S-B (Children)

In care proceedings, the identity of a person who harmed a child is determined on the simple balance of probabilities. The gravity of the allegation does not raise the standard of proof. A judge need not identify an individual perpetrator where the evidence does not permit it, but should identify the pool of possible perpetrators by asking whether each person was realistically capable of involvement. A test requiring complete exculpation sets the bar too high.

A prediction of future harm must rest on past facts proved on the balance of probabilities. Only the predicted event is assessed by asking whether there is a real possibility that cannot sensibly be ignored. Once a perpetrator has been identified, the possibility that the finding is mistaken is not itself a proven risk to the child. Percentage assessments should not supplement a finding that the perpetrator cannot be identified.

The mother appealed against care and placement orders concerning two sons. The elder child had suffered non-accidental bruising while in the care of both parents. The trial judge could not identify the perpetrator, but later expressed the view that the father was 60% likely and the mother 40% likely to have caused the injuries. The younger child had not been harmed.

The judge treated the possibility that the mother had injured the elder child as establishing a risk of future harm to the younger child. She approved plans for both children to be adopted. The Court of Appeal dismissed the mother's appeal: [2009] EWCA Civ 1048.

The Supreme Court considered the standard for identifying a perpetrator, the proper method of identifying a pool of possible perpetrators, the use of comparative percentages, and the factual foundation required for a prediction of future harm.

Lady Hale delivered the judgment of the court.

1. The appeal was allowed. The trial judge had misdirected herself on the standard of proof and had used an impermissible basis for finding a likelihood of future harm to the younger child. The entire case was remitted for rehearing before a different judge.

2. Before considering welfare, a court must be satisfied that the threshold conditions in section 31(2) of the Children Act 1989 are met. It is insufficient that professionals or the court consider that a child would be better off elsewhere. Intervention must answer a pressing social need and be proportionate.

3. The identification of a perpetrator is a factual issue governed by the ordinary balance of probabilities. The observations in Re B (Children) (Care Proceedings: Standard of Proof) [2008] UKHL 35 made clear that the same standard applies to identification as to other past facts. Seriousness does not create a heightened standard or require evidence of a special degree of cogency. Once it is established that one of two people assaulted a child, the inherent improbability of child assault does not assist in deciding which person was responsible.

4. A judge who can properly identify the perpetrator on the balance of probabilities should do so, because a finding clarifies future risks and assists work with the family. There is, however, no obligation to identify an individual where the evidence does not permit it. The court should instead identify the pool of possible perpetrators by asking whether there is a real possibility that each person was involved. Requiring a person to be wholly ruled out sets the standard too high and resembles requiring proof of innocence.

5. When a judge cannot identify the perpetrator, it is ordinarily unhelpful to add comparative percentages or other amplification. Practice had changed since Re O and another (Minors) (Care: Preliminary Hearing) [2003] UKHL 18, because the same judge will normally conduct both stages of a split hearing. The finding should be left as a finding that the perpetrator cannot be identified, accompanied by identification of the possible pool where appropriate.

6. If the court identifies a perpetrator, the possibility that the finding is mistaken is not a proven risk to the child and cannot be used as though it were one. The underlying evidence and other identified risk factors remain relevant to welfare. The court must remain alert to possible error and may revisit an earlier identification if fresh evidence warrants reconsideration.

7. A prediction of future harm must be founded on past facts proved on the balance of probabilities. Only after those facts have been established does the real-possibility standard apply to the predicted future event. The finding concerning the younger child was therefore defective: the mere real possibility that the mother had injured the elder child could not establish the factual foundation for future risk.

• United Kingdom Supreme Court: Allowed the mother's appeal and remitted the entire case for rehearing before a different judge: [2009] UKSC 17.
• Court of Appeal: Dismissed the mother's appeal against the care and placement orders: [2009] EWCA Civ 1048.
• First instance: Following a fact-finding hearing and a later final hearing, the judge found the statutory threshold crossed for both children, approved plans for adoption, and made care and placement orders.