Hacker News Reader

Comment by Tycho | Hacker News Reader

Yesterday in England, they published statistics that showed (among other things) the rate of non-covid deaths among vaccinated people greatly increased from March to August. By a factor that dwarfs the improvement in covid-related deaths. For some reason they didn't feel it was necessary to comment on or explain this surprising pattern. I can think of a number of explanations, but it makes me wonder if they are really on the ball, and would really make sound judgements on other matters such as natural immunity.

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsde...

View on HN

There are lots of experts who regularly interpret this data.

Your comment is pretty vague, so I looked at the excel sheet myself - I assume you’re referring to table 5, “Second dose”? The highest death rate in that table is 14.6 per 100k on 2-Jul-21. Meanwhile, looking at the unvaccinated population in that same table, the death rate per 100k starts out at 22 per 100k, goes as high as 66 per 100k in February, and then comes back down and is in the 30s and 20s per 100k.

A naive interpretation of this data would suggest that being vaccinated is protecting people from murder and other non-COVID causes of death. Obviously this isn’t right (I’m not immediately sure why though), but I’d suggest that your suggestion that the NHS and other experts are unaware of patterns in the data may just be the result of you being unaware of how patterns in this type of mortality data typically work.

You didn’t notice what is happening in the “21 days after first dose” column? It went from 22 to 100!

And if the NHS and other experts noticed it, they for some didn’t include an explanation along with their other comments on the data.

I had a look and have a job figuring which bit you are looking at - I assume its from "Deaths occurring between 2 January and 2 July 2021 edition of this dataset"? Though that isn't really "March to August."

Have you considered that at the start of the year few people would have died 21 days after vaccination because that had only just happened. 6 months later more would have died as they would have had 6 months to get dying of various stuff.

Yes it starts in January but I mentioned March because it took some time for a large number of people to actually get the vaccine, and also there was probably flu going around in Jan/Feb. I should have said July not August.

The numbers reflect weekly, age standardised death rates. Therefore once you have a decent population size in each cohort, the timing shouldn’t matter. These are non-covid deaths so rates shouldn’t change except for normal seasonal patterns.

However it could be something like what JoshuaDavid mentions below - that healthy people are disproportionately being moved from the 21 days cohort to the second dose cohort by some selection mechanism.

If the vaccine was dangerous and resulting in “non covid” deaths, why would the mortality be lower for people who had a second dose than for people who just had 1 dose?

Commented separately but it’s pretty clear there’s some sort of selection effect going on with the people who are not getting a 2nd dose that makes them more likely to die of non-covid causes.

The 2nd dose cohort at that point had quite a different age profile (they were rolling out doses oldest to youngest). Over 50 age bands were 90%+ double dosed but 60% or less for under 50 age bands. It’s possible that older cohorts were out-performing typical mortality because of something like lack of flu due to lockdown.

Another possibility is that the first dose is an absorbing barrier if it causes so much illness and death that victims do not get a second dose. The two-dosed cohort are the “survivors” with the vulnerable already removed from the population.

Do you know how the rate compares to 2019?

Without looking extensively, it seems that it's usually between 1000 and 1500 per 100k per annum, so between 20 to 30 per week as a direct comparison. This seems steady over recent history. This compared to a rate for the vaccinated that went from 20 to 100 over spring and summer. The only thing that springs to mind is that maybe the treatment of with-covid/from-covid deaths became less significant through the year and previously understated alternative causes start showing up in non-covid deaths. Still strange that it has risen so high in absolute terms though.

https://www.ons.gov.uk/peoplepopulationandcommunity/birthsde...

The “100” is specifically for the population of 21 days after first dose, i.e. those who didn’t get their second dose of vaccine. You can see the population who fall into the “21 days after” shrinks the whole time the “Second dose” population grows. Probably there’s something specific to those people who didn’t get or were unable to get the second dose of the vaccine that elevates their risk of mortality, like poverty, low education, or poor health.

It would need to be a heck of a big effect to drive that sort of elevation in mortality.

I suppose conceivably there is a strong adverse selection/sorting effect happening, but I’m not sure that fits with the pattern. Need to think about it. (Note also the upward trend of the second-dosed.) I think it’s more likely an issue with the way the data is collected.

I would not be surprised if there _was_ a heck of a big effect -- something along the lines of "how bad the vaccine side effects hit you is strongly correlated with general health such that people who are less healthy tend to have worse side effects, and people who have particularly bad side effects from their first dose are less likely to get a second dose".

I don't have time to look into it right now, but if that were the case we should expect the mortality rate for single-dosers to go up two weeks after their first dose in countries with a two week gap between doses and twelve weeks after their first dose in countries with a twelve week gap (i.e. people in the 12 week gap countries who are 6 weeks after dose 1 should have much lower mortality than people in the 2 week gap countries who are 6 weeks after dose 1 and have not gotten a dose 2, even after controlling for age and comorbidities).

I took a look at the data, and it looks like deaths after 1 dose go up by ~3x vs unvaccinated, while deaths after 2 doses go down ~3x vs unvaccinated. Am I reading that right? Any idea why that might be?

Are you referring to table 5, non-covid deaths? 2-dosed people have a lower absolute rate, but it still trends steadily upwards while unvaccinated trend steadily downwards.

Or are you referring to the deaths from covid?

Reply on news.ycombinator.com