When the numbers do add up.

Jamie Egan

At the climax of the farce that was the Letby trial, M'Lud Goss laid on the lash with gusto. To be fair, what else was he supposed to do? He had his part to play, too.

As an example of how successfully Goss had barred the way to all reason in the trial, he said Letby appeared to have a particular interest in twins. Three pairs of twins and one set of triplets were among her 13 victims and in babies born with vulnerabilities. All the babies, I suspect, were vulnerable, indeed the ones Letby was convicted of. That they were born in an open sewer seems not to pique M'Lud's curiosity about how legions of rampant superbugs impact vulnerability. There is no point at all, I suspect, in getting into the old correlation-causation debate with M'Lud. Letby's victims all happened to be in the highest risk groups for neonatal deaths, but they would say she had the smarts to pick these babies. Confirmation is one of the worst biases.

Death, M'Lud takes a particular interest in twins and even more so in triplets.

Mortality in neonates rises very sharply with falling weight and multiparity. In the U.K., the crude neonatal death rate per 1000 births is 1.8. When I say crude, it really is that. The number 1.8 per 1000 births does not account for anything other than the birth itself. No influences of gestation, weight, parity, place, or picture are highlighted in that number. However, things get a little more interesting when we become a little more granular.

MBBRACE is doing a good job, but its stats are difficult to navigate. Quite rightly, they concentrate on perinatal deaths, which combine stillbirths and early and late neonatal births. At the moment, we are only interested in the latter.

The CoGH had 17 deaths between June 2015 and July 2016, so a year thereabout. According to MBBRACE, 1.8 deaths are expected per 1000 births in that year. According to the RCPCH, the CoGH had 4000 births annually, so we can expect 7 deaths by nature. Previous or future years, like investment warnings, are only a guide. If 7 babies had died on the unit, the performance measured in mortality would be average.

So, we need to account for 10 deaths. How much extra activity would we need to achieve this, by chance?

Let's look at the Prediction of survival for preterm births by weight and gestational age: retrospective population-based study.

From this and other studies, we know that neonatal deaths per 1000 in the U.K. are 1.8.

According to the RCPCH report, the CoGH's admission population was 96% inborn, meaning the chance of cot occupancy was also 96%.

From the United States CDC:

Neonatal deaths per 1000 under 1500g is 178/1000.

Below 1000 grams, 267/1000.

These are absolute numbers and will account for all causes of death in those weight categories.

To move from 1.8 per 1000 to the 2015-2016 CoGH June to July period, we need 10 extra deaths. On average, how many extra deliveries under 1500 grams do we need to achieve this?

178/1000 x 57 = 10.1. Therefore, we need only one extra birth per week to get near that number if the babies are < 1500g.

267/1000 x 38= 10.1 Therefore, we need only one extra birth every 10 days if they are < 1000g.

Yet Dr Brearey said that "acuity and activity increases could not account for the difference in the death rate in the CoGH in this period.

I don't believe him, and the figures above probably mean you shouldn't either.