bas.haz: Baseline hazard for individual-level and study-level from mixed-level meta-analysis

Description

Returns survfit object for patient-level data and cumulative hazard estimates and CI based on study-level estimates.

Usage

bas.haz(
        formula,
        haz.formula,
        data,
        coef,
        var,
        alpha=.05,
        FUN=exp
)

Arguments

formula

Surv formula for patient-level model as given in mlma.

haz.formula

Hazard model component of meta.formula of mlma

data

Patient-level data frame

coef

Vector of coefficients output from mlma

var

Matrix of variances for coef estimates from mlma

alpha

Significance level for normal-based CI for study-level cumulative hazard

FUN

Function for the parameter transformation at CI endpoints

Value

List of ipd.survfit and meta.bas.haz.
ipd.survfit is a survfit object.
meta.bas.haz is a data frame with est, lower, upper and time. The time at which the study-level estimates are evaluated are the event times for the patient-level data set.

Details

The haz.formula is the component of the study-level model given in mlma that corresponds to the log-cumulative hazard, i.e. ~log(t).

Examples

Run this code

data(ipd.data)
data(meta.data)


fit <- mlma(
     Surv(time,event)~trt,
     surv~-1+log(time),
     random=~(1|group),
     ipd.groups=8,
     meta.groups=2,
     ipd.data=ipd.data,
     meta.data=meta.data,
     sigma2=meta.data$sigma2,
     study.group=meta.data$sub.group,
     max.iter=5,
     min=100
)

#CUMULATIVE HAZARD SUMMARIES

H <- bas.haz(
             Surv(time,event)~trt,
             ~-1+log(t),
             ipd.data,
             fit$coef,
             fit$var$coef,
             )

#RETURN OBJECTS
#PATIENT-LEVEL
H$ipd.survfit

#STUDY-LEVEL
head(H$meta.bas.haz)


#PATIENT-LEVEL BASELINE HAZARD

plot(H$ipd.survfit,fun="cumhaz",ylab="H(t)",bty="n")

#STUDY-LEVEL BASELINE HAZARD WITH 95\% CI

lines(x=H$meta.bas.haz$time,y=H$meta.bas.haz$lower,type="l",lty=2)
lines(x=H$meta.bas.haz$time,y=H$meta.bas.haz$est,type="l",col="blue")
lines(x=H$meta.bas.haz$time,y=H$meta.bas.haz$upper,type="l",lty=2)

legend("topleft",legend=c("Study-level","Patient-level"),lty=1,col=c("blue","black"))