# Understanding edge cases library(survival) # # this is from a user report of a problem with cumevents. When there is # a row merged in that is a censor, don't mark it as a cumevent. # base <- data.frame( id = 1:2, tstart = c(0, 0), tstop = c(10, 10), got_flu = c(0, 0), has_flu = factor(c("no", "no"), levels = c("no", "yes"))) base <- tmerge(base, base, id = id, got_flu = event(tstop, got_flu)) # add time-varying covariates vars <- data.frame(id = c(1, rep(2, 5)), time = c(0, (0:4) * 2), x = rnorm(6)) base <- tmerge(base, vars, id = id, x = tdc(time, x)) # add cumevents, using a covariate events <- data.frame( id = c(2, 2, 2), # notice the zero -- the second row should not add an event got_flu = c(1,0,2), has_flu = c("yes", "no", "yes"), time = c(3, 5, 8)) b2 <- tmerge(base, events, id = id, got_flu = cumevent(time, got_flu), has_flu = tdc(time, has_flu)) all.equal(b2$got_flu, c(0,0,1,0,0,0,3,0)) # Tied times in the merger data set # for all of them missings are essentially ignored # last obs wins for tdc and event tiedat <- data.frame(id=c(1, 1, 1, 2,2,2), time=c(3,4, 4, 3, 5, 5), x=c(1, NA,0, 2,3,4)) b3 <- tmerge(base, tiedat, id=id, x1= tdc(time, x), x2=cumtdc(time, x), x3= event(time, x), x4 = cumevent(time, x)) all.equal(b3$x1, c(NA, 1, 0, NA, NA, 2,2, 4,4,4)) all.equal(b3$x2, c(NA, 1, 1, NA, NA, 2,2, 9,9,9)) all.equal(b3$x3, c(1,0,0,0,2,0,4,0,0,0)) all.equal(b3$x4, c(1,0,0,0,2,0,9,0,0,0)) # Multiple overlapping time windows in the first step. # Should generate an error message test <- tryCatch( {tmerge(pbcseq[, c("id", "trt", "age", "sex")], pbcseq, id, death = event(futime, status==2))}, error= function(cond) { if (grepl("duplicate identifiers", cond)) cat("successful tmerge error test\n") } ) # Using a tdc that depends on more than one variable. If they are not # exactly the same class, tmerge should fail. # Happens with wide data sets tdata <- data.frame(id= 1:3, age=c(40,44,38), dtime=c(700, 600, 500), t1 = c(111, 211, 311), x1= as.integer(c(4, 5, 6)), t2 = c(120, 240, 400.3), x2=c( 9, 8, 7), t3 = c(400, 500, 450), x3=c(12,2, 0)) # This works wide1 <- tmerge(tdata[,1:2], tdata, id=id, death= event(dtime), x = tdc(t1, x1), x= tdc(t2, x2), x= tdc(t3, x3)) r1 <- data.frame(id=rep(1:3, each=4), age= tdata$age[rep(1:3, each=4)], tstart=c(0,111, 120, 400, 0, 211, 240, 500, 0, 311,400.3, 450), tstop =c(111, 120, 400, 700, 211, 240, 500, 600, 311, 400.3, 450, 500), death= rep(c(0,0,0,1), 3), x= c(NA,4, 9,12, NA, 5, 8, 2, NA, 6,7, 0)) all.equal(r1, wide1, check.attributes=FALSE) tdata$x2[2] <- 'c' # different data type test <- tryCatch( {tmerge(tdata[,1:2], tdata, id=id, death= event(dtime), x = tdc(t1, x1), x= tdc(t2, x2), x= tdc(t3, x3))}, error= function(cond) { if (grepl("tdc update does not match prior variable type: x", cond)) cat("successful tmerge error test\n") } )