What happened to Lucia de Berk
Lucia de Berk was a paediatric nurse at the Juliana Children’s Hospital in The Hague. In 2001 the hospital noticed that a number of patient deaths and medical emergencies had occurred during her shifts. An internal review identified her as the common factor. The case went to trial on the strength of that shift coincidence plus post-hoc medical reinterpretation of the deaths as suspicious. In 2003 she was convicted of four murders and three attempted murders and sentenced to life imprisonment. Further convictions followed on appeal in 2004.
The statistical centrepiece of the case was a calculation that the probability of Lucia being on shift for so many incidents “by chance” was on the order of 1 in 342 million. The medical centrepiece was re-interpretation of previously-explained deaths as unexplained, and unexplained deaths as poisonings.
How the case was overturned
Prof. Richard Gill of Leiden University, a senior Dutch statistician, led an independent examination of the evidence from 2006 onwards. His critique made three structural points that matter here:
- Selection effect. The “incidents” selected for the statistical calculation were those that looked suspicious in retrospect, and that suspicion was partly formed by noticing Lucia’s presence. The calculation was therefore retrospective, not predictive.
- Reinterpretation. Deaths that had been clinically routine at the time were re-read as suspicious after the pattern was noticed. The re-reading was not conducted blinded. Expectations of guilt shaped the retrospective diagnosis.
- Specialist framing. The case was presented by medical witnesses who were not statisticians. A qualified statistician would have identified the selection effects. None gave evidence for the prosecution.
The Dutch Supreme Court reopened the case in 2008 on the basis of Prof. Gill’s submissions, and acquitted Lucia de Berk in April 2010. She was exonerated after seven years in prison. The case is the textbook Dutch miscarriage of justice of the modern era.
Where the Letby case matches
The procedural parallel is uncomfortably close:
- Cluster noticed. Countess of Chester consultants noticed a cluster of deaths on the neonatal unit from July 2015 onwards. Juliana Hospital management noticed a cluster around Lucia de Berk from 2001.
- Common factor identified. In both cases, a nurse who worked more unsociable shifts than average was identified as the common factor.
- Retrospective medical reinterpretation. Deaths that had been explained at the time were re-diagnosed as unexplained once the nurse was identified. This reinterpretation was not done blinded.
- Statistical chart without a statistician. The prosecution produced a shift-rota chart (Letby) or a probability calculation (de Berk) and presented it to the jury/judges through non-statistician witnesses. In both cases, independent statisticians subsequently identified the selection-effect problem.
- Medical experts who had not worked in the specific acute care setting recently. In de Berk the re-interpreting pathologist had limited recent paediatric-poisoning casework. In Letby the Crown’s lead expert had not worked in routine neonatal intensive care for over a decade.
- Management pressure on the investigation frame. In both cases, hospital management had a strong institutional interest in a “bad apple” explanation rather than a systemic one.
What Richard Gill himself says about the parallel
Prof. Gill — the statistician who helped overturn the de Berk conviction — has been publicly clear since 2023 that the Letby case is structurally the same case, and has written about it extensively. In a 2024 lecture (“A tale of two Lucies”) he set out the parallel in detail. The lecture is now mirrored on lucyletby.org and is referenced in the Sir David Davis Commons debate.
His specific warning to UK readers has been: this is a reproducible structural error of criminal-justice reasoning, and the fact that it has been corrected in one European jurisdiction does not mean the same error will automatically be corrected in another. Correction requires specific institutional attention — which is what the CCRC application is requesting.
How this should change how we read the Letby verdict
The Lucia de Berk parallel is not an argument by analogy in the loose sense. It is a technical argument: the same statistical and methodological errors that produced a wrongful conviction in the Netherlands are demonstrably present in the Letby record. That does not itself entail that the Letby convictions are wrong. It does entail that they should be examined by the CCRC with the same rigour the Dutch Supreme Court applied to the de Berk case. That is what the October 2025 application asks for.