Share this comment
The age-normalized mortality risk in the Czech data is even higher for J&J than Moderna.
Relative to a baseline of unvaccinated people, the mortality risk in 2021-2022 was about 0.42 for Pfizer, 0.50 for AstraZeneca, 0.55 for Moderna, and 0.65 for J&J:
> t=fread("sars2.net/f/czbucketske…
> a=t[,.(dead=sum(dead),pop=sum(alive)),.(month,,type…
The age-normalized mortality risk in the Czech data is even higher for J&J than Moderna.
Relative to a baseline of unvaccinated people, the mortality risk in 2021-2022 was about 0.42 for Pfizer, 0.50 for AstraZeneca, 0.55 for Moderna, and 0.65 for J&J:
> t=fread("https://sars2.net/f/czbucketskeep.csv.gz")[dose<=1]
> a=t[,.(dead=sum(dead),pop=sum(alive)),.(month,,type,age=pmin(age,95)%/%5*5)]
> a[dose==0,type:="Unvaccinated"]
> a=a[type!="Other"&month%like%"202[12]"]
> a[,type:=relevel(factor(type),"Unvaccinated")]
> lm=glm(dead~type+factor(age)+month+offset(log(pop)),poisson,a)
> o=a[,.(dead=sum(dead),pyears=round(sum(pop)/365)),,type]
> coef=coef(lm)[paste0("type",o$type)]
> ci=qnorm(.975)*sqrt(diag(vcov(lm)))[paste0("type",o$type)]
> o[,rr:=exp(coef)][,low:=exp(coef-ci)][,high:=exp(coef+ci)]
> print(mutate_if(o[order(-pyears)],is.double,round,2),r=F)
type dead pyears rr low high
Unvaccinated 115901 10505884 NA NA NA # unvaccinated rr is 1.00
Pfizer 94035 8676742 0.42 0.42 0.42
Moderna 18032 825149 0.55 0.54 0.56
AstraZeneca 25548 764170 0.50 0.49 0.51
Janssen 6328 539457 0.65 0.64 0.67
Novavax 14 3648 0.44 0.26 0.75
I adjusted the mortality risk only for age group and observation month. The difference between the vaccine brands might be explained by confounders I didn't adjust for.
I assigned people to a vaccine brand based on the brand of their first dose, which was usually also the brand of their subsequent doses.
From this plot that shows ASMR by month and vaccine type, you can also see that on most months J&J had higher ASMR than the other three main vaccine types: https://sars2.net/czech2.html#ASMR_by_month_and_vaccine_type.