CompoML {Wcompo}

Fit a proportional means regression model for weighted composite endpoint of recurrent event and death

Description

Fit a semiparametric proportional means regression model for the weighted composite endpoint of recurrent event and death (Mao and Lin, 2016). (Jared D. Huling (ORCID: 0000-0003-0670-4845) contributed to the optimization of this code.)

Usage

CompoML(id, time, status, Z, w = NULL, ep = 1e-04)

Arguments

id

A vector of unique patient identifiers.

time

A vector of event times.

status

A vector of event type labels. 0: censoring; 1: death; 2, 3,..., K: different types of (possibly recurrent) nonfatal event.

Z

Covariate matrix (must be time-constant).

w

A K-vector of weights assigned to event types 1 (death), 2, ..., K (nonfatal events); If NULL, an unweighted endpoint is modeled (i.e., with w=c(1, 1, ..., 1)).

ep

Convergence threshold for the Newton-Raphson algorithm.

Value

An object of class CompoML with the following components. beta: a vector of estimated regression coefficients (log-mean ratios); var: estimated covariance matrix for beta; t: unique event times; y: estimated baseline mean function (of t).

References

Mao, L. and Lin, D. Y. (2016). Semiparametric regression for the weighted composite endpoint of recurrent and terminal events. Biostatistics, 17, 390-403.

See Also

plot.CompoML, print.CompoML

Examples

## load package and data
library(Wcompo)
head(hfmock)
## fit a weighted PM (w_D=2, w_1=1)
obj <- CompoML(hfmock$id,hfmock$time,hfmock$status,hfmock[,c("Training","HF.etiology")],
               w=c(2,1))
## print out the result
obj

oldpar <- par(mfrow = par("mfrow"))
par(mfrow=c(1,2))
## plot the estimated mean function for
## non-ischemic patients by treatment
plot(obj,c(1,0),ylim=c(0,1.5),xlim=c(0,50),
     main="Non-ischemic",
     xlab="Time (months)",cex.main=1.2,lwd=2)
plot(obj,c(0,0),add=TRUE,cex.main=1.2,lwd=2,lty=2)
legend("topleft",lty=1:2,lwd=2,c("Exercise training","Usual care"))


## plot the estimated mean function for
## ischemic patients by treatment
plot(obj,c(1,1),ylim=c(0,1.5),xlim=c(0,50),
     main="Ischemic",
     xlab="Time (months)",cex.main=1.2,lwd=2)
plot(obj,c(0,1),add=TRUE,cex.main=1.2,lwd=2,lty=2)
legend("topleft",lty=1:2,lwd=2,c("Exercise training","Usual care"))
par(oldpar)