By Boo Kok Chuon

The recent case involving the death of eight-month-old Trevor Neo Yu Hin is difficult to read.

According to the facts reported from the court proceedings, Trevor had been subjected to repeated acts of abuse during his extraordinarily short life. He was shaken. His breathing was deliberately obstructed. He was struck and pinched. On one occasion, an injury beneath his tongue was sufficiently concerning that his nanny insisted that he be taken to hospital. His grandmother subsequently noticed bruises and warned his mother against hitting him.

On 1 September 2024, Trevor was again crying. His mother repeatedly covered his face, nose and mouth. He was later placed face-down on a mattress. By approximately 11.30pm, he was motionless. He was pronounced dead at KK Women’s and Children’s Hospital at about 12.20am on 2 September 2024. His parents pleaded guilty on 12 October 2026 to offences arising from his death and the abuse preceding it, and are due to be sentenced on 21 October 2026.¹

Much can and undoubtedly will be said about the appropriate punishment. But there is another question which troubles me more.

How did an infant apparently suffer for months while moving through a world containing parents, grandparents, relatives, a nanny and healthcare professionals without the danger surrounding him being detected in time?

The question is especially important because Trevor could not have asked anyone for help. He could not tell a teacher what happened at home. He could not telephone the police. He could not approach a social worker. He could not describe where the bruises came from.

Whatever protection he had depended entirely upon adults noticing something was wrong and ensuring that what they noticed reached somebody capable of acting upon it.

We often say that it takes a village to raise a child. Trevor’s case makes me wonder whether the more important question is whether the village knows how to speak.

Megan Khung and the Dots We Failed to Connect

Singapore has recently confronted precisely this problem through another child’s death.

Megan Khung was four years old when she died in February 2020 after months of abuse. A later independent review found multiple lapses, shortcomings and miscommunication among agencies that handled her case. MSF subsequently acknowledged that there had been opportunities to pick up the abuse which might have prevented her death.²

The problem was therefore not merely whether somebody cared. It was whether the dots could be connected.

That review contributed to the development of Sensemaking Alerts for Frontline Engagement and Response, or SAFER. SAFER has been operational since the end of June 2026 and was publicly launched on 22 September 2026. It brings together relevant information from different government agencies and uses a risk algorithm to surface families whose circumstances warrant further professional review. In its first two months, about 600 cases were surfaced for review and 70 were flagged to the Protective Service and Child Protection Case Management agencies for strengthened support.³

This distinction is important.

A risk signal is not proof of child abuse. It is a reason to look.

And for a young child incapable of explaining what happens behind the doors of his home, sometimes somebody simply looking more closely may make all the difference.

SAFER Can Connect the Dots. But Who Creates Them?

Trevor’s case potentially exposes the problem immediately upstream of SAFER.

His nanny reportedly saw an injury beneath his tongue and insisted that he be taken to a doctor. Healthcare professionals documented local trauma. His grandmother later saw bruises and warned his mother to stop hitting him.⁴

Each was a dot.

But an information-sharing system can connect only information that enters the system in the first place.

This suggests that there may actually be two distinct problems in child protection. The first is information fragmentation: different institutions possess different pieces of relevant information but fail to assemble them into a coherent picture. SAFER is specifically designed to address this.

The second is information invisibility: somebody outside the institutional system observes something concerning, but that observation never enters the system at all.

No algorithm, however sophisticated, can connect a dot that was never created.

There may also have been other potential observation points which simply do not appear in the public account. During the first eight months of a child’s life, routine vaccinations, developmental assessments and other healthcare encounters ordinarily create repeated opportunities for professionals to observe both the child and the surrounding family environment. A mother may likewise encounter healthcare professionals through postnatal follow-up.

We should be careful not to reason backwards from silence. The agreed statement of facts in Trevor’s case was precisely that: an account of facts agreed between the parties for the purposes of the criminal proceedings, not a high-resolution reconstruction of every encounter during Trevor’s 269 days of life. The case was not tried. The publicly reported facts therefore do not tell us whether every routine appointment was attended, what was observed at any such appointment, or whether there was anything that reasonably ought to have raised concern. It would be wrong to infer a missed opportunity by any particular healthcare professional without that evidence.

But from a systems-design perspective, those encounters still matter. A vaccination appointment, developmental review, postnatal consultation or other routine contact may create an observation capable of being flagged where something sufficiently concerning is seen. Conversely, repeated non-attendance may, in an appropriate case, itself become relevant when combined with other risk indicators. Neither proves abuse. Each may simply be another dot which acquires meaning only when connected with others.

That raises a more difficult policy question: who should bear responsibility for creating those dots?

Borrowing an Idea from Suspicious Transaction Reporting

There may be something child protection can learn from an apparently unrelated area of law: anti-money laundering.

The suspicious transaction reporting framework does not proceed upon the assumption that every suspicious transaction constitutes money laundering. Suspicion triggers information, not conviction. That information can then be assessed alongside other information available to the authorities.

There may be value in considering a similarly calibrated architecture for child protection.

The object should not be to transform grandparents, neighbours, nannies, teachers and healthcare professionals into amateur detectives. Nor should every childhood bruise result in a government investigation.

Rather, Singapore could consider whether sufficiently serious child-protection concerns observed by prescribed persons should generate an obligation to place that concern into the system. The report would prove nothing. It would simply create a datapoint.

SAFER could then do precisely what it was designed to do: determine whether that datapoint becomes more significant when considered alongside information elsewhere.

The thresholds would require careful design. The relationship between the observer and child matters. So does the child’s age, the nature and recurrence of the injury, the explanation provided, other known family circumstances and the observer’s state of knowledge.

A bruise on an active six-year-old is one thing. Repeated unexplained injuries on an infant incapable of independent movement or explanation may be quite another.

Singapore already imposes specific reporting obligations in some settings. For example, regulation 39 of the Early Childhood Development Centres Regulations makes it an offence for specified preschool actors who have reasonable cause to suspect physical or sexual abuse to wilfully or unreasonably fail to report it to the Chief Licensing Officer. MOH has also stated that medical practitioners are legally obliged under section 424 of the Criminal Procedure Code to notify Police when they are aware of the commission, or intended commission, of abuse, unless there is reasonable excuse not to do so.⁵⁶

But those are sector-specific duties. What Trevor’s case invites us to consider is whether the architecture for creating serious child-protection datapoints should be broader and more coherent.

Most importantly, the legal consequences of silence should themselves be calibrated. Failing to escalate an ambiguous observation should not be equated with knowingly concealing serious abuse or consciously leaving a child exposed to a known danger.

The purpose is not to manufacture criminal liability. It is to make silence progressively more difficult as the warning signs become progressively harder to ignore.

But What Happens After We Detect Something?

There is an equally important danger in building a better detection system.

If every report produces an investigation, every investigation carries the threat of removing a child, and every struggling parent believes that asking for help may result in punishment, we may inadvertently encourage families to hide their difficulties.

Detection therefore requires a buffer.

And I came to this conclusion partly because of something I did as a father which still troubles me today.

During COVID, I went through a very difficult period of my life. I was in financial distress, eventually sold my first home and moved my family back into my parents’ flat.

Kiki was five.

One weekend morning, I was sleeping while she was playing with my father outside my room. I woke up furious at the noise, grabbed a clothes hanger, stormed outside and shouted at her to keep quiet.

I did not hit her. Instead, in anger, I struck the hanger against the wall so hard that it broke.

I frightened my five-year-old daughter. I have felt guilty about that moment ever since.

Looking back, however, I have sometimes wondered what intervention would actually have helped me that morning.

I do not think I needed a police officer.

I needed somebody to ask: “Are you okay?”

My father did not ask. Nobody did. We went to the Bird Park that afternoon. The atmosphere was awkward, but the outing itself helped somewhat.

Had somebody called me afterwards, perhaps a counsellor, and asked what was happening, I suspect I would simply have poured out the stress and pressure that had accumulated inside me.

I already knew that frightening my daughter was wrong. What I needed was somewhere for everything else to go.

That experience has shaped how I think about early intervention.

The first credible warning sign should not necessarily trigger punishment. It should trigger contact.

A counsellor can ask what is happening. A struggling parent can receive help. Parenting difficulties can be addressed. Financial, marital or psychological pressures can be identified. Where nothing more sinister exists, intervention may end there.

But counselling performs another protective function. It places another responsible adult within sight of the child.

If explanations stop making sense, injuries continue appearing, appointments are repeatedly missed or further reports arrive, intervention can escalate accordingly.

The State should not have to choose only between doing nothing and taking a child away. There must be useful machinery in between.

Something My Family Law Professor Said

I recently had the privilege of learning from arguably the most influential authority in Singapore’s family law scene, Emeritus Professor Leong Wai Kum.

During one of Professor Leong’s lectures, our discussion turned to Megan Khung and the importance of detecting family violence before it escalates.

I raised the problem of early detection from a slightly different direction. I recounted the example of someone who had experienced family violence as a child and manifested that distress through bullying other children at school. My question was whether apparently collateral behavioural signs observed by teachers and other adults could help identify problems occurring within a child’s home.

Professor Leong’s response has stayed with me. Singapore does not necessarily lack counsellors, social workers or protective resources. One of the difficult questions is how to guide the right families towards those resources. Teachers and other adults interacting with children therefore occupy an important position in recognising when something may be wrong.

Our exchange eventually moved to parenting programmes. I asked whether such programmes were mandatory. They generally were not.

The obvious problem occurred to me: are we preaching to the choir?

The parents most willing voluntarily to attend parenting programmes may be precisely those most receptive to improving their parenting.

Professor Leong suggested that once there is some indication of problems within a family, such interventions should perhaps become mandatory.

That exchange now seems considerably more important to me after reading Trevor’s case.

Early detection is useful only if something useful happens after detection. And that response need not begin with punishment.

A Ladder, Not a Trapdoor

The architecture I envisage is therefore graduated.

Signal → Report → Sensemaking → Check-in → Support → Reassessment → Protection → Enforcement.

Not every family travels along the entire path.

A first sufficiently credible signal might produce nothing more intrusive than an assessment or counselling session. For an exhausted parent who has momentarily lost control, that intervention may be enough.

Repeated or corroborated signals justify closer attention. Evidence of continuing danger calls for protective intervention. And deliberate abuse, concealment or knowing exposure of a child to serious danger properly engages the criminal law.

The degree of State intervention should increase with the degree of demonstrated risk.

This architecture would also preserve an important distinction which can easily disappear from debates about child protection: risk is not guilt.

A person can present several recognised risk factors and never harm his child. Conversely, an apparently respectable and stable household can conceal terrible abuse.

Risk indicators therefore cannot become substitutes for evidence. Their legitimate function is narrower: determining when it is sufficiently important for a trained human being to look more closely.

That is fundamentally what suspicious transaction reporting does. It does not convict. It alerts.

Building an Institutional Village

SAFER is an important development in Singapore’s child-protection architecture.

Megan Khung taught us why information held by different institutions must be capable of forming a picture greater than the sum of its parts.

But Trevor’s death invites us to ask the question immediately before that one.

Who ensures that the dots enter the system in the first place?

A mature answer cannot consist solely of mandatory reporting and punishment for those who remain silent.

Sometimes an early warning sign may identify a dangerous caregiver. Sometimes it may identify an exhausted parent standing frighteningly close to a boundary he never imagined crossing.

The system must be capable of responding to both.

That means detecting earlier, encouraging serious concerns to enter the system, connecting those concerns intelligently, intervening initially with the least coercive response appropriate to the risk, and escalating decisively when the danger persists.

We often say that it takes a village to raise a child.

SAFER may help Singapore build an institutional village capable of seeing what no individual member can see alone.

But that village still depends upon somebody speaking.

Trevor was eight months old.

He could not make the report himself.

His village had to speak for him.

---

References

1. CNA, “Couple pleads guilty to regularly abusing their baby, who died after being smothered”, 12 October 2026.
2. MSF, Parliament Statement on Megan Khung’s Case, 5 November 2025.
3. CNA, “New MSF unit launched to flag child abuse concerns earlier, after review into Megan Khung’s death”, 22 September 2026.
4. CNA, Trevor Neo case report, 12 October 2026.
5. Early Childhood Development Centres Regulations 2018, reg 39.
6. MOH, Guidelines on Reporting Cases, 11 May 2021.