ipdmeta (version 1.1)

ipd.data: Simulated patient-level clustered survival data

Description

Patient-level survival data with clustering by group. Meant to be an example of an incomplete patient-level survival meta-analysis dataset. The complete data set has 10 studies (group) with 200 observations per study. Covariates include treatment and a continuous covariate. There is only a treatment effect (beta = -0.5) on the log-hazard scale. There is between-group heterogeneity for baseline and treatment group, a moderate between-group heterogeneity for the covariate.

The simulated mixed data meta-analysis supposes that 2 of the 10 studies did not provide patient-level data. The evidence is instead in the form of survival estimates by treatment group within study, which are given in meta.data. Thus, there are only 8 of the set of eligible trials that have patient-level outcomes.

Intended for illustrating use of mixed data survival meta-analysis estimation.

format

ll{ group Cluster factor, 1 to 10 trt Treatment indicator, 1 or 0 x Continuous covariate time Survival time event Indicator if event was observed, 1 or 0 trt.group Factor of treatment and group }

See Also

make.clustered.ipd, meta.data

Examples

data(ipd.data)

trt <- survfit(Surv(time,event)~strata(group),
                                                ipd.data,subset=trt==1)

ctrl <- survfit(Surv(time,event)~strata(group),
                                                ipd.data,subset=trt==0)

par(mfrow=c(1,2),pty="s")

plot(ctrl,mark.=FALSE,ylab="S(t)",xli=c(0,12),xlab="time",lt=1:8,main="Control")
legend(bty="n",legend=paste("Trial",1:8),lty=1:8,"topright")

plot(trt,mark.=FALSE,ylab="S(t)",xlab="time",lty=1:8,main="Treated")
legend(bty="n",legend=paste("Trial",1:8),lty=1:8,"topright")