The "statistical" evidence, as presented, is worthless. It does not analyze the data in any quantitative way. It collects some unusual facts, waves a wand, and with a theatrical whisper says "guilty". Neither is an accepted method of statistical analysis.

To put some meat on this claim, here's some things that should have been included:

- consider whether LL was given sicker children to look after

- more generally, consider other statistical factors that could have contributed: number and experience of doctors on shift, of patients, etc.

- consider whether night and day shifts had the same mortality

- consider how many "suspicious" deaths occured when other nurses were on duty

- consider wider nationwide factors affecting child mortality. E.g. is it broadly on the rise, maybe due to general deteriorating health of the population.

- once some kind of assigning of deaths was done, how much of an outlier was LL's "count"

- ...and finally comparing to some kind of base rate of how often nurses kill their patients. If you conclude there was a 1:1000 probability of these events occuring by chance, but a base rate of nurses killing patients is 1:10000, then it's still 10:1 more likely LL is innocent.

As far as I can tell none of the above was done properly. Some things were vaguely looked at, but in an anecdotal "analysis", not something containing actual mathematics.

Of course it's possible that you do this analysis and conclude LL is guilty - I don't know that. But the analysis, as summarized, is to a proper statistical evidence what a child's drawing is to a CCTV screenshot.