These include that the tools:
In addition, we have found that there has been an over-reliance on the SDM Risk Assessment tools as proxy decision-maker on ROSH reports. In our survey of DCJ senior staff, 36% of respondents went as far as to say that an SDM tool ‘usually makes’ the decision that a child is at ROSH.
Similarly, concerns exist in relation to the adoption of SDM outputs as a de facto decision that a child is ‘in need of care and protection’, which is the trigger for formal child protection intervention.
We outline some of the ways in which DCJ was put on notice of the need to comprehensively evaluate and review the tools, and the unacceptably late and limited steps it has taken to respond to this need.
DCJ also failed, over many years, to publicly disclose or release the limited evaluative studies it did commission and failed to act on their recommendations, many of which called for thorough review and evaluation of the tools.
A determination that a child is at ROSH at the very least opens the door to a determination that a child is in need of care and protection, which in turn enlivens DCJ’s considerable statutory child protection powers over the lives of children and families. Use (or non-use) of these powers can have far-reaching effects on the day-to-day welfare and long-term outcomes for children and their families.
DCJ’s failure to ensure the SDM Risk Assessment tool was calibrated to NSW data has, according to at least one of its own internal reviews, ‘undoubtedly disadvantaged minority groups’ and especially Aboriginal children.
We do not suggest that a properly designed structured decision-making approach should not be used to inform and guide critical analysis and professional judgement. In particular, while it is acknowledged, even by DCJ, that SDM tools have had biased impacts, including with respect to Aboriginal families, a system that is reliant solely on casework professional judgment is clearly not itself without risk, including inconsistent and unfair decision-making.
However, the point is not whether SDM produces more or less biased outcomes than any alternative. Rather, it is that DCJ had a duty, which it failed at for more than a decade, to properly design, test, implement and continuously review, a system for responding to the State’s most vulnerable children and families, including to ensure that it was fit for purpose in the context of NSW and its families and avoided any avoidable bias.
The Ombudsman has made findings (Chapter 10) that certain conduct of DCJ when assessing, responding to and closing ROSH reports constitutes maladministration of the kind described by section 26 of the Ombudsman Act. This includes actions (or failures to act) that are unreasonable and, which in some cases, mean that DCJ is failing to meet its statutory responsibilities under the Care Act.
DCJ has told us that it is undertaking a major reform program of its child protection system that will, among other things, affect the system for receiving and responding to ROSH reports. Accordingly, our recommendations are made with awareness of the significant reform effort already underway.
The key recommendation (Recommendation 1) made is that DCJ should undertake a comprehensive review and reform of its ROSH related policies, practices and tools to bring them into alignment with the legislation. We have also recommended (Recommendation 2) that to support such a review, the Government should consider whether the relevant provisions of the Care Act should be reviewed and improved.
Recognising, however, that the comprehensive reforms required by Recommendation 1 will take time to develop and implement, we also made 9 further recommendations for actions that DCJ can and should take as a priority. These are actions that we consider are so clearly necessary that they should not wait, and that can be readily and practically implemented in a way that does not significantly divert effort away from the broader reforms that are needed.
1. Identifying and responding when children and young people are at risk of harm is a core element of the Department of Communities and Justice’s (DCJ) role as lead agency for the NSW child protection system.[7] This function is primarily governed by the Children and Young Persons (Care and Protection) Act 1998 (Care Act).
2. The Care Act (Chapters 2 and 3) provides for DCJ to receive, assess and investigate when it receives reports from people who reasonably suspect that a child or young person is ‘at risk of significant harm’ (at ROSH). DCJ may receive voluntary reports under s 24 and mandatory reports under s 27.
3. Section 25 of the Care Act separately provides for reports to be made to the Secretary by any person who suspects on reasonable grounds, before the birth of a child, that the child may be at risk of significant harm after their birth (Pre-natal Reports).
4. Pre-natal Reports are legally distinct from ROSH reports. Mandatory reporting under s 27 does not extend to pre-natal concerns. There are also no express obligations placed on the Secretary to do or decide anything on receipt of a Pre-natal Report. A drafting note to s 25 indicates that its intention is to allow assistance and support to be provided to an expectant parent, as well as to provide early information that a child may be at ROSH after their birth, including to facilitate mandatory ROSH reporting (after the birth) where reasonable grounds for ROSH then exist.
5. Although DCJ’s conduct in respect of Pre-natal Reports is within the scope of our current investigation, it raises a number of distinct issues. As such, and to avoid complexity and confusion, we have decided to report on that aspect of the investigation later in a separate document.
6. This document concerns our investigation into whether the conduct[8] of DCJ when assessing, responding to and closing ROSH reports may amount to or involve wrong conduct (that is, maladministration) of a kind described in s 26 of the Ombudsman Act 1974.
7. Maladministration may include conduct that is contrary to law, unreasonable, unjust or otherwise wrong. Maladministration can also include conduct that is done pursuant to a law or policy, but where the law or policy itself is unreasonable, unjust or otherwise wrong.
8. Our findings in this regard are set out in Chapter 10.
9. There has been considerable scrutiny of the child protection system, including recently in published reports from the NSW Ombudsman, NSW Audit Office and DCJ itself. [9] These reports raise serious and longstanding concerns about overall performance and stewardship of the wider child protection and out-of-home care (OOHC) system, matters which are generally outside the scope of this investigation.
10. Instead, in this investigation we are focused solely on the administrative exercise of DCJ’s functions under Chapters 2 and 3 of the Care Act to deal with ROSH reports, including triage assessment, allocation and closure processes, with particular scrutiny on DCJ’s practice of closing such reports without allocation to a caseworker and/or on the stated basis of resource constraints.
11. Concerns have been raised over many decades that DCJ closes a large proportion of reports it receives without the children who are subject of those reports receiving a field assessment by a DCJ caseworker, and particularly on the stated basis that DCJ has ‘no capacity to allocate’ the matter to a caseworker, or because the allocated caseworker has other ‘competing priorities’. As well as in NSW Ombudsman reports,[10] concerns have been repeatedly raised in Parliament,[11] in public inquiries,[12] by the NSW Auditor-General,[13] as well as in a number of coronial inquests examining the deaths of children who, before their deaths, had been reported at ROSH.[14] Examples of public reports raising concerns about the closure of ROSH reports from 2011 are summarised in Appendix A-5.
12. The issue is longstanding. DCJ’s practice of closing ‘risk of harm’ reports by citing inadequate resources or competing priorities goes back to its first ‘unallocated case closure’ policy in 1996 (known from 1997 as the Priority One Policy).
13. There have been numerous DCJ reforms that intended, at least to some extent, to address the issue.[15]
14. A particularly important change occurred following the 2008 Special Commission of Inquiry into Child Protection Services in NSW (2008 Wood Special Inquiry), which had found that some risk of harm reports that warranted further attention had been prematurely closed due to competing demands on the system. In accordance with a Commission recommendation, in 2010 Parliament raised the trigger for reporting from ‘risk of harm’ to ‘risk of significant harm’. The Commission noted:
‘Changing the reporting regime [in Chapter 3 of the Care Act], for both mandatory and voluntary reporters, to one which applies in relation to children who are suspected by the reporter, on reasonable grounds, to be ‘at risk of significant harm’, rather than ‘at risk of harm,’ should have the effect of reducing the number of reports to those children who are likely to need the powers of the State under s.34 [in Chapter 4] of the Care Act, exercised for their protection …’[16]
15. However, despite the increase in the reporting threshold, and ongoing initiatives by DCJ to trial and adopt new processes,[17] the significant proportion of reports being closed by DCJ without face-to-face assessment and/or on the stated basis of inadequate capacity has continued.
16. In 2024, the NSW Ombudsman published the Protecting Children at Risk (PCAR) report.[18] The PCAR report included unpublished data for the 2022-23 financial year showing that, of all the children DCJ considered (or ‘screened’) as potentially meeting the at ROSH definition (i.e., ‘suspected ROSH’), closure occurred during initial triage due to ‘no capacity to allocate’ for more than two-thirds (64%). Only 25% of children reported at ROSH were seen by a caseworker. The outcomes of the remaining 75% of children reported at ROSH are not known.
17. We noted in the PCAR report that we would be conducting a formal investigation into whether DCJ’s conduct in respect of its response and closure of ROSH reports may involve maladministration.
18. It is well recognised (both by external bodies, as well as by DCJ itself) that a child protection system in which more than two thirds of these reports are closed on the stated basis of there being inadequate capacity to investigate or competing priorities is a system that is not working as it should.
19. In this investigation, we have sought to move beyond that obvious conclusion. We acknowledge the practical reality that DCJ operates within a context of finite resources, with limited financial funding and workforce challenges.[19] However, in this investigation we have proceeded without making any initial assumptions about these resource challenges, and whether they explain, let alone excuse, the large volume of reports that are ‘closed’ – citing insufficient capacity or competing priorities – without children or young people ever being ‘seen’ by a caseworker. Instead, we have sought to scrutinise DCJ’s practices around triaging and assessing ROSH reports – practices which it has itself described as a ‘black hole’.[20]
20. Doing so means that much of this document is detailed and technical in nature. This has been unavoidable given the complexity of DCJ’s policies and practices. Indeed, our conclusion is that this complexity is to some extent part of the problem – DCJ’s suite of policies, manuals, guides, recordkeeping systems and tools are complicated, and do not clearly align with the statutory regime which they are intended to implement. This also impedes the meaningful reporting of data that would enable clear understanding of and better public accountability for the performance of DCJ’s functions.
21. In circumstances where there has hitherto been limited public information about how DCJ actually makes decisions and performs its functions in practice in this area, we consider there to be significant public value in setting this out in some detail in this report.
22. Finally, we note that, while on our case and document reviews we observed inefficiencies in triage, assessment and decision-making (including duplication, overlapping tools, inconsistent practice), these process issues have been examined extensively elsewhere – by the Audit Office[21] and in multiple DCJ-commissioned reviews since 2014.[22] Those reviews have already recommended streamlining intake and triage, standardising document and decision tools, and clarifying closure codes. A 2021 FACSIAR time-and-motion study also suggested DCJ could, within existing resources through better time allocation and utilisation, increase the number of children seen from about 31,600 (28.6%) to 44,900 (40.6%) of those reported at ROSH.[23] Given this body of work, we did not re-audit operational efficiency in this investigation.
23. Similarly, we have not assessed workforce planning, recruitment and retention (including vacancy management and skill mix). Although workforce settings clearly affect the capacity to act on reports, these matters are the subject of other reviews and internal initiatives.
24. There is now a plethora of public reports and commentary, dating back decades, that both:
25. In this investigation we have sought to contribute something new, focusing on the ‘missing middle’ through a close examination of what DCJ’s statutory responsibilities actually require, and by examining the policies and practices used by DCJ to perform them. While this has included a sampling review of some cases (and some of these are reflected in case studies included in this report), we have not investigated or made findings about the conduct of DCJ in respect of any particular case, and nor have we investigated the decision-making or conduct of any individual officers of DCJ.
26. In brief, in addition to a desktop review of earlier system reviews, coronial inquests, legislation and decisions of courts and statutory bodies, published DCJ data and the Ombudsman’s own complaint holdings, the investigation team undertook a wide range of information-gathering activities such as issuing formal notices requiring production of internal and unpublished documents and information, meetings and briefings with agencies and staff, and compulsory witness interviews. We also issued a survey to 510 DCJ staff (of which 131 were completed).
27. The investigation team also conducted case reviews of particular cases, including desktop sampling reviews of the first 380 triage records that were closed with the label ‘No Capacity to Allocate’ in June 2023.
28. A more detailed methodology, including survey, sampling and data limitations, can be found at Appendix A-4.
Figure 1 – Operational screen outcomes do not map 1:1 to report types
29. The focus of this investigation is reports made to DCJ under s 24 and s 27 of the Care Act – noting that it is the receipt of these reports, and only these reports, that trigger DCJ’s statutory responsibilities under s 30 of the Care Act. (This is not to say that DCJ cannot and should not act on other information it receives that may suggest a child may be in need of care and protection. Rather, the point is that it is in the case of these reports that DCJ has been placed under a duty to act, and it is how well DCJ is performing that duty that is the focus of our investigation here.)
30. However, even though the receipt of such reports triggers important and unique statutory obligations, DCJ does not currently have a mechanism to clearly and separately identify, record and report data about these particular reports.
31. DCJ’s current practice is that all ‘child protection reports’ it receives (through its Child Protection Helpline) are initially recorded as ‘concerns reports’. (Neither of these terms is used or has a meaning under the Care Act.)
32. While DCJ records who made a report and assigns them as being either a mandatory reporter or non-mandatory reporter in the ‘contactor type’[25] data field, it does not record the particular legal basis of the report itself. This means that DCJ is unable to easily quantify how many mandatory reports it receives, as the data recorded and published about reports having been received from ‘mandatory reporters’ includes all report types, including both those that are mandatory under the Act (that is, s 27 (ROSH) and s 122 (living away from home without permission)) as well as those that are not mandatory ((s 25 (pre-natal) and s 120-121 (homelessness)).
33. Concerns reports are then ‘screened’ by Helpline staff, which involves classifying and recording them into two groups: a ‘ROSH report’ or a ‘non-ROSH report’. DCJ guidance states that the Helpline screens in a report as a ROSH report if it has ‘assessed’ or ‘made a determination’ that it meets the ‘ROSH threshold’.[26]
34. Reports ‘screened-in’ by Helpline staff as ‘ROSH reports’ will include not only reports that may have been made under ss 24 and 27, but also other reports that may be made under the Care Act, including:
35. These other reports may be screened-in by the Helpline and recorded as ‘ROSH reports’ even though they do not trigger the statutory ‘ROSH response’ duties in s 30 that are triggered by ss 24 and 27 reports. (That said, a Homelessness Report about a child (under 16 years) triggers a separate statutory investigation and assessment obligation under s 120. It might also be considered a report under s 24 and 27 – that is, a report could potentially be both a report that a child is homeless and a report that the child is suspected of being at ROSH).
36. On the other hand, some reports made under ss 24 and 27 could be ‘screened-out’ as ‘non-ROSH reports’ at the Helpline stage, which means they are never entered formally into DCJ’s records as having been made under s 24 or s 27.[27] According to DCJ’s published guidance:
‘[T]here will be some reports [by mandatory reporters] made with good intention that do not meet the ROSH threshold and are screened-out by the Child Protection Hotline.
…
It is important for mandatory reporters to note that a decision by the Child Protection Helpline that a report does not meet the ROSH threshold does not necessarily mean the report should not have been made. Ultimately, a report based on your suspicion that a child/young person or class of children/young people is at ROSH is in their best interests.’[28]
37. The consequence of all this is that what DCJ includes in its records as ‘ROSH reports’ will only imperfectly align with the ss 24 and 27 reports it has received and has a specific statutory duty to deal with under s 30.
38. Nevertheless, in this document, to avoid further confusion by introducing a new term, we will generally use the term ‘ROSH report’ when referring to both:
39. This approach, while imperfect, appears reasonable in circumstances where ss 24 and 27 reports are likely to make up the bulk of the concerns reports that will be classified as ‘ROSH reports’ by DCJ’s Helpline, and where decisions made by the Helpline to screen-out a report as ‘non-ROSH’ has generally not raised any of the same concerns or criticisms as DCJ’s subsequent decisions in relation to reports that were ‘screened-in’.[29] Indeed, although not expressed in this way in DCJ’s policy or practice guidance, the Helpline screening-out process could be taken as a decision under s 30(b) of the Care Act – that is, a decision ‘to take no further action’ because ‘there is insufficient reason to believe that the child or young person is at risk of significant harm’ – see section 4.3.
40. Other key terms used in this document can be found in Appendix A-1.
41. We opened this investigation in April 2024. Unless otherwise noted, it assesses DCJ’s legal, policy and practice settings as they stood then.
42. The Care Act has not materially changed since, but in September 2024, DCJ implemented its ‘interim approach to risk assessment’. The changes relate mostly to DCJ’s statutory decisions made after a face-to-face assessment, where DCJ has ceased using the SDM Risk Assessment Tool and revised its policies. Simultaneously, DCJ made interim changes to its approach to pre-natal reports.
43. In parallel to this investigation, DCJ has begun a broader program to reposition the system across three streams – family support, child protection and out-of-home care (OOHC) – of which only the OOHC reform plan has been released.
44. DCJ is also conducting related reviews: the Prioritisation, Triage and Allocation Policy Review (since 2022), the Prenatal Policy Review (from 2024) and the Child Protection Assessment Review Project (from 2024). There are also several service redesigns and evaluations on foot.
45. Where relevant, we note post-April 2024 changes and their implications.
46. Each substantive chapter includes at the start one or more short cases studies drawn from recent matters in which DCJ received a ROSH report in 2023 or 2024. Most were identified during our review of a sample of cases where ROSH reports had been closed due to NCTA, in DCJ’s responses to our notices to produce documents and information and in material provided by DCJ staff through our survey – see section 1.4. A small number of these case studies arose from complaints to our office. All case studies are de-identified to the extent possible, including by the use of pseudonyms.
47. Each case study has been included as an example of how current policies, tools and practices have played out in a real-world situation. They are put forward by way of example rather than necessarily being typical, and each has been selected for illustrative effect rather than on any statistically representative basis. The case studies are based only on the records made by DCJ at the time relevant decisions were made, and do not imply any conclusion or prediction about the long-term outcomes for the children and families involved. We make no findings in this investigation about the conduct of DCJ in respect of any particular case.
48. We provided a copy of our provisional findings to DCJ on 26 September 2025, in line with s 24(2) of the Ombudsman Act to enable DCJ to make submissions on any adverse comments and provisional findings.
49. DCJ provided submissions in response on 31 October 2025 and 11 October 2025 and in a meeting on 24 November 2025. DCJ’s submissions have been considered in preparing this final report.
50. We provided provisional recommendations to DCJ on 11 December 2025. DCJ made submissions in response on 19 December 2025.
51. Before finalising a report under s 26 of the Ombudsman Act, in accordance with s 25(2), the Ombudsman must inform the responsible Minister that he or she intends to make a final report concerning the investigation. On 11 December 2025, a copy of the draft report was provided to the responsible Minister, the Honourable Kate Washington, Minister for Families and Communities. The Minister consulted with the Ombudsman on 19 December 2025.
52. In accordance with s 26 (3) of the Act, a copy of this final report is provided to:
53. An outline of the relevant provisions of the Care Act is in Appendix A-2.
54. The provisions of the Act dealing with ROSH reports and DCJ’s responsibility to respond to them are brief, and the basic structure is simple:
55. There are 4 elements to the regime that are of particular importance to this investigation, and which it is useful to highlight at the outset.
56. First, the Act imposes a statutory responsibility on the Secretary (or a delegated DCJ officer) to consider each and every ROSH report it receives, on its merits.
57. The Secretary, on receiving a ROSH report, ‘is to’ (that is, is under an obligation to) make ‘such investigations and assessment as the Secretary considers necessary’ to determine whether the child or young person is at ROSH. This assessment results in a decision being made as to whether the child is or is not at ROSH (in this report, we describe this as the ROSH determination).
58. However, the Secretary ‘may’ (that is, has a discretion to) take no further action if the Secretary considers that, on the information provided, there is insufficient reason to believe the child or young person is at ROSH (in this report, we describe this as an insufficient reason discretion decision).
59. The Act contains no express provisions concerning ROSH report ‘closure’.[33] Unless a ROSH report is finalised on the basis of the insufficient reason discretion, closing a ROSH report without having undertaken whatever investigations and assessment the Secretary considers necessary to make a ROSH determination will be inconsistent with the requirements under the Act. (As to the effect of such failures, see Chapter 4.)
60. The Act proceeds on the basis of a statutory expectation that all ROSH reports will be investigated and assessed as the Secretary considers necessary, except in those cases where an insufficient reason discretion decision has been made.
61. There is no express provision in the Act allowing for the triaging or prioritisation of ROSH reports, and no express statutory timeframes within which the Secretary’s decisions (including any investigations and assessment) must be made.
62. Nevertheless, all statutory functions must be exercised reasonably, which includes within reasonable timeframes.[34] In considering what is a reasonable timeframe in which to make decisions about a volume of reports, an initial triaging or prioritisation of those reports is both permissible and appropriate.
63. Third, DCJ’s investigations and assessment are directed toward determining whether the child or young person is at ROSH having regard to the circumstances known to DCJ as at the time the determination is made – that is, the question it must consider is whether the child or young person is at ROSH.
64. This may be a different question to ‘was the child or young person at ROSH when the ROSH report was made?’ and ‘did the person who made the ROSH report believe or have reasonable grounds for believing the child or young person was at ROSH?’.
65. Of course, in considering whether a child or young person is at ROSH, consideration of what has already happened will be necessary. For example, one of the reasons why current concerns may exist for the safety, welfare or wellbeing or the child or young person may be because the child or young ‘has been’ physically or sexually abused or ill-treated. Even then, the question to be asked is whether concerns are ‘current’ concerns that ‘exist’ as at the time of the decision.
66. This means that, while the circumstances surrounding the making of the ROSH report may be relevant for DCJ to consider, DCJ’s decision-making focus under the Care Act is not directed to the report itself, but rather to the actual and current circumstances of the child or young person who is the subject of the report.
67. Although the above seems evident from the legislation, timing issues around the application of the ‘at ROSH’ determination have been a matter of some confusion within DCJ, including in some of the language it uses to describe its task (such as describing its assessment as being one of deciding whether the report itself is ‘substantiated’). These issues are likely to take on more consequence when there is a significant timing gap between the making of the report and DCJ’s assessment of it. We take this issue up in Chapter 4.
68. The Secretary’s duty to take any action toward addressing the risks faced by a child or young person only arises (under Chapter 4 of the Care Act) if an opinion is formed by the Secretary that the child or young person is ‘in need of care and protection’ (INOCAP). If the Secretary forms an opinion that a child or young person is INOCAP, then DCJ must take whatever action is necessary to safeguard or promote, the safety, welfare and wellbeing of the child or young person. In doing so, DCJ has a broad discretion as to any action to be taken. The Act provides only a non-exhaustive list of possible actions that could be considered, ranging from the provision of support services for the child or young person or their family through to seeking Children’s Court orders. DCJ can also request other government departments or government-funded agencies to provide prioritised access to services (s 17).
69. Oddly – given that it is this INOCAP decision, and not the ROSH determination, which is the critical trigger for DCJ’s statutory intervention – the Act includes no definition of what ‘in need of care and protection’ means. Nor is there any clear expression of how it links to the concept of a child or young person being at ROSH.[35]
70. Moving from a ROSH determination to considering an INOCAP determination is also not expressly mandated by the Act. Clearly, a decision that a child is INOCAP could be made even if a ROSH report has not been received.[36] Such a decision could be made because concerns or intelligence about the circumstances of a child have come to the attention of DCJ otherwise than through a ROSH report.
71. However, given the purpose of the legislative regime,[37] it also seems apparent that, following a determination that a child is at ROSH, consideration should be given to whether the child or young person is INOCAP, or whether the risk can or has otherwise been addressed (for example, by a change of circumstances or through the use of powers under s 17 of the Care Act to seek services from other agencies).
72. Unless risk is resolved, it is in our view evident, albeit implicit, that a determination that a child or young person is at ROSH triggers two further decision-making obligations:
73. This absence of an explicit link between Chapter 3 of the Care Act (in which the Secretary is required to determine if a child or young person is at ROSH) and Chapter 4 (in which the Secretary must take action if a child or young person is INOCAP) has been the source of some confusion in DCJ. We suggest this be taken up in DCJ’s proposed review of the Care Act.[38]
74. In this section, we outline DCJ’s current (i.e., as at April 2024) policy framework for responding to ROSH reports. For simplicity we have not included those policies and practices that relate specifically to reports about children and young people already in out-of-home care (OOHC).[39]
75. DCJ’s policy framework for responding to ROSH reports relies on four interrelated components.
Figure 2 – DCJ’s ROSH (s 30) & INOCAP (s 34) decisions & responses framework
A suite of triage and assessment policies and guidance documents
There are operational policies and guidance documents for each stage of receipt, triage and assessment of ROSH reports. Key documents include:
Overlaying these are overarching policies such as the Aboriginal Case Management Policy and ChildStory Knowledge Articles and guides (such as the ChildStory Triage Completion Guide and ChildStory Case Closure Reasons Guide).
A US-developed Structured Decision Making (SDM)[40] system
Since at least January 2010,[41] DCJ policy has required its caseworkers to use structured decision-making tools alongside professional judgment to make decisions about assessing and responding to risks for children and young people. The tools form part of a suite owned by Evident Change, consisting of both consensus-based tools, actuarial (data) tools and blended tools (mix of consensus-based and actuarial).
The current (and former) tools used by DCJ are listed below, with the date of first use in brackets:
* Note: On 12 September 2024, DCJ replaced the SDM Risk Assessment Tool and SDM Risk Reassessment Tool with an interim approach to risk assessment until an Aboriginal-led, co-designed assessment process is developed.
Together, the tools form an SDM ecosystem with some internal dependencies and assumptions.[42] They are each supported by a manual.
A case management system, ChildStory (records and workflow management system)
ChildStory is DCJ’s core system for recording and managing child protection and out-of-home care work and is used by more than 89,700 people across NSW. It has different user access and interfaces:
DCJ’s SDM tools are integrated with ChildStory.
Professional judgment exercised by caseworkers
This includes those DCJ staff working on DCJ’s Child Protection Helpline (Helpline) and those working in Community Service Centres (CSCs). Delegates (Director, Community Services (DCS), Manager, Client Services (MCS), Manager Helpline, Manager Casework (MCW), Team Leader, Helpline) are authorised to:
76. DCJ’s usual process[44] when a concerns report is received at the Helpline, is to first assesses whether it meets the ROSH ‘threshold’ using the SDM Screening and Response Priority Tool (SDM SCRPT). If it does, it is screened-in as a ROSH report. The Helpline assigns the report a response timeframe and transfers the report to a Community Service Centre (CSC) or other business unit as appropriate.
77. At the CSC, information is gathered and relevant ROSH reports (now called ‘triage records’) are considered in triage or peer review processes, which may include weekly allocation meetings (WAMs). Decisions are made about whether to:
78. If allocated to a caseworker, the report becomes an open ‘case’. A child will generally receive a field visit (or a face-to-face assessment) to assess and determine whether or not they are at ROSH and INOCAP.
79. Each of these steps is explained in more detail below and summarised diagrammatically in Figure 3.
Figure 3 – Key steps, tools and closure reasons in DCJ’s decision-making process
80. Mandatory reporters include paid workers (and their supervisors) delivering child-related health, welfare, education, childcare, residential care or law-enforcement services, including DCJ staff, as well as religious ministers and registered psychologists. When these workers reasonably suspect a child is at ROSH, they must make a report to DCJ.
81. Before making a ROSH report to DCJ’s Helpline, mandatory reporters are encouraged to complete a screening tool themselves to help inform their suspicion that a child is at ROSH and decide whether a ROSH report should be made.
82. DCJ has adopted the SDM Mandatory Reporter Guide (SDM MRG) for this purpose. It is accessible as a public online tool[45] or within ChildStory Reporter (a module of ChildStory that is accessible by mandatory reporters who sign up to use it).
83. Based on answers to several questions, the tool generates a decision report to guide the reporter as to what action to take. This may include:
84. These reports can be made by any person and are made by phone to the Helpline (eReporting is not available to voluntary reporters).
In the 2023-24 financial year, DCJ recorded 208,093 ROSH reports from ‘contactor types’[46] mapped as ‘mandatory reporters'. This represents 84% of all ROSH reports received. In the 2023-24 financial year, DCJ recorded 38,753 ROSH reports from ‘contactor types’ it mapped as ‘non-mandatory’ reporters. This represents 16% of all ROSH reports received.  Source: DCJ Annual Statistical Report 2023-24, ROSH reports by reporter type Note: Figures reflect contactor type (the reporter's sector – eg Education, Health, Police), not the legal basis of the report. DCJ groups these sectors as 'mandatory reporters' and everyone else as 'non-mandatory reporters'. Treat these numbers as sector-of-reporter, not evidence that a report was made under s 27. |
85. DCJ receives and processes most child protection information through the Helpline (including via its 24/7 telephone service and eReports). DCJ describes any child protection-related information that it receives through the Helpline as a ‘concerns report’.
86. These concerns reports are, on receipt, classified into either ‘ROSH reports’ or ‘non-ROSH reports’. The terms ‘concerns report’, ‘ROSH report’ and ‘non-ROSH report’ are not in the Care Act. DCJ uses these terms for policy, operational and statistical purposes.
87. Concerns reports undergo a classification process referred to within DCJ as ‘screening’.
88. Reports made under ss 24 or 27 that DCJ subsequently decides do not meet the ROSH ‘threshold’ in s 23 will be screened-out by the Helpline as a ‘non-ROSH report’. (The fact that a report is screened-out (as non-ROSH) does not necessarily mean that no action is taken in respect of it. ‘Non-ROSH reports’ may be subject to different response pathways, such as being dealt with as a request for assistance).
89. Reports that do meet the ROSH ‘threshold’ will be ‘screened-in’ as a ‘ROSH report’. The decision to screen-in a report as a ROSH report is made by the Helpline using the SDM Screening and Response Priority Tools (SDM SCRPT). SDM SCRPT is built into ChildStory and consists of two separate tools – the SDM Screening Tool and the SDM Response Priority Tool. The SDM Screening Tool is supposed to help Helpline staff assess whether information received:
90. A Helpline Manager can decide that information in a report does not meet the criteria to be a ‘ROSH report’ and does not require SDM SCRPT screening. Also, a report initially screened-in as a ROSH report could be screened-out later by a Helpline Manager as a ‘non-ROSH report’.[49]
91. Reports screened-in by the Helpline as ROSH reports are transferred for further response under the Care Act by a CSC or other unit, while ‘non-ROSH reports’ are ‘screened-out’ and either closed by the Helpline, or transferred to a CSC for further action under the Care Act or other legislation.
In the 2023-24 financial year, DCJ received 426,826 ‘concerns reports’. Of the 426,826 concerns reports received, the Helpline screened-in 246,846 (58%) as ‘ROSH reports’ and 179,980 (42%) as ‘non-ROSH reports’. Of those non-ROSH reports, DCJ recorded 148,121(82%) as being made by ‘contactor types’[50] mapped as ‘mandatory reporters'.[51] |
Figure 5 – concerns reports by ROSH and non-rosh status, 2023-24
Source: DCJ Annual Statistical Report 2023-24
92. Helpline applies the SDM Response Priority Tool to ‘assess how quickly the investigation/assessment or other response must be commenced’ and assigns a priority response time. The timeframes are:
93. In most cases, the Helpline then transfers screened-in ROSH reports to a CSC. Some ROSH reports may instead be transferred to the Joint Child Protection Response Program (JCPRP) (if a criminal offence is involved) or other relevant business unit for further investigation/assessment.
94. The timeframe assigned by the Helpline provides a benchmark, but decisions about what further action under the Care Act DCJ will take rests with the CSC triage team or other receiving unit.
In the 2023-24 financial year, of the 246,846 ‘ROSH reports’ transferred by the Helpline to CSCs, the following response times were assigned:
|
Figure 6 – ROSH reports by response time assigned by Helpline
95. DCJ’s Triage Policy requires triage officers at CSCs to apply a triage process to incoming ROSH reports (and non-ROSH reports) – which, by this stage, are referred to by DCJ as ‘triage records’. The triage assessment can result in one of three outcomes for the triage record:
96. If a ROSH report screened-in by the Helpline relates to a child or young person who already has an open case, it will be triaged and usually allocated to the existing CSC caseworker. Note that cases are opened in ChildStory only if and when a decision is made to ‘allocate’ a ROSH report to a caseworker. Until that happens, the matter is not an open ‘case’ but is considered to be a ‘triage record’.
97. DCJ triage staff may conduct a range of information gathering, assessment and referral activities during the triage stage, but they do not undertake field visits.
98. other policies may elevate particular ROSH reports for priority treatment. These are known as policy overrides. The Safety in Care Mandate, for example, requires the CSC to allocate all ROSH reports about children already in OOHC to a caseworker for investigation and assessment, unless a Director Community Services approves otherwise. ROSH reports requiring an urgent response (including those that the Helpline has set a timeframe of less than 24 hours response) are required by DCJ’s policy to be brought to the attention of the Manager Casework to be considered for a face-to-face assessment.
99. Some ROSH reports are allocated to a caseworker by the CSC for field assessment within the response priority timeframe designated by the Helpline.
100. Any ROSH report that is not allocated to a caseworker within the response priority timeframe set is generally required by DCJ’s policy to be brought to a Weekly Allocation Meeting (WAM) or ‘equivalent peer review process’ for a decision about how or if DCJ will respond.
101. The Triage Policy does not define an ‘equivalent peer review process’ or set minimum steps. In practice, it usually involves a two-person check – either a discussion between two or more triage officers/managers or a ‘dual decision’ where a second officer/manager reviews and signs off the first decision – often recorded by email or in ChildStory.
102. However, the policy allows some unallocated triage records not to proceed to consideration at the next WAM or equivalent peer review, and in some cases to be closed without ever proceeding to a WAM or equivalent decision (or any further assessment).
103. The policy provides for the following matters, if not allocated to a caseworker within the priority response timeframe, to be closed without such consideration:
104. The policy also provides that, if a ROSH report has remained unallocated for more than 28 days since the Helpline received the report, the report is to be closed on the basis there is no capacity to respond or allocate.[52] Such ROSH reports can only remain open if approved by the Director, Community Services for ‘exceptional circumstances’. There are differing practices across and within CSCs in respect of the 28-day closure policy – in some cases, a ROSH report that is more than 28 days is put on the agenda at the next WAM, where it is closed; in other cases, closure takes place without the report being put to a WAM.
105. For those ROSH reports that proceed to a WAM or equivalent, the policy requires the CSC’s Manager Casework (MCW) to decide whether a ROSH report is:
In the 2023-24 financial year, the outcomes for the 246,846 ROSH reports were:
|
Source: NSW Budget Estimates, 10 March 2025, Families and Communities, Disability Inclusion (Washington), Answers to Questions On Notice, Attachment QTON#23, 24 Table, Attachment QTON#23, 24 Table
106. The Triage Policy states that decisions to close a ROSH report should involve discussion and critique by two Managers Casework (MCW) in the first instance. Where possible, the Manager Client Services (MCS) should be part of these discussions. At least two MCW must agree on the closure decision.[53] Guidance on closure decisions exists.[54]
107. In practice, decisions to close a ROSH report are made during WAMs or via equivalent peer review process. ChildStory does not have functionality to capture multiple approvers and there is no standard way to record an equivalent peer review process or its outcome.[55]
108. When a triage record is closed, a closure reason is required to be selected in ChildStory and a record made of the closure rationale. The ChildStory Triage Completion Guide provides eight closure reasons and sets out when each is to be used:
Closure Reason | Use When … |
No Capacity To Allocate | When there is no open Case AND there is no capacity to allocate AND/OR Assistance has been provided in response to a request for assistance AND no on-going casework is required AND/OR After completing a referral to a service during the Triage process |
No Assistance Provided | Contact record has been screened in by the Helpline as a request for assistance AND there is no open Case AND no assistance was provided AND/OR When closing a Class of Children report |
Reported Information not Substantiated | The reported information is known AND has previously been assessed with an outcome of Not Substantiated AND the Case is now closed |
Unable to locate | There is no open Case AND follow up on the report was done at Triage AND the location of the child or young person could not be confirmed, therefore a Field Assessment cannot be conducted as we do not know where the child/family is |
C/YP not in Jurisdiction | There is no open Case AND it has been confirmed that the child or young person is not in the state of NSW. The information must be passed onto Interstate Liaison |
Non ROSH – JIRT Rejection | Do Not Use |
C/YP Deceased | When there is no open Case AND no siblings AND the new report has a deceased child as a Subject and the information does not require further action by the CSC. |
After Care Support Concluded | Do Not Use |
In the 2023-24 financial year, of the 165,087 ROSH reports that were transferred to CSCs and not allocated to a caseworker:
|
Source: NSW Budget Estimates, 10 March 2025, Families and Communities, Disability Inclusion (Washington), Answers to Questions On Notice, Attachment QTON#23, 24 Table, Attachment QTON#23, 24 Table & 26 Table
109. Only ROSH reports allocated to a caseworker proceed to the next stage in DCJ’s process, which is for caseworkers to undertake investigations and assessment to work out whether a child is in fact at ROSH.
110. When a ROSH report is allocated to a caseworker, this is when a ‘case’ is opened on ChildStory (unless one already exists). The caseworker will usually then arrange to visit the family in person to further investigate and assess the reported issues. This is known as a ‘field assessment’. Where resources are insufficient, a caseworker may close the ROSH report before, during or after a field assessment; in ChildStory this is recorded as closure due to ‘Current Competing Priorities’.
111. DCJ’s Assessing Safety and Risk Mandate applies to field assessments. A field assessment involves one or more face-to-face home visits by a caseworker. The SDM Safety Assessment and SDM Risk Assessment tools are applied, after which a ‘substantiation’ decision is made.
112. An initial assessment of a child’s safety is conducted using either the SDM Safety Assessment Tool or DCJ’s Alternate Assessment (for a child or young person already in OOHC). The SDM Safety Assessment is supposed to assist in assessing whether a child is likely to be in ‘immediate’[56] danger of serious harm that requires a protective intervention, and in determining what interventions are needed.
113. The tool manual distinguishes the SDM Safety Assessment component from the SDM Risk Assessment component:
Safety versus Risk Assessment: It is important to keep in mind the difference between safety and risk when completing this tool. [SDM] Safety Assessment differs from [SDM] Risk Assessment in that it [SDM Safety Assessment] assesses the child/young person/unborn child’s present danger and the interventions currently needed to protect the child/young person/unborn child. In contrast, [SDM] Risk Assessment looks at the likelihood of future ill-treatment.[57]
114. The SDM Safety Assessment requires caseworkers to assess a family’s situation against pre-defined factors across three domains – child vulnerabilities, critical dangers, and caregiver and child protective abilities. Where danger exists, a list of possible safety interventions that may mitigate the dangers must be considered.
115. After collecting this information, caseworkers make a ‘safety decision’:
116. At the time our investigation commenced, the second stage was a risk assessment, usually conducted within 30 days of the SDM Safety Assessment regardless of the safety outcome, using either the SDM Risk Assessment Tool or DCJ’s Alternate Assessment (for a child or young person in OOHC).
117. The SDM Risk Assessment Tool was ‘used to determine whether a case should remain open for ongoing casework or be closed’. It sought to do this by classifying families into risk groups based on the likelihood of re-reporting within 12-18 months, using this as a proxy for future abuse or neglect, and recommending a closure decision based on the family’s corresponding risk level.[59]
118. The SDM Risk Assessment Tool required caseworkers to score families against both a neglect index and an abuse index based on answers to pre-defined questions.[60] The total score created a preliminary scored risk level based on the highest score achieved on either index. This score could be overridden manually by caseworkers to increase a score to either the top level in very serious cases (known as ‘policy override’) or by 1 risk level in others (known as ‘discretionary override’).
119. The tool’s outcomes are 4 final risk levels – low, moderate, high or very high. Caseworkers use the SDM ‘Risk-based Case Open/Close Guide’ to then make a closure decision.
120. Following a SDM Safety Assessment and SDM Risk Assessment, the caseworker is required by DCJ policy to decide whether to ‘substantiate’ harm or risk of significant harm.[61] This ‘substantiation’ decision is recorded on ChildStory in the ‘Judgements and Outcomes’ (J&O) tab.[62] Other guidance states DCJ’s ROSH determination is documented by recording whether the issues reported and considered during the field assessment are ‘substantiated, not substantiated or not assessed’.
121. DCJ’s policies state that substantiation occurs when caseworkers assess that ‘harm or risk of harm has been, or is likely to be, suffered as defined by s 23’.[63] Various documents guide caseworkers on how to make this decision.[64]
122. According to the tool manual, the SDM Risk Assessment Tool was not to be used to make the substantiation decision.[65]
In the 2023-24 financial year, of the 116,286 children and young people in screened-in ROSH reports transferred to CSCs:
 Source: DCJ Annual Statistical Report 2023-24 |
123. If significant harm or risk of significant harm is ‘substantiated’, Manager Casework (MCW) and above are also delegated the function of forming an opinion (under section 34 of the Care Act) as to whether the child is in need of care and protection (INOCAP). In most cases, the policy provides that this decision is to be made no more than 15 days after a safety and risk assessment.
124. If such an opinion is formed (i.e., that the child is INOCAP), DCJ may then decide to refer a family to a service (e.g., a family preservation service) for ongoing voluntary casework and close the case at that point – or provide ongoing statutory casework through its caseworkers. When DCJ undertakes casework, caseworkers develop a Family Action Plan for Change (FAP), usually within 45 days of determining the child is INOCAP. The FAP is a longer-term plan than the Safety Plan (which may have been developed after the initial SDM Safety Assessment). A FAP outlines the worries about a child and what needs to change to keep the child safe and future risk lower.[66]
In the 2024-25 financial year, 4,639 children and young people were determined by DCJ as INOCAP. Another 950 had an outcome of ‘was INOCAP’.[67] This was the first time such determinations were recorded as an INOCAP decision record was only introduced in ChildStory in September 2024. DCJ does not have reliable data about the relationship between ‘substantiated ROSH reports’ and the number of children and young people determined as INOCAP for the period under review. |
125. After a period of DCJ casework, a risk re-assessment of the family’s situation is done. Risk reassessments are to be conducted within 90 days of the FAP and every 90 days thereafter by reapplying either the SDM Risk Re-Assessment Tool or DCJ’s Alternate Assessment (for a child or young person in OOHC).
126. The SDM Risk Re-Assessment Tool combines the initial SDM Risk Assessment items with additional items that evaluate a family’s progress toward FAP goals. Like the SDM Risk Assessment, the SDM Risk Re-Assessment guides whether the case should remain open for ongoing casework or be closed, again using the SDM ‘Risk-based Case Open/Close Guide’.
127. Caseworkers can close an allocated ROSH report (an ‘open case’) at different stages for various reasons. One of these is closure for ‘Current Competing Priorities’, which can be done at any stage following allocation to the caseworker, including before any field assessment is done.
128. Otherwise, if they decide to close a case after a SDM Safety Assessment or SDM Risk Assessment, they are required to follow the Open/Close Guide on page 48 of the tool manual and consult with peers.[68] The guide is intended to help them determine whether to keep a case open or close it based on SDM Safety Assessment and SDM Risk Assessment outcomes.
129. For cases referred to family preservation programs, a different matrix may be used, allowing even high-risk cases to be closed before risks are fully resolved.[69]
130. Before closing a case, caseworkers are instructed to use the Peer Review Case Closure Guide to ensure the family is safe and the decision is correct. They can also refer to the Case Closure Practice Factsheet for guidance, especially if they are proposing to close a ‘high-risk case without current support’, which requires a manager's review and Director approval.
131. When closing an open ROSH report, caseworkers must record their reason in ChildStory using the Case Closure Reasons Guide. Closure reasons are grouped under three main category headings:
132. DCJ’s guidance for commonly used closure reasons includes:
Closure Reason | Use When … | ||
Assessments indicate Child/Young Person is not in need of Care and Protection | All cases where the initial risk assessment outcome is low or moderate, the outcome of the most recent safety assessment is safe, and where no follow-up intervention is needed. All cases where the initial risk assessment outcome is low or moderate, the outcome of the most recent safety assessment is safe and a referral to an external agency has been made. All cases where the outcome of an alternate assessment is safe, the level of future risk is low or moderate, protective action by DCJ can be ceased and the child or young person is not in need of care and protection. | ||
In need of Care & Protection – Family preservation program | When a family has been referred to a PSP preservation, family preservation, intensive family preservation or Aboriginal preservation program at any point during the casework journey where the initial assessment outcome was ‘in need of care and protection’. | ||
Current competing priorities | When a case is closed due to lack of allocation capacity and/or other competing priorities without response, following a triage outcome of Allocate. | ||
In the 2023-24 financial year, 5,243 (6%) of the 81,759 ROSH reports transferred to CSCs and allocated to a caseworker were closed on the stated basis of ‘Current competing priorities’. | |||
Source: Email from DCJ dated 15 October 2025, Item 1(c)
Notes: In 2023-24, 81,759 ROSH reports were allocated. At data extraction (31 August 2024), 30,280 (37%) had a recorded closure reason; the remainder were either open or had no status available in the extract. Bars labelled open case and case status not available show non-closed records at extraction; the other bars are reasons for the closed subset. Because closure reasons are linked across Helpline/CIW, some records lack a reason – figures should be read as indicative.
ROSH report closed as the ‘28-day closure period’ had already expired by the time it was transferred to the CSC – The case of Kendra and Nanette An extended family member of Kendra and Nanette contacted the Helpline raising concerns about Kendra’s non-attendance at school and chronic headlice, and that Nanette, who was under 2 years of age, was underweight. The report was screened-in and transferred to the relevant CSC by the Helpline 49 days after it was received. Under DCJ’s policy, ROSH reports not allocated for face-to-face assessment within 28 days of receipt by the Helpline are generally required to be closed – see section 4.2. Despite the CSC’s triage record noting that previous concerns had been raised about the hygiene of the home and the mother’s mental health, the ROSH report was closed by the CSC without allocation to a caseworker and without the case being considered at a weekly allocation meeting (WAM) or equivalent. The recorded rationale for closure states in its entirety: “This engagement is being closed as the report was not transferred to [this] CSC until more than 28 days after the initial report was made to the Helpline.” |
Case allocated after 10 months and 4 reports – The case of Roland and Herbie Four reports were received over a 10-month period about siblings Roland and Herbie, both under 6 years old. All raised similar serious concerns about their mother’s alleged substance abuse, violent physical outbursts against the children, educational neglect, lack of supervision of the children and lack of funds to buy food. The first 3 reports were closed due to NCTA, with the second triage record noting the report was concerning but ‘out of time’. The fourth report was allocated to a caseworker. |
Reports about sexual abuse allegations not allocated – The case of Lenore A report was received by the Helpline from a family friend of a lower primary school child, Lenore. The friend alleged that a relative by the name of Colin had been sleeping in Lenore’s bed for more than a year while he lived with the family. The report also stated that Colin had been seen shaving Lenore’s legs and exposing himself in front of her. The reporter said they had messages from another family member confirming what they were saying. After the report was screened-in, some inquiries were made by the CSC the next day with the Police and Lenore’s school but as these did not immediately confirm the concerns, the report was closed. Shortly after the closure, the school contacted DCJ twice to advise that the teachers had noticed Lenore’s shaved legs and that Lenore had said that Colin sometimes slept in her bed, but the school noted that she did not seem scared of him. No further action was taken. More than a year later, Police received a further report that Lenore had disclosed ongoing sexual abuse by Colin, after which he was charged and convicted of criminal offences, and sentenced to a term of imprisonment. |
133. As noted in Chapter 1, our 2024 Protecting Children at Risk (PCAR) report included unpublished data for the 2022-23 financial year. This showed that the number of children in screened-in ROSH reports who had their reports closed on the stated basis of insufficient capacity (and similar labels previously used) is significant. This trend has continued into:
134. While that number continues to increase, the percentage of children involved in screened-in ROSH reports allocated a DCJ caseworker for a field assessment and ‘seen’ continues to decrease (as shown in Figure 12B):
135. The Care Act does not expressly require a face-to-face assessment in all cases, and so the above observation does not of itself necessarily mean that DCJ has failed its statutory duty in all (75-80%) of the cases where it did not conduct a field assessment (see further section 4.3 below).
136. However, unless there is ‘insufficient reason to believe’ on the basis of the information in the report that the child or young person is at ROSH, the Act requires DCJ to undertake such investigation as it considers necessary to make a ROSH determination. This means that any closure of a ROSH report that where, because of inadequate resourcing, the Secretary does not undertake the investigations it considers necessary to determine ROSH, will constitute a failure by DCJ to fulfil its statutory responsibility.
137. In more than two thirds of the NCTA sample, we could not discern from the record that a view had been formed that the child was not at risk, or that a field assessment was not necessary. (Indeed, in 67 of the sample cases, our reviewers flagged that the reported information raised such manifest and substantial reason to believe that the child or young person was at ROSH that DCJ could not possibly have formed such a view.)
138. This was also borne out in our survey of relevant DCJ senior staff, with most confirming that they regularly closed ROSH reports despite not being satisfied, on the balance of probabilities (that is, it was more likely than not), that the child or young person was not-at-ROSH.[74] In terms of frequency, the most common response overall was that they took the closure action often (weekly) – 40% (52 of 131). Of those directly involved in triage-stage decision-making, an even higher proportion of staff survey respondents (48%; 39 of 82) said that they often closed such cases.
139. The survey also asked staff whether reports were closed in circumstances where they maintained serious concerns about the level of risk or safety of a child. Most (85%, 111 of 131) said that this happened. Again, in terms of frequency, the most common response indicated that this action was taken often[75] (33%, 43 of 131).
140. The issue of premature closure of ROSH reports is a longstanding problem, and there is no sign it is abating (examples of public reports raising concerns about the closure of ROSH reports from 2011 are summarised in Appendix A-5). While DCJ’s ‘children seen’ count gives a rough sense of how many matters progress beyond triage in CSCs, it is not a reliable proxy for the number of ROSH reports not closed at triage because the metric varies by unit of count (child vs report), timing and changing definitions over time. Despite the absolute number of children ‘seen’ almost doubling over two decades (from 12,700 in 2010-2011 to 24,234 in 2023-2024, after a peak of 35,241 in 2020-2021), the proportion of screened-in ROSH reports that result in a child being seen has fallen back to below its historical low of about 21% in 2010-11.[76] (see Figure 11B).
Sources: Unless noted, 2015-16 to 2023-24 are from DCJ Annual Statistical Report 2023-24. Earlier years: 2011-12 to 2014-15 from DCJ Annual Statistical Report 2021-22 and 2010-11 from FACS Annual Report 2010-11 (p.82) or FACS Annual Report 2013-14 (p.24) for the CYP series. CYP ‘seen’ figures (2010-11; 2011-12 to 2013-14; 2014-15 to 2016-17) drawn respectively on DCJ Submission to the 2017 NSW Parliamentary Inquiry (p.10), FACS Annual Report 2013-14 (p.24) and Aboriginal-led Data Sharing: Child Protection & OOHC Statistics. 2024-25 (where shown) from NSW Budget Estimates – Answers to Questions on Notice (2025) (Q4 & Q5). Seen % is calculated as CYP seen ÷ CYP in ROSH.
Notes: time-series and comparisons over time should be treated cautiously due to changes in definitions and methods (what is counted, who is included, extraction timing and system changes such as the change from KiDS to ChildStory in 2017).
141. The closure of ROSH reports (without having made an insufficient reason discretion decision or a not-at-ROSH determination) is embedded in DCJ’s Triage Policy. That policy generally provides for screened-in ROSH reports that have not been allocated for face-to-face assessment within 28 days of receipt by the Helpline to be closed. Such closure can occur without consideration at a WAM/peer review meeting.[77]
142. The policy states that the 28-day period starts from the date that the ROSH report is received at the Helpline.[78] Any additional reports about the child or young person received within the 28-day period do not restart the 28-day period from the original report.
143. The policy, which is published as a ‘mandate’, states that a ROSH report can remain open after 28 days only in exceptional circumstances and with the approval of the Director, Community Services. Exceptional circumstances include:
144. More generally under the policy, when a ROSH report comes into a CSC, staff are also to consider policy overrides for priority treatment (e.g., the sibling case coordination mandate). They must also consider if any funded services and DCJ staff are already supporting/case managing a child, and if so notify them. They have to check for children with more than 10 prior ‘reports’, and for any previous ‘reports’ closed without a response (the latter are escalated to MCS). All these considerations can operate to prioritise some children over others in the 28-day closure pool.
145. Although data weaknesses mean the number of ROSH reports closed on the basis of the 28-day rule is unknown, it is likely to be significant. DCJ has told us that its Helpline transfers most ‘ROSH reports’ to CSCs within 28 days of receipt by the Helpline (in 2022-23 it says this occurred in 85% of cases, 194,858 of 228,347).[79] However, during this investigation, eReport backlogs at the Helpline were at times up to 8,100. At least half (4,500) of these were older than 28 days (as at October 2024). In our review of NCTA closures, 28% (106 of 380) had been transferred to the CSC more than 28 days after receipt by the Helpline.
146. If CSCs receive reports that already exceed 28 days, it is more likely that these reports will be closed without investigation. In our review of NCTA closures, exceeding, or likely exceeding, timeframes was given as a reason for closure in 28% (107 of 380) of the cases.
147. DCJ could not provide us with an answer to the question of how it settled on a figure of 28 days for this policy. It did not produce any information to indicate that, if a report has not been responded to within 28 days, there is likely no need to respond to the report – for example, why its own inaction and the effluxion of 28 days make it reasonable to assume the child is not-at-ROSH.
148. In the Inquest of Z (2021), the Coroner raised the apparent inconsistency between the 28-day closure policy and DCJ’s responsibilities under section 30 of the Care Act:
DCJ’s statutory duty includes, under s 30(a) of the Care Act, the obligation to “to make such investigations and assessment as the Secretary considers necessary to determine whether the child or young person is at risk of significant harm”. The 28-day closure policy does not sit comfortably with s 30(b) of the Care Act which provides that, on receipt of a ROSH report, the Secretary “may decide to take no further action if, on the basis of the information provided, the Secretary considers that there is insufficient reason to believe that the child or young person is at risk of significant harm”.
“Insufficient reason to believe” is arguably not satisfied by inaction under s 30(a). The “information provided” is to be derived from the investigation and assessment made, not from the mere effluxion of 28 days. It ought be the case that DCJ never closes a ROSH report that is open and unallocated at the 28-day mark if the report has been assessed as requiring a response and no information has been provided warranting any change to that assessment, particularly if the report is in the “less-than-24-hour” category.[80]
149. A time-framed closure rule could in theory be acceptable, if it operated only to formally close out cases that DCJ is confident are in fact not (or no longer) ‘at ROSH’ cases.[81]
150. However, this is clearly not the case here: the 28-day rule is currently being applied to cases where DCJ has reason to believe – and in some cases has ‘serious concerns’ – that the child or young person is at ROSH and has not undertaken the necessary investigations to determine otherwise.
151. We agree with the Coroner that this policy as currently applied is not aligned to the Act, as it both provides for and results in the closure of ROSH reports without DCJ having met its statutory responsibility to undertake the investigations and assessment required by section 30.[82]
152. Furthermore (and as detailed in section 3.2), DCJ’s current triage and prioritisation decisions mean that DCJ cannot even be confident that the cases that are closed on the basis of the 28-day rule are the lowest risk/lowest priority cases.
153. The Care Act does not expressly mandate that a field assessment, where a caseworker ‘sees’ a child and their family face-to-face, is necessary to be conducted in every case where a ROSH report is made.
154. DCJ’s decision (at the Helpline) to ‘screen-out’ certain reports as ‘non-ROSH’ appears to be an example of a situation where a decision can, consistent with the legislation, be reasonably made that a field assessment is not necessary.
155. In the case of reports made under ss 24 or 27 of the Act, the screening-out process can be viewed as being in substance (even if not properly described or recorded in this way under DCJ’s procedures) a lawful application of the discretion in s 30(b) of the Care Act to not take further action if the Secretary considers that there is insufficient reason to believe that the child or young person is at ROSH. Clearly it would be better if such decisions were made overtly on this basis, with proper records of that decision being made. Nevertheless, as a matter of substance, where Helpline staff (with delegation to make decisions under s 30(b)))[83] receive a report and screen it out as ‘non-ROSH’ because they have formed a view that there is insufficient reason to believe the child or young person is at ROSH, then those decisions appear to be compatible with DCJ’s statutory responsibilities.
156. In this context it is noted, again, that when the 2008 Wood Special Inquiry recommended the trigger for reporting under ss 24 and 27 be raised to a risk of ‘significant’ harm, it was responding to concerns about too many reports being made to DCJ that did not warrant a response under the Care Act. It recommended the change expecting that it would result in those ROSH reports being reduced in future to only reports about ‘those children likely to need the powers of the State … exercised for their protection.’[84]
157. The expectation then was that all, or at least most, of the reports made under that new reporting test would warrant a caseworker fieldwork response, as such reports would be more likely to concern children and young people who genuinely required a response under the Care Act – that is, children and young people who are at ROSH and likely INOCAP.
158. However, DCJ has reported that there continues to be some s 24 and s 27 reports made to DCJ in circumstances where there are not reasonable grounds for suspecting a child or young person meets the statutory definition of being ‘at ROSH’.[85] Reports that, on their face, do not raise reasonable grounds to believe that a child or young person is at ROSH will, in most cases, be screened-out by the Helpline.
159. However, DCJ does not capture the statutory section for each report at intake. This means it cannot, for example, say precisely how many (and which) mandatory reports it received under s 27 of the Care Act. Instead, for any report received DCJ captures the reporter’s ‘contactor type’, being the sector the reporter comes from (e.g., NSW Education, NSW Health, NSW Police, Catholic school), then tags all sectors where the reporter may have reporting obligations as ‘mandatory reporters’. Everyone else is tagged as a ‘non-mandatory reporter’. However, the number of reports received from mandatory reporters is a poor proxy for the number of mandatory reports received (i.e. reports received under s 27 of the Care Act). This is because mandatory reporters can and do also make non-mandatory reports. These include s 24 reports about young people, noting that s 27 only applies to reports about children (aged under 16); ROSH reports about young people (16 and 17) are not mandatory. Mandatory reporters may also make other (non-mandatory) reports, including under s 25 (pre-natal reports) and s 120 and 121 (homelessness reports). Because most reports come from sectors that include mandatory reporters, counting as mandatory reports any reports from these sectors overstates the share of ROSH reports attributable to s 27. In short, the extent to which ROSH volumes are driven by mandatory reporting is unclear – though likely significant.
160. We have not focused on decision-making by the Helpline to screen-out concerns reports as non-ROSH (including some reports that may have been made under ss 24 and 27 of the Care Act), focusing instead on the larger volume of such reports that are being screened-in (as ROSH reports) but subsequently closed due to resource constraints. However, we note that issues have also been raised about these decisions, including in a number of coronial inquests.[86]
161. Even where a report has been screened-in as a ROSH report, the Care Act allows for the possibility that DCJ may make a considered decision not to provide a face-to-face response without that necessarily meaning that it has failed its statutory duty under the Act.
162. The Act both authorises and requires the Secretary to undertake such investigations and assessment as the Secretary ‘considers necessary’ to determine whether the reported child or young person is at ROSH. The Act also expressly permits the Secretary to decide not to take further action if they consider there is insufficient reason to believe the child or young person is at ROSH.
163. The Secretary is given discretion to decide what (if any) investigations and assessment are needed.
164. In many cases where a report has been screened-in as a ROSH report – that is, where a report has been assessed by the Helpline as raising a reasonable suspicion that a child or young person is at ROSH – a face-to-face field response will be both appropriate and necessary (and therefore required by the Care Act). Given the nature of risks that may be faced by children and young people (which frequently involve harm within families and otherwise ‘behind closed doors’), it will be essential for the caseworker’s investigations to include seeing the situation first-hand.
165. There appears, however, to be a widespread view within DCJ that, once a report has been screened-in as a ROSH report, caseworkers can never decide that a child is at-ROSH or is not-at-ROSH without a field assessment.[87] In some respects, this has been ‘baked in’ to ChildStory, which does not open the relevant ‘form’ to enable caseworkers to formally record a ROSH determination unless and until they have completed the form related to fieldwork assessment.
166. While we would expect it to be a relatively uncommon occurrence (particularly for mandatory reports under s 27), it seems to us that there could be cases where, even though a report has been screened-in as a ROSH report by the Helpline, the CSC may, following triage and further ‘desktop’ investigations[88], make a legitimate and reasonable decision that further field investigation is not necessary for it to determine whether or not the child is at ROSH.
167. This may in effect be what has happened in a small number of cases in our sample where some ROSH reports were closed because an apparent assessment was made that the child was not-at-ROSH even though this was not recorded as such in ChildStory, but closed with other labels, including the label of NCTA.
168. When finalising a ROSH report at triage, CSC staff are required in ChildStory to select a closure reason label. They can also provide additional explanation of their reasons for closure in free text commentary fields. Of the 380 triage cases we reviewed, there were at least 10 where DCJ staff had included comments in ChildStory fields that explicitly indicated that the decision to close, albeit labelled as being for NCTA, was based on a view being formed that there were ‘no current concerns’ about the child or young person being at ROSH (or words to similar effect[89]).
169. In other words, it is apparent that in at least these 10 cases, the ROSH report likely could (and if DCJ’s policies and practices had been better, would) have been closed on the basis of a decision (allowable under the Act) that no further investigation was necessary under s 30, rather than the closure being labelled and reported as a closure for NCTA.
170. Even beyond those 10 cases, we saw a number of others in our sample that had been closed with the label of NCTA where it appeared to our reviewers that it was possible (even though no record was made) that a triage officer might have formed a ‘not-at-ROSH’ opinion.
171. Given the flawed and limited recording of these decisions, it is impossible to accurately assess how many such cases may have been present in the sample. However, we estimate that it would certainly have been less than one third of the NCTA cases we reviewed.
172. Nevertheless, although it may be a minority of cases, it is important to acknowledge that the mere fact that a ROSH report has been closed without a face-to-face response (and even if DCJ ‘labels’ the closures with NCTA or a similar label) this does not necessarily entail a conclusion that DCJ has failed to meet its statutory responsibilities in all cases. DCJ may have fulfilled its statutory responsibilities, provided it had undertaken such other investigations and assessment as it considered necessary to form the view that the child or young person was not at ROSH, and it reasonably determined that no further investigation was necessary.
173. This observation should not, under any circumstances, be read as a green light to DCJ to adopt a standard practice of relying simply on ‘desktop investigations’ to make ROSH determinations. Rather, it is merely to say that, in circumstances where substantive desktop investigations have been conducted, and where the quality of the evidence before the decision maker is such that they can make a sound and confident determination as to risk, then it is open under the legislation for them to do so. However, where there remain any reasonable concerns or suspicions, a face-to-face assessment will always be necessary.
174. More generally, for so long as DCJ has sufficient reason to believe the child or young person may be at ROSH, the Secretary is required by s 30 to undertake such investigation and assessment as he or she considers necessary to determine whether the child is at ROSH. This means the ROSH report is not to be closed until DCJ has completed those investigations and assessment, determining whether the child is at ROSH and, if the child is determined to be at ROSH, then presumably deciding what further child protection action is required under the Care Act.
175. Closing a ROSH report when DCJ still has reason to believe the child or young person is at ROSH – including closing the report because of limited resources – is not consistent with DCJ’s responsibility under the Act. As the Coroner has bluntly stated on multiple occasions:
‘Where risk of significant harm has been identified, DCJ has a responsibility to act.’[90] (emphasis added)
176. A practice of closing ROSH reports without investigation – in circumstances where there are serious concerns about the safety, welfare and wellbeing of a particular child or young person that clearly warrant investigation – also sits uncomfortably (to say the least) with the objects and principles underlying the Care Act.[91] The objects of the Act include ‘that children and young persons receive such care and protection as is necessary for their safety, welfare and wellbeing, having regard to the capacity of their parents or other persons responsible for them’ (s 8(a)). The Act is also required to be “administered under the principle that, in any action or decision concerning a particular child or young person, the safety, welfare and wellbeing of the child or young person are paramount”,[92] while intervention should be the least intrusive in the life of the child and their family consistent with that paramountcy (s 9(2)(c)).
177. A final and important point to make here is that, even when reports are allocated for field assessment, DCJ has struggled to meet the response timeframes set by the Helpline for those reports.
178. DCJ Budget estimates data shows that in 2023-24, DCJ started a field assessment for 47,941 ROSH reports. The time from the Helpline assigning a response timeframe to the start of the field assessment varied by final response priority set by SDM SCRPT:
Source: NSW Budget Estimates, 10 March 2025, Families and Communities, Disability Inclusion (Washington), Answers to Questions On Notice, Attachment QTON#23, 24 Table
Note: The data source does not explain what ‘blank’ means. We assume, consistent with what DCJ has told us for similar data, it means data was not available at the time of extraction due either due to difficulties linking relevant records or the absence of data.
179. Data for another year (2022-23) showed that on average, DCJ completed a field assessment within 60 days for 57.1% of the allocated cases.[93]
180. Even if a ROSH report is allocated to a DCJ caseworker, children and families do not necessarily receive field investigation or a service or support at all. Some are closed on the basis of ‘Current Competing Priorities’ (see sections 3.2 (Step 9) and 4.4). Others fall through the cracks as caseworkers change.
181. For example, in the Taylor family case at the start of Chapter 6, reports were repeatedly made by Police, the children’s daycare provider and their grandmother. To prevent closure under the 28-day policy, the CSC MCW allocated both triage records to herself, noting that they had been sitting in triage “over time frames”. However, gaps in continuity of assigned caseworkers resulted in months of no contact with the family, prompting further reports, until eventually the grandmother removed the children herself and obtained parental responsibility through the Family Court.
Two reports by Police and NSW Health closed due to no capacity – The case of 3 siblings A ROSH report was received from the Police following a home visit prompted by the school about the condition of 3 siblings whose mother was suspected to have mental health issues. The oldest child was under 15. They were reported to be living in a garage in unhygienic conditions and appeared very pale. The report was closed due to ‘no capacity’ to allocate. More than a year later, similar concerns were raised in a ROSH report made by NSW Health that noted the children failed to attend school, the parents had mental health and substance abuse issues and were not engaging with services. This report was also closed due to ‘no capacity’ to allocate. |
Report about a child with a severe disability was closed without follow up – The case of Billy Three reports were made in quick succession about Billy, a child in primary school who had to be fed through a nasogastric tube. He lived in a remote location and was reportedly not being properly fed on weekends and failing to maintain his weight. The triage caseworker made several specific recommendations for follow up to be completed prior to the expiry of the ’28-day decision-making timeframe’. The case was closed without evidence of those recommendations having been acted on. |
182. As the data in the previous chapter shows, closing cases on the stated basis of resource limitations has become standard practice for DCJ. It is ‘business as usual’.
183. This point was made forcefully by the Coroner in the Inquest into the death of Harmony, repeating (for at least the third time)[94] the point that insufficiency of resources and the competing needs of other families does not displace DCJ’s statutory responsibility to take action where there are concerns about a child or young person at ROSH:
‘[In relation to a report that was screened in by the Helpline, made its way to the CSC weekly allocation meeting, and was closed ‘due to competing priorities’:]
I have no trouble accepting that there were competing priorities facing caseworkers tasked to allocate work at the WAM. Prioritisation decisions such as these must be almost impossible to make and undoubtedly contribute to the extreme pressure on all staff involved. What I do not accept is a departmental ethos where these kinds of decisions are seen as “business as usual”. As I have said on many occasions, DCJ is the agency in NSW tasked with a statutory responsibility for protecting children and young people from risk of significant harm. That is a responsibility that cannot be shifted by creating a culture where overworked staff can close reports, claiming a lack of resources or “competing priorities.” DCJ must be made to grapple openly with these issues at the highest level and to find solutions to the ongoing resourcing issues identified.’ [95] [emphasis added]
184. The cultural norm that exists within DCJ whereby ‘NCTA’ or similar labels are viewed as an acceptable and routine (and even preferred) basis for closing any case that does not result in a field assessment is evident in the apparent lack of any routine post-closure scrutiny of closure decisions (for example, randomised quality assurance (QA) sampling).
185. Far from raising alarm or attracting high-level scrutiny, when closure decisions are recorded as being on the basis of resource limitations, they attract effectively no attention, comment or criticism within DCJ.
186. DCJ has resource-intensive ‘group decision making’ and peer review processes at the point of deciding whether a matter should be allocated for field assessment or closed. However, there is no routine retrospective randomised quality assurance (QA) sampling, reviews, audits or other accountability or improvement-oriented scrutiny of NCTA or CCP closures.
187. The degree to which the routine closure of ROSH reports on resourcing grounds has been accepted within DCJ is also reflected in the fact that there appears to have been no attempt to ensure that, at the very least, these labels are applied only as a last resort and where the closure is genuinely and unavoidably because of resource constraints.
188. Instead, it appears that the labels (especially NCTA, but also CCP) are used as the default label when any report is closed without there having been a face-to-face field assessment, including in cases where an incapacity to conduct such a field assessment may not have been the primary reason for deciding to close the report without having conducted one.
189. As noted in sections 4.1 and 4.3, in our review of a sample of NCTA closure cases, we saw a small number where it was apparent that a view had been formed that there were no current ROSH concerns, as well as a number of other cases where, though less obvious (from the recorded comments), it also appeared possible the decision-maker had formed that view.
190. To some extent, at the CSC and individual decision-maker level, this is entirely understandable. One can readily understand why it may be more convenient to report that ‘we could not take further action’ (because of resources) than it is to say ‘we decided not to take further action’ (because we formed a view this child or young person was not likely to be at ROSH). It is also the case that, even where an opinion may be formed that the child or young person is not at ROSH with no further investigation being necessary, the decision-maker might still take the view that, with unlimited resources, they would still wish to allocate the matter for casework field assessment.
191. In any case, DCJ has repeatedly[96] failed to provide its staff with the means to accurately and consistently record these decisions (including, in particular, any decision that no further investigation is necessary because a child or young person is believed not, or is no longer, to be at ROSH).
192. For example, DCJ’s main decision record in ChildStory – the ‘Judgements and Outcomes’ (J&O) record – does not provide for the adequate recording of ROSH assessments, and is not available for use when substantive decisions are made during triage or at the Helpline, despite Helpline Team Leaders having delegated authority to do so. More generally, ChildStory includes no field at all where staff at any stage can record that a decision has been made that further action is not being taken because there is insufficient reason to believe a child or young person is at ROSH (s 30(b)).[97]
193. Recognising the importance of record-keeping in child protection, a bill was introduced to Parliament 25 years ago to change the Care Act to require DCJ to keep records of actions it takes in response to ROSH reports. However, this provision did not technically commence until 2024.[98]
194. There is, however, still no express statutory requirement to specifically record decisions made under s 30, or to provide reasons for those decisions. DCJ has made no clear provision for such records to be made in the design of ChildStory, and provides no guidance to staff on how they could or should do so. We remain concerned that, as decisions under s 30 are also not administratively reviewable, this means there is no legal requirement in the Care Act for DCJ to provide reasons for its decisions.
195. A similar observation applies in respect of reports that are recorded as having been closed for NCTA after a referral is made to a service provider. The ChildStory Triage Completion Guide in this case actually guides caseworkers to use ‘No Capacity to Allocate’ closure label in respect of ROSH reports in any case whenever ‘A referral to a service has been completed during the triage process.’ [99]
196. However, where a ROSH report is closed at triage stage after referral to a service, that decision may not necessarily mean that resource constraints have prevented an alternative course, such as allocation to a caseworker for field investigation. It is possible that, in some of these cases, the child or young person was referred to the service because a view was formed that the child or young person is not-at-ROSH, but that some assistance (by way of referral to a service) is still appropriate. That DCJ labels all such cases with the NCTA labels, indicates again how habitual the use of that label has become.
197. In short, DCJ’s practice is that staff are expected to record any closure that has not involved a face-to-face assessment as being simply a result of ‘NCTA’, with no guidance or tools provided to enable staff to accurately and comprehensively record any substantive assessment they may have made in relation to risk, their reasons for that assessment, or how the assessment has affected their decision to close the report.
198. At an agency-wide level, tolerating (or even encouraging) this approach has two significant consequences:
199. DCJ’s Prioritisation, Triage and Allocation Policy Review Project documents indicate that it was not until that review in 2023 that work was finally done, in consultation with its legal team, to map its processes to its legislative framework. It is concerning that it took DCJ so long before it critically and comprehensively appraised the lawfulness of its processes.
200. It is not clear why more definitive legal advice was never obtained (for example, from the Solicitor General) concerning complexities (real or perceived) in its duties and powers in respect of ROSH reports. It was evident to us that DCJ has for many years, from its frontline to its senior executive, laboured under confusion about various aspects of the nature and extent of its statutory functions.
201. In our view, this delay appears consistent with a culture that appears to have been insufficiently attentive to the critical importance of the legislation itself as dictating both the requirements and the limits of what DCJ can and must do. It is perhaps also, to some extent, reflective of an environment where DCJ was aware that it was chronically unable to meet its legislative responsibilities fully and in all cases, and so there may have been little incentive to focus on a detailed clarification of those responsibilities.
202. It is also relevant in this context to recognise that DCJ’s child protection workforce is primarily a caseworker cohort, with critical skills in social work, psychology and related fields necessary for providing trauma-informed and effective services to families. However, these staff are also the ones responsible for making administrative recommendations and decisions that require a nuanced understanding of statutory duties and powers.
203. In an environment such as this, the importance of clear guidance on administering these duties and powers cannot be understated. Instead, while DCJ casework mandates are drafted in a way that guides casework staff in their interactions with families, they do not clearly and adequately set out requirements and criteria for making decisions under the Care Act, and in some cases there is also a disconnect between the legislation and the framework and tools (including SDM tools – see section 9.3) that have been provided to support caseworkers in their functions.
204. It is acknowledged that, since the commencement of this investigation, DCJ has begun developing clearer decision-making guidance for assessing ROSH and INOCAP, indicating that it is starting down a path to greater clarity. It should continue on this path.
|
Face-to-face visit attempted after receipt of 10 ROSH reports – The case of the Taylor family
Over a 3-month period, multiple ROSH reports were received from extended family, the Police and a daycare centre about siblings, Robby and Fran, both under 5 years old. The reports raised concerns about the mother’s substance abuse and lack of food available for the children, who were often left unsupervised and neglected. The children had already come to DCJ’s attention before these reports were received, and an earlier SDM Risk Assessment outcome had an outcome of ‘very high risk’ although we could not confirm from the record whether ROSH was substantiated.
After the 10th report was received, all the reports were finally assessed together and a caseworker was allocated. A home visit was attempted but did not proceed. A few weeks later, a family member applied to the Family Court and was awarded parental responsibility of the children.
A request by JCPRP for ongoing case management ignored – The case of Lachlan
JCPRP requested a CSC take ongoing case management in relation to an Aboriginal child, Lachlan, as it was concerned about medical and educational neglect and mental health issues. As the case was not allocated by the CSC, JCPRP repeatedly sought action and flagged Lachlan’s isolation and absence from school. Even though a specialist review recommended ongoing casework by Aboriginal case management, the family was eventually referred to a non-Aboriginal service, and the case was closed.
Five reports about a child with a disability made by the school closed partly on the basis that the school itself could take action – The case of Clare
A school made 5 reports to DCJ within a year about Clare, a child with a disability. All 5 were closed without Clare receiving any services. The reports raised concerns about lack of school attendance, bruising, escalating behaviours and Clare not receiving the right services.
The last report was screened-out as non-ROSH by the Helpline, noting the school could take further action regarding Clare’s non-attendance. This was despite the fact the previous report (which had been screened-in as a ROSH report) had a notation that the case would likely be allocated if the school made a further report.
205. It is no doubt the case that, like all public sector agencies, DCJ operates within a resource-constrained environment. In circumstances where DCJ has for decades said that this means that it cannot investigate and respond to ROSH reports in all or even most cases, then it must be even more imperative that DCJ has processes and practices that ensure effective and efficient decisions about the allocation and prioritisation of its resources to the highest-risk cases.
206. More often than not, those ROSH reports that are allocated to a caseworker for a field assessment (i.e., face-to-face investigation) do not result in a determination being made that the child is at ROSH.
207. In 2022-23, only 48% of the children and young people (13,444 of 27,782) who were the subject of finalised field assessments following a ‘ROSH report’ had actual harm or risk of harm ‘substantiated’. This proportion was substantially the same during 2023-24, at 49% (11,896 of 24,234).[105]
208. That some ROSH reports will, after a field assessment, result in a determination that the child or young person is not-at-ROSH is to be expected. However, in theory, the better DCJ’s processes are for identifying likely risk prior to field assessment, the lower the proportion of not-at-ROSH determinations (following field assessment) should be.
209. We do not have a view about what that proportion should be. However, the apparently high proportion of ‘not-at-ROSH’ determinations made in respect of those ROSH reports that are triaged for field assessment warrants further consideration, particularly in an environment where the other two thirds of ROSH reports are closed without allocation and where court cases,[106] coronial investigations,[107] child death reviews,[108] as well as our case reviews for the purposes of this investigation repeatedly identify ROSH reports being prematurely closed despite manifestly serious concerns having been raised, often multiple times.
210. DCJ has a goal of increasing the number of children ‘seen’ in response to ROSH reports. DCJ’s ‘seeing more children’ measure sets CSC and district targets for the numbers of ‘children seen’ in a given year. On a per capita basis, the target equates to face-to-face assessments being conducted on average for approximately 20 children per caseworker per year.[109]
211. Given the circumstances described in the previous chapter of this report, it is understandable and appropriate that DCJ should seek to increase its face-to-face response rate to ROSH reports.
212. However, these internal performance targets can have the tendency to distort triage decisions in a way that means that DCJ is not always making the best risk-based prioritisation decisions.
213. This concern was foreshadowed in a DCJ Casework Study in 2021 which reported working group feedback that Key Performance Indicator (KPI) results can be affected by allocation practices to meet targets and found that existing performance targets pressured caseworkers to focus on volume-based activity rather than quality casework.[110]
214. During DCJ’s 2023 Prioritisation, Triage, and Allocation Policy Review, DCJ staff consultations highlighted (not for the first time[111]) the tension between meeting such face-to-face performance targets and prioritising responding to the 'right' children. Districts reported that:
215. Similar views were also raised with us by DCJ staff, who told us that DCJ’s performance metrics create a perverse incentive for some staff to:
216. In a compulsory interview at one district, we heard that targets influencing some allocation decisions ‘had happened’, although the practice was not thought to be widespread. We also heard that there are no QA processes in place for analysing data to mitigate or identify any distorting effects on allocation.
217. Despite this, survey results suggest these practices are a systemic issue of concerning prevalence. We asked those staff who had a direct role in allocation for their views about the impact of the targets on allocation practices, if any. Nearly half (45%; 42 of 93) responded ‘yes’ to the question whether, in their experience, the targets influenced some staff to allocate in a way that would increase the children seen statistics, instead of allocating to those children and young people assessed to be most at risk. Of these, 59% (24 of 42) reported having directly observed this occurring. In free text responses when asked what kinds of reports are allocated to improve ‘children seen’ KPIs, staff noted much the same examples as set out above. The most common response was favouring reports concerning larger families (77%; 30 of 39 of the free text responses mentioned this).[112]
218. Of course, the number of children implicated in a ROSH report might, depending on the circumstances, also be a genuinely relevant factor when triaging the report for action under the Care Act based strictly on risk. However, if action under the Care Act is biased toward large families merely for reasons of numerical advantage in children seen data, then this raises the risk that they will be prioritised over more ‘at risk’ children. Other concerns have been raised, including that the practice of preferencing larger families could result in a disproportionate impact (that is, an increased likelihood of DCJ’s active intervention) on certain cultural backgrounds, such as in Aboriginal families,[113] if larger or extended family households are more prevalent.[114]
219. A number of staff noted that the drive to improve the children seen KPI may be heightened at certain times, for example, with encouragement toward an ‘end-of-month push’ to improve those statistics.
Figure 13 – Survey responses about the influence of targets on allocation practices
Survey Question: ‘In your experience, did ‘Seeing More Children’ targets influence some staff to allocate reports in a way that increases the ‘Seeing More Children’ statistics instead of allocating CYP most at risk?’
220. During the course of our investigation, DCJ informed us that, to address concerns that the KPI only counts ‘first’ visits, it has adjusted its children seen measure so that it now also counts second and subsequent visits to children during a 12-month period.
221. In a resource-constrained prioritisation process, it is reasonable to expect DCJ to assess and compare risks, including by reference to imminence – that is, when the harm is likely to be suffered. However, while imminence is a relevant and important consideration for prioritisation, DCJ’s current approach to allocation decisions can at times result in ROSH reports being more likely to be closed on the basis of NCTA where the threat, though perhaps less immediate, is more severe and/or likely.
222. For example, some practices embed an assumption that ‘urgency’ can be a proxy for ‘importance’. This includes reliance by some staff on SDM SCPRT’s Response Priority Timeframe outputs when prioritising matters for allocation and the 28-day closure rule (which presumes that if it was not urgent enough for another report to be made in that time period then it is unlikely to be important enough to allocate now – see section 4.2).
223. This potential bias is most visible in serious neglect cases. Cases of chronic neglect – e.g., persistent failure to attend school or medical appointments, failure to thrive, unsafe supervision or compounding developmental concerns – may lack an ‘acute incident’ but nevertheless carry high severity and high likelihood of harm. If such reports are assigned longer response windows; when backlogs push beyond 28 days, they are more vulnerable to closure or a ‘refer and close’ pathway, even though the risk is ongoing and cumulative. In effect, the absence of immediacy can eclipse the gravity of the neglect.
224. There are situations where DCJ’s Triage Policy provides that a ROSH report must be allocated (for caseworker field assessment) and/or prioritised according to specific mandates. These include:
225. JCPRP is a statewide program that aims to provide a comprehensive and co-ordinated safety, criminal justice and health response to children alleged to have experienced sexual abuse, serious physical abuse and serious neglect. DCJ has a mandate concerning transfers of case management between JCPRP and CSCs.[115]
226. When JCPRP has case management it is responsible for assessing risk/safety and whether a child is INOCAP. If the case needs to remain open at the conclusion of the JCPRP assessment, case management transfers to the CSC.
227. When receiving JCPRP transfers, the mandate states CSCs:
‘are to accept all cases transferred to them by JCPRP and carry out a response that reflects the level of risk identified by JCPRP’. [emphasis added]
228. A Transfer record is created by JCPRP in ChildStory, and the CSC receiving the transfer must then create a new Triage record and allocate the case.
229. We found evidence of JCPRP-transferred cases where no further response under the Care Act was provided by the receiving CSC and, although they were allocated, allocation was explicitly and solely ‘for the purpose of enabling closure’ under ‘Current Competing Priorities’. This means that although these matters were technically allocated, they were not prioritised for further casework in accordance with the underlying intent of the policy, which is concerned with continuity of service for cases JCPRP decides need to stay open.
230. The problem of continuity of service following a JCPRP transfer was recently commented on by the Coroner in a case where ‘it was as though the work done by the previous group [JCPRP] was treated as being over, instead of forming a continuum of care provision’.[116]
231. Approximately a quarter of cases that are ‘substantiated’ as at ROSH by the JCPRP and referred to CSCs for further casework are then closed on the stated basis of current competing priorities:
Table 1 – Field assessments accepted by JCPRP and assessed as 'substantiated' and transferred to CSC in 2022-23 and 2023-24
| 2022-23 | 2023-24 |
Case closed due to current competing priorities | 244 | 206 |
Case closed due to other closure reasons | 303 | 216 |
Case still open | 428 | 446 |
Total | 975 | 868 |
Source: DCJ data provided to the NSW Ombudsman
232. Further, JCPRP assessed risk levels for cases transferred then closed under Current Competing Priorities (prior to decommissioning of the SDM Risk Assessment tool in September 2024) were generally High or Very High:
Table 2 – Field assessments accepted by JCPRP and assessed as 'substantiated', transferred to CSC and subsequently closed due to competing priorities in 2022-23 and 2023-24, by last risk assessment outcome
Last risk assessment outcome | 2022-23 | 2023-24 |
Very high | 88 | 66 |
High | 141 | 127 |
Moderate | 10 | 9 |
Low | - | <5 |
Source: DCJ data provided to the NSW Ombudsman |
233. While data shows almost all JCPRP transfers that were closed due to Current Competing Priorities were assessed by the CSC as ‘safe’ or ‘safe with plan’, it is important to note that this SDM Safety Assessment considers immediate safety, not ongoing risk to the child or young person (see section 3 (Step 5)). It is also inconsistent with the case transfer mandate’s requirement that the matters be given a response ‘that reflects the level of risk identified by JCPRP’.
234. We obtained a copy of the working registers maintained by JCPRP teams to track matters referred to CSCs. These registers are working documents, not designed for quantitative analysis, but they were valuable in identifying cases for closer qualitative analysis (see Appendix A-4).
235. Senior DCJ staff we spoke to explained that, notwithstanding the mandate that all JCPRP transfers must be allocated, those referrals go to WAM meetings to be considered for allocation. One senior manager we spoke to acknowledged that this does not sit well with the mandate for case transfers, which requires allocation of JCPRP transfers. However, they said that ethically they did not feel they could justify preferential treatment of a JCPRP transfer, so a conscious decision was made to refer them to WAMs to be considered against the potentially higher needs of other children.
236. Lachlan, an Aboriginal child, is a case where JCPRP expressed repeated concerns about its referral not being allocated. JCPRP advocated for a review by a DCJ casework specialist who recommended ongoing DCJ casework or failing that, a referral to an Aboriginal NGO. The casework specialist advised that whatever the outcome, a further SDM Safety Assessment needed to be completed prior to the case being closed. DCJ then closed without additional casework or a further SDM Safety Assessment, following a referral to a non-Aboriginal NGO.
237. The Sibling case coordination – JCRP and CSC Mandate says that if JCPRP is involved with a child, and there are risk and safety issues for siblings or other children living in the same household (but not themselves currently the victims of any alleged crime), joint planning and information sharing (i.e., a consultation between JCPRP and the CSC) must take place. The mandate explicitly notes that there will be times when such a family will need to be allocated over other matters in the CSC of equal or greater priority, but the mandate is not explicit about what times these will be.
238. DCJ staff we heard from raised concerns that the mandate’s consultation requirements are not always complied with, and that sibling cases are sometimes prematurely closed on the basis of JCPRP involvement with the other child. In our review of JCPRP (spreadsheet) records, we found a small number where Sibling Case Coordination was marked ‘Yes, unallocated’, but case consultation with the CSC/NGO was marked ‘No’. We also observed in our review of NCTA closures at CSCs that ‘JCPRP involvement’ was sometimes given as the reason for that closure where the report being closed was about the sibling of a JCPRP-allocated child (9 cases).
239. The Triage Policy also requires that ROSH reports must be allocated (for caseworker investigation) if they concern a child or young person who is already in care. However, the Safety in Care Mandate also permits caseworkers to seek Director Community Services approval for a decision not to allocate the report for caseworker investigation.
240. Despite this, not all districts have dedicated Safety in Care (SIC) teams available to respond to such reports. In those districts that do not have a SIC team (and in some districts that do), ROSH reports about children in care are referred to WAM (where allocation may be declined on the basis of ‘No Capacity to Allocate’ or ‘Current Competing Priorities’.
241. A standout issue in the cases we reviewed concerned children with disability (or whose parents or carers had a disability), falling through gaps between child protection and other services.
242. The key issues we observed were inadequate information gathering during triage, over-reliance on the involvement of other services when deciding to close, and failing to give sufficient weight to vulnerability when assessing risk to the child with disability.
243. In Ella’s case (outlined at the start of Chapter 8), a non-verbal child with severe autism, was reported as not attending school or NDIS services. In our view, DCJ placed too much weight on the involvement of school and disability services when deciding to close the report, given that the key issue in the report was non-engagement with those services.
244. Similarly, in Jarred’s case (outlined at the start of Chapter 7), Jarred was reported as a child with severe autism who had not attended school for many years and whose mother was allegedly mentally unwell and preventing access to all services, other than his psychologist. A planned Interagency Case Discussion following a ROSH report did not eventuate, and the report was closed with a note that Jarred’s case should be allocated if re-reported. Jarred’s social isolation was a significant relevant vulnerability which was not afforded sufficient weight. Jarred was only attending school sporadically for a few months while the reporter was working with the family, but prior to that had a period of 3 to 4 years of non-attendance. Based on the information in the report, his only external support was one psychologist.
245. Clare’s case (outlined at the start of this chapter), is another example of report de-prioritisation based on involvement of schools and disability services in circumstances where inadequate engagement with school and non-engagement with disability services is at the heart of the report. In this case, there was a reported pattern of escalating reports (in both severity and frequency) and repeated NCTA closures. At the point at which reports escalated to include evidence of physical injury (facial bruising), Helpline backlogs were cited as a reason for closure, with an observation that any further reports would likely be allocated.
An under-24-hour response priority report about potential risk of sexual abuse was closed after no further response received from Police – The case of the Smith children
DCJ received a ROSH report from NSW Police about a man, who was previously convicted of child-related sex offences in another state, having moved in with a woman who had 2 young children living with her. The report was transferred to a CSC with an ‘under-24-hour’ response timeframe.
The CSC attempted to contact NSW Police over the next few days to request more information. No response was received. The report was closed a month later on the basis that it had exceeded timeframes and there was no capacity to allocate it.
An under-24-hour response priority report was closed within a week due to competing priorities – The case of Melissa
DCJ Helpline received a report that Melissa, a child under 5, was not receiving medical care for burns, which became infected as a result. There were also allegations of drug use by the mother.
The Helpline assigned an ‘under-24-hour’ response time priority.
On the same day, a DCJ triage caseworker attempted to contact a hospital and Melissa’s preschool to ask relevant questions. These attempts at contact were unsuccessful.
The report was closed 6 days later on the basis of competing priorities.
Interagency Case Discussion regarding an autistic teenager did not go ahead, and report was closed – The case of Jarred
NSW Health made a ROSH report about Jarred, a teenager described as having severe autism, who had not attended school for many years and whose family was preventing him from accessing services.
An initial decision to hold an Interagency Case Discussion (ICD) did not proceed and the report was closed on the basis of no capacity to allocate, with a notation that the case should be allocated if another report were received.
246. The 2021 FACSIAR time and motion study found that more than half (64%) of caseworker time during the triage process was spent on information gathering.[117] Despite this, in the NCTA sample we looked at, underutilisation of desktop investigation options was a recurring theme.
247. Desktop investigation may, in some cases, produce information that shows a child is less likely to be at ROSH than children in other reports, and therefore improve prioritisation of field assessment to those reports involving the highest risks.
248. In some cases, desktop investigation can also help DCJ identify the most appropriate referrals and supports that may be warranted. As noted in sections 4.1 and 4.3, in our review of NCTA closures, it was also apparent that in some cases, the triage officer had formed the view based on desktop information that the child was not-at-ROSH on the basis there were no ‘current concerns’, and that this view appeared to be reasonably held.
249. This was the premise of DCJ’s Helpline Advanced Screening Pilot (HASP), which conducted early desktop enquiries (prior to running the SDM SCRPT tool) to identify relevant information, including whether there were protective factors present that meant the child was not-at-ROSH. The HASP was operated as a trial from 2018. It was repeatedly extended, but DCJ said it could not be rolled out beyond certain geographical trial sites due to resourcing. It ceased altogether in 2024, on the stated basis it was not aligned with the requirements of s 30 – that DCJ was using s 30(a) powers (making enquiries and referrals) to decide whether a report met the definition of at ROSH – that is, using s 30 to decide whether s 30 was enlivened.[118]
250. It appears that DCJ’s current policies and practices are such that a determination under s 30(a) of the Care Act (as to whether or not a child or young person is at-ROSH or is not-at-ROSH) can effectively only be made (or at least recorded as having been made) if there has been a caseworker field assessment, and never solely on the basis of desktop investigation. Even if that policy view is warranted (which we doubt – see section 4.3), in circumstances where most ROSH reports are closed on the basis of NCTA, an early, efficient and thorough desktop investigation is essential to ensuring that the lowest-risk reports are the ones that will be closed (even if they have not been subject to a formal ‘not-at-ROSH’ determination).
251. Staff reported to us that, during triage, attempts to contact the child or family occur at least sometimes (80% of survey respondents reported this; 67 of 82) and that other desktop-based activities (e.g., obtaining other information) occurs often (51% of survey respondents reported this; 42 of 82), with District Hubs more likely to report that desktop-based activities occurred always.
252. However, we also asked staff if they had been involved in the closure of cases where reasonably available information necessary for deciding if a child or young person was at ROSH was not collected. The most common response overall was sometimes (35% of survey respondents reported this; 46 of 131). See, for example, Lenore’s case at the start of Chapter 4.
253. We also observed a practice of not making reasonably straightforward enquiries (e.g., of reporters, agencies or the family) to find out what the child’s circumstances before closing reports due to the expiry of 28 days, and sometimes months after the report was received. See, for example, the three siblings case study children at the start of Chapter 6.
254. DCJ has information-gathering powers that permit it to obtain information from other agencies, for example NSW Health, Department of Education and Police.
255. In this area, the timeliness of information exchange is clearly critical. One DCJ staff member told us:
‘Limited information sharing and collaboration between agencies and professionals or delays in receiving information within timeframe allocation policies can oftentimes hinder the ability to make informed decisions’.[119]
256. The Smith children’s case and Melissa’s case at the start of this chapter illustrate a recurring theme where ROSH reports were closed before information requested from third parties had been received. In other cases, attempts to contact third parties were made, but abandoned if a connection could not be quickly established.
257. The 28-day rule in the Triage Policy likely suppresses information gathering: because keeping a report open beyond 28 days requires ‘exceptional circumstances’, the policy discourages staff from waiting for key third-party information before closing matters (e.g., only kept open if a ‘pivotal’ item is pending). This risks bias towards reports where information is fast to confirm and may deter complex, multi-agency matters that take longer. The approach may also be inconsistent with the broader intent of the Care Act’s agency exchange of information provisions.
258. At least one Allocation Hub has DCJ Helpline, triage and allocation staff co-located with staff from NSW Education CWU, the Benevolent Society and the Family Connect Support Service. We understand this facilitates rapid access to relevant information held by these other agencies. However, this arrangement is not common across the department.
259. The Triage Policy requires staff to conduct ChildStory checks to ‘form a holistic picture of risk’ beyond the current report being assessed. They must review the child’s history and that of their siblings or other children in the household, their parents and caregivers. They are also to check for previous reports (particularly recent ones) closed without a response. If there are previous unallocated ROSH reports closed without response, these are supposed to be escalated to an MCS (it does not appear that this generally happens).
260. ChildStory history checks are a key source of information about escalating risk. Our survey of staff indicates that ChildStory checks are routinely completed during triage, with 67% (55 out of 82) responding that this always happens. However, this happens less frequently (often) for reports older than 28 days.
261. Nevertheless, we observed cases where multiple (and escalating) reports were repeatedly closed on the basis that there was no capacity to allocate, even when earlier decisions had been made with closure commentary that explicitly stated that, if a further report is received, the next report should be allocated. See for example, the Roland and Herbie’s case at the start of Chapter 4.
262. We also observed cases where multiple risks or vulnerabilities had been reported (often in separate reports) but the response provided under the Care Act did not reflect an adequate assessment of all the important risk information on the file. For example, we saw cases where consideration appears to have been given only in respect of some, but not all, key risk issues. See, for example, Herbie and Roland’s case at the start of Chapter 4.
263. Interagency Case Discussions (ICDs – sometimes referred to as Interagency Panels) are important consultation processes for managing complex risk and service coordination. However, they are not often used prior to NCTA closure.
264. In our review of NCTA closures, only 4% (15 of 380) indicated in the triage record that the matter had been taken to an ICD prior to closure.
265. Staff survey responses also indicated that ICD use is infrequent and irregular.[120]
266. Perceptions about the time and effort involved in establishing an ICD may be a disincentive to their use, although a 2021 DCJ Casework Study challenged this view:
The ICD results are interesting and dispel some myths. Feedback from our [working group] is that ‘setting up an ICD takes time and resources, which may be a disincentive for CSCs when they know that they are closing these reports. However, the hub uses ICDs regularly and shows them to be an efficient use of time, over time spent gathering information for matters that still proceed to closure. An ICD is better practice than simply gathering information and closing – as they offer some level of safeguarding to the child by creating a network that knows what might be happening for a child.’[121]
Three reports about a family were closed with referral to a family preservation service – The case of Eliza A primary school made a ROSH report that Eliza had a bruised and swollen eye. After follow-up by DCJ, the school agreed to refer the family to a family preservation service and the report was closed due to no capacity to allocate. A few weeks later, another report was received from the school alleging sexualised behaviour by one of Eliza’s siblings and reporting that the family was homeless and all the children had diagnosed disabilities. This report was also closed due to no capacity to allocate, and on the basis of the school’s agreement to refer the family to a family preservation service. A few days later a third report, this time by NSW Health, stated that Eliza’s younger sibling was malnourished and not being given prescribed supplements. Although this report was allocated and a safety assessment completed, it was finalised through a referral to a family preservation service. |
Reports about non-engagement with services closed on the basis of the involvement of those services – The case of Ella A disability service made 2 reports within 10 days of each other saying that Ella, a lower primary school, autistic and nonverbal child, received NDIS funding for services, but the family was not engaging. Ella was also not attending school. This was confirmed in a separate report made by the school, which also alleged that the mother was suspected of using drugs. The 2 reports from the disability service were closed due to no capacity to allocate, exceeding timeframes, and the involvement of support services. This was despite the fact that the reports had been made by that service itself, and because there was a failure to engage with services. The report made by the school was closed on receipt as a ‘duplicate’ of the reports made by the disability service. |
A family preservation service’s attempts to get assistance from DCJ – The case of the AA Family We were contacted by a family preservation service supporting a family of 4 children, which had made 4 ROSH reports over 5 months raising their concerns about hoarding, mental health issues, safety concerns, lack of supervision and neglect. The service provider, who wished to remain anonymous, told us they made the reports in an attempt to improve collaboration with DCJ and its responsiveness. However, having made the reports, the service reported to us that they still had to ‘chase’ DCJ to attempt to have their concerns addressed. |
267. DCJ may refer families involved in ROSH reports to other agencies and service providers before closing a report. In the context of DCJ’s Care Act responsibilities, there are several ways referrals could be utilised during triage:
268. Although these uses of referrals are conceptually distinct, DCJ’s practices are blurred. DCJ does not require decision-makers to distinguish those cases (if any) where the involvement of another agency or service was such that it led to a view being formed that the child or young person was not (or no longer) at ROSH.
269. This obscurity means that we do not know which referrals were made because it was considered the most appropriate response under the Care Act to that case, and when instead it was merely done as a better-than-nothing alternative in circumstances where the more appropriate response (resources permitting) would have been to allocate the matter to a DCJ caseworker for investigation.
270. What is apparent, however, is that DCJ frequently relies on the involvement of other agencies, and service referrals it has made, when deciding that a report is to be closed on the basis of NCTA. That happens despite there being no assessment about whether the referral has or will adequately mitigate risk.
271. For example, we saw many reports closed with the label of NCTA apparently on the basis that another agency was involved (and therefore had ‘eyes on’ the child). This was despite there still being unresolved indicators of serious risk. This occurred in some cases where the other agency in question had itself been the source of the ROSH report, and where the concerns raised in its report had included the family’s refusal to engage with that agency. This kind of over-reliance on the fact that there is some ‘other agency involved’ as a basis for de-prioritising and then closing ROSH reports appears to be a recurring issue, particularly for children with disabilities (see section 6.5).
272. DCJ does not routinely follow-up to check whether, following referral, the risk has, in fact, been mitigated. In most cases it does not even check whether the referred service was even provided, let alone the outcomes for the child or young person if it was. This is an issue that the NSW Ombudsman has raised repeatedly over many years, noting that DCJ has had no process for following up referrals where acceptance of the referral is not confirmed or is declined.[122]
273. We saw cases where the triage recommendation was to make a referral (rather than to allocate to a caseworker), but the referral did not eventuate. For example, this might occur where limited service availability meant that a referral could not be effected within 28 days, and so the report was simply closed for NCTA with no referral having actually been made.
274. Most of the ROSH reports closed by DCJ on the basis of NCTA do not result in a referral to any service. It appears that up to around 80% of the children or young people who are reported at ROSH and not seen by a caseworker also do not receive a referral. However, data limitations mean the true numbers are unknown.
275. Of the children and young people reported at ROSH but not seen in 2022-23 (92,052 children), DCJ estimates that:
276. In our review of NCTA closures, the need for a referral was considered by DCJ in only 11% of cases (40 of 380). Of these, a referral was only actually made in 21 cases (DCJ made the referral in 15 cases, while the reporter or another agency made the referral in the remaining 6 cases). A ChildStory review showed that a service request (REQ) record was created on ChildStory to capture details of the referrals made by DCJ in only 10 of the 15 cases. None of these records were subsequently updated to reflect the outcome of the referral.
277. Our survey of DCJ staff also indicated that consideration is given to connecting families with services when capacity to allocate caseworkers is limited, although there is little structure and guidance as to how this should be rigorously and consistently done. When asked about the challenges of making appropriate service referrals, staff pointed to:
278. Another challenge arising from referrals being driven by lack of capacity to allocate a DCJ caseworker to the case, is that there can be little if any substantive assessment of the family’s service needs prior to referral, resulting in the potential for inappropriate or ineffective referrals with a mismatch between service and needs.
279. Children and their families can get ‘stuck’ in closure and re-report loops for long periods. These loops can also produce ‘reporting fatigue’, where reporters stop reporting because they see no action or receive little feedback.
280. It is not uncommon for DCJ staff to make a note when closing due to resource constraints that if any further reports are received, they should be allocated (e.g., Clare’s case at the start of Chapter 6 and Roland and Herbie’s case at the start of Chapter 4). But we also saw repeated examples where the subsequent re-reports were also closed due to resource constraints, or where they were closed as duplicates (see section 8.4 below). This can entrench both repeated re-reporting and reporter disengagement.
281. In a meeting with a Children’s Court Magistrate, we heard feedback about cases where there had reportedly been many (in some cases in excess of 100) earlier reports, many of which were closed citing resource constraints, before necessary action was taken.
282. Re-reporting also arises frequently in cases where reports are closed after service referral. In 2022-23, 56% of all children who exited an FPS were re-reported at ROSH within 9 months, regardless of whether they completed the program. Even among those who completed the program, 50% were re-reported.[124] As at March 2025, 71.7% of all children whose cases were closed for the closure reason ‘In need of Care & Protection – Family preservation program’ and referred to FPS were re-reported within 12 months of case closure.[125]
283. DCJ has been aware of re-report issues for some time. Multiple DCJ evaluations of services orbiting DCJ’s triage system have expressed concerns about DCJ’s closure practices, which can significantly affect their service delivery and create cyclical problems and re-report loops that can undermine provision of appropriate support and resources.[126] The Audit Office also recently made observations about re-report drivers.[127]
284. There is also a risk of under-reporting. Concerns have been raised that when there is little visible action or feedback on prior reports – and organisational messages emphasis reducing ‘re-reports’ – some mandatory reports may stop reporting or re-reporting.
285. An early evaluation of Child Wellbeing Units found that lack of closure feedback and divergent outcomes between the SDM MRG and SDM SCRPT (e.g., the Helpline screening out as non-ROSH after the SDM MRG recommended reporting) discouraged reporters; this theme appeared in 40% of case studies reviewed by the evaluators.[128]
286. A NSW Coroner (Jade’s inquest) recorded evidence that service providers were “educated” by DCJ not to make s 27 reports about families they were already supporting, reinforcing the risk of under-reporting.[129]
287. Under-reporting can mask escalation, skew performance data and shift risk to informal channels (e.g., internal noting and direct CSC contact instead of Helpline reporting).
Figure 14 – referral and re-/no report loops
288. Practice failures (and underlying policy settings) can increase re-report volumes. The following table outlines some of the practices that appear to influence or drive re-reports to DCJ.
Table 3: Some practices that influence re-reporting
Practice | Influence |
Closure, delay and inaction on earlier reports | Inaction on earlier reports can lead to re-reports from reporters trying to prompt DCJ to take action, contributing to overall report volumes.[130] Delays in allocation and feedback to reporters can mean re-reports by mandatory reporters. Casework delays (e.g., due to caseworker changes) can prompt re-reports to get DCJ to respond (see the Taylor family’s case study at the start of Chapter 6). |
Inadequate risk assessments and information exchange | If an FPS considers that they have been referred a child or young person with higher-than-assessed risk and complex needs, this can result in the FPS re-reporting back to DCJ (on the basis that they are not equipped or funded to support a family with that level of risk or complexity). We also heard that providers may sometimes lodge reports to prompt better exchange of information and co-working. |
Case closure means further ‘update’ information channels through Helpline as new reports | Closures immediately following triage referral mean that, without an allocated caseworker to contact, providers re-report via Helpline to ‘update’ DCJ as circumstances change. FPS have commented that, if DCJ kept cases open after referrals, this would help manage escalated risk more efficiently than for providers to make a re-report to DCJ.[131] This was a theme we also heard from ACWA, and we understand DCJ is considering options for keeping files open for up to a further three months. (However, it is important to note here that Aboriginal FPS had a different view on this issue, suggesting that closure after referral should occur as soon as possible after referral to reduce the pressure on families of DCJ’s continued involvement).[132] |
Double handling of reports relating to open cases or reports about the same children | In 2022-23, 3 in 10 ROSH reports (31% or 25,676 of 81,582) that were triaged and allocated to be seen by DCJ caseworkers related to open (already allocated) cases. These reports involved 10,928 children, including 3,811 children who were already in OOHC.[133] Reports for open cases funnel through Helpline and triage to the allocated caseworker and receive fresh consideration at each stage, by multiple people. They also enliven s 30 and require a decision under s 30, although arguably this may not be required if the child has already been determined to be at ROSH and that determination remains current (rather, the information in the report could be considered as part of the overall information about the ongoing risk and safety of the child). DCJ recently found that 30% of Helpline assessments related to the same 5% of children reported.[134] This is consistent with other DCJ data that shows the number of reports per child is increasing. |
Quotas and ‘reports to get supports’ | The availability of funded supports that FPS can draw on may vary depending on whether a family comes to an FPS as a ‘community referral’ (10%) or a ‘DCJ referral’ (90%). DCJ’s referrals to Family Connect and Support (FCS) services should not exceed 30% of all referrals received by the FCS without prior approval from the FCS Program Manager. Some programs also have eligibility criteria requiring evidence of DCJ involvement and risk assessment. Despite DCJ’s relatively large referral quota, we observed 10 (3%, 10 of 380) NCTA closures where DCJ’s triage response was to ask the reporter themselves to make a community referral. This means that those referrals, if made, would draw down on the smaller community referral allocation rather than the DCJ referral quota. Eliza’s case at the start of this chapter is an example of a referral we think DCJ should have made itself. We also heard that because some service funding can only be accessed via a DCJ referral following a ROSH report, providers sometimes make Helpline reports to enable families to obtain access to services the service provider cannot arrange directly. In 2024, an FCS evaluation report noted that FCS providers are more limited in the types of referrals they can make than DCJ, and observed that: The NSW Family Preservation program only allocates 10% of its capacity to community referrals. To get support for families, FCS providers reported they feel under pressure to report families via the Child Protection Helpline to reopen their case, but this is perceived as damaging trusting relationships and resulting in potential overreach in terms of statutory response.[135] |
Absence of consent and warm referrals | In the NCTA closure sample, we found 11 referrals made by DCJ which mentioned consent and/or consultation with the family prior to making the referral. However, 8 out of these 11 referrals were made without recording whether or not the family consented. Where DCJ has talked to the family and obtained prior consent to the referral, the family is more likely to engage and the referral more likely to be successful (reducing re-reports in some cases). DCJ also has access to considerably more information to inform a referral than other providers – which is particularly important for families impacted by domestic and family violence (DFV) and sexual abuse. Warm DCJ referrals mean services can receive the information they need to work with the family safely and quickly. DCJ also has statutory powers to request prioritised access to government and NGO services for children at ROSH.[136] Without DCJ’s child protection history it can be very difficult to work with families and, unless they have family consent, voluntary services can face lengthy delays obtaining family information. Families may also miss out on, or experience delays accessing services, if they are considered too high risk in the absence of DCJ information.[137] |
Over-reporting by mandatory reporters | We heard repeatedly from DCJ staff that risk aversion in the policies and practices of other agencies who employ mandatory reporters, and gaps in other government and community services, resulted in reports about issues that were not child protection concerns. The Audit Office also recently made observations about re-report drivers, including potential over-reporting. DCJ is reviewing Child Wellbeing Units (CWUs) and improving guidance for mandatory reporters to address the culture of over-reporting. Reporter feedback is a potential tool for reducing over-reporting and improving the information in reports. We heard that this was a strength of the former HASP model, which used reporter feedback to build capacity in CWUs to improve the information being provided as part of the initial report and streamline information gathering. |
289. The Helpline does not generally have a role in closing reports that are ‘screened-in’ as ROSH reports. However, the Helpline can close such a report if it is assessed to be a ‘duplicate’.
290. Helpline staff told us that duplicate reports, once closed by the Helpline, are only forwarded to the CSC if the triage record (for the original report) is still open at the CSC or there is an open case.
291. According to Helpline guidance a report can only be considered a duplicate if:
292. Despite this, we saw several cases where reports were closed as duplicates in circumstances where:
293. The most concerning potential consequence of incorrectly deciding that a ROSH report is merely a duplicate is that evidence that may help inform a view about whether there is an ongoing or escalating risk may be missed. This is particularly the case where there is new information. It is also a concern where reports may look the same or similar but are made at a materially later point in time. Later reports may be evidence of an ongoing or escalating pattern of risk, including those involving aspects of neglect.
294. Structured Decision Making or SDM comprises a proprietary suite of structured decision-making tools created and licensed by a Californian company, now known as Evident Change. DCJ describes Evident Change as ‘an American-based non-profit social research organisation, which implements SDM systems in a variety of human service fields, both in USA and in Australia.’[139]
295. DCJ has used a range of SDM tools when making key decisions under the Care Act, including when responding to ROSH reports.
296. The adoption and use of SDM tools have not been the primary focus of this investigation, and we have not conducted an in-depth review of the SDM suite of tools or other DCJ custom tools, such as the Alternate Assessment tool, which is used when responding to ROSH reports about children in OOHC. However, in examining DCJ’s response to ROSH reports, it has been necessary to give some consideration to DCJ’s use of these tools as they have been applied to make decisions about those reports.
297. Prior to the adoption of SDM, DCJ staff used what is referred to as the ‘professional judgement model’ to make child protection-related decisions. There was, however, little written guidance or formal decision-making criteria, particularly to assist Helpline staff in making judgements about matters such as response times and urgency.[140]
298. DCJ had considered the viability of incorporating structured decision-making tools into assessment processes as early as 2005 but concluded that the benefits were not sufficiently significant to warrant investment at that time. A key issue was that DCJ did not have full information to measure the extent of error in its existing ‘professional judgment model’ against which to compare the structured decision-making approach. DCJ said it would instead improve its assessment system and monitor the implementation of SDM in other jurisdictions.
299. The 2008 Wood Special Inquiry considered the contemporary ‘risk assessment debate’ around the use of structured decision making and the professional judgment model and noted that discourse had begun to move away from an either/or approach. It noted that, while some structured decision making tools were considered to more accurately classify risk, this did not rule out the need to also use other approaches such as consensus-based and clinical judgement to determine what services would help to ameliorate risk and to engage families with services.
300. At the 2008 Wood Special Inquiry, DCJ submitted that it had concluded there would be benefit in introducing a structured analysis approach in NSW, based on experience in other jurisdictions. DCJ acknowledged it would need clearly defined and consistently applied decision-making criteria to support Helpline staff, and that caseworkers would be expected to complement any structured analysis outcomes with professional judgment. DCJ further observed that:
301. Although, as outlined in submissions to the 2008 Wood Special Inquiry, the NSW Ombudsman saw merit in the potential for structured risk assessment tools, we also noted that evidence was lacking on their current effectiveness, and that an evaluation by Deakin University suggested that the available tools did not support consistency in decision making. In that context, the NSW Ombudsman said:
‘…we would favour an initial trial of the [structured decision making] tool to ascertain whether it improves assessments, and addresses some of the fundamental weaknesses associated with the current assessment system. We also believe that there should be a considered analysis of the results of Deakin University’s evaluation of the tool.’[141]
302. The 2008 Wood Special Inquiry agreed that such testing of the tool was warranted at the Helpline and at CSCs in relation to assessments and interventions, including decisions about family restorations.[142]
303. With funding from the Keep Them Safe reforms, DCJ began introducing SDM tools to support screening and assessment of reports from 2010 after a trial in 2009.[143]
304. The SDM tools are a mix of actuarial, consensus and blended tools. The suite of tools comprises:
Name of tool | Description | In use | Last updated (as at Sept 2024)[144] |
SDM Mandatory Reporter Guide | A tool to help mandatory reporters assess whether they need to make a report | 2010 to present | 2016 |
SDM Screening and Response Priority Tools | Tools to help Helpline workers assess whether a report meets the ‘threshold’ of ROSH requiring assessment/ investigation and how quickly the assessment/investigation should happen | 2010 to present | 2017 |
SDM Safety Assessment Tool | A tool to help caseworkers assess the immediate safety of a child during a face-to-face visit | 2011 to present | 2012 & 2024 |
SDM Risk Assessment Tool | A tool to help caseworkers assess whether a child is now safe enough for a case to be closed | 2011 to 12 September 2024 (replaced during this investigation by a new ‘Family Based Assessment Tool’) | 2012 |
SDM Risk Re-Assessment Tool | A tool to help caseworkers re-assess whether a child is safe enough for a case to be closed | 2011 to 12 September 2024 (replaced during this investigation by a new ‘Measuring Change Assessment Tool’). | 2012 |
SDM Restoration Assessment Tool | A tool to help caseworkers assess whether a child should be restored to their family | 2017 to present | 2018 |
SDM Family Strength and Needs Assessment Tool | A tool to help caseworkers assess caregiver’s needs and strengths and services that would help the family reduce risk to inform a Family Action Plan | DCJ trialled the tool in its Short-Term Court Orders pilot (2014-2015) and Brighter Futures program (2011-2012) but stopped using the tool in 2013 | Not adopted after trial |
SDM Carer Assessment Tool | A tool to help caseworkers assess allegations against carers for children in OOHC. | N/A never adopted | Not adopted |
305. DCJ’s website contains the following diagram which summarises the questions that each SDM tool is intended to help answer:
Figure 15 – Questions each SDM tool is intended to help answer  Source: DCJ Website, ‘The NSW Structured Decision Making Framework’ (accessed 27 August 2025). Note: This diagram, which appears in on DCJ’s website, depicts only those tools in use by DCJ. It is noted that the diagram summarises the purpose of each tool in slightly different terms from the descriptions of them in their relevant manuals. |
306. While we have not undertaken a comprehensive audit or evaluation of the SDM tools in use, in Appendix A-3 we outline some of the limitations in design (and related flaws in use) that are known to DCJ or have become apparent to us in the course of this investigation.
307. We generally focus only on the following tools, as these are of most relevance to the triage response to ROSH reports:
308. It is important to note that the limitations in design and configuration are not themselves flaws (in the sense of errors) in the tools themselves. However, it is the way in which the tools are then used which may be considered flawed or wrong, if that use is inappropriate in the circumstances (for example, by applying the tool to a situation it was not designed for) or if it is used in a way that does not otherwise properly account for the tool’s limitation (e.g., by overlaying professional judgement to interpret and consider the tool’s output).
309. Some of the more obvious, but significant, design limitations with the SDM tools, which are well known to DCJ (and are discussed in Appendix A-3) include:
The SDM Risk Assessment tool, for example, is designed to predict the likelihood that a child will be re-reported within 12-18 months, not to decide current ROSH or INOCAP for a child. It scores families (low to very high) from large datasets (NSW used the default Californian model), using future system contact as a proxy for harm. It does not assess present harm – the tool’s name has also caused confusion on this issue.
The tool is not applying any NSW statutory term or definition, such as whether a child is ‘at ROSH’ or ‘INOCAP’. Nor does it consider the particular statutory duties of DCJ under the Care Act, when responding to that question.
This is one reason why any outputs of the tool can only be considered as a factor in decision-making, and never determinative of the statutory decision that DCJ is required to make.
Instead, the SDM Risk Assessment tool, for example, is built on Californian child protection data sets.
This means the tool’s items (the scored checklist factors e.g. prior reports or domestic violence), weights (the points each factor contributes based on how strongly it predicts re-reporting) and cut-points (the total score thresholds that assign low to very high risk bands) reflect California’s patterns and base rates of future re-reporting, not NSW’s.
Because NSW differs in law and practice, reporter obligation and mix, triage/allocation models, service pathways, demographics and Aboriginal community context, a California-calibrated model can be misaligned here – over- or under-classifying families, distorting priorities and producing uneven error rates across groups. Without local validation and recalibration to NSW data, the tool’s risk bands may not reliably map to NSW likelihoods and should be treated as one input, not a determinative measure.
The SDM suite of tools includes a tool to help decide whether or not a child protection report should be screened-in and a timeframe for response set (the SCRPT tool), as well as a tool to help caseworkers later decide whether and when the case can be closed after their investigation and any action (the SDM Risk Assessment tool). There is no tool designed to assist when deciding which screened-in reports should be prioritised to be allocated to a caseworker for investigation. This is because the SDM tools assume no triage, allocation or prioritisation process exists – that all screened-in reports will be investigated and responded to (as they should be).
Clearly that is not the reality in NSW, where most screened-in reports are closed without allocation to a caseworker for investigation and any further response under the Care Act. The application of SDM tool outputs (directly or indirectly) when deciding which such reports are to be closed is inherently flawed and can lead to perverse prioritisation decisions (as explained in more detail in section 6).
310. When SDM tools were being introduced, the Ombudsman was assured by DCJ that the SDM Risk Assessment tool was intended to lead to more consistent decision making but that it would not replace clinical judgement.[145]
311. A DCJ working group observed:
The direct attribution of the SDM Risk Assessment outcome to the statutory determinations that DCJ needs to make is not supported by the owners of the tool – Evident Change. Both the SDM Safety and Risk Assessment Tools should inform these decisions but not make them.[146]
312. The determination as to whether a child or young person is, or is not, at ROSH is a determination that requires evaluative judgement. It is a decision that must be made by a person with appropriate delegation and who has engaged in a ‘mental reasoning process’ to come to that conclusion. The decision-maker may be assisted by considering the outputs of a structured decision making tool (assuming those outputs are valid and relevant). However, those outputs cannot be simply taken by the decision-maker as the decision, without themselves having engaged in a mental reasoning process in a way as would ensure that the decision is genuinely their decision, and not just the output of the structured decision making tool.[147] Furthermore, in making that decision, the decision-maker must consider and apply the particular statutory test of ‘at ROSH’ under the Care Act, and not just a general sense of ‘at risk of significant harm’.
313. The SDM tool manual says “The risk level [identified by the tool] is not used to make the substantiation decision [i.e. the decision of whether the child or young person is or is not-at-ROSH].”[148] DCJ’s Assessing Safety and Risk Mandate, which provides guidance on how to make a substantiation decision, also refers to the need for critical analysis, with links to professional knowledge, experience and theory.
314. Despite this, DCJ’s Decisions Count report found in 2019 that ‘practitioners struggled with, and were at times unable to grasp, the ‘logical transition from actuarial scoring and clinical judgment’. Overreliance on tools and a lack of critical reflection was found to be an issue: ‘Fear of making a wrong decision combined with a lack of confidence and skill means that some practitioners see structured decision making tools as deterministic and rely on them to tell them what to do – rather than incorporating their professional wisdom alongside the voice of children, young people and families.’ [149] In our survey of DCJ senior staff (Manager, Casework level and above), 36% (47 out of 131 responses) of respondents went as far as to respond that an SDM tool ‘usually makes’ the decision that a child is at ROSH.[150]
Figure 16 – DCJ Staff Survey responses to question 25  |
315. At some point during customisation, a policy decision was made to use certain SDM Risk Assessment outputs not just for the purpose of assessing risk (relevant to the determination of whether a child or young person is at ROSH), but explicitly as a proxy for the decision as to whether a child is INOCAP.[151]
316. This is despite the fact that, under the Act, there is a distinction between making an at ROSH determination (under s 30) and forming an opinion that a child or young person is INOCAP (under s 34), and uncertainty about the relationship between those two decisions (see section 2 and Appendix A-2). Under the legislation as currently drafted, however, it does not follow as a matter of legal necessity that a child or young person determined to be at ROSH is inherently also INOCAP.
317. The formation of an opinion as to whether a child or young person is INOCAP is one which also requires evaluative human judgement – this is perhaps even more clearly so than in respect of the determination that a child or young person is at ROSH. There is no statutory definition of INOCAP, and the Act does not prescribe any express mandatory considerations to be taken into account in making the determination as to whether a child or young person is INOCAP. More importantly, it is the INOCAP decision that generally triggers both the power and obligation on DCJ to take action, including potentially removing a child or young person from their family and taking them into OOHC.
318. It appears the explicit coupling of the SDM Risk Assessment tool and INOCAP decision occurred in around August 2012,[152] when we understand DCJ inserted the following INOCAP matrix into the tool’s manual:
Figure 17 – INOCAP Matrix  Note: ‘RA’ refers to the SDM Risk Assessment. |
319. While it is likely that INOCAP decisions would have been influenced by the SDM Risk Assessment outputs (risk levels) even without explicit coupling, doing so means that direct automation has effectively been introduced into decision-making about whether a child or young person is INOCAP. A DCJ report found:
This INOCAP determination is currently made automatically, populated by the outcome of SDM Safety and Risk Assessments. It is most commonly understood as being determined when the SDM Risk Assessment category produces a High or Very High outcome. This outcome currently drives which families DCJ will continue to work with, as well as which funded services a family can be referred to [153] [emphasis added]
320. More succinctly, DCJ advised us that:
the Risk Assessment outcomes were understood by caseworkers to be a proxy for the INOCAP decision, [and] in this sense the INOCAP decision has been automated. [emphasis added][154]
321. Despite the DCJ guidelines prescribing an INOCAP matrix that directly and automatically links SDM outputs to particular INOCAP opinions, and despite DCJ’s admissions above that the opinion has therefore been effectively automated, we have considered whether it might still be the case that DCJ caseworkers have nevertheless been making INOCAP decisions themselves, informed – rather than dictated – by the SDM tools’ ratings.
322. The results of our DCJ staff survey are ambivalent. Most managers did not point to the SDM tools as usually making the INOCAP decision, but a quarter (25%; 32 of 123) did.
323. INOCAP decisions are, in practice, impenetrable. Given the important consequences of this decision, it is remarkable that DCJ can have operated for so long without a clear understanding of precisely how, when and by whom these decisions are made.
324. Since the introduction of ChildStory in 2017, there has been no field for caseworkers even to record that a decision was made that a child or young person is ‘in need of care and protection’, let alone to provide any reasons for that decision.[155] While DCJ identified the issue earlier, it was only during this investigation, in September 2024, that DCJ adjusted its practice so that a new INOCAP record now explicitly records whether a decision has been made that the child or young person is INOCAP or not, as well as a mandatory field for recording the decision rationale.
325. Until that point, the only records in ChildStory that reference any INOCAP decisions are in the ChildStory Closure Reasons, where the reasons a caseworker can select for closing a case included that a child is not (or is no longer) in need of care and protection, or that a child is in need of care and protection but will be referred to an external agency.
326. This means that even the fact that an INOCAP opinion has been formed can generally only be assumed from the subsequent conduct of DCJ in either taking action that is only authorised when such an opinion has been formed or closing a case and ticking a label that refers to the child or young person having been INOCAP.
327. DCJ, in its advice, to us referred to these ChildStory closure records and suggested the following:
However, [even though the Risk Assessment outcomes were understood by caseworkers to be a proxy for the INOCAP decision] caseworkers can decide whether to take any actions and whether to keep a case open or not, separate to the Safety Assessment and/or Risk Assessment outcomes with the appropriate delegation approval (Manager or Director). In this way, the INOCAP decision is tied to case closure. That is, if the case is closed without action, then the decision has been that the child or young person is not INOCAP. A case could also be closed noting that DCJ believes the child is INOCAP but another action has been taken (e.g. referral to family preservation) as indicated in this closure reason. A case can only be closed when the child is at high/very high risk (aligned to INOCAP) if approved by a higher delegation (Director). If the case is kept open, then decision has been that the child or young person is INOCAP. [156]
328. In other words, after explaining that the SDM output has effectively automated the statutory INOCAP decision, DCJ suggested that because caseworkers still retain freedom of action to close a case, their actions (i.e. either closing the case or keeping it open) are to be taken as implying that they themselves are making a genuine decision as to whether the child is or is not INOCAP, and not merely relying on the SDM output.
329. We do not accept any such implication:
330. DCJ’s claim that closure/action decisions by caseworkers imply they are also making independent INOCAP decisions also appears inconsistent with ChildStory functionality and the decision guidance.
331. ChildStory itself ties INOCAP decisions to SDM outputs. ChildStory is configured in such a way that staff can select the case closure reason ‘Assessments indicate the child/young person is not in need of care and protection’ only when the SDM tool already shows Safety = ‘Safe’ and Risk = ‘Low’ or ‘Moderate’. If those SDM results are not recorded, the INOCAP option is disabled. In practice, this means caseworkers cannot record an INOCAP decision based solely on professional judgment; the system requires – and effectively compels – reliance on SDM outputs to close case on that basis.[157]
332. Also, DCJ’s guidance says that when ‘Deciding whether to close or continue working with the family’, staff are to ‘[b]e guided by the risk level and the SDM open or close decision matrix ...’ – that is, without mentioning the need or possibility of them forming a threshold opinion themselves about whether the child or young person is even INOCAP.[158]
333. In summary, DCJ has, since at least 2012, failed to:
with the result that the responses to an unknown number of ROSH reports are likely to have proceeded on the basis of ‘automatic’ acceptance of SDM outputs without decision-makers undertaking a requisite mental reasoning process.
334. The SDM tools were trialled in 2009.[159] There was reported to be positive feedback from caseworkers, and 75% of SDM Safety Assessment and SDM Risk Assessment trial caseworkers were reported as saying the tools ‘made their job easier’.[160]
335. From 2010, the tools were adopted at scale (see table in section 9.2). In the years following the introduction of the SDM tools, there were many internal and external calls for review, whether of particular SDM tools or generally.
336. In this section, we outline some of the ways in which DCJ was put on notice of the need to comprehensively evaluate and review the assessment tools, and the unacceptably late and limited steps it has taken to respond to this need.
337. In our assessment DCJ failed, for more than a decade from 2011, to properly review and evaluate the effectiveness of SDM tools for use in NSW, despite being repeatedly put on notice, including by the owner of the tools, of the need to do so. This includes the SDM Risk Assessment tool: in August 2023 DCJ resolved and began taking implementation steps to retire the tool, with its use finally ending in September 2024. DCJ also failed, over many years, to publicly disclose or release the limited evaluative studies it did commission. It also failed to act on their recommendations, many of which called for thorough review and evaluation of the tools.
338. Evident Change reports having conducted the following ‘evaluations’ of SDM in NSW. None of these have been published. It is not known whether any of these included analysis of Deakin University’s evaluation of the tool:
Table 4: List of evaluations of NSW SDM tools by Evident Change (2009-2020)
State | Year of SDM implementation | Types of evaluation conducted by Evident Change |
NSW | 2009 |
|
Source: Evident Change, Statement to the Australian Royal Commission into Violence, Abuse, Neglect and Exploitation of People with Disability (Exhibit 8-071), Royal Commission into Violence, Abuse, Neglect and Exploitation of People with Disability – Public Hearing 8 (November 2020), pp.2 and 3.
Note: The source does not identify which SDM tool was subject to each evaluation
339. On their adoption 15 years ago, Evident Change recommended the SDM tools be reviewed and updated at intervals no longer than every 5 years.[161] An earlier 2009 presentation by Evident Change to DCJ also noted a ‘preliminary risk calibration study’ as part of the implementation process.[162]
340. In 2013, Evident Change was commissioned by DCJ to conduct a ‘risk validation study’ on the SDM Risk Assessment tool. The existence of that study was not made public until January 2024.
341. The study itself has still not been published. DCJ says that it showed overall that the tool ‘validly classified families’ but that, to further improve equity of the assessment, it recommended that DCJ reduces the risk categories from 4 to 3 and conduct periodic evaluations to ensure the risk assessment was calibrated to the cultural and demographic mix of families in NSW.[163]
342. DCJ implemented neither recommendation.[164]
343. In 2018, the SDM SCRPT tool, which had had only had 3 minor amendments since its introduction in 2010, was the subject of a ‘case reading study’ by Evident Change (commissioned by DCJ) to review its efficacy.[165]
344. Internal DCJ documents indicate that Evident Change’s recommendations included: ‘A way to address the high number of screened-in-but-unallocated reports and the lack of practical attention given to the response priority system in the allocation of assessment responses’. However, they also indicate that, while individual recommendations were responded to by Helpline leaders:
[t]he review of SCRPT remained an outstanding item which was outside the Helpline Operations team.[166]
345. In August 2021, DCJ commissioned Evident Change to review all tools. Evident Change began to do a risk validation study in May 2022 to consider the SDM Risk Assessment tool and its accuracy, fairness and future purpose. In the early stages of that process (around September 2022), Evident Change recommended that the SDM Risk Assessment tool be decoupled from any ‘in need of care and protection’ decision (an issue that is discussed in section 9.4). That did not happen until September 2024.
346. As at November 2023, DCJ has incurred $1.5 million in costs to Evident Change for the review of all tools. The validation study was not completed.
347. It is doubtful that review of the tools by the tool’s owner and creators (Evident Change) is sufficiently arms-length to satisfy recommendations for an independent review.[167]
348. In 2015, University of Melbourne researchers, commissioned by DCJ, conducted a ‘stage one’ review of SDM tools. This work was only made publicly known when produced by DCJ to the Disability Royal Commission in 2020.
349. The researchers recommended that the department evaluate the overall performance of the SDM Risk Assessment tools and ‘rigorously validate the safety and risk assessment instruments using an independent evaluator.’[168]
350. DCJ’s evidence to the Disability Royal Commission (in November 2020) was that it had: ‘undertaken reviews in consultation with [Evident Change] and the University of Melbourne to review and revise the [MRG] and the Screening and Response Priority and Procedures Manual’, with the SDM MRG reviewed ‘in 2015/16’ and the Manual last updated in 2017.[169] It gave evidence that it was now (in 2020) reviewing the Restoration Assessment tool.
351. Notably absent in the response was reference to any review having been undertaken, being undertaken, or planned of the SDM tools for:
352. There has been no ‘rigorous validation’ of the SDM Safety Assessment and SDM Risk Assessment tool by an independent evaluator, as recommended by the University of Melbourne report in 2015.
353. In 2017, the Australian Institute of Family Studies identified potential structured decision-making weaknesses, including actuarial tools not considering unusual or context-specific factors, and being insufficiently flexible to incorporate professional judgement. Further, there was concern about potential for trust in structured decision making to impact whether a user rejects or accepts an output – in particular, bias may occur where a user assumes the structured decision making is always accurate.[170]
354. Also in 2017, the NSW Legislative Council General Purpose Standing Committee No 2 Inquiry into Child Protection questioned the effectiveness of DCJ’s SDM tools and recommended an independent review.[171] It raised concerns that the SDM Safety Assessment and SDM Risk Assessment tool are not suited to a wide variety of factors and circumstances, including to assess risk in Aboriginal families and communities, or to assess child protection concerns in the context of domestic or family violence. It also appeared that existing tools may be inadequate in assessing cumulative harm.
355. Also in late 2019, the Family is Culture review found considerable deficiencies in the way SDM was being utilised in safety and risk assessments for Aboriginal children and families, and raised concerns about a lack of Aboriginal consultation during the use of SDM Safety Assessment and SDM Risk Assessment tools and the cultural bias of caseworkers.[172] It recommended that DCJ commissions an independent review of its SDM tools and processes to identify how they could be improved to enhance objectivity within child protection assessments and that the review should be undertaken in partnership with Aboriginal community and stakeholders (Recommendation 56).
356. The Government response was that the Office of the Children’s Guardian (OCG) and Aboriginal Deputy Children’s Guardian would prepare a special report for the Minister by 30 June 2021. The resulting OCG report of March 2022 formulated actions for DCJ to take to ensure Aboriginal input to enhance assessment tools and better incorporate the Aboriginal and Torres Strait Islander Child Placement Principle and the Aboriginal Case Management Policy.[173]
357. In its 2020 hearings, the Disability Royal Commission identified that structured decision making relies, in part, on static factors to identify risk levels in a way that presents a risk of contributing to the overrepresentation of First Nations people with disability in child protection systems, particular First Nations people with mental illness. This was accepted by the NSW Government.[174]
358. The Commission considered further research was required to investigate the nature and extent of possible bias in SDM tools for First Nations people and families and reviews or inquiries should be commissioned to consider how best to validate SDM for First Nations families to minimise the risk of future overrepresentation.[175]
359. This prompted DCJ in 2021 to commence its Child Protection Assessment Review Project (later known as the ‘Better Decisions for Children’ project) with the aim of redesigning its suite of SDM tools.
360. In August 2022, the Queensland child protection department ceased using SDM tools after a study by Dr Brian Jenkins and Professor Clare Tilbury in 2021 substantiated racial bias within the tool.[176] DCJ continued to use the tools.
361. In 2016 and 2017, DCJ’s Office of the Senior Practitioner undertook an internal study known as ‘Decisions Count’. The Decisions Count report (when finalised in 2019) was highly critical of DCJ’s assessment decision-making processes generally. It observed that ‘practitioners struggled with, and were at times unable to grasp, the ‘logical transition from actuarial scoring and clinical judgement’.’ Other factors that were found to undermine the use, effectiveness and reliability of SDM assessment tools included inconsistent use by practitioners, overreliance on tools (creating overconfidence and a lack of critical reflection), and skewing of results where practitioner misgivings cause them to defer to their intuition or professional judgement. It also identified issues with ‘different weightings or interpretation of some criteria within the same tool’ and mistrust or disbelief among some practitioners that the tool captured the complexities of the lives of many of their clients. It observed:
A focus on nuclear family members within assessments excludes the perspectives and role extended family or community members can play in keeping a child safe. This is especially important in Aboriginal communities where child rearing and ensuring connection to country and culture is a community responsibility.
362. The report also observed that ‘divergence and discrimination when assessing safety can have enormous and lifelong implications for children and families’.[177] (See also Appendix A-3 (Risk Assessment tool counting issues) in relation to the application of the SDM Risk Assessment tool to Aboriginal families.)
363. A recent internal DCJ briefing note (which is consistent with comments made in a 2020 NSW Auditor-General report)[178] asserts that in 2018 and 2019 DCJ scoped a review of the SDM Safety Assessment, SDM Risk Assessment and SDM Risk Reassessment tools, and developed a business case to consider changes to the SDM SCRPT and the SDM MRG. However, this work was put on hold without a plan following a decision by DCJ investment committees not to submit the case as part of the 2020-21 budget in late 2019. The briefing note states:
DCJ has been in support of a review of Safety and Risk Assessments for some years, however, was unable to pursue this review due to organisational and resource constraints.[179]
364. In January 2021, when DCJ commenced its Child Protection Assessment Review/Better Decisions for Children project, the scope of the project initially only included the SDM Safety Assessment and SDM Risk Assessment and DCJ Alternate Assessment. However, from March 2021 project management and governance of the full suite of SDM assessments transferred to the OSP, after which the review was expanded to include the SDM MRG, SDM SCRPT and SDM Risk Reassessment tool (and the SDM Family Strengths and Needs Assessment tool, which has not yet been fully implemented in NSW).
365. In June 2021, DCJ advised the NSW Ombudsman that, as part of the project, a ‘Quality Services Review’ of its SDM tools would commence with Evident Change later that month. DCJ stated that the focus would be:
… on co-designing updates with Aboriginal people, practitioners and researchers to improve racial equity, validity and accuracy to NSW population data, practice and legislative and policy settings. Implementation of the updated tools will focus on workforce and leadership development and bolstering systems to safeguard practice and decisions. These factors are pivotal requirements to ensure any assessment tool is used effectively, accurately and consistently.[180]
366. In November 2021, the NSW Ombudsman expressed continued concern about DCJ’s use of SDM in a report we published about government use of machine technology, observing that it was not clear what (if any) validation of SDM had occurred since 2011.[181]
367. In December 2021, the NSW Ombudsman asked DCJ to provide evidence that use of SDM tools over time produces accurate and reliable results, including quality assurance, verification and validation testing, and monitoring against performance metrics. In response, DCJ advised:
The SDM risk assessment tool currently used by DCJ is the Californian risk assessment model and based on Californian actuarial data. The risk assessment model is currently subject to a validation study and as a result will be calibrated to NSW actuarial data. The project has commenced and will ready (sic) for implementation into ChildStory in the final quarter of 2022.[182] [emphasis added]
368. It also went on to advise in its response that:
Efforts throughout 2015-2020 have focused on reviewing the application of the tools, rather than testing the efficacy of the tools.
369. Despite its advice in December 2021 that the SDM Risk Assessment model was ‘currently subject to a validation study’ and as a result would be calibrated to NSW actuarial data, we now understand that no such study was being undertaken at that time.
370. In July 2022, DCJ advised the NSW Ombudsman that, as part of its Quality Service Review it commenced a validation study in May 2022 for the SDM Risk Assessment. It further advised that:
371. During 2023, a contract DCJ had entered with AbSec for consultation on the 2021 Quality Services Review and redesign was not fulfilled, in part due to AbSec’s view that the SDM Risk Assessment tool was racially biased and that it would not be satisfied with any tool that was not designed by and for Aboriginal people. DCJ made a commitment to review its use of the SDM Risk Assessment tool.
372. In August 2023, DCJ executives resolved to stop using the SDM Risk Assessment and SDM Risk Reassessment tools and to co-design replacement tools with Aboriginal sector partners.[184]
373. Implementation, however, took 12 months because the SDM Risk Assessment outputs had been hard-wired into other decisions and processes – especially the decision about whether a child is ‘in need of care and protection’ (INOCAP) and the referral rules for family preservation services. Removing the tool required an interim statutory decision framework, ministerial approvals, policy and guidance changes, ChildStory re-configuration and staff training.
374. Ministerial approval to develop the interim approach was given in February 2024 and the interim model was approved in April 2024. The SDM Risk Assessment and SDM Risk Reassessment tools were formally withdrawn in September 2024, with caseworkers directed to use a professional judgment-based, DCJ custom ‘Family Based Assessment’ tool. This new tool ‘has a basis’ in the former SDM Risk Assessment tool.[185]
375. In parallel, DCJ advised Parliament in January 2024 that all SDM tools were under review and in May 2025, entered a new formal partnership with AbSec and the Aboriginal Legal Service (NSW/ACT) (ALS) to further system reform. As at November 2025, DCJ continues to use four SDM tools – SDM MRG, SDM SCPRT, SDM Safety Assessment and SDM Restoration Assessment.
376. A determination that a child is at ROSH, at the very least, opens the door to a determination that a child is INOCAP, which in turn enlivens DCJ’s considerable statutory child protection powers over the lives of children and families. Use (or non-use) of these powers can have far-reaching effects on day-to-day welfare and long-term outcomes for children and their families.
377. As noted above, DCJ was advised at least as far back as 2013 of the need to recalibrate the SDM Risk Assessment tool with NSW data and was then repeatedly put on notice (including by the owner of the tools) of the need for a comprehensive evaluation and review.
378. Its failure to do so has, according to at least one of its own internal reviews, ‘undoubtedly disadvantaged minority groups’, especially Aboriginal children of NSW:
The SDM Risk Assessment tool was introduced using a Californian child protection data set. This was because DCJ did not hold comparable data at the time. The intent was always for DCJ to use the SDM Risk Assessment tool for a period of two years and then re-engage with developers to input local risk data into the tool’s algorithms. For reasons not known, this did not eventuate and the SDM Risk Assessment tool used in NSW continues to rely on population data from the United States. This has undoubtedly disadvantaged minority groups, most notably NSWs Indigenous population.[186] [emphasis added]
379. The DCJ Child Protection Decision Making Operating Working Group accepted that the two main problems it identified (failure to calibrate for NSW data, demographics and context, and coupling of the INOCAP decision) had also produced inherent racial bias within the tool:
Both of these problems have significant links to the racial bias that exists within the tool:
380. These observations should not be taken to suggest that a structured decision-making approach, if properly designed and used to inform and guide critical analysis and professional judgement, should not have been pursued at all, or that it should not be pursued in future. There appears no reason to believe that the reasons that prompted the 2008 Wood Special Inquiry and this office to support the testing of structured decision-making approaches are not still valid reasons today.
381. In particular, while it is acknowledged, even by DCJ, that SDM tools have had biased impacts, including with respect to Aboriginal families, a system that is reliant solely on casework professional judgment is clearly not itself without risk of racial bias. However, the point is not whether SDM is productive of more or less racially biased outcomes than any alternative. It is that DCJ had a duty, which it failed at for more than a decade, to properly design, test, implement and continuously review, a system for responding to the State’s most vulnerable children and families, including to ensure that it was fit for purpose in the context of NSW and its families and avoided any avoidable bias.
382. This section sets out our findings and recommendations in light of the conclusions set out above.
383. Having regard to the analysis set out in the preceding chapters, I find that the following conduct of DCJ is unreasonable and as such constitutes maladministration of a kind referred to in section 26 of the Ombudsman Act – DCJ:
1. engaged, over a protracted period of time, in a practice of closing a large (but indeterminate) number of ROSH reports in circumstances where serious concerns persist about whether a child or young person is at ROSH and where investigations and assessment are necessary – which conduct also means it is failing to meet its statutory responsibilities under the Care Act.
2. adopted and applied a policy directive that DCJ staff are to close ROSH reports where a case worker had not been allocated within 28 days of receipt – which policy is also inconsistent with its statutory responsibilities under the Care Act.
3. tolerated a culture where the closure of ROSH reports on the stated basis of ‘no capacity’ or ‘competing priorities’ has become normalised as ‘business as usual’.
4. failed to provide clear and comprehensive policies, guidance and support to its staff to empower them to properly make decisions under section 30 of the Care Act.
5. failed to provide for the clear and comprehensive making and keeping of records in relation to decisions, and reasons for decisions, under section 30 (ROSH decisions) as well as under section 34 (INOCAP decisions) of the Care Act.
6. failed to ensure that, where a face-to-face response was necessary but not possible due to resource limitations, decisions were consistently made based on highest risk and in accordance with mandated policy priority cases.
7. failed to ensure that its desktop information gathering and referral processes were adequately calibrated to ensure all at-risk children are identified and appropriately supported.
8. failed, from 2011 and for more than a decade:
9. failed to:
10. failed to adequately respond to community concerns about its SDM tools, including that they were not calibrated to the NSW population, and accordingly failed to take adequate steps to protect against the risk of potentially biased impacts.
384. We sent DCJ a draft investigation report on 11 December 2025, which also contained provisional recommendations for comment.
385. On 19 December 2025 the Secretary of DCJ wrote with submissions on the proposed recommendations. That submission included a request for additional time to ‘work with your office over the coming weeks to ensure the recommendations can be implemented safely and effectively to achieve the intended system improvements’.
386. In light of those comments and the request for extension, we decided to issue the Investigations (Findings) Report omitting recommendations, and to separately prepare and issue an Investigations (Recommendation) Report following further consultation with DCJ.
387. To that end, a draft report was provided to DCJ on 12 January 2026.
388. On 14 January 2026 we met with DCJ executives to consult further on our provisional recommendations. Following the meeting DCJ also provided an update on relevant reform work in train. The update has been incorporated where necessary in below.
389. DCJ has told us that it is undertaking a major reform program of its child protection system led by a ‘System Reform Division’ that will, among other things, affect the system for receiving and responding to ROSH reports.
390. DCJ has informed us about the status of the program, including the following elements which appear to be most directly relevant to this investigation.
391. On 8 May 2025, a partnership agreement was executed between DCJ, The Aboriginal Child, Family & Community State Care Secretariat (AbSec) and Aboriginal Legal Services (ALS) to review and redesign the frameworks that govern child protection assessment within the context of current child protection and out-of-home care reform. The Partnership aims to deliver a safer, fairer and more culturally responsive child protection system, and to reduce the number of Aboriginal children in out-of-home care.
392. As part of the reform agenda the following policy projects are underway:
393. DCJ is now using three Structured Decision Making (SDM) tools to support key decisions in assessing the safety of children who have been reported as at risk of significant harm. These are the SDM Mandatory Reporter Guide, SDM Screening and Response Prioritisation Tool (SCPRT) and the SDM Safety Assessment. DCJ implemented interim changes and retired the SDM Risk Assessment, Risk Re-assessment and Restoration assessment tools.
394. The interim assessment approach was implemented on 12 September 2024. DCJ advised that it addresses the inherent cultural biases within the SDM Risk Assessment by removing the scoring on family history and dynamics and promotes consideration of family strengths support networks.
395. Although the above projects represent significant reform momentum, there remains considerable uncertainty as to the precise nature and scope of the final changes, and of course the outcomes are not yet known. Critical questions therefore remain about how DCJ’s responsibilities to respond to ROSH reports, its policies and practices, may change.
396. Accordingly, while our recommendations below have been made with awareness of a significant reform effort already underway, we made recommendations that seek to ensure that identified gaps and deficiencies in DCJ’s current policies and practices related specifically to receiving, assessing and responding to ROSH reports will be addressed. Recognising that comprehensive reforms will take time to develop and implement, we also made recommendations for actions that DCJ can and should take as a priority. These are actions that we consider are so clearly necessary that they should not wait, and that can be readily and practically implemented in a way that does not significantly divert effort away from the broader reforms that are needed. To the extent that any of our recommendations have already been identified and committed to under DCJ’s existing reform agenda, then this can be reflected in DCJ’s response to the recommendation.
397. We recommend the following:
1. DCJ should undertake a comprehensive review and reform of its ROSH and related policies, practices and tools.
This should include to:
2. To support the above, the Government should consider whether the relevant provisions of the Children and Young Persons (Care and Protection) Act 1998 (Care Act) should be reviewed and improved.
This should include the statutory definition of the concept of ‘at risk of significant harm’ and the absence of any definition of, or express link with, the concept of ‘in need of care and protection’.
Consideration should be given to who is best placed to undertake this work, including whether the NSW Law Reform Commission or similar law reform experts should be commissioned to conduct such a review.
3. DCJ should then roll out comprehensive training to all relevant DCJ staff in the revised policies, and on the principles and practices of good decision-making, that is aligned with the Care Act.
Administrative law expertise should be obtained when designing training.
4. In the meantime, DCJ should:
398. In accordance with section 26(5) of the Ombudsman Act, the Ombudsman has requested that DCJ:
Paul Miller
NSW Ombudsman
Key terms used for the purposes of this report
Term | Definition |
Actuarial tool | A structured instrument built from statistical models (large historical datasets) that scores fixed items to estimate the probability of a specified outcome (e.g., that a child about whom a ROSH report is made will be the subject of another ROSH report in the future). |
Blended tool | Combines an actuarial core (standardised items, risk score) with structured professional judgement (defined overrides/context factors) so workers can adapt decisions transparently when case specifics warrant it. |
Care Act | |
Caseworker | DCJ employs caseworkers to perform both triage and casework functions with DCJ. To distinguish more clearly between staff involved in triage versus casework on open cases, we refer to DCJ staff as follows:
|
Child/young person | Under the Care Act (s 3) a ‘child’ is under 16 years and a ‘young person’ is 16 or 17 years old. This is different in other legislation, which refers to a child as anyone under 18 years old (see e.g., Community Services (Complaints, Reviews and Monitoring) Act 1993). For simplicity in this report, references to a ‘child’ or ‘children’ include young person/s, unless otherwise stated. |
Child Protection Helpline | DCJ’s central contact point for receiving, screening and prioritising reports about children who may be at risk of significant harm. |
Child protection report | This is a term currently used by DCJ (which is not found in any legislation) to describe any report it receives that relates to a child’s safety, welfare and wellbeing. |
ChildStory | DCJ’s case management system for child protection and out-of-home care since 2017, giving authorised users a shared, real-time view of case information. |
Community Services Centre (CSC) | DCJ locally based community services offices. There are approximately 80 CSCs across NSW. |
Concerns report | This is a term currently used by DCJ (which is not found in any legislation) to describe any child protection report it receives that raises concerns about a child’s safety, welfare and wellbeing. Concerns reports may include:
|
Consensus tool | A checklist or guidelines based on expert/practice consensus that prompts information gathering but leaves weighting and the final judgment to the practitioner. |
Current Competing Priorities (CCP) | A ChildStory closure reason, for use after a triage decision has been made to allocate to a caseworker and open a case, if there are insufficient resources for a further response. A face-to-face assessment may or may not have occurred before closure. |
Department of Communities and Justice NSW (DCJ) | The lead agency in the NSW Government Communities and Justice portfolio, which aims to create safe, just, inclusive and resilient communities through its services. DCJ is the statutory child protection agency in NSW under the Care Act. The Minister for Families and Communities and the DCJ Secretary are entrusted with special roles under the Care Act (see ss 15 and 16). Any reference to DCJ includes a reference to its predecessor agencies. |
Director Community Services (DCS) | Senior DCJ leader for a district/region, accountable for Community Services operations and practice oversight. |
District | A geographic region used by DCJ for planning, service delivery and local decision-making. Districts are led by 7 Executive District Directors and boundaries are aligned to the 15 Local Health Districts. |
Early intervention/Targeted Early Intervention (TEI) Services | Early intervention (child protection) Services in the child protection context for families showing early signs of risk, aiming to prevent abuse, neglect or escalation of issues. Typically for families with non-ROSH reports. Targeted Early Intervention (TEI) Services DCJ-funded NGO programs supporting vulnerable families through community strengthening and wellbeing/safety services. |
Evident Change | US-based non-profit organisation that designs the Structured Decision Making® (SDM®) model used in child protection and other social services. Any reference to Evident Change includes a reference to its predecessor, the Children’s Research Centre. |
Family Connect and Support (FCS) Services | DCJ-funded NGO voluntary services for families, aiming to address issues early. Under current arrangements, 30% of FCS capacity can be used by DCJ to refer families who are the subject of a ROSH report with a less than 10-day response timeframe. |
Family preservation/ Family Preservation Services (FPS) | Family preservation Programs designed to help at-risk families keep children safely at home and avoid entering out-of-home care. Family Preservation Services (FPS) DCJ-funded NGO voluntary programs for family preservation. These services typically support families with ROSH reports, who have undergone DCJ safety and risk assessments and/or who have a child at imminent risk of entering OOHC. As at 2025, the Family Preservation programs are:
As at 2025, DCJ funded 4,500 places per year for up to 12,500 children, and was redesigning and recommissioning FPS services. |
INOCAP determination decision | The decision under s 34 of the Care Act as to whether a child or young person is in need of care and protection. |
Insufficient reason discretion decision | The decision made under s 30(b) of the Care Act where the Secretary (or delegate) ‘may’ (that is, has a discretion) take no further action if the Secretary considers that, on the information provided, there is insufficient reason to believe the child or young person is at ROSH |
Joint Child Protection Response Program (JCPRP) | Joint DCJ-NSW Police-Health program responding to serious child abuse or neglect cases. DCJ’s Joint Referral Unit (JRU) receives reports from the Helpline to determine program eligibility and response pathways. |
Manager Casework (MCW) | Frontline team leader. Supervises caseworkers, allocates work within the team, quality-assures assessments, guides court and safety/risk decisions under delegation and supports professional development. |
Manager Client Services (MCS) | Operational manager overseeing 1 or more CSC teams. Leads triage and weekly allocation processes, manages performance and risk escalations and provides practice supervision to MCWs and senior caseworkers. |
Mandatory (section 27) report | A report made by a mandatory reporter who has reasonable grounds to suspect that a child is at ROSH (Care Act, s 27). |
Mandatory reporter | A person prescribed in s 27(1) of the Care Act, who has a duty to report if they have reasonable grounds to suspect that a child is at ROSH. This includes any person who, in the course of their professional work or other paid employment, delivers health care, welfare, education, children’s services, residential services, or law enforcement, wholly or partly, to children. |
No Capacity to Allocate (NCTA) | A ChildStory closure reason for use during the triage stage when a ROSH report is closed without allocation to a caseworker on the basis of insufficient resources to do so. |
Non-ROSH report | This is a term currently used by DCJ to describe concerns reports that have been screened-out by the Helpline as not being a ROSH report. |
Office of the Senior Practitioner (OSP) | DCJ practice leadership unit that sets and supports practice standards, provides expert advice on complex matters and conducts practice reviews, audits and capability building across districts. |
Out-of-home care (OOHC) | The Care Act provides for 2 types of out-of-home care: (1) statutory out-of-home care (statutory OOHC), which requires a Children’s Court care order; and (2) supported out-of-home care (supported OOHC) which provides either temporary or longer-term support for a range of other care arrangements made, provided or supported by DCJ without the need for a care order. A prerequisite common to both types is generally that a child must first be considered to be ‘in need of care and protection’ (INOCAP). |
Paramountcy principle | Under the Care Act, a child’s safety, welfare and wellbeing are the most important consideration in decision-making (s 9(1)). |
Pre-natal Report | A report about an unborn child suspected to be at risk of significant harm after birth (Care Act, s 25). |
Pre-natal ROSH report | This is a term currently used by DCJ to describe a report about a child not yet born that the Helpline screens-in as a ROSH report. |
Policies | DCJ currently employs a complex, structured policy hierarchy that includes:
Ultimately, these are all policies (whether high-level policy, or operational policy guidance), and we refer to them generally as such. Where necessary, we cite the specific document. |
Report | Under the Care Act, ‘report’ is defined to mean only information given to DCJ by way of a mandatory (section 27) report, a voluntary (section 24) report or a pre-natal (s 25) report: s 3. However, DCJ uses the term more generally to refer to any report made that relates to a child’s safety, welfare and wellbeing (see definition of ‘child protection report’ above). |
Request for Assistance (RFA) | A request to DCJ for advice, material assistance or referral under ss 20 or 21 made by a child, their parent or funded NGO on a child’s behalf. |
ROSH (risk of significant harm) | The definition of ‘at ROSH’ is in s 23 of the Care Act. A child or young person is at ROSH if ‘current concerns exist for the safety, welfare or wellbeing of the child or young person because of the presence, to a significant extent, of any one or more of’ the circumstances set out in that section. Circumstances include where the child or young person has been, or is at risk of being, physically or sexually abused or ill-treated. |
ROSH assessment | Such investigations and assessment as the Secretary considers necessary to determine whether the child or young person is at ROSH. |
ROSH determination | A determination under s 30(a) of the Care Act as to whether a child who was the subject of a report is at risk of significant harm. |
ROSH report | This is a term currently used by DCJ to describe concerns reports that have been screened-in by the Helpline as a ROSH report. A concerns reports is screened-in by the Helpline as a ROSH report if the Helpline considers that the report suggests the child may be at ROSH. |
SDM Risk Assessment | An SDM tool used by DCJ’s child protection teams to predict likelihood of future child protection reports in 12 to 18 months and inform case closure or continuation. Outcomes are: low, moderate, high or very high risk. It does not predict whether a child is currently at ROSH under the statutory definition in s 23 of the Care Act. |
SDM Safety Assessment and Risk Assessment (SARA) | Term for using both the SDM Safety Assessment and SDM Risk Assessment tools together. |
SDM Safety Assessment and Risk Assessment (SARA) | Term for using both the SDM Safety Assessment and SDM Risk Assessment tools together. |
SDM Screening and Response Priority Tool (SDM SCRPT) | An SDM tool used by the Helpline to decide if a report is to be screened-in as a ROSH report, and how quickly it needs a response. |
Screened-in | This is a term currently used by DCJ to describe concerns reports that have been screened-in by the Helpline as a ROSH report. A concerns reports is screened-in by the Helpline as a ROSH report if the Helpline considers that the report suggests the child may be at ROSH. |
Screened-out | This is a term currently used by DCJ to describe concerns reports that have been screened-out by the Helpline as not being a ROSH report. A concerns reports is screened-out by the Helpline as not being a ROSH report if the Helpline considers that the report does not suggest the child may be at ROSH or does not meet other jurisdictional requirements (such as NSW residency or age). |
Statutory response duty | Section 30 of the Care Act requires that, on receipt of a voluntary (section 24) report or a mandatory (section 27) report:
(Note that s 30 does not apply to pre-natal reports.) |
Structured decision making (general concept) | Applying clear criteria and evidence in a prescribed step-by-step process with a view to making consistent decisions. |
Suspected ROSH | This is a term currently used by DCJ to describe a situation where a concerns report has been screened-in as a ROSH report, but DCJ has not completed such investigations and assessment as it considers necessary to determine whether or not the child or young person is at ROSH. |
Triage | Process of prioritising and deciding actions for ROSH reports against resources at a local CSC. |
Triage officers | DCJ employs caseworkers to perform both triage and casework functions with DCJ. To distinguish more clearly between staff involved in triage versus casework on open cases, we refer to DCJ staff according to the function that is being performed as follows:
|
Voluntary (section 24) report | A report made by any person who has reasonable grounds to suspect that a child or young person is at ROSH (Care Act, s 24), and that is not a mandatory report. |
Terminology changes over time – ‘Competing priorities’ and ‘No capacity to allocate’
Over time, DCJ has changed the labels it uses to record report closures due to resource constraints, and applies different terms depending on where in the process the decision is made:
These changes can cause confusion when understanding terminology and analysing long-term trends. Illustrating this issue, DCJ recently described No Capacity to Allocate data as ‘closed due to competing priorities’ data when answering questions on notice during budget estimates.[190]
In this report, we use DCJ’s capitalised labels (eg No Capacity to Allocate, Current Competing Priorities) when referring to post-2018 data. This reflects the label recorded by DCJ, without implying that we accept it as an accurate or appropriate reason for closure in any individual case.
1. The key legislation that governs DCJ’s responsibilities for the child protection system in NSW is the Children and Young Persons (Care and Protection) Act 1998 (Care Act).
2. The Act sets out escalating processes for promoting and ensuring the safety, welfare and wellbeing of children. Within a framework of overarching principles, the Act:
3. This report is focused on how DCJ responds to reports under s 24 and s 27 that a child is at ROSH (ROSH reports). In assessing how DCJ responds to any given ROSH report, it can be necessary to also consider DCJ’s response to any requests for assistance after the ROSH report, any relevant pre-natal reports, and whether the child is also INOCAP.
4. For this reason, we have set out below a summary of key provisions relating to ROSH reports and INOCAP decisions (up to, but not including, the making of applications to the Children’s Court).
5. This is a high-level summary only, and relevant provisions are examined in greater depth where necessary throughout the report.
6. The Care Act contains objects (s 8) and principles (ss 9-14 and 36) that govern how DCJ responds to reports. Principles include:
7. A person who has reasonable grounds to suspect a child or young person is at risk of significant harm may make a report to the Secretary (s 24). Similarly, a person who has reasonable grounds to suspect, before the birth of a child, that the child may be at risk of significant harm after the birth may make a report (s 25).
8. Certain professionals such as police, health workers and teachers are mandatory reporters and, as such, must make a report if they have reasonable grounds to suspect a child under the age of 16 years is at risk of significant harm (s 27).
9. However, alternative reporting arrangements exist that enable certain mandatory reporters to refer to their agency’s assessment officer matters that would otherwise be required to be reported to the Secretary of DCJ (s 27A).
10. As separate categories, requests for assistance may be made (ss 20, 21 and 113) and reports about homelessness may be made (ss 120, 121 and 122). Each of these sections, except for s 122, set out what DCJ must do upon receipt – see s 22, 113(2) and 120(2) and (3).
11. Section 23 deems a child or young person to be ‘at risk of significant harm’ if current concerns exist for the safety, welfare or wellbeing of the child because of the presence, to a significant extent, of any one or more the circumstances listed in s 23(1). This includes a child’s physical and psychological needs not being met or the child has been or is at risk of being physically or sexually abused. It also includes certain children who, prior to their birth, were the subject of a pre-natal report.
12. There are two additional circumstances found outside s 23(1).[191]
13. Section 30 provides that:
30 Secretary’s investigations and assessment
On receipt of a report that a child or young person is suspected of being at risk of significant harm:
14. If the Secretary determines that a child or young person is at ROSH, the Secretary may request a government- or non-government-funded service to provide prioritised access to services for children at risk of significant harm and their family (s 17(2)).
15. The Secretary may also respond to requests for assistance and exercise other powers as described at para 22 below.
16. Section 34 provides that:
34 Taking of action by Secretary
(1) If the Secretary forms the opinion, on reasonable grounds, that a child or young person is in need of care and protection, the Secretary is to take whatever action is necessary to safeguard or promote the safety, welfare and wellbeing of the child or young person.
(2) Without limiting subsection (1), the action that the Secretary might take in response to a report includes the following [...]
17. Actions listed include providing or arranging support services, developing care plans, developing parental responsibility contracts, exercising emergency protection powers or seeking appropriate orders from the Children’s Court.
18. Section 35 then provides for circumstances where the Secretary may decide not to take action:
35 Decision against taking action
(1) The Secretary may decide to take no action if the Secretary considers that proper arrangements exist for the care and protection of the child or young person and the circumstances that led to the report have been or are being adequately dealt with.
(2) If the Secretary decides to take no action, the Secretary must make a record of the reasons for the decision.
19. Section 36 requires DCJ to have regard to three principles of intervention when deciding the appropriate response to a report concerning a child or young person.
20. If DCJ determines that a child or young person is in need of care and protection, DCJ may also:
21. DCJ may also exercise emergency removal powers where there is a risk of serious harm.[192]
22. DCJ has a range of other statutory powers to respond to the circumstances of a child or family that raise safety concerns. The exercise of these powers is not conditional on DCJ having made a ROSH or INOCAP determination as such. However, each power may be subject to specific preconditions, which in some cases will require DCJ to form a view about the risks or needs of the child. For example, if relevant preconditions are met, DCJ may:
23. A child or young person can seek assistance from the Secretary (s 20). A parent may seek assistance to obtain services that will enable their child to remain in, or return to, their care (s 21) and a funded service may seek assistance to obtain other services for a child (s 21).
24. If a person or service seeks assistance from the Secretary, regardless of whether or not a child is suspected of being in need of care and protection, the Secretary must (s 22):
25. However, this does not require the Secretary to take any action other than assessing the request for assistance (s 22(2)).
26. Where there is serious or persistent conflict between parents and children or if parents are unable to supervise their child, the parent, child or young person may ask the Secretary for assistance (s 113(1)). In response, the Secretary may provide or arrange for advice or assistance to be provided (s 113(2)).
27. Anyone may report the homelessness of a child to the Secretary (s 120(1)). On receipt, the Secretary must conduct such investigation and assessment concerning the child as the Secretary considers necessary and may provide or arrange for the provision of services, including residential accommodation (s 120(2) and (3)). The homelessness of a young person may also be reported, but only with the consent of the young person (s 121). However, the section does not provide for a response to these reports.
28. While this investigation focuses only on reports under ss 24, 25 or 27, information in requests for assistance or reports about homelessness may also fall within ss 24, 25 or 27 (for example, if a parent contacts to ask for assistance because they are concerned that, without the assistance, a child is at risk of significant harm).
1. While we have not undertaken a comprehensive audit or evaluation of the SDM tools in use, this Appendix outlines some of the limitations in design (and related flaws in use) that are known or have become apparent to us during the course of this investigation.
2. We generally focus only on the following tools, as these are of most relevance to the response to ROSH reports (i.e. s 24 and s 27 reports):
3. The SDM SCRPT manual states that the tool is used for reaching the following decision:
Whether the information received meets the threshold of risk of significant harm for a child protection report requiring an investigation.[194]
4. DCJ’s 2023 Better Decisions for Children project paper (SDM Screening Assessment Paper) summarised the two key questions SDM SCRPT is designed to answer as follows:
5. The SDM Screening Assessment Paper noted that the SDM SCRPT tool assumes that all reports that meet the statutory threshold for ROSH will be responded to at the CSC. This assumption was, and continues to be, not valid. In 2010-11, only 21% of children reported at ROSH were seen by caseworkers to make a ROSH determination. In 2023-24, a majority of ROSH reports are closed with the label ‘No Capacity to Allocate’, and our sampling confirmed that many cases closed are without any substantive response.[196]
6. The assumption means that the SDM SCRPT’s output of ‘Response Priority Timeframes’ reflects only its recommendation as to how quickly the investigation/assessment or other response should be commenced,[197] not whether or not a response is required, or the overall importance or priority to be given to that response. That is, the timeframes cannot be used to reliably predict overall level of risk.[198]
7. For example, the level of risk and need for action in a case with a response priority timeframe of less than 72 hours may be the same, more or less than another case with a response priority timeframe of 24 hours:
A child may potentially be at immediate risk and need a rapid response (e.g., because a violent parent may be returning to the house tomorrow) or at longer-term risk (e.g., because a violent parent may be returning to the home in 6 weeks). The level of risk may in fact be the same (or even higher in the second case), but the immediacy of the first warrants a <24 hours response priority whereas the second may receive a longer (<72 hours or <10-day) response.
8. However, in CSCs where decisions are being made to close cases on the basis of ‘no capacity to allocate’, some decision-makers use the SDM Response Priority timeframe (<10 days, <72 hours and <24 hours) to inform which cases do and do not get a response or referral, in particular by prioritising cases based on their having a shorter response priority timeframes – effectively taking this to mean they must have a high risk level.[199] If the CSC triage process prioritises reports that require a shorter response timeframe, other reports, with the same or high levels of risk, may be closed without action under the Care Act. This is a particular concern in cases of neglect risk, where that risk may be very high but response times will generally be longer than cases of abuse risk.
9. An assumption may also be made at CSCs that a longer SDM Response Priority timeframe indicates not only lower risk levels, but consequently also the type of referral a family needs. Again however, these timeframes were not designed for this use and do not necessarily correlate to family need or referral appropriateness.
10. Further, DCJ has been using the SDM Response Priority timeframes in determining when families with a ROSH report can be referred to Family Connect and Support (FCS) services. FCS primarily serves non-ROSH families, but DCJ’s program specifications requires them to accept referrals for families with “a low-level ROSH report (less than 10 days response)”.
11. An interim FCS evaluation found that DCJ stakeholders, including FCS service providers, believed the screening process was inaccurate in assessing risk, as they observed that cases requiring a response within 10 days often involved greater risks than those screened at high risk with shorter response times.[200] As a result, FCS providers say they have found themselves working with families facing more complex needs and higher risks than expected or appropriate.
12. The SDM Screening Assessment Paper noted that this risked children or young people being re-reported to DCJ by those FCS services, which are obliged to take the referral but subsequently assess the family as being ineligible for services because they are too high risk:
Many families who are referred to a [service] because the report about them screened a suspect ROSH requiring a response within 10 days, are often re re-reported to DCJ by services because the service considers the family ineligible because they too high risk.
13. It concluded that practices associated with these incorrect assumptions:
reflect a misuse of the response priority timeframe that has led to inconsistent practices and decision making across NSW and significant and unintended consequences.
14. Reports about a child being harmed by a person outside of their home are screened for ROSH in the same manner as reports about harm caused by a parent. SDM SCRPT does not consider the protectiveness of a parent in cases where the harm is caused by someone outside the home (e.g., alleged assault by a coach).
15. It also fails to consider any Aboriginal cultural supports and strengths that may have a protective effect for a child, resulting in the potential for over-prediction of risk where children are in protective Aboriginal kinship groups and communities.
16. DCJ’s 2022 Quality Service Review of SDM SCRPT, which included consultation with Aboriginal stakeholders, led to 6 recommendations, including that SDM SCRPT be revised to explicitly consider culture and protective factors, and alter response priority settings accordingly.[201] That recommendation has not yet been implemented.
17. The Helpline uses the SDM SCRPT tool to identify issues in reports, and these issues are automatically populated into the Judgement and Outcomes (J&O) record that caseworkers later use to record their ROSH determination (or substantiation) decision after a field assessment.[202] This means SDM SCRPT plays a key role in shaping the later ROSH determination.
18. However, the SDM SCRPT Manual and the J&O record do not explicitly reference the statutory ROSH criteria set out in s 23 and some of the SDM SCRPT issues used in the J&O record do not clearly map to s 23. For example, SDM SCRPT lists the issues ‘risk of significant neglect’ and ‘risk of significant psychological harm’ as standalone issues but to meet the statutory definition in s 23 require an additional linking factor – such as a parent’s act or omission or domestic violence. Some s 23 circumstances are missing altogether (e.g., continued risk after a pre-natal report or extended stays in specialised substitute residential care),[203] while others, such as parental risk factors, are included despite not being ROSH criteria, even if they may still need to be relevant.
19. If SDM SCRPT’s issue definitions do not align with the statutory definition of ROSH, this misalignment can influence which issues and families are prioritised and the final ROSH determination.
20. We have seen numerous examples of cases where reports have been screened-in by the Helpline as ‘sexual abuse’ or ‘suspicious indicators consistent with sexual abuse’, but subsequent triage decision-makers have indicated that the cases were better characterised as involving ‘problematic/harmful sexual behaviour’. We also saw the reverse.
21. It is likely that these matters are screened-in as ‘sexual abuse’ because SDM SCRPT does not otherwise have a specific category or guidance for the kinds of behaviour in question, and Helpline staff consider ‘sexual abuse’ to be the closest SDM SCRPT category of relevance. Examples we have seen include:
Example 1: a report concerning alleged sexual behaviour between two 5-year-olds was screened-in as a ROSH report by the Helpline as “sexual abuse – sexual act or exploitation”.[204]
Example 2: a report concerning a child being physically assaulted by peers at school, after which they were then told to open their legs to be filmed, was screened-in as a ROSH report as “suspicious indicators consistent with sexual abuse”.[205]
22. On the other hand, some cases are screened-in using the category of ‘risk-taking behaviours’ which also appeared inapt to the circumstances:
Example 3: a report concerning an older child having subjected a child with a cognitive impairment to penetrative sexual assault was screened-in as a ROSH report as “serious risk-taking behaviours”.
23. In each of these cases (and others we reviewed), the case was closed with a label of ‘No Capacity to Allocate’. However, it appeared possible that in at least some of these cases the decision-maker may have formed a view that there was insufficient reason to believe the child was at ROSH.
24. The SDM Risk Assessment tool is designed to predict the likelihood a child will be re-reported to the child protection system within 12-18 months based on data about previous caseworker decisions on all reports – not whether a child is currently at ROSH or ‘in need of care and protection’.[206] It uses statistical analysis of large datasets to identify family characteristics linked to repeat system involvement and classifies cases as low, moderate, high or very high risk. In NSW, the default dataset from California was used.
25. While possibly useful for prioritising cases, the tool does not assess actual harm or current risk in individual cases. It uses future system contact as a proxy for harm, which may be misleading. Despite this, it has been used inappropriately to support statutory decisions about individual children – decisions it was not designed to inform. Its name has also caused confusion, leading workers and others to believe it helps caseworkers determine ROSH.[207]
26. The SDM Risk Assessment tool is designed to assess and classify the overall risk profile of a family, not the safety or welfare of a specific child. However, it is used by DCJ to decide whether individual children are ‘in need of care and protection’.[208]
27. Evident Change has reiterated concerns that the use of risk classification of families determined by the tool should not be conflated with making a critical statutory decision about individual children.
28. DCJ policy, and the design of the SDM Risk Assessment tool, is such that staff only have discretion to increase but never to decrease the SDM Risk Assessment outcome.[209] In other words, staff could not manually override an SDM Risk Assessment tool outcome if they consider that the risk has been assessed as too high, including because of new information.
29. This issue was noted by the District Court in a case where a child had been removed and taken into care. That action had been taken following a SDM Risk Assessment output that DCJ staff had been unable to override, even though they had become aware that it was based, among other things, on incorrect medical evidence. The Court noted:
Specifically, as will appear from my review of the evidence of [the Manager Casework and caseworker] the departmental management of data on matters of child risk had a rigidity which could not be overridden by caseworkers and managers who came into possession of contrary information that ought to have served to dispel some crucial recorded departmental notions of risk.[210] (emphasis added)
30. Four questions in the SDM Risk Assessment tool have the potential, when combined, to drive an outcome score of high or very high.[211] These are N1 (current report is for neglect), N2 (number of prior screened-in reports), A1 (current report is for abuse) and A2 (number of prior screened-in reports of abuse).
31. Unless manually overridden, counts N2 and A2 include reports that were screened-in but later closed, without being assessed and determined, including due to insufficient resources. Counting these undetermined earlier reports increases the likelihood that a high or very high risk rating – and therefore an INOCAP result – may be unduly influenced by localised factors driving report volumes or (non) responses under the Care Act (e.g., caseworker vacancies, reporter practices, or biases). DCJ staff observed:
‘In a system where the over-surveillance of Aboriginal people and racist assumptions made about the parenting of some Aboriginal children is acknowledged, this negatively skews outcomes for Aboriginal children.’[212]
32. The case scenario of John below (not a real case) has been put together to illustrate the potential effects of this count issue. We have developed it using the guidance that was available at the time the SDM Risk Assessment tool was still operational:
Case scenario: John (Aboriginal child) | |
Application of the tool | Unfairness to John |
John has been the subject of 5 prior ROSH reports (neglect). None of these were determined, all were closed due to ‘no capacity to allocate’. A new ROSH report concerning neglect is made and this time the report is allocated. A caseworker visits and decides that right now John is Safe (i.e., there is no immediate danger, although there may be ongoing risks). Next, the SDM Risk Assessment tool is run:
A High risk result means that, under the INOCAP matrix, John is INOCAP. DCJ is directed by the Care Act to take whatever action is necessary to safeguard or promote John’s safety, welfare and wellbeing. |
|
33. Like SDM SCRPT, the SDM Safety Assessment and SDM Risk Assessment tools also do not consider protective factors, such as family strengths. Nor does it assess family needs.
34. At the time the SDM Safety Assessment and SDM Risk Assessment tools were implemented in 2011, the SDM Family Strength and Needs Assessment Tool (SDM FSNA) was not included in the suite of tools for roll out, despite being a usual part of the SDM process. SDM FSNA is designed to be completed after a SDM Risk Assessment to inform development of a Family Action Plan. According to DCJ, it guides practitioners to identify family strengths (which are not identified in assessments before this point) and identify what the child and family actual needs are, helping caseworkers to:
35. The SDM FSNA was used in the Brighter Futures early intervention program (2011-2012) before a decision by DCJ Executive to stop using the tool in 2013, apparently on the basis that practitioners were still attempting to adjust to the implementation of the SDM Safety Assessment and SDM Risk Assessment tools in 2012. It was also used in a Short-Term Court Orders pilot (2014-2015) but is no longer in use in DCJ due to the cessation of the program. [214]
36. Our investigation does not examine the SDM FSNA, and we make no comments or findings regarding the decision not to adopt it.
37. However, it is important to note that, without SDM FSNA, caseworkers now attempt to translate the safety and ‘risk’ focus of SDM Safety Assessment and SDM Risk Assessment tools into a Family Action Plan, without any of the structure, guidance and focus on strengths and needs supported by the SDM FSNA tool.[215]
38. On 15 September 2022, Evident Change confirmed that the SDM Risk Assessment is not the best tool to identify items for the Family Action Plan. It recommended that the SDM Risk Assessment Tool be used to identify which families would benefit from applying the SDM FSNA, and that the SDM FSNA, along with the family and worker, then be used help select Family Action Plan items.[216]
39. As at April 2023, DCJ said it was working towards a custom-built SDM FSNA, noting that:
‘….we are the only jurisdiction that uses SDM and doesn’t have the Family Strengths and Needs Assessment as a compulsory part of our assessment process … it’s probably the most important stage because its really how you focus in on the families [sic] own perspective, children’s perspective about what their strengths are and what they need in order to change and to do safety assessment and Family Action Plans without the FSNA is highly questionable really and you could argue that it has led us to quite deficit-based practices.’[217]
40. The investigation team collected information from a range of sources and conducted activities as described below:
Source data
41. DCJ provided triage records[218] for the first 380 ROSH reports closed during June 2023 with the closure reason ‘No Capacity to Allocate’. These closures happened to be spread across 1 and 2 June 2023.
42. The figure of 380 was calculated based on figures for the June 2023 quarter as a sample size that achieves a confidence level of 95% and confidence interval of 0.05.
43. The 380 covered a spread of CSCs and Districts – more than half of all CSCs (55%, 44 of 79) from nearly all Districts (93%, 15 of 16).
44. Early June was chosen as a point in time that was both close to the end of the (then) most recent financial year and outside the school holidays.
Sample analysis and limitations
45. Our initial review of all 380 cases looked only at the source data provided by DCJ. This review identified total counts, issues of concern and potential case studies.
46. We then reviewed ChildStory records of 237 cases (62%, 237 of 380) to obtain deeper insight as to reasons for closure and adequacy of response (at a minimum by inspecting the Triage record and Contact record, and where relevant the Notes and Attachments or other records). This included comprehensive review of the ChildStory records of all potential case studies and all records where referral was considered in triage record commentaries.[219]
47. We also reviewed ChildStory records of all 380 to identify or confirm:
48. Full review and quality assurance (QA) of sample results was conducted by 2 Senior Investigation Officers and 1 Principal Investigator.
49. Further ChildStory reviews were conducted during QA for all records where reviewers had:
50. The number of reports closed on the basis of capacity or exceeding 28 days was ascertained from triage records provided by DCJ (not from ChildStory).
51. We considered decisions through the lens of DCJ policies and procedures, and the Care Act, as applicable at the time of the triage decisions in June 2023. Note that at this time, the principle of ‘active efforts’ was not in effect.
Evaluating not-at-ROSH cases
52. Our reviewers assessed whether it was open to DCJ to determine that a child is not-at-ROSH and close each triage record.
53. A not-at-ROSH determination uses evaluative judgement to apply an objective test. There will be considerable scope for reasonable minds to differ as to whether or not a child is at ROSH.
54. Assessors considered whether it was open to the DCJ decision maker, on the information in the triage data supplied by DCJ, to reach a determination that the child is not currently[220] at ROSH (immediately prior to or at the time of the record closure). [221]
Necessary inferences and underlying data quality
55. It was necessary to draw inferences from content in the narrative, summary and case closure commentaries. For example, in some cases, the category of screened-in risk was not always apparent, and inferences were drawn where required.
56. It is accepted that:
57. As such, total numbers are to be used as general indicators of possible prevalence of relevant decision-making practices, not as definitive practice indicators.
Response selections
58. Reviewers selected one of the following:
Relevant assumptions
59. It was necessary for reviewers to apply some general assumptions for the purposes of:
60. These assumptions therefore take a conservative approach to DCJ’s role in responding to ROSH, focusing on the minimum that could be expected of it in a given scenario.
61. It is important to note that in any given case, the assumptions described below may not apply (or be determinative) – e.g., if there are other circumstances which, taken together, mean it would not be open to DCJ to determine ROSH and close (e.g., due to the presence of multiple risk factors, signs of escalation or problematic gaps in information).
Cases where it was reasonably open to DCJ to determine ROSH and/or close
62. Assessments about whether it was reasonably open for DCJ to make a not-at-ROSH determination and/or close assumed that IF reasonably attainable relevant information has been gathered, THEN in the absence of other information to the contrary, it is GENERALLY reasonable for DCJ:
63. It was NOT generally considered reasonable for DCJ to make a ROSH determination if:
Cases where DCJ appeared to decide that the child was not-at-ROSH
64. In some cases it appeared that the decision maker reached a concluded view that the child was not-at-ROSH as the decision maker explicitly recorded that there were no concerns (or words to that effect) for the child, and we were able to rely on reasonably sound inference that this view was reached.
65. There were also a significant number of matters where it appeared to us that the decision maker may well have reached a view that the child was likely not-at-ROSH. However, without interviewing the relevant staff about the decisions they had made more than 12 months prior (to test the inferences we had drawn), reliable data could not be obtained.
Cases where more action would have been preferable
66. The investigation and assessment that is necessary to determine whether a child is not-at-ROSH is a subjective test (as the Secretary sees fit). This is a very broad discretion but must be exercised reasonably. The Minister (and, where delegated, the Secretary) also has general powers to provide assistance under the Community Welfare Act 1987 (NSW).[222]
67. Reviewers considered more action preferable in circumstances such as:
Pre-natal and duplicate records
68. Pre-natal reports were only recorded as ‘No’ if the report was closed without an adequate pre-natal response (on the basis ROSH should not be determined prior to birth).
69. Reports containing duplicate information were:
Other sample limitations
70. Not all ChildStory records had Aboriginality recorded. Records where Aboriginality was ‘not stated’ were included when deriving overall statistics unless otherwise stated.
71. Because the sample spans only 2 days (i.e., closures that occurred on 1 and 2 June 2023) it may not reflect a given CSC's overall annual performance. Results may be affected by point-in-time circumstances, such as fluctuations in resources available on a given day, rostering of particular staff or certain events. For example, there was a teen party that gave rise to multiple peer-on-peer reports.
72. Because the sample includes only triage records, there may be other information (not available to reviewers) that was available to DCJ decision makers in ChildStory that influenced decisions. This was mitigated by ChildStory review of at least 237 of 380 records (62%). Note that ChildStory was accessible only to one Senior Investigation Officer on the investigation team.
73. We looked at a small but targeted number of cases closed due to ‘Current Competing Priorities’ after allocation.
74. The first cohort were selected from JCPRP cohorts, due to the generally serious nature of the risks and policy about allocation of these matters. We had also heard allegations that due to resource constraints CSCs:
75. JCPRP teams maintain a spreadsheet that assists in the tracking and management of its matters. In the later stages of our investigation, we sought the 2022-2023 and 2023-2024 register data for three JCPRP teams (Western, Tamworth and Central Metropolitan) areas.
76. We looked at 6 JCPRP cases that were closed with the CCP label following a JCPRP referral to a CSC. We obtained a copy of the working spreadsheets JCPRP teams use to help manage JCPRP matters, namely Western, Tamworth and Central Metropolitan spreadsheets for 2022-2023 and 2023-2024 financial years that met the following parameters:
77. We also asked DCJ to provide data on cases closed for the reason ‘Current Competing Priorities’ by SARA outcome. From this, we inspected ChildStory records for:
78. At our request, DCJ agreed to distribute an online survey via email to all Managers Casework (MCW), Managers Client Services (MCS) and Director Community Services (DCS). The survey comprised 31 questions.
79. The email with a link to the survey was sent to approximately 510 DCJ staff members on Thursday, 6 March 2025, requesting that they complete the NSW Ombudsman survey by Friday, 14 March 2025.
80. The email included advice that the survey responses were anonymous, and that individual responses would return directly to a small number of staff at the NSW Ombudsman and would not be provided to DCJ.
Limitations
81. The completed survey response rate (131 of 510) was very good at 25% of all staff involved in responding to ‘ROSH reports’ at MCW level and above. However, completing the survey was not compulsory, so risks of self-selection bias are present.
82. Mandatory and Voluntary questions, and progression rules as follows:
83. Staff without direct triage experience were invited to answer questions 9-25 as they may have visibility of earlier triage allocation practices and decisions further along the casework continuum and therefore valuable insights to contribute. However, a notable proportion of CSC MCWs (36-45%) staff reported no direct triage/allocation role (Note: Direct vs indirect role data (Q3 and Q26) was taken into account when analysing triage/allocation responses (Q9-25). There were no significant differences in overall results except for Q10, Q14-3, Q14-5, Q14-6 and Q15 – these disaggregated were duly considered in our primary analysis).
84. Survey limitations are noted in footnotes in the text of this report, where relevant.
This inquiry documented progress on the Government’s reform program – ‘Keep Them Safe: A shared approach to child wellbeing’ – which had stemmed from the recommendations of the Wood Special Commission of Inquiry.
A key finding of the Ombudsman’s inquiry was that, although the number of risk of significant harm (ROSH) reports received by the Department had halved following the reforms (which had included raising the reporting threshold to risk of significant harm), a very high number of serious child protection reports still received no response, and there were fewer children recorded as receiving a comprehensive risk assessment.
The available data showed that only around one fifth of all reports assessed by Community Services as indicating ROSH were receiving a face-to face response. This included a quarter of reports closed on the express basis of ‘competing priorities’.
The report noted:
‘These children [who are the subject of a ROSH report], at the very least, need Community Services to check on their circumstances and when required, intervene to protect them.
On any measure, it is unacceptable that 25 per cent of all reports assessed by Community Services as indicating risk of significant harm received no response in the first 12 months of Keep Them Safe. All of these reports were closed on the basis of ‘competing priorities’.
Where decisions to take no action are made, it is essential that these decisions are clear and justifiable. At a minimum, this requires the system to be able to identify those children who are most at risk, and ensure that case closure decisions are made with regard to the best information available from the totality of relevant agency holdings. In our child death review work and our investigations into well over 100 matters involving inadequate responses to serious child abuse and neglect, we have consistently identified a lack of rigour in decisions to close cases due to competing priorities.’
Annual Report 2010-2011 – NSW Ombudsman – October 2011
In his annual report, the Ombudsman noted that longstanding concerns about child protection cases continuing to be closed ‘due to competing priorities’ had become even more acute in the context of Community Services limiting its statutory responsibility to children at ROSH.
The report included examples, and noted ongoing investigations, where significant numbers of ROSH reports had been generated for certain children, but little or no casework had been done before the matter was closed. It noted that often cases were repeatedly opened in response to new ROSH reports and then closed again ‘due to competing priorities’ – with little recognition of the increasing evidence that demonstrated escalating risk.
Annual Report 2012-2013 – NSW Ombudsman – October 2013
Noting the Government progress in implementing its response to the Ombudsman’s Keep Them Safe? report, the Ombudsman observed some improvement in the number of children at ROSH who received a face-to-face assessment the two years since 2010-11. However, the report noted more needed to be done, including in developing and implementing an “intelligence-driven approach to child protection work to allow Community Services and partner agencies to identify children at most risk of experiencing significant harm”.
Review of the NSW Child Protection System: Are things improving? – NSW Ombudsman – April 2014
This report was a follow-up to the 2011 Keep Them Safe? report.
It found some improvement in the annual rate of face-to-face assessments of ROSH reports (from 21% to 28%), but that there was still a significant gap to bridge before the response rate was at an acceptable level. It also noted a significant improvement in Community Services’ capacity to measure, monitor and report on issues which impact on its ability to respond appropriately to ROSH reports.
Further recommendations were made, including to reduce caseworker vacancies, as well as to enhance the capacity to record, and report on, the nature of responses being provided to all children the subject of ROSH reports – not just those that result in a face-to-face assessment by Community Services.
Annual Audit Reports on Family and Community Services (2016) and (2017) – NSW Audit Office – November 2016 and December 2017
These annual financial audit reports by the Audit Office’s noted the Department’s objective to provide more children at ROSH with a face-to-face child protection response from a caseworker.
They reported that the number of ROSH reports responded to was increasing (e.g., 31% of reports in 2016 received a face-to-face response), but the number of children being reported was also increasing significantly.
Child Protection – NSW, Legislative Council, General Purpose Standing Committee No. 2 – 2017
The Parliamentary Committee conducted a broad-ranging inquiry into the role of the Department of Family and Community Services in child protection, including the capacity and effectiveness of systems, procedures and practices to notify, investigate and assess reports of children and young people at risk of harm.
It observed that there had been an increase in the overall number of child protection reports over the last few years. While the number of face-to-face assessments had slightly increased, it was still at an unacceptably low rate of about 30%.
The inquiry also questioned whether screening tools were fit for purpose (especially for Aboriginal families, and cases involving domestic and family violence, and cumulative harm).
It called for more objective and consistent triage, smarter and more innovative use of limited resources and a stronger shift to early intervention, rather than relying on more crisis responses.
The Committee added:
‘One has to question what is happening with the remaining number of children who are simply not being seen, those cases that are simply closed due to ‘competing priorities’.
It seems that decisions about which cases will be investigated or allocated appear to be based more on resource and capacity issues of the department, rather than on the outcomes of a safety assessment. While we accept that these are complex and difficult issues, we believe that the system needs to work far more effectively and intelligently to ensure that all children identified at risk of significant harm receive an adequate response from the department.’
Inquest into the death of BLGN and DG – Coroner’s Court (Deputy Coroner, Magistrate Harriet Grahame) – 2018
This inquest concerned the deaths of two sisters, aged 3 months and 19 days, in 2014 and 2015 respectively. Both were sudden unexpected deaths in infancy, and in both cases the cause of deaths remained ‘underdetermined’ after inquest. However, the inquest found ‘a picture of bureaucratic failure and an ongoing inadequate response to a family in genuine need’.
Several ROSH reports had been made in relation to the children, including some that were closed due to ‘competing priorities’, with the inquest finding that these ‘should have been allocated for immediate response’.
It observed that the 'competing priorities' category was a catch-all that prevented proper assessment of a report’s seriousness.
The Deputy State Coroner said that ‘[DCJ’s] statutory responsibility for protecting children and young people from risk of significant harm… cannot be shifted by creating a culture where overworked staff can close reports, claiming a lack of resources or “competing priorities”.’
The inquest recommended, among other things, that DCJ should undertake a review of the types of ROSH reports being allocated, referred to services or “closed for competing priorities” at triage (including during weekly allocation meetings), so that the DCJ Executive can better monitor, consider and review resource allocation and address the need for any procedural changes.
Inquest into the death of Z – Coroner’s Court (Deputy Coroner, Magistrate Harriet Grahame) - March 2021
This inquest concerned the death of a toddler who died in 2012 from suspected abuse-related injuries after 13 prior ROSH reports to DCJ were not allocated for a ‘statutory response’. All but one were closed due to ‘competing priorities’.
The Deputy Coroner found that a proper risk-based assessment against other reports the CSC was dealing with at the time had not been done to determine whether it was doing its best to allocate the highest risk matters within available resources. The report noted flawed decision-making at important junctures. For example, weekly allocation meetings (WAMs) did not consider Z’s full child protection history because it had not been included in full on the WAM form; unallocated reports were not escalated; and physical-abuse reports were not referred to the Joint Child Protection Response Program (JCPRP).
Biennial report of the deaths of children in New South Wales: 2018 and 2019 – Child Death Review Team/NSW Ombudsman – December 2021
The report identified common themes arising from the deaths of children who died due to abuse or neglect and who had been the subject of a ROSH report in the 12 months preceding their deaths. These included:
The Ombudsman recommended, among other things, that DCJ detail the actions it was taking at a strategic level to address the premature closure of ROSH reports due to competing priorities, including cases closed without comprehensive assessment or face-to-face contact, and where referrals are made (in place of an assessment), as well as the outcomes of the policy and practice reviews that had been recommended in the coronial inquest into the death of Z. It also recommended that DCJ consistently record and follow up on outcomes when a child reported at ROSH is referred to a service provider – including with prompts to identify when a provider declines or cannot allocate a child for service.
Inquest into the death of SG – Coroner’s Court (Deputy Coroner, Magistrate Kennedy) - August 2023
The Deputy Coroner examined a case of a 15-year-old First Nations student who died by suicide in 2020 after DCJ, NSW Health and the Department of Education each closed their files, largely assuming another agency was providing support.
The Deputy Coroner recommended DCJ ensure suicide risk – especially prior attempts – is explicitly given weight during its Prioritisation, Triage and Allocation Policy Review to ensure better identification and prioritisation of children most at risk.
Inquest into the death of BW – Coroner Court (Deputy Coroner, Magistrate Harriet Grahame) - September 2023
The Deputy Coroner examined a case in which a father died by suicide in 2020 hours after police urgently removed his daughter during a welfare check. In the preceding 6 months, multiple ROSH reports were closed without a home visit due to ‘competing priorities’ and existing NGO involvement. The latest report had been allocated for a visit 2 days before the removal.
Caseworkers said closures reflected capacity restraints, not absence of risk. The Deputy Coroner characterised long-standing closures on capacity grounds as ‘dangerous’, stressed DCJ’s unique statutory duty to protect children, noted the difficulty of allocating neglect cases before crisis, and that NGOs cannot substitute for a statutory response. Although the closures had been peer-reviewed and did not breach policy, the Deputy Coroner found some of the decisions ‘wrong’ in context and symptomatic of a recurring systems failure driven by inadequate resourcing.
Inquest into the death of Baby Q – Coroner’s Court (Deputy Coroner, Magistrate Harriet Grahame) - April 2024
This inquest concerned a 9-month-old killed by the father in 2018 during a psychotic episode after months of homelessness and fragmented agency contact.
The inquest criticised the premature closure of reports, failure to consider cumulative neglect risk and system gaps (e.g., the absence of any automatic alerts on ChildStory when Child Wellbeing Units (CWUs) added entries), and noted evidence of systemic resourcing problems.
Oversight of the NSW Child Protection System – NSW Audit Office – 6 June 2024
This audit assessed the effectiveness of DCJ in planning, designing, and overseeing the NSW child protection system. The overall findings included that the system is inefficient, ineffective, and unsustainable. The audit found that, despite recommendations from numerous reviews, DCJ had not redirected its resources from a ‘crisis driven’ model to an early intervention model that supports families at the earliest point in the child protection process.
It reported that 75 per cent of children presumed to be at ROSH were not receiving a home-based safety and risk assessment, and that DCJ does not know the outcomes for these children. It also reported that DCJ was not meeting most of its timeframes to assess child protection reports and visit children who are presumed to be at ROSH.
The audit found that DCJ did not have clear policies, mandates and processes to ensure caseworkers could make consistent and equitable statutory decisions about children and families across NSW, and that its ROSH report assessments were repetitive and ineffective.
Recommendations included for DCJ to “address inefficiencies and duplications in child protection reporting and risk assessment processes”.
Protecting Children at Risk: An assessment of whether the Department of Communities and Justice is meeting its core responsibilities – NSW Ombudsman – 5 July 2024
This inquiry sought to assess whether DCJ was meetings the 3 elements of its core child protection responsibilities: (1) statutory assessment and response to children reported at risk, (2) supporting children in OOHC and exiting them to a permanent home, and (3) prevention, early intervention and family preservation. Overall, the inquiry concluded that DCJ could not demonstrate it was meeting any of these responsibilities.
In relation to its responsibility to assess and respond to ROSH reports, the inquiry It found DCJ does not collect or publish enough information to show what response children actually receive; whether they are safe and whether DCJ’s statutory duties are met. Although about three-quarters of children reported at ROSH did not receive a face-to-face visit, DCJ could not demonstrate the extent to which this reflected decisions being made that children were ‘not-at-ROSH’ and did not require further investigation, or whether it was a result merely of capacity constraints. Nor could it demonstrate how urgent and high-risk cases were identified and prioritised, or what support (if any) was provided when reports were closed. Nor could DCJ account for the outcomes for the 84,810 children in 2022-23 who had no face-to-face casework response.
It was in the context of this inquiry that the Ombudsman decided to commence the current investigation.
Inquest into the death of BG – Coroner’s Court (Deputy Coroner, Magistrate Kasey Pearce) – January 2025
The inquest examined the 2020 death of a baby from head injuries in a domestic violence context. Before the baby’s birth, there were 3 DV-related ROSH reports (2 of which were screened as ROSH). After the birth, a further ROSH report alleged serious neglect and risk factors. All were considered at a weekly allocation meeting (WAM) and closed by the CSC for the reason ‘no capacity to allocate’. A person later identified as the perpetrator had been recorded by DCJ in 2013 as a Person Causing Harm, but that link was not identified during triage and so was not considered at the WAM.
DCJ’s internal case review found that more triage activities could have been undertaken, record-keeping was insufficient to explain decisions and DV was not well understood or recorded.
The Deputy Coroner noted that DCJ’s 2022 Triage Assessment Mandate updates after BG’s death did not clearly amount to material practice change, and heard evidence that workforce capacity was the primary driver of closures – this particular CSC had only been able to fill 9 of 19 funded caseworker positions.
DCJ conceded it could and should have done more in response to the last ROSH report. The Deputy Coroner observed that, while some lessons were said to have been learned, it was unclear that changes by the time of the hearing (August 2024) adequately addressed the issues.
Inquest into death of Harmony – Coroner’s Court (Deputy Coroner, Magistrate Harriet Grahame) – March 2025
This inquest concerned Harmony, a teenager in temporary accommodation, who died by suicide in 2020. DCJ knew that 2013 court orders gave her father sole parental responsibility, restricted the mother’s contact and restrained the mother’s partner from unsupervised contact. However, at the time of her death, Harmony was staying with her mother and mother’s partner.
The Deputy Coroner found DCJ’s overall response to reports in the 5 years leading up to Harmony’s death had been ‘inadequate’. Several Helpline reports were wrongly screened as non-ROSH; others went to WAMs but were closed due to ‘competing priorities’ despite notes that further investigation was required.
The Deputy Coroner highlighted that, once ROSH is identified, DCJ must act, and closures on capacity grounds cannot be treated as ‘business as usual’. The decision to close despite known protective court orders was particularly criticised, with the Deputy Coroner urging DCJ to confront resourcing issues openly and fix prioritisation, triage and allocation practices.
Inquest into the death of Jade – Coroner’s Court (Deputy Coroner, Magistrate Joan Baptie) – 24 June 2025
Jade was a 3-year-old First Nations child, who died in 2018 from bronchopneumonia against a background of family neglect. Jade and her family were ‘known to’ DCJ, including as the result of multiple ROSH reports. DCJ had previously referred the family to the ‘Brighter Futures Program’, a family preservation program run by a non-government provider.
The Deputy Coroner noted various decisions made by DCJ to close ROSH reports because of a lack of capacity. The Deputy Coroner also considered whether the provider of the program should have made a further mandatory ROSH report – although she also observed that, had such a report been made, it was unlikely DCJ would have acted on it in any event. The inquest report raised concerns about both issues, noting that:
“The closure of a verified [i.e. screened in] ROSH report due to lack of capacity was never within the contemplation of the legislation. It is a practise which appears to have evolved within DCJ and CSC offices, with no legislative basis or justification.”
The Deputy Coroner further observed that:
“Agencies such as BF [Brighter Futures] have become [a] defacto type of child protection agency, undermining the intent of the legislation.
This situation is exacerbated when those defacto agencies are “educated” not to comply with [their obligation to report to DCJ when they suspect a child is at ROSH]... The potential consequences of such arrangements mean that appropriate cases are not escalated…”
Recommendations were made, including to the Minister to ‘review these findings and examine the caseworker capacity of the [relevant] CSC and consider the other issues canvassed in these findings.’
s 23 Care Act.
The Care Act defines a child as under 16, and a young person as under 18. For simplicity, we also use child or children to refer to any person aged under 18.
In some cases, a ROSH report may be transferred to a specialist unit, such as the Joint Child Protection Response Program (JCPRP).
See above: 25% in 2022-23; 21% in 2023-24; 19.5% in 2024-25.
Our view is that there would be some cases where DCJ could make legitimate and reasonable decisions following further ‘desktop’ investigations that a face-to-face assessment is not necessary for it to determine whether the child is at ROSH. This observation should not be read as a green light to DCJ to be relying primarily on ‘desktop investigations’ to make ROSH determinations. Where there remain any reasonable concerns, a face-to-face assessment will always be necessary.
Inquest into the death of Jade (2025), at [153]-[158].
DCJ identifies other core elements as: (1) supporting children in OOHC and exiting them to a permanent home; (2) providing prevention, early invention and family preservation (to avoid children entering the OOHC system). See NSW Ombudsman, Protection children at risk: an assessment of whether the Department of Communities and Justice is meeting its core responsibilities (2024)(PCAR Report (2024)), p.2.
Conduct may include action and inaction.
NSW Ombudsman, PCAR Report (2024); NSW Audit Office, Oversight of the Child Protection System (6 June 2024)(Oversight Report (2024); NSW Audit Office, Safeguarding the rights of Aboriginal children in the child protection system (2024); DCJ, System review into OOHC (2024).
E.g.,: DOCS; Critical Issues (2002), p.10; Reviewable Deaths Annual Report 2003-2004 (2004) – Recs 1, 2, 6, 9 and 10; Keep Them Safe? (2011) – Rec 1; Review of the NSW Child Protection System: Are things improving? (2014) – Rec 2(f) and 3; Rec 16; Biennial report of the deaths of children in New South Wales: 2018 and 2019 (2021) – Recs 4, 5, 6. This issue has also been raised in Annual Reports in 2010-2011 (p.68 – competing priority case studies) and in 2011-2012 (p.74) and 2012-2013 (p.84) due to concerns about reduction in face-to-face assessments.
Eg,: NSW, Parliamentary Debates, Legislative Council, 28 June 1999 (Lee Rhiannon) and Legislative Council Questions and Answers, 30 November 1999 (Mr Corbitt), pp.351-352; Standing Committee on Social Issues, Inquiry – Child Protection Services (2002), pp.87-90 – Recs 28, 29 and 30; Report of the joint DOCS / PSA Working Party (Kibble Committee)(2002), pp.14, 16-17 – Recs 5.1 and 5.2; NSW, Legislative Council, General Purpose Standing Committee No. 2, Child Protection (2017), pp.48-52 – Rec 4.
The Hon James Wood, Report of the Special Commission of Inquiry into Child Protection Services in NSW (2008) (Wood Inquiry Report (2008)), Vol 1, pp.147, 277-279 – Recs 6.1, 10.2; David Tune, Independent Review of Out of Home Care in NSW – Final Report (2015)(Tune Report (2015)), p.43 – Rec 4.3, 4.4.
Eg,: Report on FACS (2016), pp.25, 31; Report – FACS (2017), p.16; Oversight Report (2024), pp.5, 6, 54.
Eg,: Inquest into the death of BL GN and DG (2018) – Rec 1, 2 and 5; Inquest into the death of AP (2020) – Rec 1; Inquest into the death of Z (2020) – Rec 1; Inquest into the death of BW (2023), para.131-132; Inquest into the death of SG (2023) – Rec 6; Inquest into the death of Baby Q (2024), [168] and [171]; Inquest into the death of BG (2025), [10.11]; Inquest into death of Harmony (2025), [80]; Inquest into the death of DB (2025); Inquest into the death of Jade (2025); Inquest into the death of Jacob (2025).
Pre-2010 include: Priority One Project (1998-2003); Report of the joint DOCS / PSA Working Party (2002), pp.14, 16-17 – Recs 5.1 and 5.2; Unallocated Case Closure Policy and Procedures Steering Committee and Trial (2003-2005); KiDS Business Rule 179 – Unallocated Case Closure (2003-2016); 2008 Wood Special Inquiry (2007-2008); Intake Assessment Guidelines Trial (2010-2011). For post-2010, see n 17.
Wood Inquiry Report (2008), Vol 1, pp. ii-iii and 185-186.
E.g.: Keep Them Safe (KTS) Reforms (2010-2014); Practice First (2011-2016); Seeing More Children Project (December 2014-2020); Keep Them Safe Outcomes Evaluation (2014); Child Wellbeing Units Evaluation (June 2014); Tune Review (2015); C Eastman, A Shlonsky & J Ma, NSW Community Services Structured Decision Making Review - Analytical Report (2015)(Analytical Report (2015)); Professor Megan Davis, Family Is Culture Final Report: Independent review of Aboriginal Children and Young People in Out of Home Care (October 2019)(FIC Review (2019))(2016-2019); Their Futures Matters (TFM) Reforms (2017); Limebridge Australia Review of Helpline Procedures and Allocation Systems (2018-19); Northern NSW Streamlined Response Pilot (renamed to the Helpline Advanced Screening Program (HASP)) (2018-2024); Their Futures Matter system transformation project (May 2017-2019); Collaborative Support Pathways Pilot (renamed to the South West Sydney Allocation Hub)(2019-2021); Greenfields Project at Blacktown and Nowra CSCs (2020-unknown); Assessment Review Project (Jan 2021-Mar 2023), renamed to the Better Decisions for Children Project, including the Safety in Care Assessment Design Project (Mar 2023-2024); Protecting Our Most Vulnerable Children Project (2021-2022); Prioritisation, Triage and Allocation Policy Review Project (July 2022-present).
NSW Ombudsman, PCAR Report (2024).
Media Release “Rebuilding services for young people in out-of-home care” (24 June 2025), announcing $191.5 million funding to recruit more caseworkers; DCJ, System review into out-of-home care (2024), pp.108-09; PSA ‘Child Protection Crisis’ Campaign announced in April 2024.
Email from DCJ dated 7 July 2025.
E.g.: Oversight Report (2024), pp.2, 5, 6 and 35.
E.g.: Keep Them Safe Outcomes Evaluation (2014), p.49; Analytical Report (2015), p.82; Tune Report (2016), p.45; Their Futures Matter, Diagnostic and High-Level Opportunity Design for the Child Protection Helpline and CSC Allocation Process (2018), pp.53, 56 and 67; FACSIAR, DCJ Casework Study (2021).
FACSIAR, DCJ Casework Study (2021), pp.8 and 15.
The NSW Ombudsman has a function of monitoring and reviewing certain deaths of children and young people, being deaths that were due to (or occurred in circumstances suspicious of) abuse or neglect, and deaths of children and young people who had been living in care or in detention: Community Services (Complaints, Reviews and Monitoring) Act 1993, Part 6.
In ChildStory, the ‘mandatory reporter’ flag is derived from the ‘contactor type’ field and the individual themselves may not identify as a mandatory reporter or be making a mandatory report. In ChildStory data, ‘contactor type’ is defined as ‘the type of contact entity (person/organisation) that is contacting the Helpline, represented by a label (e.g.: family, NSW Health, GP, parent and NSW Department of Education): see Service Event: Contactor Type, Label X (50), Metadata.NSW. These are then grouped by type of organisation (e.g., NSW Health, NSW Health etc): see Service Event: Contactor Type, Grouped Label X (45), Metadata.NSW. ‘Mandatory reporter’ is defined as a flag indicating whether the contactor type is mapped as a mandatory reporter: Service Event: Mandatory Reporter, Yes/No Label X (3), Metadata.NSW.
See, e.g., DCJ Website, ‘The MRG and reporting to the Child Protection Helpline’.
It may be that a decision to screen-in the report as a ‘non-ROSH report’ could (as a matter of substance, even if not formally expressed this way by DCJ) be characterised as a decision under s 30(b) of the Care Act. That is, a decision is made that the report provides insufficient reason to believe the child or young person is at ROSH. Be that as it may, the point we are making here is that DCJ’s records do not identify such a report as having been a s 24 or s 27 report.
DCJ Website, ‘The MRG and reporting to the Child Protection Helpline’.
Note also that the definition of ‘risk of significant harm report’ for the purposes (only) of s 172A of the Care Act (concerning DCJ’s function of reporting certain ‘reportable deaths’ known to DCJ) appears to be consistent with the way DCJ has been using the term ROSH report generally. ↑
In summary, this includes a person who delivers, or who manages an entity that delivers, health care, welfare, education, children’s services, residential services, law enforcement or disability services, wholly or partly, to children. It also includes those providing religion-based activities to children, and registered psychologists: s 27(1) Care Act and Care Act Regulation 2022, cl 7.
Non-mandatory reports may be made anonymously: s 26 Care Act.
If the report is about a young person (i.e. aged 16 or 17) and it is known that the young person did not wish the report to be made about them, then the Secretary must have regard to that wish when deciding how to make investigations and assessment (taking into account the age of the young person and the extent to which they or others appear to be at ROSH): s 31 Care Act.
Indeed, the concept of ‘closing’ a report is an operational one that has been created by DCJ that has no foundation in the Act. Rather, under the Act, a report is made (at a point in time), and that immediately triggers the statutory obligation for DCJ to make certain decisions. A report cannot be ‘open’ or ‘closed’ under the Act. It is simply made. It is DCJ’s operational ‘file’ or ‘matter’ about that report which is opened, and which might later be closed once DCJ has discharged its responsibilities under the Care Act in respect of the report.
The ‘ordinary’ rule, where a statute does not provide a timeframe within which a statutory function is to be done, is that it is implied that it is required to be done within a reasonable time: Koon Wing Lau v Calwell (1949) 80 CLR 533, 573–4; Minister for Immigration and Citizenship v Li [2013] HCA 18; (2013) 249 CLR 332 (Li) at [102].
This contrasts with corresponding legislation in other jurisdictions. E.g., under the Victorian Children, Youth and Families Act 2005, a person makes a report if they have a ‘significant concern’ (s 28) – a term that is not defined – and if the department considers the child may be ‘in need of protection’, it can classify the report as a ‘protective intervention report’ (s 34) to be investigated (s 205), and then take action if the child is ‘in need of protection’ (s 240) – a term that is defined (s 162).
See Appendix A-2.
The Wood Special Inquiry also recommended changing the reporting trigger to ‘significant’ harm on the basis this would result in the reports DCJ receives being limited to ‘those children who are likely to need the powers of the State under s.34 ... exercised for their protection’: Wood Inquiry Report (2008), Vol 1, p.185.
NSW Budget Estimates, 3 September 2024, Transcript, p.32 (Paul O’Reilly).
We also exclude policies and practices for Pre-natal Reports, Homelessness Reports or Requests for Assistance (or other non-ROSH contacts).
‘Structured Decision Making®’ and ‘SDM®’ are registered trademarks of Evident Change – Structured Decision Making® (SDM) Model – Evident Change. These terms differ from the general concepts of ‘structured decision making’ and ‘SDM’.
Structured decision making was introduced by DCJ in response to Rec 9.1 from the Wood Inquiry Report (2008), Vol. 1, p.xv.
E.g.: SDM intake assessment assumes all reports that meet a statutory threshold will be responded to, which appears to be partly why the SDM Risk Assessment tool counts all reports as risk factors, whether substantiated or not (N2 and A2).
DCJ, Care Act Delegations Manual (Effective: 1 March 2023; Accessed: 15/07/2025).
Different processes may apply in some circumstances – e.g., where DCJ is required to quickly exercise emergency removal powers.
See, the SDM MRG website (accessed 12/08/2025).
See footnote 25 for DCJ’s data definition of ‘contactor type’.
SDM Screening and Response Priority Tool Manual (2017)(SDM SCRPT Manual), p.3; p.1 describes the threshold differently as ‘a child protection report’.
Such as age and residency: s 4 Care Act.
SDM SCRPT Manual (2017), p.9 – such as under tool Part 1A (age/residence not met) or Part 1B (no ROSH but DCJ response still required).
See footnote 25 for the data definition of ‘contactor type’.
Email from DCJ dated 15 October 2025, Item 1(a).
Example ‘capacity’-oriented languge in the Triage Policy is: ‘where CSC management will prioritise reports for a face-to-face child protection assessment, against available resources’ (p.2) and ‘those that cannot be allocated due to a lack of capacity’ (p.5). DOCS’s draft ‘Unallocated Case Closure Policy’ included a requirement that "Where the case is not immediately allocated or closed, the priority must be reconsidered prior to closure by following 1(a) to 1(d) above” (at p.3). This is important as circumstances can change in 28 days, but this not required by current policy.
ChildStory Knowledge Article, Close a Triage, p.1.
Caseworkers can access the ChildStory Knowledge Article Close a Triage for instructions on how to record a ‘dual decision’ before submitting the record to a Manager Casework for approval. When a Manager Casework submits the record for approval themselves, the record is automatically approved. A Safeguard Practice intranet article also exists for guidance on conducting peer reviews and WAMs.
Email from DCJ dated 27 August 2025, Response to 1(c).
The Assessing Safety and Risk Mandate states ‘immediate’ is within the next 48-72 hours (p.5).
SDM Safety, Risk and Risk Reassessment Manual (2012)(SDM SARA Manual (2012)), p.28.
The safety plan is not the case plan. A safety plan permits a child/young person to remain home during the course of the investigation/assessment/ongoing work: SDM SARA Manual (2012), p.34.
SDM SARA Manual (2012), pp.47-48.
SDM SARA Manual (2012), pp.38-46 – each index lists 10 factors. The N1-N10 neglect factors include current and prior screen-in reports; previously received child protection casework; number of children in the household; physical care consistent with need. The A1-A10 abuse factors include current and prior screen-in reports; prior injuries; presence of domestic or family violence; parental drug and alcohol misuse.
Assessing Safety and Risk Mandate, p.17.
Assessing Safety and Risk Mandate, p.16; ChildStory Knowledge Article, Complete Judgements & Outcomes. In ChildStory, a Judgement and Outcome (J&O) is automatically created based on issues identified in Helpline assessments. One J&O is created for each issue and each subject of the Helpline assessment and any issues that are selected as ‘not assessed’ transfer to subsequent field assessments when they are launched.
Assessing Safety and Risk Mandate, p.17.
Assessing Safety and Risk Mandate, p.16; ChildStory Knowledge Article, Complete Judgements & Outcomes; SDM SARA Manual (2012), p.3 definitions of ‘substantiation of significant harm’ and ‘substantiation of risk of significant harm’.
SDM SARA Manual (2012), p.48.
DCJ Website, The Family Action Plan for Change (accessed 14/7/2025).
Email from DCJ dated 30 September 2025, Item 2(i).
Assessing Safety and Risk Mandate, pp.18 and 20.
The matrix ‘Preservation service matrix for referral pathways for family preservation programs where the safety decision is safe or safe with plan’ is hyperlinked in the Assessing Safety and Risk Mandate (p.19). Despite the title, it applies to all SDM outcomes.
Count reflects ROSH reports only. DCJ advised that CSCs may occasionally ‘re-classify’ a ROSH report as a non-ROSH Request for Assistance (RFA) and close it under ‘No Capacity to Allocate’ (NCTA) as NCTA also applies where a RFA is made and fulfilled without ongoing casework – but these instances are likely rare.
A further 2.7% (6,264) of ROSH reports were allocated to a caseworker but closed with the label ‘Current Competing Priorities’: Item 3, Table 2.
NSW Budget Estimates, 10 March 2025, Families and Communities, Disability Inclusion (Washington), Answers to Questions On Notice, Attachment QTON#15, 25, 26 Table; and DCJ Annual Statistical Report 2023-24, Supporting children in statutory protection (Number of ROSH reports). While the table in Attachment QTON#15 describes the data as for ‘competing priorities’, it relates to ‘No Capacity to Allocate’ (NCTA) decisions as the 2022-23 data quoted in the table corresponds to NCTA data provided to the NSW Ombudsman. A further 2% (5,243) of ROSH reports were allocated to a caseworker but closed with the label ‘Current Competing Priorities’: Email from DCJ dated 15 October 2025, Item 1(c).
Email from DCJ dated 16 January 2026. DCJ noted: ‘That among the 159,180 ROSH reports with triage decision to close with closure reason "No capacity to allocate/competing priorities", 134,036 (84.2%) were from children who had no other ROSH report with triage decision to allocate during the FY. However, 25,144 (15.8%) were from children who had another ROSH report with triage decision to allocate during the FY. Depending on when that other report was made, it is possible that some of these reports were actioned as part of an action to a related ROSH report. This means that the proportion of reports closed at triage as "no capacity to allocate" could be over-estimated. Nevertheless, taking this into consideration still shows that of the 244,710 reports, at least 54.8% (134,036/244,710) received no response from DCJ.’
The survey asked whether they were satisfied on the balance or probabilities. The survey did not define what it meant to be ‘satisfied’. It should be noted that the duty to make a determination in s 30(a) does not expressly use the language of ‘satisfaction’.
‘Often’ was not defined for this question.
That is, 21% ‘seen’ in 2010-11 to 18.5% ‘seen’ in 2023-24.
Triage Assessment Mandate, p.9.
Our survey of DCJ staff showed that in practice this timeframe is generally understood by most managers, although 18% (22 of 123) thought the 28-day period started from the time the report reaches the CSC, including 23% of CSC MCWs (15 of 64) and 50% of Executive (2 of 4). All hub staff correctly identified Helpline receipt. Managers were Managers Casework (MCW), Manager Client Services (MCS) and Executive (Director Community Services and above) (DCS & up) – excludes 6 caseworkers, 1 casework specialist and 1 ‘not stated’.
The NSW Auditor General reported in 2024 that 23.4% of ‘child protection reports’ received at the Helpline were not assessed and transferred to CSCs within 28 days (2022-23). The difference in those numbers from ours may be due to ROSH reports being only one type of child protection (that is, concerns) reports.
Inquest into the death of Z (2021), [138]-[139].
That said, if such assessment or determination has been made, and in circumstances where DCJ says it does not have sufficient resources even to respond to all children and young people whom it assesses may be at ROSH, then these cases should be closed immediately (on the basis of ‘assessed/determined not at ROSH’) rather than remaining open to be closed later based on a time-based cut-off rule.
A senior officer at DCJ put to us that, while the Triage Policy allows closure within 28 days, it should not be characterised as a specific rule ‘designed’ for that purpose: Email from DCJ dated 11 November 2025. They said the intent of the policy was to promote timely triage decisions and avoid reports remaining ‘open in name only’, and that it was not intended to create a rule designed to close reports that had never been the subject of decision at a WAM or other decision-making process. We saw no information to support this narrower ‘intent’, for example, in communications to staff about the policy. The policy’s express terms and structure – closure at 28 days unless both ‘exceptional circumstances’ apply and a Director, Community Services, approves an extension – operates clearly as a rule, with a limited exception. DCJ’s own reference to ‘closure timeframes’ (DCJ, Prioritisation, Triage and Allocation Policy Review – Discussion Paper (2024), Discussion Question 10) and the mandatory nature of the rule reinforce how it is understood and applied in practice, including as a basis for closing unallocated ROSH reports. Furthermore, whatever the intention may have been, as noted above, closure under the policy can and does occur in some cases without any decision having been made, and in some cases without the report having even been considered at a WAM or other decision-making meeting: Triage Policy, p.9.
Staff include Manager Helpline and Team Leader, Helpline and above: DCJ, Care Act Delegations Manual, Delegation 22.10 (Effective: 1 March 2023; Accessed: 15/07/2025).
Wood Inquiry Report (2008), Vol 1, p.185.
DCJ data since 2010-11 shows that, on average, around 47% of concerns reports are screened out as non-ROSH and that this proportion has remained relatively stable over time (see Figure 2). DCJ has also advised that 83% (148,121 of 179,980) of non-ROSH reports in 2023-24 were made by “mandatory reporters” (Email from DCJ, 15 October 2025, Item 1(a)). Helpline staff told us that over-reporting can be driven by factors such as agency policies that require reporting regardless of risk, failure to consult Child Wellbeing Units, multiple mandatory reports about a single event, and repeated reports when a child repeatedly leaves and returns to OOHC (Helpline staff meeting, 9 May 2025). That said, concerns have also been raised about under-reporting. This includes because mandatory reporters may consider there would be no use reporting (or re-reporting) if no action has been taken by DCJ on previous reports. In the Inquest into the death of Jade, the Coroner noted evidence given by a provider to the effect that DCJ had been discouraging service providers, who are mandatory reporters, from making ROSH reports in respect of families they were already engaged with. The coroner referred to agencies having been “educated” not to comply with s 27 of the Care Act: [158].
E.g., Inquest into the death of Harmony (3 March 2025), at [60]; Inquest into the death of Baby Q (2024), [169]-[171]; Inquest into the death of AP (2020), [24]-[26], [31] & [32].
Our DCJ staff survey showed that 65% (53 of 82) of respondents believed they were not permitted to decide a child is not at ROSH based only on desktop assessment and investigation (Question 7). DCJ’s working and operational consultation groups collectively supported a recommendation that ROSH can only be determined after speaking with a child and their family: (Final draft) Working Group Report: Child Protection Decision-Making Operating Model (October 2023), p.35. Elsewhere, DCJ and its managers advised the auditor in 2024 that a ROSH determination can only be made after a child has had two visits by a caseworker – one for a safety assessment, the other for a risk assessment: Oversight Report (2024), p.5.
Desktop triage work can include examination of the report and records held in relation to the child or young person as well as information gathering activities such as contacting a family, support services, schools, health providers, police or other services. It can also include making of referrals and use of mandatory reporter feedback.
E.g., words such as ‘[CYP] not at risk’ and ‘Case to be closed, no further concerns’. A further example is, after describing factors that point to safety, ‘this matter is not appropriate for a statutory response’.
Inquest into the death of Harmony (3 March 2025), at [69].
While these do not confer any right or entitlement on any person that is enforceable at law, they are intended to give guidance and direction in the administration of the Act: Care Act, s 7. Also, an interpretation of the Act that would promote its purposes or objects shall be preferred over one that does not: Interpretation Act 1987 (NSW), s 33.
This is often referred to as the ‘paramountcy principle’: s 9(1) Care Act.
DCJ District and CSC Performance Tables June 2023, p.2, Theme 1.2, NSW total.
Inquest into the death of Z (2021), at [5]; and 2 years later in the Inquest into the death of BW (2023), at [132].
Inquest into the death of Harmony (3 March 2025), at [69].
Issues with the quality of DCJ’s record keeping is not new. Before ChildStory was introduced in November 2017, researchers advised DCJ to better define its closure reasons, particularly for ‘competing priorities’, so decisions could be better understood: Analytical Report (2015), p.82.
Similarly, although various policy documents refer to the INOCAP decision, at the time this investigation commenced, there was no specific INOCAP decision record on ChildStory.
The Children and Young Persons (Care and Protection) Miscellaneous Amendments Bill 2000 introduced a requirement to record actions taken in response to reports (which, along with the requirement to record reports, did not commence until 2024). The Second Reading Speech stated: ‘Record keeping and the documenting of why and when decisions are taken is a vital role in any system of child protection. It is good practice. This new provision does not alter this. What it does do is make this best practice an express and clear requirement for all departmental officers. In doing this, it recognises that a lack of record keeping is one of many areas in which the department is seeking to improve the way in which it serves our children and young people’: NSW, Parliamentary Debates, Legislative Assembly, 21 June 2000, 7372 (Mrs Lo Po’).
It is worth reiterating here our repeated concerns about DCJ’s failure to follow up, record and report outcomes for children and young people reported at ROSH whose case is closed for NCTA after referral to a service, including even just to know whether that service was, in fact, provided: see e.g., Review of the NSW Child Protection System: Are things improving? (April 2014), p.8; Rec 2(f). A year later, DCJ’s progress update on this issue referred to its response to another recommendation that ChildStory was being developed to ‘collect and share relevant important data’ – FACS April 2015 Progress Report. We highlighted this lack of data again in 2021 – see Submission No. 57 (January 2021) to NSW Parliamentary Inquiry, p.3. And made 2 more recommendations about it in our biennial child death reports in 2021 (Rec 6) and 2023 (p.189, Rec 2). DCJ advised the intent of our 2023 recommendation would be considered in its Prioritisation, Triage and Allocation Policy Review.
Although it is noted that in 2021 FACSIAR found DCJ could be seeing 44,900 more children (40% of CYP reported at ROSH) within existing resources by mostly making some adjustments to time allocation and utilisation uplift: FACSIAR, DCJ Casework Study (2021).
Inquest into the death of BW (2023), at [132]
See s 24 and 27 and this language in s 30.
Under the definition of ‘at ROSH’, the ‘concerns’ that exits must be ‘because of’ the presence, to a significant extent, of one of more of a listed set of circumstances. Accordingly, where the Act refers to there being reasonable grounds to suspect a child or young person is at ROSH, this could be read as referencing the need for reasonable grounds to suspect one or more of those circumstances, rather than reasonable grounds to suspect there are concerns. However, that could be much clearer, in either the legislation or the policy (e.g., by stating explicitly that a person is to make a report if they have reasonable grounds to suspect that one or more of those circumstances exist, and listing those circumstances.)
Compare this with the corresponding Victorian legislation, for example, which defines INOCAP (in a manner somewhat similar to the NSW definition of ‘at ROSH’) but leaves undefined the term used to trigger reporting to the department – which is that someone has a ‘significant concern for a child’: see n 30.
2022-23 and 2023-24 data taken from DCJ Annual Statistical Report 2023-24. At least 14% of allocated ROSH reports (11,063 of 81,582) were closed during 2022-23 for the reason ‘Assessments indicate Child/Young Person is not in need of Care and Protection’ (i.e., the Safety Assessment outcome was ‘Safe’ and the Risk Assessment was ‘low’ or ’moderate’ risk): DCJ data provided to NSW Ombudsman.
E.g.,: NSW Children’s Court: Re Malakhai [2022] NSWChC 6, where since June 2020, DCJ received 44 reports about Malakhai, then another 11 after birth; Finn, Lincoln, Marina and Blake Hughes [2022] NSWChC 4, where several reports were received about 4 children in interim care.
See cases listed at n 14.
E.g.,: NSW Ombudsman biennial reports of the deaths of children in 2021 (Rec 6) and 2023 (p.189, Rec 2).
Calculated by dividing the YTD Goal for the benchmark ‘1. CYP at ROSH that have received a face to face assessment’ (37,428) by the NSW total for number of funded caseworkers (1,905) = 19.64, rounded to 20. A 2011 Ernst & Young study commissioned by DCJ calculated caseloads of 11.5 to 23.1 per year (excluding court cases) depending on experience. The study did not consider efficiencies possibly generated by SDM and/or ChildStory: Ernst & Young, Child Protection Caseload Review (2011).
FACSIAR, DCJ Casework Study (2021), p.188. A similar point was made by the NSW Audit Office in Oversight Report (2024), p.35.
DCJ’s earlier Decisions Count (2019) study found that caseworkers often struggled to balance achieving targets with building relationships with families, leading to a perception that children were treated as 'throughput' to meet productivity goals (p.75).
Others included: lower-risk families (38%; 15 of 39); reports otherwise viewed as likely to be faster or easier to deal with (28%; 11 of 39); children not yet counted in KPI period (10%; 4 of 39).
In 2021, NSW Aboriginal and Torres Strait Islander households averaged 3.1 people vs 2.6 for NSW overall: for NSW Aboriginal households, see ABS, 2021 Census Aboriginal and/or Torres Strait Islander People QuickStats (Area code 1RNSW), Households; for NSW overall, see ABS, Snapshot of NSW (2021), Household and families. Aboriginal and Torres Strait Islander women also have higher fertility than the population average, pointing to more children per household (Total Fertility rate 2.17 vs 1.50 in 2023): see ABS, Births, Australia (2023), Summary statistics.
DCJ has also noted that, while seeing children in larger families might improve performance against the ‘seeing more children’ metric, larger families can impact casework hours if intervention is required, given complexities of sibling groups: FACSIAR, DCJ Casework Study (2021), p.151.
Transfer of a child or family between teams, CSCs, Interstate OOHC and JCPRP Mandate (18 June 2024). Although if the child is in OOHC, the CSC may be assigned case plan tasks relating to their care and support.
Inquest into the death of Jacob (2025), at [434].
FACSIAR, DCJ Casework Study (2021), p.130 – defined to mean gathering information from reporter, call reporter/school/hospital/police etc, conduct chronological file review, prepare case file notes (p.214).
Briefing Note – Closure of the Helpline Advanced Screening Program (11 March 2024), pp.1-2; DCJ, DRAFT Rapid review: Key decisions in the child protection system – legislative basis and policy, tool and process alignment (March 2023), p.17. Lack of legal alignment was also cited in DCJ’s Child Deaths 2023 Annual Report (p.59)
DCJ Staff Survey Results, p.13.
The most common survey response about whether ICDs occur before closure of reports older than 28 days was sometimes (35%, 46 of 131), followed closely by rarely (31%, 41 of 131).
FACSIAR, DCJ Casework Study (2021), p.137
See n 99.
NSW Parliament, Legislative Council, Budget Estimates Hearings, 3 September 2024, p.58 (Anne Campbell).
DCJ, Family Preservation Annual Snapshot 2022-23, pp.2 and 4 – data for ‘post-exit’.
Interviewee #1, Re-report rates by closure type
Example evaluations that identify issues with DCJ’s closure practices that affect other services: FCS Program Evaluation interim Reporthttps://familyconnectsupport.dcj.nsw.gov.au/documents/family-connect-and-support-fcs-evaluation-interim-report-july-2023.pdf (July 2023), p.39 - 41; TEI Program Evaluation – Interim Report (Final) (30 May 2024), p.9; SHS Program Evaluation – Final Report (October 2023), p.73; HYAPs Program Evaluation: Final Report (June 2020), pp.18, 61, 64; CWU Evaluation Report (June 2014), pp.10, 20, 43, 81 and 88. ↑
Oversight of the Child Protection System (2024), p.6.
Ernst & Young, CWU Evaluation Report (June 2014), p.88.
Although it may not be possible to ascertain the precise causal link between re-reports and earlier triage failures, service provider staff also told us re-reports are made to prompt DCJ into action where a family’s needs and risks are too complex for the service to address.
DCJ, What we heard: Key themes from the Family Preservation Recommissioning Stakeholder Workshops (September 2023), p.9.
DCJ, What We Heard: Aboriginal families, communities and service providers and their experiences of family preservation in NSW (March 2024), p.13.
DCJ data provided to the NSW Ombudsman.
DCJ internal document, Item 5, 5A.06, p.2.
Final FCS Evaluation Report (October 2024), p.24.
Care Act, s 17(2).
Conversely, there may be times when recent DCJ involvement can lead to families being considered or assessed as too high risk for referral or support by universal or targeted services (eg. playgroups, parenting programs, some family support services (TEI)) which may need to consider safety of other families, staff etc in context of violence or AVOs – they may not be able to offer the intensity of support/staffing to manage these risks. In other cases, DCJ not making a referral can be taken as an indicator that the family is suitable for lower-level services – which may or may not be correct.
Child Protection Helpline Practice Guidance, Duplicate Reports (16 August 2018), p 1.
DCJ Website, ‘The NSW Structured Decision Making Framework’ (accessed 27/08/2025).
Wood Inquiry Report (2008), Vol. 1, pp.331-333 (at 9.32).
NSW Ombudsman, Submission to Wood Special Inquiry – Part 1: Children’s Court (10 March 2008), p 11.
Wood Inquiry Report (2008), Vol. 1, Rec 9.1: ‘DoCS should test the use of Structured Decision Making tools at the Helpline and at CSCs in relation to assessments and interventions including restoration.’
SDM SCRPT monitoring and review included case readings (performed by Evident Change). SDM Safety Assessment and SDM Risk Assessment underwent a 10-week field test in 8 CSCs, and a 12-week trial with a JIRT team. There was also a 3-month trial of the SDM Restoration Assessment Tool.
Minister for Families and Communities, Answers to Question on Notice 1654, 24 January 2024, p.2012 (answers 5 & 6).
Ombudsman File Note – DoCS briefing on Structured Decision Making – 17 March 2009.
(Final draft) Working Group Report: Child Protection Decision-Making Operating Model (October 2023), p.19. Evident Change said: ‘the SDM risk assessment is the wrong tool to help determine child ‘in need of care and protection‘: DCJ, ‘Overview RA Working Group – Sept 2022’, p.5.
When a decision maker is required by statute to consider a claim or other mandatory criteria, the decision maker must engage in an active intellectual process directed a that claim or criteria: Carrascalao v Minister for Immigration and Border Protection (2017) FCAFC 107at [45]. A “decision” involves “reaching a conclusion on a matter as a result of a mental process having been engaged in”: Pintarich v Deputy Commissioner of Taxation (2018) FCAFC 79 at [141] and [143], quoting Semunigus v Minister for Immigration and Multicultural Affairs [1999] FCA 422 at [19].
SDM SARA Manual (2012), p.48.
DCJ, Decisions Count (2019), p.117.
We cannot discount that some may have been referring to the SCRPT tool rather than the Risk Assessment tool as the ‘decision maker’ for ROSH determinations (although this is unlikely given there was only one Helpline staff member among the 47 who responded that an SDM tool usually makes the ROSH determination).
DCJ, (Final draft) Working Group Report: Child Protection Decision-Making Operating Model (October 2023), pp.18-19.
The INCOCAP matrix did not appear in the initial July 2011 version of the SARA Manual provided to the NSW Ombudsman. However, DCJ updated the manual in August 2012 to include the matrix. In October 2012, the NSW Ombudsman received a summary document of the changes, which noted definitions for ‘substantiation’ and ‘in need of care and protection’ had been added, but did not refer to a matrix.
DCJ, (Final draft) Working Group Report: Child Protection Decision-Making Operating Model (October 2023), p.9.
Email from DCJ dated 19 September 2024.
DCJ, (Final draft) Working Group Report: Child Protection Decision-Making Operating Model (October 2023), p.37: ‘The explicit linking of the SDM tools to the INOC&P determination is currently DCJs only way to demonstrate that this determination has been made. There is no INOC&P record in ChildStory (this record existed in the previous IT system KiDS but was not carried over to the build of ChildStory).’ (emphasis added)
Email from DCJ dated 19 September 2024.
ChildStory Case Closure Reasons Guide – see the ‘*’ note.
Assessing Safety and Risk Mandate, p 18.
See n 143.
DCJ, Child Deaths 2011 Annual Report (2012), p.59.
Briefing Note to the Minister re Advice and update on Better Decisions for Children Project (March 2023), p.2.
Children’s Research Centre, An Introduction to the Structured Decision Making (SDM) System (PowerPoint Slide, March 2009), p.51.
Minister for Families and Communities, Answer to Question on Notice 1654, 24 January 2024, p.2011.
See answers to Q4 and Q7 above. The 4 low, moderate, high and very high risk categories also remained in the SARA Manual (2012).
This followed the Limebridge Australia Diagnostic Final Report (2018). See also, Children’s Research Centre [Evident Change] Helpline’s use of the SCRPT assessments: Case reading results (2018)
DCJ, A paper to seek endorsement of the SDM Screening assessment as part of the Better Decisions for Children Project (June 2023), p.11.
Recommendations were made in 2015 by the Analytical Report (2015), p.13; in 2017 by the NSW Legislative Council General Purpose Standing Committee No 2 Inquiry into Child Protection (Rec 4); and in 2019 by the FIC Review (2019), pp.219-220 (Rec 56).
Analytical Report (2015), p.13. The Tune Review (2015) also recommended DCJ redesign its intake, assessment and system architecture.
Royal Commission into Violence, Abuse, Neglect and Exploitation of People with Disability (Disability Royal Commission), Public Hearing 8 (November 2020), Statement of Michael Coutts-Trotter dated 28 October 2020, paragraphs 106-109.
Australian Institute of Family Studies (AIFS), Risk assessment instruments in child protection (Resource Sheet, June 2024).
NSW Legislative Council General Purpose Standing Committee No 2 Inquiry into Child Protection, Final Report (2017), Rec 4, para 3.77-3.131.
FIC Review (2019), pp.219-220.
OCG, Special Report under section 139(2) of the Children’s Guardian Act 2019 – Family is Culture Review (March 2022), p.7.
Disability Royal Commission, Public Hearing 8 (Day 3) (23-27 November 2020). A final hearing report was not released, but Counsel Assisting proposed findings that the SDM tools used by each jurisdiction present a risk of contributing to the overrepresentation of First Nations people with disability in child protection systems (p.90). This was accepted by the State of NSW: Submissions in reply by the State of NSW, p.34 at [45]. ↑
Submissions in reply by the State of NSW, p.35 at [48]-[49]; Submissions of Counsel Assisting, [276]-[279].
Jorge Branco, “Queensland child safety department ditches ‘racially biased’ screening tool”, 9News, 1 December 2022 (accessed 22 July 2025). For the study, see Brian Q Jenkins, Clare Tilbury, ‘An evaluation of the racial equity of the actuarial Family Risk assessment instrument used in Queensland, Australia’, Children and Youth Services Review 164 (2024)(accessed 22 July 2025).
DCJ, Decisions Count (2019), pp.97; 116-119.
NSW Auditor-General, Performance Audit – Their Futures Matter (July 2020), p.24.
Briefing Note to the Minister – Advice and update on Better Decisions for Children Project (March 2023), p.3.
Letter from DCJ dated 21 June 2021.
The new machinery of government: using machine technology in administrative decision-making (2021), pp.65-68.
DCJ response to NSW Ombudsman dated 7 December 2021, Question 13.
Email from FACS dated 11 July 2022, Attachment ‘Tab F – Item 4 – Structured Decision Making QSR’, p.3.
DCJ, (Final draft) Working Group Report: Child Protection Decision-Making Operating Model (October 2023), p.4. This decision immediately followed a workshop with mainly Aboriginal DCJ staff from various divisions to consider options for interim changes to attend to cultural bias across child protection practice. Staff provided strong feedback that the use of the risk assessment tool by DCJ should be discontinued.
DCJ, ‘An interim approach to assessing risk in child protection’ (August 2024), p.8. It notes the FBAT differs from the SDM Risk Assessment tool as scoring removed; no outcomes; strengths opportunities to consider strengths, protection and network; 13 consolidated questions, linear flat assessments (no weighting); questions focused on how adult behaviour impacts child; professional judgment tool; critical analysis of information to understand the child’s experience; and links to s 23.
DCJ, (Final draft) Working Group Report: Child Protection Decision-Making Operating Model (October 2023), p.18.
DCJ, (Final draft) Working Group Report: Child Protection Decision-Making Operating Model (October 2023), pp.3-4.
This recommendation should not be read as condoning the continued practice of making such decisions. A core finding of this investigation is that closing a ROSH report for reasons other than as a result of DCJ having made a decision, in respect of the child or young person who is the subject of the report, under s 30(a) or s 30(b) of the Care Act, is not consistent with the Care Act and is contrary to DCJ's statutory responsibilities.
However, as discussed in our Investigation (Part 1 – Findings) Report, this practice is longstanding and deeply entrenched. DCJ has told us that – even accepting our findings – it is not possible or desirable for it to immediately adopt a practice of never closing a ROSH report except following a requisite s 30 decision. Attempting to do so in isolation from holistic reform (as contemplated by recommendation 1 above) would, in DCJ's submission, result in a large backlog of ROSH reports remaining technically 'open' in ChildStory with no real effect on any decisions or actions being made about them and/or result in the system-wide diversion of resources with the result that DCJ would fail to meet its responsibilities elsewhere.
Having regard to this submission, we have not made a recommendation that DCJ must immediately cease its practice of closing ROSH reports without a decision having been made under s 30. However, we note for so long as this practice continues, DCJ will be failing its responsibilities, and we urge acceptance and implementation of recommendation 1 of this report as soon as possible.
Previously known as ‘SAS1 completed – closed competing priorities’.
Care Act, s 154(2) – a child is at ROSH if someone other than the Secretary or an authorised carer is providing OOHC to the child; and Children’s Guardian Act 2019, s 8ZA – a child under 18 years remains in specialised substitute residential care (SSRC) for longer than permitted.
Care Act, ss 43(1), 44(1) and s 233(1)(warrant).
The prequisities for this order are in ss 91A-91I.
SDM SARA Manual (2012), p.3.
DCJ, A paper to seek endorsement of the SDM screening assessment as part of the Better Decisions for Children Project (June 2023)(SDM Screening Assessment Paper (June 2023)), p.11.
See Sections 4.1 and 8.2 above, and Answers to Questions on Notice
SDM SCRPT Manual (2017), p.63.
DCJ’s example in the SDM Screening Assessment Paper is that a child may have been sexually assaulted by someone who no longer has access to them. This report would potentially screen as requiring a <10-day response. Whereas a report about a child who is in contact with a known sex offender would potentially screen as requiring a <24-hour response (p.12, n 29).
DCJ, SDM Screening Assessment Paper (June 2023), p.12.
FCS Evaluation Interim Report (July 2023), p.40.
DCJ, SDM Screening Assessment Paper (June 2023), p.20.
DCJ, (Final draft) Working Group Report: Child Protection Decision-Making Operating Model (October 2023), p.29.
Care Act, s 23(1)(f); Children’s Guardian Act 2019, s 8ZA.
The triage officer at the CSC closed the file on the basis that the 5-year-olds were under the age of criminal responsibility (14973707).
Closed on the basis the report was more than 28 days old, had limited information and the parent was aware of the incident (14927104).
DCJ, ‘Workshop Paper – SDM and pre-reading pack’ (August 2023), pp.3 and 13; (Final draft) Working Group Report: Child Protection Decision-Making Operating Model (October 2023), pp. 15 and 19; ‘SDM risk assessment options paper for the EGG (March 2023)’, p.8.
E.g.: a DCJ Secretary gave evidence to the Disability Royal Commission that the SDM Risk Assessment tool “is used by Departmental caseworkers to assess the immediate safety and future risk of significant harm to a child or young person” (emphasis added): Public Hearing 8 (November 2020), Statement of Michael Coutts-Trotter dated 28 October 2020, paragraph 63.
DCJ, ‘SDM risk assessment options paper for the EGG (March 2023)’, p.6; SDM SARA Manual (2012), p.47.
The two overrides available are: (1) policy override – caseworkers can increase the risk level to the top level (‘very high’) where required by policy; (2) discretionary override – caseworkers can increase the risk level by one level (e.g. from low to moderate) where the caseworker believes the determined risk level is too low: SDM SARA Manual (2012), p.50. In contrast, later, the discretionary override for the SDM Risk Reassessment allows caseworkers to increase or decrease the risk level by one level: SDM SARA Manual (2012), p.60.
J&T v Department of Communities and Justice [2023] NSWDC 78, at [66].
DCJ, ‘Workshop Paper – SDM and pre-reading pack’ (August 2023), p.3.
DCJ, (Final draft) Working Group Report: Child Protection Decision-Making Operating Model (October 2023), p.18.
DCJ, Issues Paper Overview of current issues SARA and Alternate Assessment (January 2021), p.18.
DCJ, (Final draft) Working Group Report: Child Protection Decision-Making Operating Model (October 2023), pp.16-17; Issues Paper – Child Protection Assessment Tools (January 2021), p.18.
DCJ, Issues Paper – Overview of current issues SARA and Alternate Assessment (January 2021), p.18.
DCJ, ‘Evident Change, SDM Risk Validation Stakeholder Group Background’ (September 2022), p.6.
DCJ, Minutes Executive Governance Group (24 April 2023), p.3.
When staff close a report using the closure reason ‘No Capacity to Allocate’, they can enter factual information and a description of their reasons into free text fields in ChildStory. DCJ provided: Record/identifier; District/CSC that closed the record; Summary; Narrative; Allocation supporting commentary; Transfer supporting commentary; Closure supporting commentary.
To ascertain whether referrals/consultations were considered or occurred, we checked both linked REQ records and the ChildStory Timeline for REQ records around the same time.
As opposed to at the time the report was made or when the reported event occurred.
That is, at the time the determination was made, current concerns did not exist for the safety, welfare or wellbeing of the child or young person due to the presence, to a significant extent, of any one or more of the circumstances listed in s 23.
Community Welfare Act 1987 (NSW), s 36. Power of delegation: s 5.
The Hon James Wood, Report of the Special Commission of Inquiry into Child Protection Services in NSW (2008) (Wood Inquiry Report (2008)).
We acknowledge the traditional custodians of the land on which we work and pay our respects to all Elders past and present, and to the children of today who are the Elders of the future.
Artist: Jasmine Sarin, a proud Kamilaroi and Jerrinja woman.