The Beverley Allitt framing effect — investigation confirmation bias
Prosecution claim
Operation Hummingbird was presented at trial as a neutral, evidence-led police investigation that followed the evidence to its conclusion. The Crown framed the case in court as if the hypothesis of a 'killer nurse' had been tested and confirmed by objective methods.
Counter-evidence
The Hummingbird investigation was framed from its May 2017 opening by explicit analogy to the Beverley Allitt case. The anonymous 150-page Hummingbird whistleblower report hosted on lucyletby.org, and published Thirlwall Inquiry evidence from former DCS Nigel Wenham, together indicate that the investigation proceeded on a 'suspect-first' rather than 'cluster-first' model. That framing had operational consequences: expert instruction was shaped by the hypothesis, natural-causes and systems-failure evidence was structurally deprioritised, and confirmation bias became procedurally hard to correct. Once an 'Allitt' frame is adopted, it organises how every subsequent piece of evidence is read.
A suspect-first investigation does not test the hypothesis that a suspect is responsible. It tests the hypothesis that particular pieces of evidence can be read to fit the suspect. Those are not the same investigation.
What the jury heard
The jury heard the results of the investigation but was not systematically walked through its framing assumptions. The default assumption of impartial investigation was not challenged in the detail that post-conviction expert analysis now permits.
What the Panel says
The Panel does not opine on police-investigation scoping directly, but its case-by-case medical review concludes that in every indicted case the clinical deterioration is explicable by natural causes or sub-optimal clinical care. That conclusion is inconsistent with the narrative Hummingbird was built on.
What independent experts add
- The Allitt case (1991) involved direct forensic-standard evidence: anomalous potassium and insulin values confirmed by proper laboratory testing, a stolen Kardex recovered from the suspect's flat, eyewitness accounts of proximity to each event. None of these direct-evidence elements is present in the Letby case at equivalent standard.
- The patient population is structurally different: Allitt's patients were older children with relatively low baseline mortality risk; Letby's were extremely preterm neonates with high baseline mortality risk.
- The cluster context is structurally different: Allitt's Grantham ward was not experiencing a concurrent outbreak, infrastructure failure, or level-of-care mismatch; the Countess of Chester neonatal unit was experiencing all three.
- The Clothier Inquiry's recommendations after Allitt focused on hospital escalation procedures. Those recommendations do not, on any reading, support the later use of the Allitt analogy to frame a criminal investigation of a different clinical context.
- Prof. Richard Gill's 'tale of two Lucies' lecture (March 2024) and the Hummingbird whistleblower report both argue the Allitt framing led investigators to stop testing alternative explanations too early.