Looking at the first table, the vaccinated cohort was obviously WAY less healthy to begin with. As expected; sick people were more inclined to protect themselves. Especially with the cutoff for inclusion being May 2021, presumably before younger, healthier people were being vaccinated.
Am I missing something in the methods that would control for this or... is it just that dumb?
Not sure about that. While there is a small bias, I'm not sure it sufficient to significantly nudge the result. For example, hypertension comorbidity is 262,659 / 842,974 in vax vs. 207,862 / 842,974 in unvax.
I'm not sure I'd call that hypertension difference small, and it's by far not the largest difference. There's ~30% more diabetes in the vaccinated cohort. 2x the rate of kidney failure.
Note that the data was based on vaccinations up until May 2021. Most European countries (I'm assuming Sweden was the same) prioritised vacinating high risk groups first, and only opened up vaccinations to the general population in June or later. So the vaccinated group is most likely a group of older people who are more at risk than the general population.
The data seems to reflect that as it only breaks down age ranges into above and below 80. Even the Israel data from a few months ago showed that someone who is 80 and fully vaccinated is at a higher risk than someone who is 25 and unvaccinated.
Great point, thanks for the clarification. Median age is 53 (Table 1) and they do breakdown by VE against symptomatic infection in <50, 50-64, 65-79, >= 80 age groups plus adjust for comorbidities. But for VE against severe infection they only breakdown in <80 and >=80, and VE in <80 holds at 82% at 6 months. Regrettably the respective table has no data for the 9 month data point, which is the most interesting moving forward.
Am I missing something in the methods that would control for this or... is it just that dumb?