Other parts: rootclaim.html, rootclaim2.html, rootclaim3.html, rootclaim4.html, rootclaim5.html.
Kirsch posted the plot below and wrote: "Every time you compare the two cohorts, the more vaccinated cohort will have mortality even though they tracked at the start. This is NOT NORMAL. If they track at the start, they should continue to track. If the shots worked, the more vaxxed cohort would exhibit LOWER mortality, not higher mortality." [https://kirschsubstack.com/p/japan-cmrr-data-website-shows-clear]
By "mortality", Kirsch meant the cumulative number of deaths multiplied by a constant factor, which he chose so that his two lines were roughly aligned at the start of the x-axis. But if by mortality you mean ASMR instead, then the 4-dose cohort has lower mortality than the 3-dose cohort even at the end of the x-axis.
At the start of the x-axis, people with 4 doses have reduced mortality because of the healthy vaccinee effect, and people with 3 doses have elevated mortality because of the straggler effect. But over time the ASMR of the two cohorts converges.
Kirsch aligned the lines so that they matched at the start of the x-axis, but when the line for dose 4 subsequently rose above the line for dose 3, Kirsch attributed the increase to harm caused by vaccines. But if you align the lines so that they are matched at the end of the x-axis instead, then it rather looks like dose 3 has high mortality at the start of the x-axis relative to dose 4, which can be explained by the straggler effect: [https://medicalfacts.info/kcor.rb]
Kenji Fujikawa has now published CSV files of FOI data from 7 cities here: https://fujikawa.org/pub/kkcor/. I used his files to make the next plot, which shows in late 2022 when the 4th dose is rolled out, the mortality of people with 3 but not more doses shoots up because of the straggler effect. But by late 2023 the straggler effect has weakened, so people with 3 but not more doses have almost as low mortality as people with 4 or more doses:
system("wget -rl1 -np -nd -nH -Acsv https://fujikawa.org/pub/kkcor")
ma=\(x,b=1,f=b){x[]=rowMeans(embed(c(rep(NA,b),x,rep(NA,f)),f+b+1),na.rm=T);x}
t=do.call(rbind,lapply(Sys.glob("jp*WKA.csv"),fread,fill=T))
date1=as.Date("2021-2-21");date2=as.Date("2024-6-30")
d=t[,.(.I,dead=date_death,age,dose=rep(0:4,each=.N),vax=c(rep(date1,.N),as.IDate(unlist(.SD,,F)))),.SDcols=patterns("date_dose[1-4]")]
d=d[!is.na(vax)]
d=d[dose!=2][dose>2,dose:=dose-1]
e=d[,.(pop=.N),.(date=vax,age,dose)]
e=rbind(e,e[dose>0][,.(pop=-pop,dose=dose-1,date,age)])
dead=d[!is.na(dead)][.N:1][rowid(I)==1,.(dead=.N),.(age,dose,date=dead)]
e=rbind(e,dead[,.(age,dose,date=date+7,pop=-dead)])
e=merge(e,dead,all=T)
e=e[date>=date1&date<=date2]
a=merge(do.call(CJ,lapply(e[,1:3],unique)),e,all=T);a[is.na(a)]=0
a[,pop:=cumsum(pop),.(dose,age)]
a[,`:=`(dead=ma(dead,2),pop=ma(pop,2)),.(dose,age)]
a[,base:=pop*a[,sum(dead)/sum(pop),age]$V1[factor(age)]]
a=rbind(a,a[dose>0][,dose:=4],copy(a)[,dose:=5])
a=a[,.(dead=sum(dead),pop=sum(pop),base=sum(base)),.(dose,age,date)]
p=a[,.(y=ifelse(sum(pop)<1e3,NA_real_,sum(dead)/sum(base))),.(x=date-3,z=dose)]
lab=c("Unvaccinated","Dose 1-2","Dose 3","Dose 4+","All vaccinated","All people")
color=c(hsv(c(0,27/36,21/36,1/3),c(.4,.4,.7,.7),c(.9,.9,.7,.7)),"gray60","black")
p[,z:=factor(z,,lab)]
xstart=as.Date("2021-1-1");xend=as.Date("2025-1-1");xbreak=seq(xstart+182,xend,"year")
ystart=0;yend=max(p$y,na.rm=T);ybreak=0:yend
ylim=p[,.(max=max(y,na.rm=T))]
ggplot(p)+
geom_hline(yintercept=ybreak,color="gray91",linewidth=.4)+
geom_rect(data=ylim,aes(ymax=max),xmin=xstart,xmax=xend,ymin=ystart,lineend="square",linejoin="mitre",fill=NA,color="gray72",linewidth=.4)+
geom_hline(yintercept=1,color="gray72",linewidth=.4)+
geom_vline(xintercept=seq(xstart,xend,"year"),color="gray72",linewidth=.4)+
geom_line(aes(x,y,color=z),linewidth=.65)+
labs(x=NULL,y=NULL,title="Weekly ratio of observed to expected deaths in Japan FOI data\n(±2-week moving average)")+
scale_x_date(limits=c(xstart,xend),breaks=xbreak,labels=year(xbreak))+
scale_y_continuous(limits=c(0,yend),breaks=ybreak)+
scale_color_manual(values=color)+
coord_cartesian(clip="off",expand=F)+
guides(color=guide_legend(row=2,byrow=F))+
theme(axis.text=element_text(size=11,color="gray50"),
axis.ticks.length=unit(0,"pt"),
legend.background=element_blank(),
legend.box.background=element_blank(),
legend.box.spacing=unit(0,"pt"),
legend.key=element_blank(),
legend.key.height=unit(12,"pt"),
legend.key.spacing.x=unit(2,"pt"),
legend.key.spacing.y=unit(0,"pt"),
legend.key.width=unit(24,"pt"),
legend.margin=margin(,,3),
legend.position="top",
legend.spacing.x=unit(2,"pt"),
legend.spacing.y=unit(0,"pt"),
legend.text=element_text(size=11,vjust=.5,margin=margin(,,3)),
legend.title=element_blank(),
panel.background=element_blank(),
panel.grid=element_line(),
plot.title=element_text(size=11,hjust=.5,face=2,margin=margin(,,3)))
ggsave("1.png",width=5.5,height=3.2,dpi=300*4)
system("mogrify -trim 1.png;magick 1.png \\( -size `identify -format %w 1.png`x -font Arial -interline-spacing -3 -pointsize $[42*4] caption:'Source: fujikawa.org/pub/kkcor. Includes about 2.3 million people from the cities of Koganei, Sagamihara, Hamamatsu, Makinohara, Kasugai, Toyokawa, and Saiki. The expected number of deaths for each week was calculated by multiplying the number of people for each age group with the total mortality rate of the age group over the observation period (week 7 of 2021 to week 26 of 2024). Weeks with population size below 1,000 people are not shown. People were assigned a fixed 10-year age group so that aging over time was not accounted for, which causes the observed-to-expected ratio to be too high in 2024 relative to 2021.' -splice x$[16*4] \\) -append -trim -resize 25% -bordercolor white -border 20 -dither none -colors 256 1.png")