Introduction

This appeal is a significant restatement of principle in public law children cases about how courts must assess risk and balance welfare. Three siblings, R (8), S (6) and L (5), had lived safely and successfully for nearly three years with their paternal grandparents after suffering neglect, domestic abuse and parental substance misuse. The Family Court, nevertheless, made care orders with a plan for long‑term foster care, based largely on historic sexual convictions of the grandfather from 1975–1984 and certain incidents concerning the paternal uncle, coupled with concerns about the grandparents’ “insight.”

The Court of Appeal allowed the appeal, discharged the care orders, substituted a 12‑month supervision order, restored the interim child arrangements order, and remitted the grandparents’ special guardianship (SGO) application. The court condemned reliance on unproven allegations and “background concerns” to inflate risk, warned against calibrating decisions to achieve “100% safety,” underscored the proportionality discipline required by Article 8 ECHR, and emphasized that foster care is not risk‑free.

In doing so, the Court draws a clear line between proved facts and suspicions in risk assessment, rejects the transplantation of the Supreme Court’s approach in Pearce (Parole Board) into Children Act decision‑making, and re‑affirms the centrality of holistic welfare balancing in the light of real-world outcomes for children.

Summary of the Judgment

• Outcome: Appeal allowed; care orders set aside. Court of Appeal substituted a 12‑month supervision order, restored the prior interim child arrangements order, and remitted the SGO application to the Family Court.
• Core holdings:
  • Risk assessments in Children Act proceedings must be founded on proven facts and proper inferences from those facts, not on unproven allegations, suspicions, or “background concerns.”
  • It was procedurally unfair to weigh disputed allegations against the uncle where the local authority had not sought findings and where his statement was untested and unchallenged.
  • The recorder’s welfare evaluation was overwhelmed by the risk factor and implicitly demanded “100% safety,” giving patently excessive weight to unproven or marginal risks over the certain and severe harm of removal.
  • Proportionality: Long-term foster care, with its inherent uncertainties and potential for harm, was a disproportionate interference with the children’s and grandparents’ Article 8 rights when the children had thrived for nearly three years in the family placement.
  • Case management: If unproven matters are material to risk, they must be identified and determined; courts cannot “import” them as part of the risk canvas without findings.
• Additional observations: The local authority’s execution of removal during the appeal window, the inadequately planned and rapidly altered placements, and the reduction of contact after permission to appeal was granted, drew strong judicial censure and illustrated the non‑trivial risks of foster care.

Background and Key Issues

Following significant harm in the parental home (substance misuse, domestic abuse, neglect), the children moved to their paternal grandparents in July 2022. They flourished: improved school attendance, better behaviour, close attachments, and positive professional feedback. Safety plans were implemented and complied with. The uncle moved out by March 2025.

The Family Court nevertheless removed the children, citing sexual risk arising from:

• Grandfather’s convictions (1975–1984, aged 14–22 at the times), with no offending for 40 years.
• Uncle’s accepted incidents: inappropriate touching at age 11; underage sexual intercourse with a 15‑year‑old girlfriend (he believed she was 16).
• Unproven allegations about the uncle (2006–2015) on which no findings were sought or made.
• School reports of sexualised behaviour by R in 2023/2024, about which no finding was sought or made, and where the source was unestablished.

Expert evidence (Dr Parsons; ISW Ms Sensicle) broadly assessed the grandfather’s risk as low/very low, rising to low‑medium due to opportunity in a carer role, and the uncle’s risk as low/very low absent unsupervised access. The local authority and Guardian nonetheless pursued removal, with the Guardian expressing concern that only “24/7” support would keep the children “100% safe.”

Issues on appeal included:

• Whether the recorder erred in law by relying on unproven allegations and unresolved concerns (including R’s school behaviour) as part of the risk calculus.
• Whether the recorder’s welfare analysis was disproportionately weighted to risk, implicitly calibrating to “zero risk,” and inadequately considered less intrusive orders (e.g., supervision).
• Whether the overall decision was reasonably open on the evidence.

Detailed Analysis

1) Precedents and Authorities Cited

The Court of Appeal anchors its reasoning in the leading authorities and a recent synthesis by Baker LJ, reaffirming the architecture of risk analysis in children cases:

• Re H (Minors) (Sexual Abuse: Standard of Proof) [1996] 1 AC 563: The cornerstone. Courts act on evidence. Future risk must be founded on proved present facts and proper inferences. Non‑proven allegations cannot be used as facts to ground likelihood.
• Re M and R (Child Abuse: Evidence) [1996] 4 All ER 239; Re O and N [2003] UKHL 18; Re B [2008] UKHL 35: The “real possibility” test for future harm and consistency of approach across threshold and welfare stages.
• Re S‑B (Children) [2009] UKSC 17: Three settled propositions restated by Baroness Hale: (i) actual past harm requires proof; (ii) likelihood findings must be grounded in proved facts; (iii) future harm need only reach the “real possibility” standard — but still on a foundation of proved facts.
• Re A (Fact‑Finding: Disputed Findings) [2011] EWCA Civ 12: Courts may refer to unproven evidence for narrative context but may not rely upon it as a basis for risk evaluation.
• Re G (A Child: Scope of Fact‑Finding) [2025] EWCA Civ 1044 (Baker LJ): A recent definitive restatement, adopted and applied here.
• Devon CC v EB [2013] EWHC 968 (Fam): The court can review a wide evidential canvas, but may not base risk assessment on unproven facts. The Court distinguished and confined its use here.
• R (Pearce) v Parole Board [2023] UKSC 13: The Court of Appeal rejected attempts to import the Parole Board’s risk-assessment approach into Children Act proceedings. Pearce was a different statutory and factual context; it cannot dilute the Re H / Re S‑B constraints.

Key doctrinal clarification: The “fact in issue” at welfare/risk stage is not the abstract “level of risk,” but the proved facts from which any risk assessment is drawn. This resolves the Guardian’s submission: courts cannot evade fact‑finding by treating “level of risk” itself as the operative fact and smuggling in unproven allegations as “background factors.”

2) The Court’s Legal Reasoning

a) The evidential bedrock requirement

Peter Jackson LJ reiterates that public law decisions carry grave consequences; the discipline of proof for past facts (balance of probabilities) and the “real possibility” test for future harm are complementary, not alternatives. A welfare decision can reference unproved material for context, but may not use it as a foundation for a risk conclusion. Here, the recorder materially relied on unproven matters:

• Uncle’s disputed history: The judge did not distinguish between accepted and disputed behaviours, despite the LA not seeking findings and the uncle not being called or cross‑examined. That was procedurally unfair.
• R’s school behaviour: No finding was sought or made as to whether, what, when, or from where any sexualised behaviour arose; yet it was repeatedly weighed as risk‑indicative and folded into a “three generations” pattern. That was impermissible.
• “Generational patterns” construct: References to a three‑generation pattern presupposed a probative link that was neither alleged nor proved.

b) Proportionality and avoiding the “100% safe” trap

The recorder accepted a catalogue of powerful pro‑kinship factors (attachment, sustained progress, settled schooling, sibling unity, parental support for the placement, proven cooperation, and absence of any harm for three years) but let perceived sexual risk dominate the welfare balance. The Court of Appeal held this was a miscalibration:

• Welfare decisions do not require the elimination of all risk. A de facto “zero risk” threshold (or reliance on formulations such as “100% safe” and “24/7” supervision) improperly privileges a single factor over the holistic welfare evaluation demanded by the Children Act and Article 8.
• Foster care does not guarantee safety and carries well‑known risks (placement breakdowns, moves, separation, stigma; here, an allegation of slapping by a foster carer; and the court’s reminder, informed by the grandfather’s history, that fostered children can be mistreated).
• Given the children’s thriving placement with grandparents, any plan to remove them to long‑term care required compelling justification. It was not present.

c) The role of “insight”

The recorder over‑relied on perceived deficits in the grandparents’ insight. The Court of Appeal drew important lines:

• Lack of insight may complicate professional engagement but does not, of itself, equate to an inability to protect, especially against a track record of compliance, successful safety planning, and evident child progress.
• It is unduly rigorous to insist that “following rules and boundaries” is insufficient absent full internalization of professional risk models.

d) Case management and fairness

• If unproven matters are relied upon as risk‑critical, courts must identify and determine them, using flexible procedural tools where appropriate. Time and resource pressures are not a lawful reason to bypass proof.
• Standing down the uncle as a witness and then having disputed allegations effectively count against the family was unfair.

e) Remedies and proportional response

Recognizing the extreme delay (five times the 26‑week target) and the narrowness of the legal error, the Court declined to remit for rehearing and instead substituted a 12‑month supervision order, restored the interim child arrangements order, and remitted the SGO application. This spared the children further delay and uncertainty and provides a structured, proportionate, and reviewable protective framework.

3) Impact and Significance

Immediate doctrinal effect

• K‑H decisively reconfirms that unproven allegations cannot lawfully form part of the risk calculus in care cases. It closes the door on attempts to invoke Pearce to dilute the Re H / Re S‑B framework in the children context.
• It recalibrates the use of “insight” and rejects the idea that absence of full insight trumps demonstrable protective behaviour over time.
• It underscores that proportionality requires serious, realistic consideration of less intrusive orders (e.g., supervision), especially where a kinship placement is working in practice.

Practical ramifications for future cases

• For courts: Identify case‑critical disputed facts at an early stage; hold tailored fact‑findings where needed; resist risk assessments built on suspicion or patterns not anchored in proof.
• For local authorities and Guardians: When the core “pillars” are limited (e.g., historic convictions and low‑level admitted incidents), do not inflate risk through untested “concerns” or generational narratives. If additional allegations matter, seek findings. Treat “lack of insight” with nuance, against real-world compliance and outcomes.
• For kinship carers: Sustained adherence to safety plans, positive engagement, and children’s progress carry significant weight. “Doing the right things” can suffice even if carers articulate risks differently from professionals.
• On implementation post‑judgment: Avoid precipitate removals during appeal windows where there is no urgency; maintain contact levels proportionate to children’s welfare, particularly after permission to appeal is granted.

Complex Concepts Simplified

• Threshold vs welfare
  • Threshold (s.31(2) Children Act): the gateway to care/supervision orders. Past harm must be proved on the balance of probabilities, or future harm must be grounded on proved facts showing a real possibility.
  • Welfare (s.1(1) and s.1(3)): once threshold is crossed (often on parental harm), the court decides what order best serves the child’s welfare, considering the welfare checklist and proportionality.
• Standards of proof
  • Past events: balance of probabilities (more likely than not).
  • Future harm: “real possibility” standard — but crucially, predicated on proved facts.
• Proven facts vs unproven allegations
  • Courts can mention unproven material for context but cannot rely on it to assess risk. If a point matters, it must be proved or not taken into account as a risk foundation.
• Proportionality (Article 8 ECHR)
  • Interference (e.g., removal into foster care) must be necessary and proportionate to the risks. The court cannot insist on zero risk if less intrusive measures can safely manage concerns.
• Care order vs supervision order
  • Care order: corporate parenting; extensive LA powers; often entails removal; intrusive and long‑term oversight.
  • Supervision order: child remains with family; LA “advises, assists and befriends”; can be tailored and extended; less intrusive, suitable where cooperation is demonstrated.
• Special Guardianship Order (SGO)
  • Gives carers enhanced parental responsibility; supports permanence in kinship care; requires robust assessment of capacity to meet needs and manage any risks.
• “Insight”
  • Helpful, but not a legal precondition to caring. Protective behaviour over time, adherence to safety plans, and actual child outcomes are key.

Practice Guidance and Lessons

• Fact‑finding discipline
  • Case manage early: identify disputed matters that are material to risk; if they matter, list them for findings; if not, exclude them from the risk foundation.
  • Do not rely on “generational patterns” unless there are proven links with probative value.
• Expert evidence
  • Experts may consider wide information; the court must filter their conclusions through the proved facts only rule.
  • Ensure experts are explicit about which elements of their analysis depend on allegations that have not been proved.
• Risk calibration
  • Avoid “100% safety” or “24/7 monitoring” rationales. Calibrate to “safe enough” in the context of real-world risks and protective factors.
  • Remember that foster care has its own risks; removing to eliminate a low/low‑medium familial risk may be disproportionate.
• Remedies and alternatives
  • Where kinship placements work, consider supervision orders and structured support as proportionate alternatives to care orders.
  • If LA resists “care at home,” judges should still weigh non‑public law routes (e.g., child arrangements + supervision, or SGO) where they meet welfare needs proportionately.
• Appeal period conduct
  • Avoid rushed removals absent urgency; consider pausing implementation where appeals are mooted; preserve contact continuity, particularly once permission to appeal is granted.

Conclusion

K‑H (Children) is a powerful recalibration of risk and proportionality in public law children cases. It restates with clarity that:

• Risk assessments must be built on proved facts; unproven allegations, however concerning, cannot lawfully underpin removal decisions.
• The court must resist the gravitational pull of risk‑dominant reasoning that implicitly demands “zero risk.” Welfare is holistic, and the Article 8 proportionality discipline requires close attention to the certain harms of removal and the real‑world success of kinship care.
• “Insight” is relevant but not determinative. Sustained compliance, effective safety planning, and demonstrable child progress matter more than rhetorical alignment with professional language.
• Foster care is not a panacea. Its inherent uncertainties and potential harms must be confronted honestly in the welfare balance.

On the facts, the Family Court’s decision, heavily influenced by unproven matters and an elevated risk threshold, could not stand. The Court of Appeal’s substitution of a supervision order, restoration of child arrangements, and remit of the SGO application exemplify a proportionate, child‑centered outcome. Going forward, K‑H will be a touchstone for advocates and judges: prove what matters, balance what is real, and choose the least interventionist, effective order that safeguards children’s welfare.