Factual and Procedural Background
Child A was born in May 2007. At four weeks old he was discovered with bruising to his arms and face which medical professionals considered to be non-accidental. He was immediately removed from the care of his parents. Both the mother and the father had attended to the baby that morning, and each denied responsibility. The bruises were recent and not so severe that the non-perpetrating parent must necessarily have known of them, making the case a classic “whodunit.”
Child B was born in July 2008 while care-proceedings relating to Child A were still pending. By that time the parents had separated, the father had disengaged from the proceedings, and Child B was taken into foster care soon after birth. He has never suffered harm; the asserted basis for his removal was the risk of future harm.
The trial judge initially intended a split (fact-finding then welfare) hearing but, for practical reasons, the process became fragmented. The father did not attend the fact-finding stage. After hearing evidence in January and March 2008, the judge gave a note-form judgment in April 2008—prior to the House of Lords decision in Re B—stating that the allegations were “very serious and unusual” and that “cogent evidence equivalent to the gravity of the allegations” was required.
The judge concluded that both parents remained possible perpetrators, but later, in an “Adjunct to Judgment,” indicated a 60% likelihood that the father was responsible and 40% that the mother was. At the welfare stage in January 2009 the judge declined to treat the father as the identified perpetrator, held that the statutory threshold was met for both children, approved care plans for adoption, and made care and placement orders. Permission to appeal was granted.
The Court of Appeal dismissed the mother’s appeal, after which the matter came before the Supreme Court (this judgment).
Legal Issues Presented
- Whether the civil standard of proof (balance of probabilities) applies to identifying the individual who perpetrated non-accidental injuries to a child.
- Whether, when a judge cannot identify a perpetrator on that standard, the judge should delineate a “pool” of possible perpetrators and/or express comparative likelihoods between them.
- Whether a predicted risk of future harm to a second, unharmed child can be based merely on a “real possibility” that the mother injured the first child.
- Whether the trial judge misdirected herself by applying a heightened evidential standard inconsistent with Re B.
Arguments of the Parties
The opinion does not contain a detailed account of the parties' legal arguments.
Table of Precedents Cited
| Precedent |
Rule or Principle Cited For |
Application by the Court |
| Re H (Minors) (Sexual Abuse: Standard of Proof) [1996] AC 563 |
Balance-of-probabilities for past facts; “real possibility” test only for future predictions. |
Reaffirmed as governing threshold findings and distinguished from future-risk analysis. |
| Re B (Children) (Care Proceedings: Standard of Proof) [2008] UKHL 35 |
Single civil standard; seriousness of allegation does not raise the standard. |
Used to show trial judge’s misdirection and to confirm the same standard applies to identifying perpetrators. |
| B v Chief Constable of Avon & Somerset [2001] 1 WLR 340 |
Instances where criminal standard applies in ostensibly civil proceedings. |
Cited to illustrate categories causing confusion over evidential standards. |
| R (McCann) v Crown Court at Manchester [2002] UKHL 39 |
Same category as above regarding criminal standard in civil context. |
Referenced in explaining doctrinal background. |
| Secretary of State v Rehman [2001] UKHL 47 |
Confirmation that only one civil standard exists. |
Shown as consistent authority supporting Re B. |
| Re D (Secretary of State for Northern Ireland intervening) [2008] UKHL 33 |
Application of the single civil standard. |
Cited for doctrinal coherence. |
| Re M & R (Minors) (Sexual Abuse: Expert Evidence) [1996] 4 All ER 239 |
Welfare stage must rely on proved facts, not suspicions. |
Approved as correct approach to risk assessment. |
| Re O (Minors) (Care: Preliminary Hearing) [2003] UKHL 18 |
Guidance where perpetrator cannot be identified; need to consider risk on basis someone in pool caused harm. |
Quoted to demonstrate necessity of fact-finding and utility of same judge hearing both stages. |
| Lancashire County Council v B [2000] 2 AC 147 |
“Attributability” includes care by anyone undertaking the child’s care when responsibility cannot be isolated. |
Used to clarify threshold where care is shared between households. |
| Re D (Care Proceedings: Preliminary Hearings) [2009] EWCA Civ 472 |
Judges should not strain to identify a perpetrator merely because of Re B. |
Adopted to caution against unnecessary perpetrator findings. |
| Re K (Non-Accidental Injuries) [2004] EWCA Civ 1181 |
Benefits for professionals and child of clear perpetrator findings; child’s right to truth. |
Cited to explain why findings, where possible, are valuable. |
| North Yorkshire County Council v SA [2003] EWCA Civ 839 |
Pool of perpetrators should be limited to those with a “likelihood or real possibility” of involvement. |
Referenced in discussing identification of the possible perpetrator group. |
| Re I (A Child) [2009] UKSC 10 |
Court may revisit perpetrator identification when fresh evidence emerges. |
Used to illustrate flexibility to correct mistaken findings. |
Court's Reasoning and Analysis
1. Standard of proof. The Court reaffirmed that the balance of probabilities is the sole civil standard for all past-fact findings, including the identification of a perpetrator. References to “heightened” or “cogent” evidence based on gravity were rejected as misconceived.
2. Misapplication by the trial judge. The judge equated the seriousness of allegations with a need for “evidence of reliability and cogency equivalent to the gravity of the allegations,” thereby importing a heightened threshold and misdirecting herself.
3. Percentage likelihoods. The practice of expressing comparative probabilities (e.g., 60/40 split) where the perpetrator cannot be identified on the balance of probabilities was discouraged. Such indications risk confusing the legal test and provide little assistance at the welfare stage.
4. Pool of possible perpetrators. Where the perpetrator cannot be identified, the judge must still define a realistic pool based on a “real possibility” of involvement, aiding both threshold analysis (“attributability”) and risk assessment.
5. Risk to an unharmed child. The Court held that the trial judge erred in applying a mere “real possibility” that the mother injured Child A to establish a likelihood of future harm to Child B. Predictions of future harm must rest upon past facts proved on the balance of probabilities; only after facts are proved does the “real possibility” test apply to future events.
6. Need for rehearing. Because the factual findings were tainted by the wrong standard and an impermissible basis for the future-risk threshold, justice required the entire matter to be reheard by a different judge.
Holding and Implications
APPEAL ALLOWED; CASE REMITTED FOR A COMPLETE REHEARING BEFORE A DIFFERENT JUDGE.
Direct consequences include setting aside the existing care and placement orders and requiring fresh fact-finding and welfare determinations. The judgment clarifies that (1) the balance-of-probabilities standard governs identification of perpetrators; (2) judges should avoid percentage likelihood expressions when the standard is not met; and (3) future-risk findings for an unharmed child must be grounded in facts proved on that standard. While reinforcing existing principles rather than creating new law, the decision offers practical guidance likely to influence how lower courts handle “whodunit” child-protection cases.