What the jury was shown
The prosecution showed the jury a selection of Facebook searches Lucy Letby had made for the surnames of parents whose babies had been on the neonatal unit. The selection was presented as pattern evidence: a nurse obsessively checking the families of her alleged victims. In opening and closing, the Crown framed these searches as sinister. The particular searches were timed at the point of particular events, and the implication put to the jury was that this was a form of grooming or post-event fixation.
What the full record looks like
The seized digital record from Letby’s phone and laptop contains tens of thousands of searches over a multi-year period. In that record are searches for:
- Hundreds of neonatal-unit colleagues, nurses, doctors and their families.
- Neighbours and acquaintances.
- Ex-partners and their families.
- School friends from years earlier.
- Celebrities, public figures, and random names encountered in news stories.
- Her own name and her own photographs.
- Routine lifestyle and holiday searches.
The searches for families of babies on her unit are a small fraction of the total. The fraction is consistent with what one would expect from a young woman with active social media habits. The Crown’s selection picked out those that followed a charged event. The defence position was that if the same selection lens were applied to any nurse with comparable Facebook habits, a similarly “sinister” pattern could be constructed.
The base-rate problem
This is the same structural error as the shift-rota chart. A base-rate-neutral reading of the record would ask: given N searches per day on Facebook, and given M babies on the unit, how many search-for-family events would you expect purely from random variation over the year? The answer is not zero. Once the jury was shown only the “suspicious” subset, the base rate was out of the conversation.
Prof. Richard Gill and Prof. Peter Green have written publicly on how the Facebook evidence replicates the selection-effect pattern of the shift chart. The point is not that the searches did not happen. The point is that presenting a curated subset without the denominator is, in statistical terms, meaningless.
What neonatal nurses actually do
Several experienced neonatal nurses have since written and spoken publicly about the professional reality of working on a NICU. Nurses on a high-acuity unit form attachments to the babies and families they care for. After a death, they think about the family. They sometimes attend funerals. They check whether the family is doing alright. Some look them up on social media. This is not universal practice but it is a recognisable one, and it is not inherently pathological.
The Crown’s Facebook framing presented it as pathological. The independent nursing commentary — including from nurses who had worked on the Countess of Chester unit themselves — presents it as within normal range, particularly for a young nurse without established distancing practices.
The “birthdays” sub-claim
A particular prosecution sub-argument was that some searches were timed around children’s birthdays or anniversaries of deaths. The defence response addressed two things. First, these are the points in the year when families themselves are active on social media (memorial posts, anniversary posts) and therefore when a search might naturally surface them. Second, a nurse thinking of a family on an anniversary is not evidence of criminal intent; it is evidence of memory.
Letby’s own testimony addressed this directly. See our summary of her testimony.
Why this matters
The Facebook evidence is of the same character as the Post-it notes and the shift chart: a body of ambiguous material from which the prosecution selected the subset that supported its narrative. The defence had to address each selection on the merits. The jury was not systematically walked through the denominator. In a post-Sally-Clark criminal justice system, the denominator is not optional.