trastu/tranothing.r
In chjackson/adverse: Data on adverse events of adjuvant breast cancer treatment
library(readxl)
library(tidyverse)
filter <- dplyr::filter
matches <- dplyr::matches
library(devtools)
datf <- "DataTrastuzumab-Meta-analysis-doublechecked.xlsx"
datf <- sprintf("~/Dropbox/Breast cancer therapies meta-analysis/Trastuzumab/%s", datf)
## Read event names
row1 <- unlist(read_excel(datf, sheet="Trastuzumab v nothing", col_names=FALSE, range="A1:CBP1"))
aetype <- data.frame(row1=row1, stringsAsFactors=FALSE) %>% fill(row1) %>% unlist %>% unname
## Cleaning up event names by hand
aetype[aetype %in%
c("CARDIOVASCULAR DISORDERS (not reduced LVEF or congestive heart failure) GRADE 1 (or grade 1/2)",
"CARDIOVASCULAR DISORDERS (not reduced LVEF or congestive heart failure) GRADE 2",
"CARDIOVASCULAR DISORDERS (not reduced LVEF or congestive heart failure) GRADE 3 (or grade 3/4)",
"CARDIOVASCULAR DISORDERS (not reduced LVEF or congestive heart failure) GRADE 4"
)] <- "Cardiovascular disorders (not low LVEF/CHF)"
aetype[aetype=="Cardiac events, including decreased LVEF (not congestive heart failure)"] <- "Cardiac events"
aetype[aetype=="Infections (general, no neutropaenia)"] <- "Infections"
## Read grade names
row2 <- unlist(read_excel(datf, sheet="Trastuzumab v nothing", col_names=FALSE, range="A2:CBP2"))
grade <- data.frame(aetype=aetype, grade=row2, stringsAsFactors=FALSE) %>%
group_by(aetype) %>% fill(grade) %>% pull(grade)
eids <- 37:length(grade)
grade[eids][is.na(grade[eids])] <- "unknown"
grade <- gsub("WHO ","",grade)
grade <- tolower(grade)
grade <- gsub("grade 1 \\(or 1\\-2 where undifferentiated\\)","grade 1 or 1/2",grade)
grade <- gsub("grade 3 \\(or 3\\-4 where undifferentiated\\)","grade 3 or 3/4",grade)
grade <- gsub("grade 1 or 1/2 \\(when not differentiated\\)","grade 1 or 1/2",grade)
grade <- gsub("grade 1 \r\nor 1/2 when not differentiated","grade 1 or 1/2",grade)
grade <- gsub("grade 1 or 1/2 \\(when not differentiated\\)","grade 1 or 1/2",grade)
grade <- gsub("grade 3 \r\n\\(or 3\\+ where not subdivided\\)","grade 3 or 3/4",grade)
grade <- gsub("grade 3 \\(or 3\\+ where not subdivided\\)","grade 3 or 3/4",grade)
grade <- gsub("grade 3 or 3\\+ where not subdivided","grade 3 or 3/4",grade)
unique(grade)
## Not done anything with grades so far.
## For bisphosphonates, we just counted proportion of events that are serious.
## Leave until later, merge all grades for now
## For table we want number of grade 3 or above among those where grade was recorded
c("grade 3 or 3/4","grade 4", "grade 3-4")
eventgrade <- paste(aetype, grade, sep=" ; ")
aetypes <- unique(aetype[eids])
## Read arm in E / C format
row3 <- unlist(read_excel(datf,
sheet="Trastuzumab v nothing", col_names=FALSE, range="A3:CBP3"))
arm <- data.frame(row3=row3, stringsAsFactors=FALSE) %>% unlist %>% unname
arm[1:36] <- NA
varname <- data.frame(row3=row3, stringsAsFactors=FALSE) %>% unlist %>% unname
varname[eids] <- NA
fullnames <- ifelse(is.na(varname), eventgrade, varname)
fullnames[eids] <- paste(fullnames[eids], arm[eids], sep=" ; ")
# trtlabs <- c("Trastuzumab", "Nothing")
# trtlabs <- c("Trastuzumab_low", "Trastuzumab_mid", "Trastuzumab_high", "Nothing")
armnos <- c("E1","E2","E3","C1","C2")
fullnames[match(c("N Exp1", "N Exp2", "N Exp3", "N Control", "N Control2"), fullnames)] <-
paste("N",armnos,sep=";")
tranothing <- read_excel(datf, sheet="Trastuzumab v nothing", range="A4:CBP27", col_names=fullnames)
## label comparison pairs of 2x2-arm studies distinctly
tranothing$`Trial name`[tranothing$`Trial name` == "FinHer"] <- c("FinHer_do","FinHer_vi")
tranothing$`Trial name`[tranothing$`Trial name` == "GeparQuinto"] <- c("GeparQuinto_cy","GeparQuinto_do")
tranothing$`Trial name`[tranothing$`Trial name` == "FNCLCC PACS-04"] <- c("FNCLCC PACS-04_FEC","FNCLCC PACS-04_ED")
tranothing$`Trial name`[tranothing$`Trial name` == "H0649g -1"] <- c("H0649g -1a","H0649g -1b")
## get data of form 1 row per trial, event and arm, with N, r. DO FIRST
## on one row per trial and event, with risk and risk difference of each event labelling trt and control arms.
## for now do for any adverse events
aecolnames <- names(tranothing)[37:ncol(tranothing)]
aencolnames <- c(grep("N;", names(tranothing), value=TRUE), aecolnames)
varnames <- c("Trial name", "Trastuzumab arm", "Control arm", "Trastuzumab dose","metasteses",
aencolnames)
traarmtypegrade <- tranothing %>%
select(all_of(varnames)) %>%
gather(vname, x, all_of(aencolnames)) %>%
extract(vname, c("aetype","grade","armno"), "^(.+) ; (.+) ; (.+)$", remove=FALSE) %>%
extract(vname, c("aetypeN","armnoN"), "(.+);(.+)$") %>%
mutate(aetype = ifelse(aetypeN=="N", aetypeN, aetype)) %>%
mutate(armno = ifelse(aetypeN=="N", armnoN, armno)) %>%
select(-aetypeN, -armnoN) %>%
mutate(trt = ifelse(`Trastuzumab arm` == armno, "Trastuzumab",
ifelse(`Control arm` == armno, "Control",NA)))
## Events assigned to wrong control arm: C2 instead of C1
#c2ind <- traarmtypegrade$`Trial name`=="NCCTG N9831" & traarmtypegrade$aetype != "N" &
# traarmtypegrade$armno == "C2"
#c1ind <- traarmtypegrade$`Trial name`=="NCCTG N9831" & traarmtypegrade$aetype != "N" &
# traarmtypegrade$armno == "C1"
#traarmtypegrade$x[c1ind] <- ifelse(!is.na(traarmtypegrade$x[c2ind]),
# traarmtypegrade$x[c2ind], traarmtypegrade$x[c1ind])
#traarmtypegrade$x[c2ind] <- NA
# No these are correct. So for these events we don't have any data.
bugind <- traarmtypegrade$`Trial name`=="BCIRG 006" & traarmtypegrade$aetype=="Endometrial cancer"
c1ind <- bugind & traarmtypegrade$armno=="C1"
c2ind <- bugind & traarmtypegrade$armno=="C2"
traarmtypegrade$x[c1ind] <- 0
traarmtypegrade$x[c2ind] <- NA
traarmtypegrade <- traarmtypegrade %>%
filter(!is.na(x)) %>%
mutate(`Trastuzumab dose` = factor(`Trastuzumab dose`)) %>%
mutate(`Trastuzumab dose` = replace_na(as.character(`Trastuzumab dose`), "0")) %>%
mutate(`Trastuzumab dose` = recode(`Trastuzumab dose`, `0`="unknown dose", `
1`="mid", `2`="low", `3`="high")) %>%
mutate(trtdose = ifelse(trt=="Trastuzumab", paste(trt, `Trastuzumab dose`), trt)) %>%
mutate(trtdose = ordered(trtdose, levels=c("Control",
paste("Trastuzumab",
c("unknown dose","low","mid","high"))))) %>%
mutate(count = as.numeric(x)) %>%
rename(study = `Trial name`) %>%
mutate(studyid = match(study, unique(study)))
## Drop extra arms that Alex excludes for this analysis, e.g.
## NCCTG N9831 was sequential vs concurrent trastuzumab
## FNCLCC PACS-04 : trastu arms distinguished as FEC vs ED chemo
traarmtypegrade <- traarmtypegrade %>%
filter(!is.na(trt)) %>%
mutate(grade3 = ifelse(grade %in% c("grade 3-4","grade 3 or 3/4","grade 4"), "Grade 3+",
ifelse(grade %in% c("grade 1-2","grade 1 or 1/2","grade 2"), "Grade 1-2",
NA)))
saveRDS(traarmtypegrade, file="tranothing_byarmtypegrade.rds")
## merge all grades
notevents <- c("N","NA","Withdrawals because of AEs")
traarmtype <- traarmtypegrade %>%
group_by(study, studyid,`Trastuzumab arm`,`Control arm`, aetype, armno, trt, trtdose, metasteses) %>%
summarise(count = sum(count, na.rm=TRUE)) %>%
ungroup
## add denominator for each count.
denoms <- traarmtype %>% filter(aetype=="N") %>% select(studyid, armno, count) %>% rename(N = count)
traarmtype <- traarmtype %>%
arrange(aetype, study, armno) %>%
filter(! aetype %in% notevents) %>%
filter(!is.na(count)) %>%
droplevels %>%
left_join(denoms, by = c("studyid", "armno"))
tracon <- traarmtype %>%
filter(trt == "Control") %>%
select(study, studyid, aetype, count, N) %>%
rename(rcon = count, ncon = N)
trastudytype <- traarmtype %>%
filter(trt == "Trastuzumab") %>%
left_join(tracon, by = c("study", "studyid", "aetype")) %>%
droplevels()
## data by arm and event type, with all event grades merged
saveRDS(tracon, file="tracon.rds")
saveRDS(traarmtype, file="tranothing_byarmtype.rds")
## Fix this by hand, Gasparini, Alopecia, see
## https://core.ac.uk/download/pdf/53842161.pdf Table 4
## assume it is double counting people who have both grade 1-2 and grade 3-4 events
## since without this fudge we get num > denom
trastudytype[21,]$count <- 59
trastudytype[21,]$rcon <- 50
## Raw data on frequency of each type of event, aggregated over all studies
trastudytype <- trastudytype %>%
mutate(oddsact = ((count+0.5)/(N+1)) / (1 - (count+0.5)/(N+1)),
oddscon = ((rcon+0.5)/(ncon+1)) / (1 - (rcon+0.5)/(ncon+1)),
est = oddsact / oddscon,
selogor = sqrt(1/(rcon+0.5) + 1/(ncon-rcon+0.5) + 1/(count+0.5) + 1/(N-count+0.5)),
or = est,
orlower = exp(log(est) - qnorm(0.975)*selogor),
orupper = exp(log(est) + qnorm(0.975)*selogor),
a = count, b=N-count, c=rcon, d=ncon-rcon,
serd = sqrt(a*b/(a+b)^3 + c*d/(c+d)^3),
pact = count/N,
pcon = rcon/ncon,
riskdiff = pact - pcon,
rdlower = riskdiff - qnorm(0.975)*serd,
rdupper = riskdiff + qnorm(0.975)*serd,
brisk = rcon / ncon,
brisklower = qbeta(0.025, 0.5 + rcon, 0.5 + ncon - rcon),
briskupper = qbeta(0.975, 0.5 + rcon, 0.5 + ncon - rcon),
aetype = factor(aetype, labels=str_to_sentence(levels(factor(aetype))))
)
## data by study and event type
saveRDS(trastudytype, file="tranothing_bystudytype.rds")
chjackson/adverse documentation built on June 16, 2022, 4:53 p.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
library(readxl)
library(tidyverse)
filter <- dplyr::filter
matches <- dplyr::matches
library(devtools)
datf <- "DataTrastuzumab-Meta-analysis-doublechecked.xlsx"
datf <- sprintf("~/Dropbox/Breast cancer therapies meta-analysis/Trastuzumab/%s", datf)
## Read event names
row1 <- unlist(read_excel(datf, sheet="Trastuzumab v nothing", col_names=FALSE, range="A1:CBP1"))
aetype <- data.frame(row1=row1, stringsAsFactors=FALSE) %>% fill(row1) %>% unlist %>% unname
## Cleaning up event names by hand
aetype[aetype %in%
c("CARDIOVASCULAR DISORDERS (not reduced LVEF or congestive heart failure) GRADE 1 (or grade 1/2)",
"CARDIOVASCULAR DISORDERS (not reduced LVEF or congestive heart failure) GRADE 2",
"CARDIOVASCULAR DISORDERS (not reduced LVEF or congestive heart failure) GRADE 3 (or grade 3/4)",
"CARDIOVASCULAR DISORDERS (not reduced LVEF or congestive heart failure) GRADE 4"
)] <- "Cardiovascular disorders (not low LVEF/CHF)"
aetype[aetype=="Cardiac events, including decreased LVEF (not congestive heart failure)"] <- "Cardiac events"
aetype[aetype=="Infections (general, no neutropaenia)"] <- "Infections"
## Read grade names
row2 <- unlist(read_excel(datf, sheet="Trastuzumab v nothing", col_names=FALSE, range="A2:CBP2"))
grade <- data.frame(aetype=aetype, grade=row2, stringsAsFactors=FALSE) %>%
group_by(aetype) %>% fill(grade) %>% pull(grade)
eids <- 37:length(grade)
grade[eids][is.na(grade[eids])] <- "unknown"
grade <- gsub("WHO ","",grade)
grade <- tolower(grade)
grade <- gsub("grade 1 \\(or 1\\-2 where undifferentiated\\)","grade 1 or 1/2",grade)
grade <- gsub("grade 3 \\(or 3\\-4 where undifferentiated\\)","grade 3 or 3/4",grade)
grade <- gsub("grade 1 or 1/2 \\(when not differentiated\\)","grade 1 or 1/2",grade)
grade <- gsub("grade 1 \r\nor 1/2 when not differentiated","grade 1 or 1/2",grade)
grade <- gsub("grade 1 or 1/2 \\(when not differentiated\\)","grade 1 or 1/2",grade)
grade <- gsub("grade 3 \r\n\\(or 3\\+ where not subdivided\\)","grade 3 or 3/4",grade)
grade <- gsub("grade 3 \\(or 3\\+ where not subdivided\\)","grade 3 or 3/4",grade)
grade <- gsub("grade 3 or 3\\+ where not subdivided","grade 3 or 3/4",grade)
unique(grade)
## Not done anything with grades so far.
## For bisphosphonates, we just counted proportion of events that are serious.
## Leave until later, merge all grades for now
## For table we want number of grade 3 or above among those where grade was recorded
c("grade 3 or 3/4","grade 4", "grade 3-4")
eventgrade <- paste(aetype, grade, sep=" ; ")
aetypes <- unique(aetype[eids])
## Read arm in E / C format
row3 <- unlist(read_excel(datf,
sheet="Trastuzumab v nothing", col_names=FALSE, range="A3:CBP3"))
arm <- data.frame(row3=row3, stringsAsFactors=FALSE) %>% unlist %>% unname
arm[1:36] <- NA
varname <- data.frame(row3=row3, stringsAsFactors=FALSE) %>% unlist %>% unname
varname[eids] <- NA
fullnames <- ifelse(is.na(varname), eventgrade, varname)
fullnames[eids] <- paste(fullnames[eids], arm[eids], sep=" ; ")
# trtlabs <- c("Trastuzumab", "Nothing")
# trtlabs <- c("Trastuzumab_low", "Trastuzumab_mid", "Trastuzumab_high", "Nothing")
armnos <- c("E1","E2","E3","C1","C2")
fullnames[match(c("N Exp1", "N Exp2", "N Exp3", "N Control", "N Control2"), fullnames)] <-
paste("N",armnos,sep=";")
tranothing <- read_excel(datf, sheet="Trastuzumab v nothing", range="A4:CBP27", col_names=fullnames)
## label comparison pairs of 2x2-arm studies distinctly
tranothing$`Trial name`[tranothing$`Trial name` == "FinHer"] <- c("FinHer_do","FinHer_vi")
tranothing$`Trial name`[tranothing$`Trial name` == "GeparQuinto"] <- c("GeparQuinto_cy","GeparQuinto_do")
tranothing$`Trial name`[tranothing$`Trial name` == "FNCLCC PACS-04"] <- c("FNCLCC PACS-04_FEC","FNCLCC PACS-04_ED")
tranothing$`Trial name`[tranothing$`Trial name` == "H0649g -1"] <- c("H0649g -1a","H0649g -1b")
## get data of form 1 row per trial, event and arm, with N, r. DO FIRST
## on one row per trial and event, with risk and risk difference of each event labelling trt and control arms.
## for now do for any adverse events
aecolnames <- names(tranothing)[37:ncol(tranothing)]
aencolnames <- c(grep("N;", names(tranothing), value=TRUE), aecolnames)
varnames <- c("Trial name", "Trastuzumab arm", "Control arm", "Trastuzumab dose","metasteses",
aencolnames)
traarmtypegrade <- tranothing %>%
select(all_of(varnames)) %>%
gather(vname, x, all_of(aencolnames)) %>%
extract(vname, c("aetype","grade","armno"), "^(.+) ; (.+) ; (.+)$", remove=FALSE) %>%
extract(vname, c("aetypeN","armnoN"), "(.+);(.+)$") %>%
mutate(aetype = ifelse(aetypeN=="N", aetypeN, aetype)) %>%
mutate(armno = ifelse(aetypeN=="N", armnoN, armno)) %>%
select(-aetypeN, -armnoN) %>%
mutate(trt = ifelse(`Trastuzumab arm` == armno, "Trastuzumab",
ifelse(`Control arm` == armno, "Control",NA)))
## Events assigned to wrong control arm: C2 instead of C1
#c2ind <- traarmtypegrade$`Trial name`=="NCCTG N9831" & traarmtypegrade$aetype != "N" &
# traarmtypegrade$armno == "C2"
#c1ind <- traarmtypegrade$`Trial name`=="NCCTG N9831" & traarmtypegrade$aetype != "N" &
# traarmtypegrade$armno == "C1"
#traarmtypegrade$x[c1ind] <- ifelse(!is.na(traarmtypegrade$x[c2ind]),
# traarmtypegrade$x[c2ind], traarmtypegrade$x[c1ind])
#traarmtypegrade$x[c2ind] <- NA
# No these are correct. So for these events we don't have any data.
bugind <- traarmtypegrade$`Trial name`=="BCIRG 006" & traarmtypegrade$aetype=="Endometrial cancer"
c1ind <- bugind & traarmtypegrade$armno=="C1"
c2ind <- bugind & traarmtypegrade$armno=="C2"
traarmtypegrade$x[c1ind] <- 0
traarmtypegrade$x[c2ind] <- NA
traarmtypegrade <- traarmtypegrade %>%
filter(!is.na(x)) %>%
mutate(`Trastuzumab dose` = factor(`Trastuzumab dose`)) %>%
mutate(`Trastuzumab dose` = replace_na(as.character(`Trastuzumab dose`), "0")) %>%
mutate(`Trastuzumab dose` = recode(`Trastuzumab dose`, `0`="unknown dose", `
1`="mid", `2`="low", `3`="high")) %>%
mutate(trtdose = ifelse(trt=="Trastuzumab", paste(trt, `Trastuzumab dose`), trt)) %>%
mutate(trtdose = ordered(trtdose, levels=c("Control",
paste("Trastuzumab",
c("unknown dose","low","mid","high"))))) %>%
mutate(count = as.numeric(x)) %>%
rename(study = `Trial name`) %>%
mutate(studyid = match(study, unique(study)))
## Drop extra arms that Alex excludes for this analysis, e.g.
## NCCTG N9831 was sequential vs concurrent trastuzumab
## FNCLCC PACS-04 : trastu arms distinguished as FEC vs ED chemo
traarmtypegrade <- traarmtypegrade %>%
filter(!is.na(trt)) %>%
mutate(grade3 = ifelse(grade %in% c("grade 3-4","grade 3 or 3/4","grade 4"), "Grade 3+",
ifelse(grade %in% c("grade 1-2","grade 1 or 1/2","grade 2"), "Grade 1-2",
NA)))
saveRDS(traarmtypegrade, file="tranothing_byarmtypegrade.rds")
## merge all grades
notevents <- c("N","NA","Withdrawals because of AEs")
traarmtype <- traarmtypegrade %>%
group_by(study, studyid,`Trastuzumab arm`,`Control arm`, aetype, armno, trt, trtdose, metasteses) %>%
summarise(count = sum(count, na.rm=TRUE)) %>%
ungroup
## add denominator for each count.
denoms <- traarmtype %>% filter(aetype=="N") %>% select(studyid, armno, count) %>% rename(N = count)
traarmtype <- traarmtype %>%
arrange(aetype, study, armno) %>%
filter(! aetype %in% notevents) %>%
filter(!is.na(count)) %>%
droplevels %>%
left_join(denoms, by = c("studyid", "armno"))
tracon <- traarmtype %>%
filter(trt == "Control") %>%
select(study, studyid, aetype, count, N) %>%
rename(rcon = count, ncon = N)
trastudytype <- traarmtype %>%
filter(trt == "Trastuzumab") %>%
left_join(tracon, by = c("study", "studyid", "aetype")) %>%
droplevels()
## data by arm and event type, with all event grades merged
saveRDS(tracon, file="tracon.rds")
saveRDS(traarmtype, file="tranothing_byarmtype.rds")
## Fix this by hand, Gasparini, Alopecia, see
## https://core.ac.uk/download/pdf/53842161.pdf Table 4
## assume it is double counting people who have both grade 1-2 and grade 3-4 events
## since without this fudge we get num > denom
trastudytype[21,]$count <- 59
trastudytype[21,]$rcon <- 50
## Raw data on frequency of each type of event, aggregated over all studies
trastudytype <- trastudytype %>%
mutate(oddsact = ((count+0.5)/(N+1)) / (1 - (count+0.5)/(N+1)),
oddscon = ((rcon+0.5)/(ncon+1)) / (1 - (rcon+0.5)/(ncon+1)),
est = oddsact / oddscon,
selogor = sqrt(1/(rcon+0.5) + 1/(ncon-rcon+0.5) + 1/(count+0.5) + 1/(N-count+0.5)),
or = est,
orlower = exp(log(est) - qnorm(0.975)*selogor),
orupper = exp(log(est) + qnorm(0.975)*selogor),
a = count, b=N-count, c=rcon, d=ncon-rcon,
serd = sqrt(a*b/(a+b)^3 + c*d/(c+d)^3),
pact = count/N,
pcon = rcon/ncon,
riskdiff = pact - pcon,
rdlower = riskdiff - qnorm(0.975)*serd,
rdupper = riskdiff + qnorm(0.975)*serd,
brisk = rcon / ncon,
brisklower = qbeta(0.025, 0.5 + rcon, 0.5 + ncon - rcon),
briskupper = qbeta(0.975, 0.5 + rcon, 0.5 + ncon - rcon),
aetype = factor(aetype, labels=str_to_sentence(levels(factor(aetype))))
)
## data by study and event type
saveRDS(trastudytype, file="tranothing_bystudytype.rds")
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.