Clear guidance from the expert who was dropped at the start of the Letby investigation, Prof. Jane Hutton. Three lightly edited excerpts: (1) "I'm frankly ashamed, as a British citizen, by the outright failure of English doctors, police and lawyers to follow basic principles of collecting evidence and evaluating experience. Faster misapplication of statistics will not improve care or prevent harm. The elephant in the room is that the hospital inquiry, the police inquiry and the court judgments have all avoided expert analysis of the variation in patterns of deaths and collapses. Lady Justice Thirlwall has ignored warnings from the House of Commons from almost 20 years ago. The Department of Health, Public Health England and the Forensic Science Regulator have all produced clear guidance, and it has been ignored. There is no reference that I can find in the report to the fact that reliable methods already exist. We do not need another set of documents. What we need to do is use the existing documents rather than repeat the mistakes. The flaws in the statistical evidence have not been tracked from the beginning. Doctors with no expertise in statistics are allowed to give statistical opinions." (2) "One of the consultants knew that he should involve statisticians but didn't. All of the consultants and the management should have been familiar with Department of Health guidance from 2011. That basically says that if you have an outlier, you must consult a qualified statistician, and you must ensure good data collection. That was ignored. And indeed, the report is contradictory by its own standards. The report claims that the deaths of triplets O and P might have been avoided. Sadly, those deaths were predictable before the children were even born. They were identical triplets. By the criteria of Thirlwall's own report, where she says that a probability of one in ten is "almost certain", we can be almost certain that at least one of those babies would die. Even if there had been no problem during the pregnancy – which was not the case – in those pregnancies there is a death of at least one baby about one time in four. [Editor's note: We understand Hutton to be referring to the elevated prior mortality risk in monochorionic triplet pregnancies, not asserting that O and P's individual deaths were predictable or necessarily natural. In a multicentre cohort (Sileo et al., 2023 [*]), among 67 monochorionic-triamniotic pregnancies without fetal anomalies or antenatal chorionicity-related complications, all three infants survived beyond the neonatal period in only 74.6%; in 4.5%, only one survived. Because the three infants share the same pregnancy and, in monochorionic pregnancies, the same placenta, their outcomes are correlated rather than statistically independent.] This exercise has simply been an exercise in ignoring the fundamental problem, which is people avoiding science that has been developed over a very long period. That is why the correct action was not taken." (3) "Unlike, say, the national air-traffic-control system, NHS culture is not conducive to improving quality. I remember pointing this out when quality management was introduced in the late 1980s. The NHS and government are not interested in well-established methods. Deming established these ideas in Japan in the 1950s, contributing to the Japanese quality revolution. The Royal Statistical Society has published Performance Measurement: The Good, the Bad and the Ugly. Governments of all shapes, sizes and colours continue to ignore this. It is not that statistics has been tried and failed. Statistics is found difficult and therefore ignored. The recommendations will not solve this. They do not include statisticians. The existing advice on investigating clusters makes it clear that you should have an epidemiologist or two involved. Yet the recommendation for the panel of experts does not mention epidemiologists or statisticians. Recommendation 9 proposes developing a protocol and two-page guidance. I would rather my taxes were not wasted reinventing the wheel. The protocol was published in 2019. Figure 1 on page 12 gives a very clear overview of what should be done. I have a detailed written discussion of all the mistakes that were made when judged against these already-established protocols. It is a lack of willingness to confront statistical illiteracy." ------------------- Alas, these are not new criticisms. On 22 August 2024, The Economist ran an article titled "The trial of Lucy Letby has shocked British statisticians." Its deeper point was cultural: lawyers and other decision-makers may not merely lack statistical knowledge; they may also lack awareness of what they do not know. Arguably, the most depressing point is therefore not simply that statistics may have been mishandled in the Letby case. It is that, even after Thirlwall, the proposed expert machinery still does not explicitly require statisticians or epidemiologists. [*] Sileo FG, Accurti V, Baschat A, et al. "Perinatal outcome of monochorionic triamniotic triplet pregnancy: multicenter cohort study." Ultrasound in Obstetrics & Gynecology. 2023;62(4):540–551. doi:10.1002/uog.26256. Open Access paper. obgyn.onlinelibrary.wiley.co… #LucyLetby