Complex medical fact-finding requires a written, evaluative judgment engaging with parental evidence and resolving key factual disputes

1. Introduction

This appeal concerned fact-finding determinations made in public law care proceedings about two young children, “B” (now 5) and “G” (now 2). B had multiple bruises in September 2024; G had a bruise over the ribcage and imaging showing chronic subdural haematomas. The local authority alleged the injuries were inflicted (including an alleged shaking mechanism for the intracranial bleeding) and that the perpetrator failed to seek medical help.

A significant complicating feature was haematology: both children were found to have (i) a heterozygous variant in the Factor V gene (with mild/borderline Factor V deficiency) and (ii) a heterozygous HPS1 pathogenic variant associated (in the biallelic state) with Hermansky-Pudlak Syndrome. The combined significance of these variants was described as “uncharted territory” with no research base for co-inheritance.

The judge made “pool” findings that the injuries were inflicted by the mother and/or father and found that the perpetrator of the shaking injury failed to seek medical attention. The parents appealed those injury-related findings; other threshold matters (domestic abuse, substance misuse, mental health) were not challenged.

2. Summary of the judgment

• The Court of Appeal allowed the appeal and set aside the findings that the injuries were inflicted by one or both parents and that the perpetrator failed to seek medical attention.
• The Court held the fact-finding judgment was inadequately reasoned for a complex medical case: it did not properly evaluate the parents’ evidence, misinterpreted key expert evidence, and left a central factual dispute unresolved.
• The Court expressed disquiet about the trial judge making substantive transcript amendments after seeing appeal materials, emphasising the need for reasons to reflect the judge’s reasoning at the time of decision.
• The case was remitted to the Family Presiding Judge (Arbuthnot J) to decide whether a re-hearing on causation is necessary and, if so, its scope.

3. Analysis

3.1 The core appellate failure: lack of evaluative engagement with parental evidence

Baker LJ treated the absence of a clear credibility/reliability assessment of the parents as a central deficiency. The parents’ oral evidence occupied around half the hearing, yet the judgment contained little analysis of what the judge made of their accounts, demeanour, consistency, or plausibility, and no clear findings on whether either parent was truthful, mistaken, minimising, or evasive on the critical injury issues.

The Court stressed that it is not enough to recite that parents “do not have to prove anything” and that their evidence “should be considered”; the court must actually evaluate it in a way that allows parties (and an appellate court) to see how it affected the outcome—especially where the medical evidence is complex and less than diagnostic.

3.2 Misinterpretation of the neurosurgical evidence on G’s subdural haematomas

The Court found the judge wrongly treated the neurosurgeon (Mr Jayamohan) as supporting, on the balance of probabilities, a non-accidental explanation. Properly read, his evidence identified multiple potential aetiologies—birth-related trauma (with forceps in a caesarean delivery), accidental trauma, medical cause/bleeding predisposition, and inflicted injury (including shaking)—but he could not say which was most likely.

This error mattered because (i) the other experts deferred to the neurosurgeon on the imaging interpretation and (ii) the local authority’s “shaking” case depended on nuanced assessment of risk factors and alternative explanations. The Court accepted it can still be open to a court to find non-accidental injury even where medical evidence is not diagnostic (see further below), but here the misreading of the expert position undermined the reasoning.

The Court also highlighted the significance (as part of the “totality”) of features said to weaken an inflicted/shaking conclusion: an asymptomatic course, absence of two elements of the “triad” (no retinal haemorrhages; no encephalopathy), routine healthcare contacts without concern, and the obstetric history including forceps use.

3.3 Bruising, “disproportionality”, and the unresolved foster care dispute

On bruising, the trial judge accepted a “disproportionate” pattern in September 2024 compared with other periods, relying materially on a foster carer’s logs and the paediatric expert’s analysis of those logs. The Court of Appeal held the judgment left unresolved a factual dispute whether bruising continued in foster care in a way supportive of an underlying tendency.

The Court criticised the reliance on incomplete recording: the foster carer logged only “large” or “significant” bruises, which risks circularity where the legal/medical question is precisely whether apparently minor knocks yield disproportionate bruising. The judgment noted the criticism but did not explain how it was discounted when accepting the “no disproportionate pattern in foster care” conclusion.

Importantly, the treating haematologist (Dr Dickens) had formed a clinical view that the children bruised very easily across settings, and that the genetic findings meant one could not be sure the injuries could not be explained medically. The Court held the trial judge did not adequately explain why she discounted this clinical evidence as a mere “snapshot”, nor did she adequately engage with a substantial bruise chronology prepared for the parents.

3.4 Genetic uncertainty increases the importance of the “clinical picture”

The Court emphasised that the co-existence of two rare genetic variants—without research guidance on combined effect—meant that the observed pattern over time (“clinical picture”) took on enhanced importance. Against that backdrop, the Court considered the trial judge failed to resolve the factual questions about bruising persistence and pattern across different care environments, and therefore could not safely reason to an inflicted conclusion.

The Court also noted that recommended further steps (parental genetic testing; possible instruction of a geneticist) were not taken. While not framed as a discrete procedural ground of appeal, this formed part of the context of “forensic limitations” requiring careful judicial evaluation rather than an assumption that lack of an alternative explanation equates to proof of infliction.

3.5 Ex tempore delivery and post-judgment transcript amendments: open justice and legitimacy

A notable feature was that judgment was reserved for six weeks but delivered ex tempore from notes in a case involving dense and uncertain medical evidence. The Court regarded that as an “unwise” course in such a case, because it increases the risk of omission of important matters.

The Court accepted judges may correct transcripts (including, in some circumstances, substantively), but expressed “considerable disquiet” here because substantive changes were made after the judge had been sent (or had requested) (i) a note of her questions to the father and (ii) the grounds of appeal. Baker LJ stressed that the purpose of giving reasons is to show what was in the judge’s mind when the decision was made, and amendments must not become an ex post facto rationalisation. Andrews LJ endorsed the observation that for complex cases the safer course is a written judgment, circulated in draft under embargo so omissions can be corrected before hand-down.

3.6 Precedents cited and how they shaped the Court’s approach

Re B (A Child) (Adequacy of Reasons) [2022] EWCA Civ 407

This decision provided the framework for assessing whether reasons are adequate. The Court adopted the emphasis on evaluation and explanation as the “engine” of a judgment: setting out evidence is not enough; the judge must show why key issues were resolved as they were. The Court’s criticism here was that the judgment did not make clear how competing explanations were weighed, particularly in light of the parents’ evidence.

Re C, D and E (Care Proceedings: Adequacy of Reasons) [2023] EWCA Civ 334

Cited to reinforce that where deficiencies are fundamental, they cannot necessarily be repaired by later “clarification”. This was relevant to the argument that the judgment was not merely missing a few sentences but lacked the evaluative structure necessary to sustain serious findings.

Re BR (Proof of Facts) [2015] EWFC 41

The trial judge cited this case for a list of “risk” and “protective” factors. The Court of Appeal did not criticise the use of the framework as such, but held that the finding that most protective factors were present made it all the more important to conduct a careful analysis of the parents’ credibility and reliability before concluding that one of them harmed the children.

Devon County Council v EB & Ors (Minors) [2013] EWHC 968 (Fam)

Used to underscore that in alleged abuse cases the carers’ evidence is “of the utmost importance” and the court must form a clear assessment of credibility and reliability. The Court treated this as directly engaged on the facts: absent a clear evaluative assessment, the findings were unsafe.

A County Council v K, D and L [2005] EWHC 144 (Fam) and A County Council v A Mother and others [2005] EWHC Fam 31

These authorities anchored the “totality of the evidence” principle: medical evidence is important but not determinative; neither medical nor non-medical evidence has automatic precedence; context matters. Crucially, Baker LJ rejected a submission that medical uncertainty automatically defeats the local authority case—non-accidental injury can still be found if the whole evidential picture supports it. However, the Court concluded that the trial judge’s analysis of the whole picture was incomplete and flawed, and therefore the local authority had not been shown to have discharged the burden on the reasons given.

3.7 Impact

The decision is likely to be cited in three recurring situations in family fact-finding:

1. Complex medical causation with scientific uncertainty: where experts cannot be definitive, courts must be meticulous in resolving the “clinical picture” and in demonstrating how uncertainty is handled rather than treated as a gap filled by inference.
2. Bruising cases where record-keeping is incomplete or filtered: the judgment signals the forensic risk of relying on selective logging (e.g. “large bruises only”) when the legal question concerns pattern and disproportionality.
3. Adequacy of reasons and judgment method: while not creating a formal rule, Andrews LJ’s endorsement strongly encourages written reserved judgments (with draft circulation) in cases where complexity makes omissions foreseeable, to reduce appellate vulnerability and enhance transparency.

Procedurally, the case also serves as a cautionary illustration: extensive post-hoc transcript amendments—especially after exposure to appeal grounds—risk undermining confidence that the judgment reflects the contemporaneous reasoning required by open justice.

4. Complex concepts simplified

“Balance of probabilities”

The civil standard: a fact is proved if it is more likely than not (>50%). It does not require certainty, but the reasoning must show how the court reached that probability judgment on the evidence.

“Pool of perpetrators”

A finding that an injury was inflicted by one of a limited group (here, either parent) without identifying which. Such findings demand careful reasoning, particularly where the evidence might differentiate opportunities, timing, or credibility.

Subdural haematomas and the “triad” in shaking allegations

A subdural haematoma is bleeding between the brain and its outer covering. In some abusive head trauma cases, clinicians may see a cluster of findings historically described as a “triad” (subdural bleeding, retinal haemorrhages, encephalopathy). The absence of elements of the triad does not rule out inflicted injury, but it is a relevant contextual factor in assessing likelihood.

Genetic “heterozygous variant”, “carrier status”, and “variant of uncertain significance”

“Heterozygous” means one altered copy and one normal copy of a gene. “Carrier status” often implies the person does not have the full disease but may carry a trait. A “variant of uncertain significance” means science has not yet established whether that change causes clinical problems. Where such findings are present, longitudinal observation (“clinical picture”) becomes especially important.

“Disproportionate bruising”

Bruising that appears more extensive or in unusual locations than would be expected from the described (or observed) level of trauma. Determining disproportionality depends heavily on complete and unbiased recording over time.

5. Conclusion

B and G (FACT-FINDING) reaffirms that in public law fact-finding—particularly where medical evidence is complex, uncertain, and dependent on longitudinal pattern—adequate reasons require (i) a clear evaluation of parental credibility and reliability, (ii) accurate treatment of expert evidence, and (iii) resolution of key factual disputes underpinning “clinical picture” assessments (such as whether bruising persisted across placements).

The decision also delivers a strong practical message about judgment craft: when complexity makes omissions foreseeable, a written reserved judgment (ideally circulated in draft) is not mere best practice; it may be essential to ensuring the parties, and any appellate court, can see why the findings were made.