library(survival) load("/data/SEER/lymyleuk.RData") # this loads in DF canc=c("NHL","Hodgkins","MML","ALL","CLL","AML","CML") code=list(c(2000,2008),c(2015,2016),c(2030,2030),c(2040,2040), c(2041,2041),c(2050,2050),c(2051,2051)) DF$dwd <- ((DF$COD>=20010)&(DF$COD<=37000)) summary(as.factor(DF$numPrim)) DF <- DF[DF$ageRC!=99,] ageRC.labels <- c( "Age 00","Ages 01-04","Ages 05-09","Ages 10-14","Ages 15-19","Ages 20-24", "Ages 25-29","Ages 30-34","Ages 35-39","Ages 40-44","Ages 45-49","Ages 50-54", "Ages 55-59","Ages 60-64","Ages 65-69","Ages 70-74","Ages 75-79","Ages 80-84", "Ages 85+") DF$ageRC <- factor(DF$ageRC, 0:18, ageRC.labels) DF$surv <- DF$surv+0.1 DF$sex <- factor(DF$sex, 1:2, c("M","F")) DF$aDiag5 <- DF$aDiag / 5; DF$yDiag5 <- DF$yDiag / 5; DF <- DF[DF$yDiag <= 2004 & DF$aDiag>=20,] DF <- DF[(DF$ICD9 >=code[[1]][1])&(DF$ICD9 <=code[[1]][2]),] #Non-Hodgkins Lymphoma #DF <- DF[DF$ICD9 %in% code[[1]],] #Non-Hodgkins Lymphoma gCox <- coxph(Surv(surv,dwd) ~ sex+ns(aDiag,3)+poly(yDiag5,3)+numPrim, data=DF) coef(gCox) confint(gCox) # more primaries correlates with longer survival since more target time to get one if living longer