arxiv: 2605.07421 · v1 · submitted 2026-05-08 · 📊 stat.AP

Recognition: no theorem link

There to care; not to kill: medical settings, statistics and wrongful convictions

Richard D. Gill

Pith reviewed 2026-05-11 01:53 UTC · model grok-4.3

classification 📊 stat.AP

keywords wrongful convictionsstatistical evidencemedical settingsnurse prosecutionshospital mortalitycriminal justice

0 comments

The pith

Statistical spikes in deaths during a nurse's shifts can lead to wrongful convictions when no direct evidence exists.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper examines wrongful convictions of nurses in hospitals where the main evidence consists of statistical associations between the nurse's presence and clusters of patient deaths or collapses. It notes the consistent absence of direct proof such as eyewitness accounts, DNA from tampering, CCTV footage, confessions, or prior suspicious behavior, with initial medical records instead attributing events to natural causes. A sympathetic reader would care because these cases show how statistics can substitute for missing evidence of actual harm, potentially sending innocent caregivers to prison. The author emphasizes that police investigations are often driven by the same hospital consultants responsible for the patients' care, which may introduce bias into the interpretation of the numbers.

Core claim

In medical settings, nurses can be convicted of harming or killing patients primarily on the basis of a statistical spike in adverse events coinciding with their shifts, even though coroners issued natural-death certificates at the time, hospital staff saw nothing suspicious, and no physical or direct evidence of wrongdoing by the nurse is available.

What carries the argument

The central mechanism is the statistical association between a particular nurse's duty periods and elevated rates of deaths or collapses, treated as the primary indicator of deliberate harm.

If this is right

Convictions can rest on reinterpreted private writings or ordinary behavior as supposed confessions without establishing motive.
Forensic findings may implicate the nurse in only one or two incidents while the statistical pattern is used to allege a broader series.
Investigations led by the clinically responsible consultants may overlook alternative explanations for the events.
Similar cases may arise whenever statistical clustering substitutes for missing direct evidence of causation.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same statistical approach could affect other roles involving high patient contact, such as doctors or care assistants.
Legal teams might benefit from routine consultation with statisticians to evaluate whether observed clusters exceed expected natural variation.
Hospitals could track baseline mortality rates by shift and ward to provide context for any later investigations.

Load-bearing premise

Statistical patterns of increased deaths or incidents during one nurse's shifts reliably indicate that nurse's deliberate wrongdoing rather than coincidence or other medical or staffing factors.

What would settle it

A reanalysis of one of the cases that accounts for variables such as total patient numbers, staff levels, or seasonal effects and finds the statistical association no longer holds, or an independent investigation that uncovers direct non-statistical proof of the nurse's guilt.

read the original abstract

This paper discusses wrongful convictions in a medical setting, focusing on nurses. Common features are lack of strong direct evidence: the nurse was never seen doing anything wrong. There is no DNA evidence of tampering of apparatus or medications by the nurse. There is no CCTV footage showing suspicious actions. Analysis of medical records at the time led coroners to issue certificates of natural deaths, and most events were not, at the time, thought suspicious by hospital staff. There is no confession and the nurse consistently asserts they are completely innocent. There is no evidence of earlier psychopathic behaviour. Instead, private writings (e.g., in a diary) are interpreted by the prosecution as a confession; mundane behaviour is given a sinister interpretation. Motive remains speculation. The main evidence is statistical: a spike in deaths or collapses and a statistical association with a particular nurse. There is forensic evidence which suggests one or two patients might have been harmed by administration of medication much used in the hospital, and even legitimately used earlier in the care of the alleged victims. Police investigations are driven by the hospital consultants who were clinically responsible for the patients allegedly killed or harmed by the nurse.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Gill's paper usefully catalogs recurring patterns in nurse-related medical conviction cases driven by statistical spikes but offers no independent verification that the cases were wrongful rather than genuine crimes.

read the letter

The main thing here is that this paper describes a familiar set of features in cases where nurses face murder charges in hospitals: no direct evidence like DNA or CCTV, initial natural-death rulings by coroners, no staff suspicion at the time, and statistical clustering as the central link to one individual. Gill walks through how diaries get read as confessions, mundane notes turn sinister, and motive stays speculative while investigations often start from the responsible doctors themselves. That summary is clear and direct. It gives a reader a compact checklist of what these prosecutions tend to look like without burying the point in jargon. For someone who has not followed the specific cases, the description of how the statistical association becomes the main evidence is the part that lands cleanly. The paper does not claim new math or fresh data, and it does not pretend to. It stays at the level of pulling together observed patterns from existing incidents. The soft spot is exactly the one the stress-test flags. The argument treats the listed features as reliable markers of wrongful conviction, yet the text supplies no post-conviction outcomes, successful appeals, or re-analyses that would separate those patterns from cases where the nurse was actually guilty. Without that anchor the same description could fit either side, so the cautionary force rests on the reader's prior acceptance of the premise rather than on evidence presented here. It is a limitation, not a fatal one, because the piece is framed as discussion rather than proof. This is for statisticians or legal researchers who already care about evidence standards in medical settings. A reader who wants concrete examples of how base-rate and clustering issues play out in real prosecutions will get something from it, though they will still need to look up the individual cases for full details. I would send it for peer review. The topic is worth referee attention and the description is coherent enough to be useful once the evidential gap is noted.

Referee Report

2 major / 2 minor

Summary. The paper discusses common features of wrongful convictions involving nurses in medical settings, noting the lack of direct evidence (no DNA, CCTV, confessions, or observed wrongdoing), reinterpretation of mundane records and private writings as sinister, speculative motives, and the central role of statistical spikes in deaths or collapses associated with a particular nurse. It argues that police investigations driven by hospital consultants can lead to miscarriages of justice based primarily on these statistical associations rather than forensic proof.

Significance. If the patterns described reliably distinguish wrongful convictions from undetected guilt, the work could contribute to applied statistics literature by illustrating risks of overinterpreting associations in forensic medical contexts and the prosecutor's fallacy in low-base-rate events. However, the absence of any quantitative analysis, verification of case outcomes, or comparison to control data limits its potential impact to a descriptive cautionary note.

major comments (2)

[Abstract] Abstract: the assertion that statistical spikes and associations constitute the 'main evidence' leading to convictions is presented without any verification that the referenced cases were in fact wrongful (e.g., via exonerations, successful appeals, or re-analyses); this is load-bearing for the central claim that the listed features reliably signal miscarriages rather than actual guilt, as the same pattern could describe either outcome.
[Full text] Full text (discussion of statistical evidence): no quantitative details, baseline rates, p-values, or error analysis are supplied for the 'spike in deaths' or nurse associations, nor is there comparison to expected variation in similar hospital wards; this leaves the critique of statistical misuse unsupported by the tools of the field.

minor comments (2)

The manuscript would benefit from explicit section headings (e.g., 'Case Features', 'Statistical Critique') to improve readability and separate narrative from analysis.
Add citations to relevant statistical literature on base-rate neglect or forensic misuse of coincidence (e.g., works on the prosecutor's fallacy) to ground the discussion.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive comments. Our manuscript is a discussion paper examining patterns in convictions involving nurses in medical settings where statistical associations formed the primary evidence in the absence of direct proof. We address the major comments below and will make revisions to improve clarity and scope.

read point-by-point responses

Referee: [Abstract] Abstract: the assertion that statistical spikes and associations constitute the 'main evidence' leading to convictions is presented without any verification that the referenced cases were in fact wrongful (e.g., via exonerations, successful appeals, or re-analyses); this is load-bearing for the central claim that the listed features reliably signal miscarriages rather than actual guilt, as the same pattern could describe either outcome.

Authors: We agree that the abstract should more precisely describe the evidential basis of the cases. The paper examines convictions in which statistical associations with a particular nurse were central to the prosecution case, despite the absence of direct evidence such as DNA, CCTV, or confessions, and where medical records at the time indicated natural causes. To address this, we will revise the abstract to state that we discuss cases in which convictions relied primarily on such statistical evidence and where expert and public debate has raised concerns about potential miscarriages of justice. We will note the status of referenced cases (e.g., appeals or ongoing scrutiny) without claiming universal exoneration. This clarifies that the features are presented as raising risks of overinterpretation rather than as definitive proof of wrongful conviction in every instance. revision: partial
Referee: [Full text] Full text (discussion of statistical evidence): no quantitative details, baseline rates, p-values, or error analysis are supplied for the 'spike in deaths' or nurse associations, nor is there comparison to expected variation in similar hospital wards; this leaves the critique of statistical misuse unsupported by the tools of the field.

Authors: The manuscript is a qualitative discussion of how statistical evidence has been used in legal contexts rather than a quantitative statistical analysis. We intentionally avoid p-values and formal error analysis because the core argument concerns the legal weight given to associations without supporting forensic evidence or proper contextual baselines. We will add a short paragraph in the full text referencing established medical literature on typical mortality variation in hospital wards and the known difficulties of interpreting small-number spikes (including the prosecutor's fallacy in low-base-rate settings). This will support the critique without shifting the paper to a data-driven study, for which individual-level hospital data are not publicly available. revision: partial

Circularity Check

0 steps flagged

No significant circularity; paper is a descriptive commentary on external cases with no derivations, fits, or self-referential reductions.

full rationale

The paper is a discussion of patterns in medical wrongful-conviction cases, emphasizing reliance on statistical spikes and associations in the absence of direct evidence. No equations, parameters, predictions, or ansatzes appear in the provided text or abstract. The central observations are drawn from external cases and do not reduce by construction to the paper's own inputs or self-citations. The argument therefore contains no load-bearing steps of the enumerated kinds and is self-contained as commentary.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

The paper is a discussion of existing legal and statistical practices in medical conviction cases and introduces no new parameters, axioms, or postulated entities.

pith-pipeline@v0.9.0 · 5495 in / 993 out tokens · 46410 ms · 2026-05-11T01:53:34.304442+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

13 extracted references · 13 canonical work pages

[1]

World Health Organisation

URL https://www.njad vocates.com/2023/08/07/the-most-common-causes-of-death-in-hospitals-and-healthcare-set tings. World Health Organisation. Patient safety,

work page 2023
[2]

Ania Zylbersztejn, Ruth Gilbert, Anders Hjern, Linda Wijlaars, and Pia Hardelid

URL https://www.who.int/news-room/fact-sheets/detail /patient-safety. Ania Zylbersztejn, Ruth Gilbert, Anders Hjern, Linda Wijlaars, and Pia Hardelid. Child mortality in england compared with sweden: a birth cohort study.The Lancet, 391(10134):2008–2018, May

work page 2008
[3]

doi: 10.1016/s0140-6736(18)30670-6

ISSN 0140-6736. doi: 10.1016/s0140-6736(18)30670-6. URLhttp://dx.doi.org/10.1016/S0140-6736(18)30670-6. Thirlwall Inquiry. Thirlwall Inquiry: Evidence,

work page doi:10.1016/s0140-6736(18)30670-6
[4]

doi: 10.1111/j.1556-4029.2006.00273.x

ISSN 1556-4029. doi: 10.1111/j.1556-4029.2006.00273.x. URL http://dx.doi.org/10.1111/j.1556-4029.2006 .00273.x. Elizabeth Yardley and David Wilson. In search of the ’angels of death’: Conceptualising the contemporary nurse healthcare serial killer.Journal of Investigative Psychology and Offender Profiling, 13(1):39–55, November

work page doi:10.1111/j.1556-4029.2006.00273.x 2006
[5]

doi: 10.1002/jip.1434

ISSN 1544-4767. doi: 10.1002/jip.1434. URLhttp://dx.doi.org/10.1002/jip.1434. R. Meester, M. Collins, R. Gill, and M. van Lambalgen. On the (ab)use of statistics in the legal case against the nurse lucia de b.Law, Probability and Risk, 5(3–4):233–250, February

work page doi:10.1002/jip.1434
[6]

doi: 10.1093/lpr/mgm003

ISSN 1470-840X. doi: 10.1093/lpr/mgm003. URLhttp://dx.doi.org/10.1093/lpr/mgm003. Richard D. Gill, Piet Groeneboom, and Peter de Jong. Elementary statistics on trial—the case of lucia de berk. CHANCE, 31(4):9–15, October

work page doi:10.1093/lpr/mgm003
[7]

doi: 10.1080/09332480.2018.1549809

ISSN 1867-2280. doi: 10.1080/09332480.2018.1549809. URL http: //dx.doi.org/10.1080/09332480.2018.1549809. Francesco Dotto, Richard D Gill, and Julia Mortera. Statistical analyses in the case of an italian nurse accused of murdering patients.Law, Probability and Risk, 20(3):169–193,

work page doi:10.1080/09332480.2018.1549809 2018
[8]

doi: 10.1093/lpr/mgac007

ISSN 1470-840X. doi: 10.1093/lpr/mgac007. URLhttp://dx.doi.org/10.1093/lpr/mgac007. Richard D. Gill, Norman Fenton, and David Lagnado. Statistical issues in serial killer nurse cases.Laws, 11(5):65, August

work page doi:10.1093/lpr/mgac007
[9]

doi: 10.3390/laws11050065

ISSN 2075-471X. doi: 10.3390/laws11050065. URL http://dx.doi.org/10.3390/laws11050

work page doi:10.3390/laws11050065 2075
[10]

Royal Statistical Society

URL https://rss.org.uk/news-publication/news-p ublications/2022/section-group-reports/rss-publishes-report-on-dealing-with-uncertain ty-i/. Royal Statistical Society. N. Ross.A Picture of Beverley Allitt

work page 2022
[11]

doi: 10.1177/00258024251404604

ISSN 2042-1818. doi: 10.1177/00258024251404604. URLhttp://dx.doi.org/10.1177/00258024251404604. S. Knapton and P. Elston. Spike in deaths at letby hospital ’could be explained by how small and premature babies were’; and: The scientific case against lucy letby (part two): the explicable spike.The Telegraph,

work page doi:10.1177/00258024251404604 2042
[12]

Dominik Rozkrut, Włodzimierz Okrasa, Oleksandr H

URL https://www.telegraph.co.uk/news/2024/09/06/spike-in-deaths-at-letby-hospital-could-b e-explained/. Dominik Rozkrut, Włodzimierz Okrasa, Oleksandr H. Osaulenko, Misha V . Belkindas, and Ronald L. Wasserstein. The post-conflict reconstruction of the statistical system in ukraine. key issues from an international perspective. Statistics in Transition ne...

work page 2024
[13]

doi: 10.59170/stattrans-2023-001

ISSN 2450-0291. doi: 10.59170/stattrans-2023-001. URLhttp://dx.doi.org/10.59170/stattrans-2023-001. 12

work page doi:10.59170/stattrans-2023-001 2023