Child death reviews
The death of a child is a tragedy in any circumstance, let alone when the cause is abuse. Literature on child deaths from abuse says that it is difficult to predict and impossible to prevent all deaths. However, putting support around , and at risk can reduce the likelihood of serious harm and death. Thorough reviews of what led to the death of tamariki, involving multiple agencies and using a robust systems framework, are also opportunities to make changes to the system to prevent serious harm in the future.
Malachi died in November 2021. Between December 2021 and June 2025, Police and Oranga Tamariki completed 17 reviews into the deaths of 14 tamariki who died from abuse by a person meant to be caring for them. These 17 reviews are opportunities to help close the critical gaps identified in the Poutasi report. They are 17 missed opportunities to deliver systemic change.
While not every review was robust or followed a systemic framework and not every gap was evident in every review, we found that, before their death:
- one child was moved by whānau to an unsafe environment after their parents were imprisoned and processes were not followed by Oranga Tamariki, which also did not sight the child or assess their needs
- three tamariki had had reports of concern from health professionals where their risk of future harm may not have been adequately addressed
- the CPP was not followed by Police for one child
- one child had a “significant but unknown number of reports of concern” incorrectly entered by Oranga Tamariki
- two tamariki had caregivers that were not adequately checked by Oranga Tamariki
- for four tamariki, Oranga Tamariki did not effectively sight them, where practice guidance indicated they should have
- for at least four tamariki who were sighted by either Police or Oranga Tamariki, the agencies did not share information to understand the overall, and potentially increasing, risks to the child.56
Notwithstanding the limitations of current child death review processes, areas needing improvement are well documented. As we outline in this review, action is needed, and there is a long way to go before the gaps in the safety net are closed to prevent child abuse occurring.
The way child death reviews are conducted also reflects the critical gaps in child safety. For example, across the 17 child death reviews, we saw the following.
- Reviews are not child-centred so do not consider what the child needed but instead look at how the agency responded.
- Agencies are not collaborating on child death reviews to learn from each other and make changes that improve the system. Reviews are focused on the agency’s requirements rather than an assessment of the risk to the child and putting what the child needed to be kept safe at the centre of analysis.
- While some child death reviews include information from other agencies, suggesting information was shared to help inform the death review, others do not. We only saw limited evidence of information sharing between the Police and Oranga Tamariki to inform the child death reviews they undertake.
- Findings and recommendations of child death reviews do not appear to be shared with relevant agencies. This means agencies may not be aware that findings or recommendations have been made for them to action.
- Even when agencies are deciding whether to review the deaths of , some tamariki seem to be invisible. This is evident in decisions by Oranga Tamariki not to undertake reviews of some deaths, even where the PFVDR appears to show a clear opportunity for Oranga Tamariki to address systemic issues when a child dies.
In completing death reviews, kaimahi follow the policies and guidance set by their agency. However, the actions required by those policies do not necessarily equate to the actions that are necessary to achieve systemic change. For example, Police advised that PFVDRs are not specifically designed to be child-centred but instead to assess Police’s response to family violence concerns. Police told us that assessing the broader needs of children is not within the subject matter expertise or statutory function of Police.
This is just one example of how the policies do not always support systemic change. If policies are not designed to help bring about system change, we cannot be surprised that death reviews are not achieving this.
Oranga Tamariki does not undertake reviews of all deaths of and known to it. Decisions on whether or not to undertake a review are made by the Decision-Making Forum (DMF), a group of Oranga Tamariki senior leaders. Sometimes, particularly if the DMF considers Oranga Tamariki had little involvement with the child and their , a decision is made not to undertake a review.
Tamariki aged under 1 were the largest group to die from abuse by a person who was meant to be taking care of them. Considering their short lifetime and the relatively fewer opportunities for others such as teachers and the public to suspect abuse, the decision not to review deaths where there was little or no previous involvement may be too narrow. We question how much involvement, in a life spanning less than 1 year, would be required for it not be considered “little”. Oranga Tamariki is missing opportunities to identify, learn from and address risks, especially for babies.
We received information from Oranga Tamariki that included six deaths where its DMF had decided “a review is not required due to little or no previous involvement” with Oranga Tamariki. However, information also provided by Oranga Tamariki identifies that five of these six tamariki or one of their siblings had had previous involvement with Oranga Tamariki.
In determining whether to undertake a child death review, Oranga Tamariki should give consideration to the wider risks the child faced, including their surrounding environment and notifications about siblings or other tamariki and rangatahi in the household.
PFVDRs for these same tamariki identify family violence notifications. Where Police attend family violence incidents, these are referred to a family violence multi-agency table. The table assesses whether a report of concern is required. When no action is required, this information is still passed to Oranga Tamariki to record, either as a contact record or a no further action report of concern. Either way, Oranga Tamariki should have a record of the incident or incidents. In cases such as these, an Oranga Tamariki review could consider whether it should have assessed the information in the way it did or if it missed an opportunity to intervene and provide support.
Where death reviews are completed by Oranga Tamariki, responses often do not address root causes
We asked Oranga Tamariki whether any recommendations, practice and/or policy changes were implemented as a result of the death reviews it had undertaken and whether any analysis had been done on the impact of any changes made.
In response, it provided a summary of changes made to respond to thematic insights. These included changes to how it records reports of concern at the NCC. It told us thematic insights had also informed updates to the CPP – but not what those updates were. It told us that insights had been used to embed the Practice Approach and Practice Systems – but again not what the changes were. It told us insights had informed the use of Practice Notes from the Chief Social Worker – but not what those Practice Notes were in relation to. It did not tell us whether it knew what impact these changes are having.
Oranga Tamariki further told us that the findings from reviews, reports and practice opinions are consistently shared with site and regional leaders, actions are agreed, including practice-focused discussions and planning with sites, and they are supported by the Office of the Chief Social Worker/Professional Practice regional and residential quality practice teams.
While Oranga Tamariki tells us it is applying learnings from reviews, as we have highlighted in this review, the critical gaps remain unaddressed. The problem may be that the learnings Oranga Tamariki is applying are not what is needed.
Our 2024 review identified that actions that had been taken by agencies – including but not limited to Oranga Tamariki – were addressing symptoms but not the underlying causes of the child protection system not keeping tamariki safe. For example, after Malachi was killed, the Chief Social Worker issued a Practice Note to require that Oranga Tamariki social workers have at least 12 months’ practising experience before undertaking initial assessments. This was because the Chief Social Worker’s report identified that the social worker who assessed the report of concern for Malachi had less than 12 months’ experience and the initial assessment should have resulted in a decision to undertake a core assessment rather than a no further action decision. However, a less-experienced social worker was doing the initial assessment on their own because the site was under-resourced for the volume of work it had to respond to. As we have seen in our review this year, this problem persists.
The action of issuing the Practice Note may have been measurable, but it did nothing to address the root cause of the issue.
Unless the root causes are addressed, the critical gaps are unlikely to close.
As described above, agency death review processes are internally focused on the work of the agency rather than on understanding the systemic response that would have been required to prevent the death occurring. While there are independent, systemic, multi-agency mortality review processes such as those undertaken by the Family Violence Death Review Subject Matter Experts on behalf of the National Mortality Review Committee, they are not sufficiently resourced to ensure all deaths are reviewed.
Police told us it supports strengthened co-ordination and oversight of child death reviews to ensure system-wide learning and that PFVDRs are routinely provided to the Family Violence Death Review Subject Matter Experts to support this goal. However, restrictions on the use of the information collated for the purpose of reviews by the National Mortality Review Committee prevent the public release of findings. As a result, recommendations from individual reviews of the National Mortality Review Committee are collated at the national level and become the subject of substantive reports.
Recommendations from reviews are released by the National Mortality Review Committee to individual agencies. However, there is no accountability mechanism built into the review process and agencies are not required to report back to the National Mortality Review Committee on their progress in implementing recommendations made.
Oversight, potentially at ministerial level, of the findings and issues arising from national and agency death reviews could drive a more co-ordinated response across agencies to prioritise actions and address findings at a system level.
"There have been numerous reviews of previous cases of child abuse that have drawn very similar conclusions to mine. In reality, a number of the findings and recommendations of this review have largely been made before. Some recommendations have been implemented but fallen away as the spotlight moves on and the process defaults to what it knows; others have not been attempted because the environment was not seen as ready. As a society, we cannot continue to allow a cycle of abuse, review, outrage and distress – and then retreat from the difficult challenges. It is not acceptable we give up because it is too difficult. There must be sustained, determined and bold change. As difficulties arise in implementation, solutions focused on the protection of children and must be found. This requires regular monitoring and accountability for change."
Dame Karen Poutasi on child abuse and death reviews conducted prior to the death of Malachi Subecz
56 Police advised that this is based on the PVFDRs and, where cases were investigated under the CPP, additional information about interagency information sharing is held in the respective case files. It told us that, in this respect, the statement that “agencies did not share information to understand the overall, and potentially increasing, risks to the child” may not reflect the full extent of information sharing in CPP cases. Police did not explain why this information would not be reflected in the PFVDR if it exists.