Lucy Letby case: Did probability evidence shape public belief?

The Lucy Letby case continues to stir debate because statistical arguments presented at trial helped shape how the public understood the evidence. Prosecutors showed jurors a shift chart that placed the nurse at every incident, and experts later warned that this framing invited the Texas sharpshooter fallacy. The question now is whether those probability claims hardened public belief before the medical record could be fully examined.

Shift chart in the courtroom

The prosecution used a simple grid to show that Lucy Letby was the only staff member on duty for each of the 25 incidents. Jurors saw a visual that looked decisive, even though the chart did not compare her presence against other nurses across the full roster. The graphic became shorthand for guilt in media summaries and social media threads that followed the verdict.

Defense lawyers objected that the table ignored the larger pool of shifts when nothing happened. Statisticians later noted that selecting only the bad shifts and then highlighting one person within them creates a classic selection bias. The Royal Statistical Society had already warned about this pattern in its 2022 guidance on healthcare investigations.

Despite the objections, the chart remained central to closing arguments. Coverage in outlets on both sides of the Atlantic treated the visual as near-proof. Public discussion online quickly adopted the same framing, repeating phrases such as “always there” without context about how the data had been filtered.

Prosecutor’s fallacy risks

Early police briefings included the line that the chance of one nurse being present for every death was “very low.” That phrasing invited the prosecutor’s fallacy, the error of equating a low random probability with proof of deliberate action. Jurors were not given a base rate that included staffing levels or the hospital’s overall mortality spike.

Richard Gill, a Dutch statistician, published Bayesian re-analyses in early 2025 that attempted to correct for these missing numbers. His models suggested that once alternative explanations such as understaffing and equipment failures were included, the posterior odds shifted sharply. Gill’s posts circulated widely on statistics forums and true-crime podcasts in the United States.

Critics of Gill’s work argued that his assumptions about systemic failures were speculative. Supporters countered that the original trial presentation had been equally selective. The exchange kept the statistical debate alive long after the verdicts, feeding ongoing public skepticism about how the numbers were sold.

Expert panel enters the picture

In February 2025 a panel of fourteen neonatologists led by Dr. Shoo Lee reviewed the medical records and concluded that none of the deaths showed evidence of murder. The panel attributed the outcomes to natural causes or substandard care. Their findings did not directly address the shift chart, yet the announcement reignited questions about whether the statistical evidence had carried disproportionate weight.

Media reports quickly paired the panel’s medical conclusions with earlier statistical critiques. Headlines asked whether the jury had been shown a misleading pattern before the clinical picture was complete. Public discussion on platforms such as Reddit and X treated the two strands of evidence as intertwined, even though the panel’s remit was strictly medical.

The timing mattered. The panel report arrived while the Thirlwall Inquiry was still hearing testimony about hospital conditions. That overlap kept probability arguments in the news cycle and encouraged readers to revisit the original shift table with fresh doubts.

Royal Statistical Society guidance

The Royal Statistical Society released a 2022 report that warned investigators against cherry-picking clusters of deaths and then calculating the odds of any single employee appearing in that cluster. The society cited earlier cases such as Lucia de Berk in the Netherlands, where similar statistical framing later contributed to overturned convictions.

In 2024 the RSS issued a brief statement acknowledging member concerns about the Lucy Letby trial. It did not declare the nurse innocent, but it reiterated the need for independent statistical review in future healthcare inquiries. The statement received modest coverage in the UK but spread quickly among data-science communities online.

American readers familiar with the Sally Clark case recognized the pattern. In that earlier UK miscarriage, flawed probability evidence about sudden infant death helped secure a conviction that was later reversed. The parallel reinforced the view that statistical presentation can steer public and judicial belief even when medical facts remain contested.

Public amplification online

After the verdicts, social media threads distilled the shift chart into simple graphics that omitted the surrounding roster data. Users shared the image with captions such as “30 times more likely,” a figure drawn from a later media comment by prosecution expert Dewi Evans rather than from trial testimony. The number traveled faster than any correction.

Substack writers and independent statisticians posted detailed breakdowns that reached thousands of readers within days. Some analyses labeled the original chart an example of the Texas sharpshooter fallacy, others focused on base-rate neglect. The volume of commentary created an impression that the statistical foundation of the case was under broad professional attack.

At the same time, supporters of the convictions pointed out that the medical evidence, not the chart alone, had driven the guilty verdicts. They argued that public focus on probability questions risked overshadowing the clinical testimony. The split in online discourse mirrored the larger debate over how much weight any single piece of evidence should carry.

Media framing and perception

UK broadsheets initially presented the shift chart as compelling corroboration of the medical narrative. Later coverage in the same outlets began to include caveats from statisticians and the 2025 expert panel. The shift in tone tracked the release of new information rather than any change in editorial stance.

US outlets, less invested in the original story, picked up the statistical controversy as a cautionary tale about data in courtrooms. Podcasts compared the case to American wrongful-conviction stories where flawed forensics had once seemed decisive. The narrative angle moved from “nurse on every shift” to “how did the numbers influence belief.”

Thirlwall Inquiry transcripts show that hospital managers also relied on the same probability language when deciding to alert police. Internal emails quoted in inquiry documents used phrases such as “highly unlikely to be chance.” Those early assumptions fed the investigation and later surfaced in public discussion of whether the statistical premise had narrowed the scope of inquiry from the start.

Comparisons to past miscarriages

Jane Hutton, a statistician who gave evidence to the inquiry, stressed that proper analysis requires looking at all deaths during the period, not only the subset linked to one nurse. Her point echoed the RSS guidance and the lessons drawn from Lucia de Berk. The comparison helped frame the Lucy Letby case as part of a recurring problem rather than an isolated event.

Each past case involved a cluster of unexpected deaths, a single suspect identified through presence data, and later statistical re-evaluation that challenged the original odds. Public memory of those reversals primed readers to question whether the same pattern was repeating. The parallel kept probability evidence in the foreground even as medical debates continued.

Advocacy groups in both the UK and the US circulated summaries that paired the Letby chart with earlier flawed testimony. The material reached audiences already attuned to true-crime coverage of miscarriages of justice. The repetition strengthened the perception that statistical presentation can lock in belief before alternative explanations receive equal scrutiny.

Inquiry testimony on probability

During the Thirlwall Inquiry, statisticians testified that the mortality spike at the Countess of Chester carried an estimated probability of roughly 0.008 when measured across the full patient population. That figure was presented without the selective filtering used in the shift chart shown to jurors. The difference highlighted how the choice of data set alters perceived likelihood.

Inquiry counsel asked whether hospital managers had received adequate statistical advice before contacting police. Witnesses answered that no independent statistician had reviewed the data package. The admission added weight to arguments that probability claims had driven investigative decisions without sufficient checks.

Transcripts also revealed that some staff had raised concerns about understaffing and equipment issues during the same period. Those contextual details were not part of the original shift-chart presentation. Their later emergence in inquiry evidence encouraged readers to reconsider how much the simplified probability narrative had shaped early public understanding.

Bayesian re-interpretations

Gill’s 2025 analyses applied Bayes’ theorem to the trial data by assigning likelihood ratios to both the prosecution theory and alternative explanations. The resulting posterior probability of innocence exceeded 99 percent under his chosen priors. While the calculation drew criticism for its assumptions, it demonstrated how different framing of the same numbers produces sharply different conclusions.

Other statisticians published shorter responses that avoided full Bayesian modeling yet still flagged the absence of base-rate information in the original chart. Their shorter pieces spread more widely on social platforms and kept the technical critique accessible. The combination of detailed and simplified critiques sustained public engagement with the probability questions.

These re-interpretations did not claim to prove innocence; they argued that the statistical evidence presented at trial had overstated certainty. The distinction mattered for readers tracking the case as an example of how courts handle complex data. It also kept pressure on the Criminal Cases Review Commission to consider statistical as well as medical grounds for appeal.

Statistical literacy going forward

The Lucy Letby case illustrates how a single visual and a few probability phrases can anchor public belief long before expert panels or inquiries revisit the record. Future healthcare investigations will likely face stricter requirements for independent statistical review, partly because of the attention this case has drawn. Whether that change prevents similar framing issues remains to be seen.