Simulation of recurrents events

We start by simulating some recurrent events data with two type of events with cumulative hazards

\(\Lambda_1(t)\) (rate among survivors)
\(\Lambda_2(t)\) (rate among survivors)
\(\Lambda_D(t)\)

where we consider types 1 and 2 and with a rate of the terminal event given by \(\Lambda_D(t)\). We let the events be independent, but could also specify a random effects structure to generate dependence.

When simulating data we can impose various random-effects structures to generate dependence

Dependence=0: The intensities can be independent.
Dependence=1: We can one gamma distributed random effects \(Z\). Then the intensities are
- \(Z \lambda_1(t)\)
- \(Z \lambda_2(t)\)
- \(Z \lambda_D(t)\)
Dependence=2: We can draw normally distributed random effects \(Z_1,Z_2,Z_d\) were the variance (var.z) and correlation can be specified (cor.mat). Then the intensities are
- \(\exp(Z_1) \lambda_1(t)\)
- \(\exp(Z_2) \lambda_2(t)\)
- \(\exp(Z_3) \lambda_D(t)\)
Dependence=3: We can draw gamma distributed random effects \(Z_1,Z_2,Z_d\) were the sum-structure can be speicifed via a matrix cor.mat. We compute \(\tilde Z_j = \sum_k Z_k^{cor.mat(j,k)}\) for \(j=1,2,3\).
Then the intensities are
- \(\tilde Z_1 \lambda_1(t)\)
- \(\tilde Z_2 \lambda_2(t)\)
- \(\tilde Z_3 \lambda_D(t)\)

We return to how to run the different set-ups later and start by simulating independent processes.

The key functions are

simRecurrent
- simple simulation with only one event type and death
simRecurrentII
- extended version with possibly multiple types of recurrent events (but rates can be 0)
- Allows Cox types rates with subject specific rates
simRecurrentIII
- lists are allowed for multiple events and cause of death (competing risks)
- Allows Cox types rates with subject specific rates
sim.recurrent to simulate from Cox-Cox (marginals) or Ghosh-Lin-Cox

In addition we can simulate data from the Ghosh-Lin model and where marginals of the rates among survivors are on on Cox form

simGLcox
- can simulate data from Ghosh-Lin model (also simRecurrentCox)
- with frailties
  - where survival model for terminal event is on Cox form
- can simulate data where rates among survivors are are con Cox form
  - with frailties

see examples below for specific models.

Utility functions

We here mention two utility functions

tie.breaker for breaking ties among jump-times which is expected in the functions below.
count.history that counts the number of jumps previous for each subject that is \(N_1(t-)\) and \(N_2(t-)\).

Marginal Mean

We start by estimating the marginal mean \(E(N_1(t \wedge D))\) where \(D\) is the timing of the terminal event. The marginal mean is the average number of events seen before time \(t\).

This is based on a rate model for

the type 1 events \(\sim E(dN_1(t) | D > t)\)
the terminal event \(\sim E(dN_d(t) | D > t)\)

and is defined as \(\mu_1(t)=E(N_1^*(t))\) \[\begin{align} \int_0^t S(u) d R_1(u) \end{align}\] where \(S(t)=P(D \geq t)\) and \(dR_1(t) = E(dN_1^*(t) | D > t)\)

and can therefore be estimated by a

Kaplan-Meier estimator, \(\hat S(u)\)
Nelson-Aalen estimator for \(R_1(t)\)

\[\begin{align} \hat R_1(t) & = \sum_i \int_0^t \frac{1}{Y_\bullet (s)} dN_{1i}(s) \end{align}\] where \(Y_{\bullet}(t)= \sum_i Y_i(t)\) such that the estimator is \[\begin{align} \hat \mu_1(t) & = \int_0^t \hat S(u) d\hat R_1(u). \end{align}\]

Cook & Lawless (1997), and developed further in Gosh & Lin (2000).

The variance can be estimated based on the asymptotic expansion of \(\hat \mu_1(t) - \mu_1(t)\) \[\begin{align*} & \sum_i \int_0^t \frac{S(s)}{\pi(s)} dM_{i1} - \mu_1(t) \int_0^t \frac{1}{\pi(s)} dM_i^d + \int_0^t \frac{\mu_1(s) }{\pi(s)} dM_i^d, \end{align*}\]

with mean-zero processes

\(M_i^d(t) = N_i^D(t)- \int_0^t Y_i(s) d \Lambda^D(s)\),
\(M_{i1}(t) = N_{i1}(t) - \int_0^t Y_{i}(s) dR_1(s)\).

as in Gosh & Lin (2000)

Generating data

We start by generating some data to illustrate the computation of the marginal mean

library(mets)
set.seed(1000) # to control output in simulatins for p-values below.

 data(CPH_HPN_CRBSI)
 dr <- CPH_HPN_CRBSI$terminal
 base1 <- CPH_HPN_CRBSI$crbsi 
 base4 <- CPH_HPN_CRBSI$mechanical

rr <- simRecurrent(200,base1,death.cumhaz=dr)
rr$x <- rnorm(nrow(rr)) 
rr$strata <- floor((rr$id-0.01)/100)
dlist(rr,.~id| id %in% c(1,7,9))
#> id: 1
#>   entry time  status dtime fdeath death start stop  x     strata
#> 1 0     132.6 0      132.6 1      1     0     132.6 1.386 0     
#> ------------------------------------------------------------ 
#> id: 7
#>   entry time  status dtime fdeath death start stop  x      strata
#> 7 0     141.3 0      141.3 1      1     0     141.3 0.6628 0     
#> ------------------------------------------------------------ 
#> id: 9
#>     entry time  status dtime fdeath death start stop  x       strata
#> 9     0.0 433.5 1      558.7 1      0       0.0 433.5 -0.7528 0     
#> 204 433.5 558.7 0      558.7 1      1     433.5 558.7 -0.6565 0

The status variable keeps track of the recurrent evnts and their type, and death the timing of death.

To compute the marginal mean we simly estimate the two rates functions of the number of events of interest and death by using the phreg function (to start without covariates). Then the estimates are combined with standard error computation in the recurrentMarginal function

#  to fit non-parametric models with just a baseline 
xr <- phreg(Surv(entry,time,status)~cluster(id),data=rr)
xdr <- phreg(Surv(entry,time,death)~cluster(id),data=rr)
par(mfrow=c(1,3))
plot(xdr,se=TRUE)
title(main="death")
plot(xr,se=TRUE)
# robust standard errors 
rxr <-   robust.phreg(xr,fixbeta=1)
plot(rxr,se=TRUE,robust=TRUE,add=TRUE,col=4)

# marginal mean of expected number of recurrent events 
out <- recurrentMarginal(Event(entry,time,status)~cluster(id),data=rr,cause=1,death.code=2)
plot(out,se=TRUE,ylab="marginal mean",col=2)

We can also extract the estimate in different time-points

summary(out,times=c(1000,2000))
#> [[1]]
#>     new.time     mean        se  CI-2.5% CI-97.5% strata
#> 245     1000 1.929941 0.1179801 1.712020 2.175601      0
#> 384     2000 4.070540 0.2243678 3.653708 4.534926      0

The marginal mean can also be estimated in a stratified case:

xr <- phreg(Surv(entry,time,status)~strata(strata)+cluster(id),data=rr)
xdr <- phreg(Surv(entry,time,death)~strata(strata)+cluster(id),data=rr)
par(mfrow=c(1,3))
plot(xdr,se=TRUE)
title(main="death")
plot(xr,se=TRUE)
rxr <-   robust.phreg(xr,fixbeta=1)
plot(rxr,se=TRUE,robust=TRUE,add=TRUE,col=1:2)

out <- recurrentMarginal(Event(entry,time,status)~strata(strata)+cluster(id),
             data=rr,cause=1,death.code=2)
plot(out,se=TRUE,ylab="marginal mean",col=1:2)

Further, if we adjust for covariates for the two rates we can still do predictions of marginal mean, what can be plotted is the baseline marginal mean, that is for the covariates equal to 0 for both models. Predictions for specific covariates can also be obtained with the recmarg (recurren marginal mean used solely for predictions without standard error computation).

# cox case
xr <- phreg(Surv(entry,time,status)~x+cluster(id),data=rr)
xdr <- phreg(Surv(entry,time,death)~x+cluster(id),data=rr)
par(mfrow=c(1,3))
plot(xdr,se=TRUE)
title(main="death")
plot(xr,se=TRUE)
rxr <- robust.phreg(xr)
plot(rxr,se=TRUE,robust=TRUE,add=TRUE,col=1:2)

out <- recurrentMarginalPhreg(xr,xdr)
plot(out,se=TRUE,ylab="marginal mean",col=1:2)


#### predictions witout se's 
###outX <- recmarg(xr,dr,Xr=1,Xd=1)
###plot(outX,add=TRUE,col=3)

Here I simulate multiple types and two causes of death causes of death

rr <- simRecurrentList(100,list(base1,base1,base4),death.cumhaz=list(dr,base4),cens=3/5000,dependence=0)
dtable(rr,~status+death,level=2)
#> 
#>      status
#> death   0   1   2   3
#>     0  31 120 109  10
#>     1  56   0   0   0
#>     2  13   0   0   0
mets:::showfitsimList(rr,list(base1,base1,base4),list(dr,base4))

Improving efficiency

We now simulate some data where there is strong heterogenity such that we can improve the efficiency for censored survival data. The augmentation is a regression on the history for each subject consisting of the specified terms terms: Nt, Nt2 (Nt squared), expNt (exp(-Nt)), NtexpNt (Nt*exp(-Nt)) or by simply specifying these directly. This was developed in Cortese and Scheike (2022).

rr <- simRecurrentII(200,base1,base4,death.cumhaz=dr,cens=3/5000,dependence=4,var.z=1)
rr <-  count.history(rr)

rr <- transform(rr,statusD=status)
rr <- dtransform(rr,statusD=3,death==1)
dtable(rr,~statusD+status+death,level=2,response=1)
#> 
#>       statusD
#> status   0   1   2   3
#>      0  79   0   0 121
#>      1   0 275   0   0
#>      2   0   0  35   0
#> 
#>      statusD
#> death   0   1   2   3
#>     0  79 275  35   0
#>     1   0   0   0 121

##xr <- phreg(Surv(start,stop,status==1)~cluster(id),data=rr)
##dr <- phreg(Surv(start,stop,death)~cluster(id),data=rr)
# marginal mean of expected number of recurrent events 
out <- recurrentMarginal(Event(start,stop,statusD)~cluster(id),data=rr,cause=1,death.code=3)

times <- 500*(1:10)
recEFF1 <- recurrentMarginalAIPCW(Event(start,stop,statusD)~cluster(id),data=rr,times=times,cens.code=0,
                   death.code=3,cause=1,augment.model=~Nt)
with( recEFF1, cbind(times,muP,semuP,muPAt,semuPAt,semuPAt/semuP))
#>       times       muP      semuP     muPAt    semuPAt          
#>  [1,]   500 0.6613768 0.08161137 0.6608857 0.08146438 0.9981990
#>  [2,]  1000 1.0744212 0.12757420 1.0656868 0.12598264 0.9875244
#>  [3,]  1500 1.4043440 0.18482988 1.4104172 0.17775750 0.9617357
#>  [4,]  2000 1.6558938 0.23411074 1.6507507 0.22529350 0.9623373
#>  [5,]  2500 1.9655087 0.31221814 1.9690466 0.29506881 0.9450726
#>  [6,]  3000 2.1616770 0.37981206 2.2020438 0.35291111 0.9291730
#>  [7,]  3500 2.4425027 0.51943450 2.5131440 0.45781796 0.8813776
#>  [8,]  4000 2.7545313 0.66543528 2.8476086 0.58085220 0.8728906
#>  [9,]  4500 2.8247378 0.66133467 2.9107949 0.59608745 0.9013401
#> [10,]  5000 2.8247378 0.66133467 2.9107949 0.59608745 0.9013401

times <- 500*(1:10)
###recEFF14 <- recurrentMarginalAIPCW(Event(start,stop,statusD)~cluster(id),data=rr,times=times,cens.code=0,
###death.code=3,cause=1,augment.model=~Nt+Nt2+expNt+NtexpNt)
###with(recEFF14,cbind(times,muP,semuP,muPAt,semuPAt,semuPAt/semuP))

recEFF14 <- recurrentMarginalAIPCW(Event(start,stop,statusD)~cluster(id),data=rr,times=times,cens.code=0,
death.code=3,cause=1,augment.model=~Nt+I(Nt^2)+I(exp(-Nt))+ I( Nt*exp(-Nt)))
with(recEFF14,cbind(times,muP,semuP,muPAt,semuPAt,semuPAt/semuP))
#>       times       muP      semuP    muPAt    semuPAt          
#>  [1,]   500 0.6613768 0.08161137 0.667650 0.08124058 0.9954566
#>  [2,]  1000 1.0744212 0.12757420 1.065307 0.12509251 0.9805471
#>  [3,]  1500 1.4043440 0.18482988 1.393495 0.17474086 0.9454145
#>  [4,]  2000 1.6558938 0.23411074 1.623926 0.22154326 0.9463182
#>  [5,]  2500 1.9655087 0.31221814 1.841123 0.28399825 0.9096148
#>  [6,]  3000 2.1616770 0.37981206 2.044395 0.33651778 0.8860113
#>  [7,]  3500 2.4425027 0.51943450 2.132034 0.41402348 0.7970658
#>  [8,]  4000 2.7545313 0.66543528 2.383191 0.52860666 0.7943773
#>  [9,]  4500 2.8247378 0.66133467 2.333195 0.53618544 0.8107626
#> [10,]  5000 2.8247378 0.66133467 2.333195 0.53618544 0.8107626

plot(out,se=TRUE,ylab="marginal mean",col=2)
k <- 1
for (t in times) {
    ci1 <- c(recEFF1$muPAt[k]-1.96*recEFF1$semuPAt[k],
             recEFF1$muPAt[k]+1.96*recEFF1$semuPAt[k])
    ci2 <- c(recEFF1$muP[k]-1.96*recEFF1$semuP[k],
             recEFF1$muP[k]+1.96*recEFF1$semuP[k])
    lines(rep(t,2)-2,ci2,col=2,lty=1,lwd=2)
    lines(rep(t,2)+2,ci1,col=1,lty=1,lwd=2)
    k <- k+1
}
legend("bottomright",c("Eff-pred"),lty=1,col=c(1,3))

In the case where covariates might be important but we are still interested in the marginal mean we can also augment wrt these covariates

n <- 200
X <- matrix(rbinom(n*2,1,0.5),n,2)
colnames(X) <- paste("X",1:2,sep="")
###
r1 <- exp( X %*% c(0.3,-0.3))
rd <- exp( X %*% c(0.3,-0.3))
rc <- exp( X %*% c(0,0))
fz <- NULL
rr <- mets:::simGLcox(n,base1,dr,var.z=0,r1=r1,rd=rd,rc=rc,fz,model="twostage",cens=3/5000) 
rr <- cbind(rr,X[rr$id+1,])

dtable(rr,~statusD+status+death,level=2,response=1)
#> 
#>       statusD
#> status   0   1   3
#>      0  92   0 108
#>      1   0 556   0
#> 
#>      statusD
#> death   0   1   3
#>     0  92 363   0
#>     1   0 193 108

times <- seq(500,5000,by=500)
recEFF1x <- recurrentMarginalAIPCW(Event(start,stop,statusD)~cluster(id),data=rr,times=times,
                   cens.code=0,death.code=3,cause=1,augment.model=~X1+X2)
with(recEFF1x, cbind(muP,muPA,muPAt,semuP,semuPA,semuPAt,semuPAt/semuP))
#>            muP     muPA    muPAt      semuP     semuPA    semuPAt          
#>  [1,] 1.145071 1.144032 1.143158 0.08446448 0.08438871 0.08433747 0.9984963
#>  [2,] 1.992395 1.987834 1.967759 0.16163097 0.16021919 0.15987853 0.9891577
#>  [3,] 2.987070 3.041386 2.975609 0.29841261 0.29352422 0.29276652 0.9810796
#>  [4,] 3.750338 3.781183 3.742977 0.38089153 0.37347550 0.37254937 0.9780983
#>  [5,] 4.401481 4.363984 4.379360 0.45643233 0.44836832 0.44476899 0.9744467
#>  [6,] 5.058667 4.943227 4.948666 0.58230638 0.56486481 0.55682137 0.9562344
#>  [7,] 5.797576 5.777539 5.609930 0.75658794 0.72607368 0.71043739 0.9390017
#>  [8,] 6.500899 6.740989 6.188770 0.89938615 0.86946649 0.84664142 0.9413547
#>  [9,] 7.160265 7.277316 6.575000 1.02814316 1.02039663 0.99768282 0.9703734
#> [10,] 7.709737 7.297200 6.832030 1.13548187 1.11392250 1.09078010 0.9606319

out <- recurrentMarginal(Event(start,stop,statusD)~cluster(id),data=rr,cause=1,death.code=3)
summary(out,times=times)
#> [[1]]
#>     new.time     mean         se   CI-2.5%  CI-97.5% strata
#> 201      500 1.145071 0.08446448 0.9909346  1.323184      0
#> 310     1000 1.992395 0.16163097 1.6995055  2.335760      0
#> 397     1500 2.987070 0.29841261 2.4558915  3.633136      0
#> 448     2000 3.750338 0.38089153 3.0734121  4.576359      0
#> 483     2500 4.401481 0.45643233 3.5919438  5.393469      0
#> 509     3000 5.058667 0.58230638 4.0369534  6.338966      0
#> 529     3500 5.797576 0.75658794 4.4891486  7.487362      0
#> 543     4000 6.500899 0.89938615 4.9569146  8.525806      0
#> 552     4500 7.160265 1.02814316 5.4038714  9.487531      0
#> 557     5000 7.709737 1.13548187 5.7766426 10.289721      0

Regression models for the marginal mean

One can also do regression modelling , using the model \[\begin{align*} E(N_1(t) | X) & = \Lambda_0(t) \exp(X^T \beta) \end{align*}\] then Ghost-Lin suggested IPCW score equations that are implemented in the recreg function of mets.

First we generate data that from a Ghosh-Lin model with \(\beta=(-0.3,0.3)\) and the baseline given by base1, this is done under the assumption that the death rate given covariates are on Cox form with baseline dr:

n <- 100
X <- matrix(rbinom(n*2,1,0.5),n,2)
colnames(X) <- paste("X",1:2,sep="")
###
r1 <- exp( X %*% c(0.3,-0.3))
rd <- exp( X %*% c(0.3,-0.3))
rc <- exp( X %*% c(0,0))
fz <- NULL
rr <- mets:::simGLcox(n,base1,dr,var.z=1,r1=r1,rd=rd,rc=rc,fz,cens=1/5000,type=2) 
rr <- cbind(rr,X[rr$id+1,])

 out  <- recreg(Event(start,stop,statusD)~X1+X2+cluster(id),data=rr,cause=1,death.code=3,cens.code=0)
 outs <- recreg(Event(start,stop,statusD)~X1+X2+cluster(id),data=rr,cause=1,death.code=3,cens.code=0,
        cens.model=~strata(X1,X2))
 summary(out)$coef
#>      Estimate      S.E.    dU^-1/2   P-value
#> X1  0.5194368 0.3797930 0.07990766 0.1714110
#> X2 -0.3358640 0.3803112 0.07976036 0.3771663
 summary(outs)$coef
#>      Estimate      S.E.    dU^-1/2   P-value
#> X1  0.5762836 0.3571995 0.08035274 0.1066710
#> X2 -0.2775462 0.3520266 0.08007337 0.4304488

 ## checking baseline
 par(mfrow=c(1,1))
 plot(out)
 plot(outs,add=TRUE,col=2)
 lines(scalecumhaz(base1,1),col=3,lwd=2)

We note that for the extended censoring model we gain a little efficiency and that the estimates are close to the true values.

Also possible to do IPCW regression at fixed time-point

 outipcw  <- recregIPCW(Event(start,stop,statusD)~X1+X2+cluster(id),data=rr,cause=1,death.code=3,
            cens.code=0,times=2000)
 outipcws <- recregIPCW(Event(start,stop,statusD)~X1+X2+cluster(id),data=rr,cause=1,death.code=3,
            cens.code=0,times=2000,cens.model=~strata(X1,X2))
 summary(outipcw)$coef
#>               Estimate   Std.Err       2.5%     97.5%      P-value
#> (Intercept) 1.19494027 0.2038193  0.7954619 1.5944187 4.552780e-09
#> X1          0.18509341 0.2825398 -0.3686743 0.7388612 5.123997e-01
#> X2          0.09805584 0.2805028 -0.4517195 0.6478312 7.266601e-01
 summary(outipcws)$coef
#>               Estimate   Std.Err       2.5%     97.5%      P-value
#> (Intercept) 1.18811953 0.2024682  0.7912891 1.5849499 4.406099e-09
#> X1          0.20055180 0.2761747 -0.3407407 0.7418443 4.677301e-01
#> X2          0.09514999 0.2729203 -0.4397640 0.6300640 7.273622e-01

We can also do the Mao-Lin type composite outcome where we both count the cause 1 and deaths for example \[\begin{align*} E(N_1(t) + I(D<t,\epsilon=3) | X) & = \Lambda_0(t) \exp(X^T \beta) \end{align*}\]

 out  <- recreg(Event(start,stop,statusD)~X1+X2+cluster(id),data=rr,cause=c(1,3),
        death.code=3,cens.code=0)
 summary(out)$coef
#>      Estimate      S.E.    dU^-1/2   P-value
#> X1  0.4896610 0.3403976 0.07609363 0.1502931
#> X2 -0.3056174 0.3395604 0.07595286 0.3680998

Also demonstrate that this can be done with competing risks death (change some of the cause 3 deaths to cause 4) \[\begin{align*} E(w_1 N_1(t) + w_2 I(D<t,\epsilon=3) | X) & = \Lambda_0(t) \exp(X^T \beta) \end{align*}\] and with weights \(w_1,w_2\) that follow the causes, here 1 and 3.

 rr$binf <- rbinom(nrow(rr),1,0.5) 
 rr$statusDC <- rr$statusD
 rr <- dtransform(rr,statusDC=4, statusD==3 & binf==0)
 rr$weight <- 1
 rr <- dtransform(rr,weight=2,statusDC==3)

 outC  <- recreg(Event(start,stop,statusDC)~X1+X2+cluster(id),data=rr,cause=c(1,3),
         death.code=c(3,4),cens.code=0)
 summary(outC)$coef
#>      Estimate      S.E.    dU^-1/2   P-value
#> X1  0.5069982 0.3651533 0.07814845 0.1649991
#> X2 -0.3255327 0.3653288 0.07801576 0.3728929

 outCW  <- recreg(Event(start,stop,statusDC)~X1+X2+cluster(id),data=rr,cause=c(1,3),
          death.code=c(3,4),cens.code=0,wcomp=c(1,2))
 summary(outCW)$coef
#>      Estimate      S.E.    dU^-1/2   P-value
#> X1  0.4955643 0.3521275 0.07650295 0.1593256
#> X2 -0.3160290 0.3519907 0.07638286 0.3692744

 plot(out,ylab="Mean composite")
 plot(outC,col=2,add=TRUE)
 plot(outCW,col=3,add=TRUE)

Predictions and standard errors can be computed via the iid decompositions of the baseline and the regression coefficients. We illustrate this for the standard Ghosh-Lin model and it requires that the model is fitted with the option cox.prep=TRUE

out  <- recreg(Event(start,stop,statusD)~X1+X2+cluster(id),data=rr,cause=1,death.code=3,cens.code=0,
           cox.prep=TRUE)
summary(out)
#> 
#>    n events
#>  740    640
#> 
#>  100 clusters
#> coeffients:
#>     Estimate      S.E.   dU^-1/2 P-value
#> X1  0.519437  0.379793  0.079908  0.1714
#> X2 -0.335864  0.380311  0.079760  0.3772
#> 
#> exp(coeffients):
#>    Estimate    2.5%  97.5%
#> X1  1.68108 0.79856 3.5389
#> X2  0.71472 0.33917 1.5061
baseiid <- iidBaseline(out,time=3000)
GLprediid(baseiid,rr[1:5,])
#>          pred    se-log    lower     upper
#> [1,] 9.117785 0.3788898 4.338882 19.160237
#> [2,] 3.876474 0.3881948 1.811361  8.296003
#> [3,] 3.876474 0.3881948 1.811361  8.296003
#> [4,] 3.876474 0.3881948 1.811361  8.296003
#> [5,] 6.516666 0.2461914 4.022230 10.558056

The Ghosh-Lin model can be made more efficient by the regression augmentation method. First computing the augmentation and then in a second step the augmented estimator (Cortese and Scheike (2023)):

 outA  <- recreg(Event(start,stop,statusD)~X1+X2+cluster(id),data=rr,cause=1,death.code=3,
         cens.code=0,augment.model=~Nt+X1+X2)
 summary(outA)$coef
#>      Estimate      S.E.    dU^-1/2    P-value
#> X1  0.5716809 0.3461524 0.08022919 0.09863056
#> X2 -0.3447143 0.3480247 0.07981423 0.32193577

We note that the simple augmentation improves the standard errors as expected. The data was generated assuming independence with previous number of events so it would suffice to augment only with the covariates.

Two-stage modelling

We simulate data with a terminal event on Cox form and recurrent events satisfying the Ghosh-Lin model or having a rate on Cox form.

type=3 is Ghosh-Lin model for recurrent events and Cox for terminal event.
type=2 is Cox model for recurrent events among survivors and Cox for terminal event.
simulations based on time-grid to make linear approximations of cumulative hazards

Now we fit the two-stage model (the recreg must be called with twostage=TRUE)

set.seed(100)
n <- 200
X <- matrix(rbinom(n*2,1,0.5),n,2)
colnames(X) <- paste("X",1:2,sep="")
###
r1 <- exp( X %*% c(0.3,-0.3))
rd <- exp( X %*% c(0.3,-0.3))
rc <- exp( X %*% c(0,0))
fz <- NULL
## type=3 is cox-cox and type=2 is Ghosh-Lin/Cox model 
rr <- mets:::simGLcox(n,base1,dr,var.z=1,r1=r1,rd=rd,rc=rc,fz,cens=1/5000,type=3) 
rr <- cbind(rr,X[rr$id+1,])
###
out  <- phreg(Event(start,stop,statusD==1)~X1+X2+cluster(id),data=rr)
outs <- phreg(Event(start,stop,statusD==3)~X1+X2+cluster(id),data=rr)
## cox/cox
tsout <- twostageREC(outs,out,data=rr)
summary(tsout)
#> Cox(recurrent)-Cox(terminal) intensity model
#> 
#>  200 clusters
#> coeffients:
#>             Estimate Std.Err    2.5%   97.5% P-value
#> dependence1  1.14651 0.14126 0.86964 1.42339       0
#> 
#> var,shared:
#>             Estimate Std.Err    2.5%   97.5% P-value
#> dependence1  1.14651 0.14126 0.86964 1.42339       0
###
rr <- mets:::simGLcox(n,base1,dr,var.z=1,r1=r1,rd=rd,rc=rc,fz,cens=1/5000,type=3,share=0.5) 
rr <- cbind(rr,X[rr$id+1,])
###
out  <- phreg(Event(start,stop,statusD==1)~X1+X2+cluster(id),data=rr)
outs <- phreg(Event(start,stop,statusD==3)~X1+X2+cluster(id),data=rr)
#
tsout <- twostageREC(outs,out,data=rr,model="shared")
summary(tsout)
#> Cox(recurrent)-Cox(terminal) intensity model
#> 
#>  200 clusters
#> coeffients:
#>             Estimate Std.Err    2.5%   97.5% P-value
#> dependence1  1.07333 0.14096 0.79707 1.34960       0
#> share1       0.67344 0.12832 0.42193 0.92495       0
#> 
#> var,shared:
#>             Estimate    2.5%  97.5%
#> dependence1  1.07333 0.79707 1.3496
#> share1       0.67344 0.42193 0.9250
###
rr <- mets:::simGLcox(n,base1,dr,var.z=1,r1=r1,rd=rd,rc=rc,fz,cens=1/5000,type=2) 
rr <- cbind(rr,X[rr$id+1,])
outs  <- phreg(Event(start,stop,statusD==3)~X1+X2+cluster(id),data=rr)
outgl  <- recreg(Event(start,stop,statusD)~X1+X2+cluster(id),data=rr,twostage=TRUE,death.code=3)
##
## ghosh-lin/cox
glout <- twostageREC(outs,outgl,data=rr,theta=1)
summary(glout)
#> Ghosh-Lin(recurrent)-Cox(terminal) mean model
#> 
#>  200 clusters
#> coeffients:
#>             Estimate Std.Err    2.5%   97.5% P-value
#> dependence1  1.04401 0.08988 0.86785 1.22017       0
#> 
#> var,shared:
#>             Estimate Std.Err    2.5%   97.5% P-value
#> dependence1  1.04401 0.08988 0.86785 1.22017       0
###
glout <- twostageREC(outs,outgl,data=rr,model="shared",theta=1,nu=0.9)
summary(glout)
#> Ghosh-Lin(recurrent)-Cox(terminal) mean model
#> 
#>  200 clusters
#> coeffients:
#>             Estimate  Std.Err     2.5%    97.5% P-value
#> dependence1 1.192629 0.094129 1.008140 1.377117       0
#> share1      1.643844 0.176676 1.297566 1.990121       0
#> 
#> var,shared:
#>             Estimate   2.5%  97.5%
#> dependence1   1.1926 1.0081 1.3771
#> share1        1.6438 1.2976 1.9901
glout$gradient
#> [1] -1.823666e-08  2.506179e-08

Standard errors are computed assuming that the parameters of out and outs are both known, and therefore propobly a bit to small. We could do a bootstrap to get more reliable standard errors.

Simulations with specific structure

The function simGLcox can simulate data where the recurrent process has mean on Ghosh-Lin form. The key is that \[\begin{align*} E(N_1(t) | X) & = \Lambda_0(t) \exp(X^T \beta) = \int_0^t S(t|X,Z) dR(t|X,Z) \end{align*}\] where \(Z\) is a possible frailty. Therefore \[\begin{align*} R(t|X,Z) & = \frac{Z \Lambda_0(t) \exp(X^T \beta) }{S(t|X,Z)} \end{align*}\] leads to a Ghosh-Lin model. We can choose the survival model to have Cox form among survivors by the option model=“twostage”, otherwise model=“frailty” uses the survival model with rate \(Z \lambda_d(t) rd\). The \(Z\) is gamma distributed with a variance that can be specified. The simulations are based on a piecwise-linear approximation of the hazard functions for \(S(t|X,Z)\) and \(R(t|X,Z)\).

n <- 100
X <- matrix(rbinom(n*2,1,0.5),n,2)
colnames(X) <- paste("X",1:2,sep="")
###
r1 <- exp( X %*% c(0.3,-0.3))
rd <- exp( X %*% c(0.3,-0.3))
rc <- exp( X %*% c(0,0))
rr <- mets:::simGLcox(n,base1,dr,var.z=0,r1=r1,rd=rd,rc=rc,model="twostage",cens=3/5000) 
rr <- cbind(rr,X[rr$id+1,])

We can also simulate from models where the terminal event is on Cox form and the rate among survivors is on Cox form.

\(E(dN_1 | D>t, X) = \lambda_1(t) r_1\)
\(E(dN_d | D>t, X) = \lambda_d(t) r_d\)

underlying these models we have a shared frailty model

rr <- mets:::simGLcox(100,base1,dr,var.z=1,r1=r1,rd=rd,rc=rc,type=3,cens=3/5000) 
rr <- cbind(rr,X[rr$id+1,])
margsurv <- phreg(Surv(start,stop,statusD==3)~X1+X2+cluster(id),rr)
recurrent <- phreg(Surv(start,stop,statusD==1)~X1+X2+cluster(id),rr)
estimate(margsurv)
#>    Estimate Std.Err    2.5%  97.5% P-value
#> X1   0.2535  0.2553 -0.2469 0.7538  0.3208
#> X2  -0.2595  0.2637 -0.7764 0.2574  0.3252
estimate(recurrent)
#>    Estimate Std.Err     2.5%   97.5% P-value
#> X1   0.6675  0.3019  0.07583 1.25921 0.02702
#> X2  -0.4356  0.2637 -0.95247 0.08135 0.09864
par(mfrow=c(1,2)); 
plot(margsurv); lines(dr,col=3); 
plot(recurrent); lines(base1,col=3)

We can simulate data with underlying dependence fromm the two-stage model (simGLcox) or using simRecurrent random effects models, for Cox-Cox or Ghosh-Lin-Cox models.

Here with marginals on - Cox- Cox form - Ghosh-Lin - Cox form

Draws covariates from data and simulates data that has the marginals given.

simcoxcox <- sim.recurrent(recurrent,margsurv,n=10,data=rr)

recurrentGL <- recreg(Event(start,stop,statusD)~X1+X2+cluster(id),rr,death.code=3)
simglcox <- sim.recurrent(recurrentGL,margsurv,n=10,data=rr)

Other marginal properties

The mean is a useful summary measure but it is very easy and useful to look at other simple summary measures such as the probability of exceeding \(k\) events

\(P(N_1^*(t) \ge k)\)
- cumulative incidence of \(T_{k} = \inf \{ t: N_1^*(t)=k \}\) with competing \(D\).

that is thus equivalent to a certain cumulative incidence of \(T_k\) occurring before \(D\). We denote this cumulative incidence as \(\hat F_k(t)\).

We note also that \(N_1^*(t)^2\) can be written as \[\begin{align*} \sum_{k=0}^K \int_0^t I(D > s) I(N_1^*(s-)=k) f(k) dN_1^*(s) \end{align*}\] with \(f(k)=(k+1)^2 - k^2\), such that its mean can be written as \[\begin{align*} \sum_{k=0}^K \int_0^t S(s) f(k) P(N_1^*(s-)= k | D \geq s) E( dN_1^*(s) | N_1^*(s-)=k, D> s) \end{align*}\] and estimated by \[\begin{align*} \tilde \mu_{1,2}(t) & = \sum_{k=0}^K \int_0^t \hat S(s) f(k) \frac{Y_{1\bullet}^k(s)}{Y_\bullet (s)} \frac{1}{Y_{1\bullet}^k(s)} d N_{1\bullet}^k(s)= \sum_{i=1}^n \int_0^t \hat S(s) f(N_{i1}(s-)) \frac{1}{Y_\bullet (s)} d N_{i1}(s), \end{align*}\] That is very similar to the “product-limit” estimator for \(E( (N_1^*(t))^2 )\) \[\begin{align} \hat \mu_{1,2}(t) & = \sum_{k=0}^K k^2 ( \hat F_{k}(t) - \hat F_{k+1}(t) ). \end{align}\]

We use the esimator of the probabilty of exceeding “k” events based on the fact that \(I(N_1^*(t) \geq k)\) is equivalent to \[\begin{align*} \int_0^t I(D > s) I(N_1^*(s-)=k-1) dN_1^*(s), \end{align*}\] suggesting that its mean can be computed as \[\begin{align*} \int_0^t S(s) P(N_1^*(s-)= k-1 | D \geq s) E( dN_1^*(s) | N_1^*(s-)=k-1, D> s) \end{align*}\] and estimated by \[\begin{align*} \tilde F_k(t) = \int_0^t \hat S(s) \frac{Y_{1\bullet}^{k-1}(s)}{Y_\bullet (s)} \frac{1}{Y_{1\bullet}^{k-1}(s)} d N_{1\bullet}^{k-1}(s). \end{align*}\]

To compute these estimators we use the prob.exceed.recurrent function

rr <- simRecurrentII(200,base1,base4,death.cumhaz=dr,cens=3/5000,dependence=4,var.z=1)
rr <- transform(rr,statusD=status)
rr <- dtransform(rr,statusD=3,death==1)
rr <-  count.history(rr)
dtable(rr,~statusD)
#> 
#> statusD
#>   0   1   2   3 
#>  93 287  32 107

oo <- prob.exceed.recurrent(Event(entry,time,statusD)~cluster(id),rr,cause=1,death.code=3)
plot(oo,types=1:5)

We can also look at the mean and variance based on the estimators just described

par(mfrow=c(1,2))
with(oo,plot(time,meanN,col=2,type="l"))
with(oo,plot(time,varN,type="l"))

Multiple events

We now generate recurrent events with two types of events. We start by generating data as before where all events are independent.

rr <- simRecurrentII(200,base1,cumhaz2=base4,death.cumhaz=dr)
rr <-  count.history(rr)
dtable(rr,~death+status)
#> 
#>       status   0   1   2
#> death                   
#> 0             32 651  84
#> 1            168   0   0

Based on this we can estimate also the joint distribution function, that is the probability that \((N_1(t) \geq k_1, N_2(t) \geq k_2)\)

# Bivariate probability of exceeding 
## oo <- prob.exceedBiRecurrent(rr,1,2,exceed1=c(1,5),exceed2=c(1,2))
## with(oo, matplot(time,pe1e2,type="s"))
## nc <- ncol(oo$pe1e2)
## legend("topleft",legend=colnames(oo$pe1e2),lty=1:nc,col=1:nc)

Looking at other simulations with dependence

Using the normally distributed random effects we plot 4 different settings. We have variance \(0.5\) for all random effects and change the correlation. We let the correlation between the random effect associated with \(N_1\) and \(N_2\) be denoted \(\rho_{12}\) and the correlation between the random effects associated between \(N_j\) and \(D\) the terminal event be denoted as \(\rho_{j3}\), and organize all correlation in a vector \(\rho=(\rho_{12},\rho_{13},\rho_{23})\).

Scenario I \(\rho=(0,0.0,0.0)\) Independence among all efects.

 data(CPH_HPN_CRBSI)
 dr <- CPH_HPN_CRBSI$terminal
 base1 <- CPH_HPN_CRBSI$crbsi 
 base4 <- CPH_HPN_CRBSI$mechanical

  par(mfrow=c(1,3))
  var.z <- c(0.5,0.5,0.5)
  # death related to  both causes in same way 
  cor.mat <- corM <- rbind(c(1.0, 0.0, 0.0), c(0.0, 1.0, 0.0), c(0.0, 0.0, 1.0))
  rr <- simRecurrentII(200,base1,base4,death.cumhaz=dr,var.z=var.z,cor.mat=cor.mat,dependence=2)
  rr <- count.history(rr,types=1:2)
###  cor(attr(rr,"z"))
###  coo <- covarianceRecurrent(rr,1,2,status="status",start="entry",stop="time")
###  plot(coo,main ="Scenario I")

Scenario II \(\rho=(0,0.5,0.5)\) Independence among survivors but dependence on terminal event

  var.z <- c(0.5,0.5,0.5)
  # death related to  both causes in same way 
  cor.mat <- corM <- rbind(c(1.0, 0.0, 0.5), c(0.0, 1.0, 0.5), c(0.5, 0.5, 1.0))
  rr <- simRecurrentII(200,base1,base4,death.cumhaz=dr,var.z=var.z,cor.mat=cor.mat,dependence=2)
  rr <- count.history(rr,types=1:2)
###  coo <- covarianceRecurrent(rr,1,2,status="status",start="entry",stop="time")
###  par(mfrow=c(1,3))
###  plot(coo,main ="Scenario II")

Scenario III \(\rho=(0.5,0.5,0.5)\) Positive dependence among survivors and dependence on terminal event

  var.z <- c(0.5,0.5,0.5)
  # positive dependence for N1 and N2 all related in same way
  cor.mat <- corM <- rbind(c(1.0, 0.5, 0.5), c(0.5, 1.0, 0.5), c(0.5, 0.5, 1.0))
  rr <- simRecurrentII(200,base1,base4,death.cumhaz=dr,var.z=var.z,cor.mat=cor.mat,dependence=2)
  rr <- count.history(rr,types=1:2)
###  coo <- covarianceRecurrent(rr,1,2,status="status",start="entry",stop="time")
###  par(mfrow=c(1,3))
###  plot(coo,main="Scenario III")

Scenario IV \(\rho=(-0.4,0.5,0.5)\) Negative dependence among survivors and positive dependence on terminal event

  var.z <- c(0.5,0.5,0.5)
  # negative dependence for N1 and N2 all related in same way
  cor.mat <- corM <- rbind(c(1.0, -0.4, 0.5), c(-0.4, 1.0, 0.5), c(0.5, 0.5, 1.0))
  rr <- simRecurrentII(200,base1,base4,death.cumhaz=dr,var.z=var.z,cor.mat=cor.mat,dependence=2)
  rr <- count.history(rr,types=1:2)
###  coo <- covarianceRecurrent(rr,1,2,status="status",start="entry",stop="time")
###  par(mfrow=c(1,3))
###  plot(coo,main="Scenario IV")

SessionInfo

sessionInfo()
#> R version 4.5.1 (2025-06-13)
#> Platform: aarch64-apple-darwin24.5.0
#> Running under: macOS Sequoia 15.6.1
#> 
#> Matrix products: default
#> BLAS:   /Users/kkzh/.asdf/installs/R/4.5.1/lib/R/lib/libRblas.dylib 
#> LAPACK: /Users/kkzh/.asdf/installs/R/4.5.1/lib/R/lib/libRlapack.dylib;  LAPACK version 3.12.1
#> 
#> locale:
#> [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
#> 
#> time zone: Europe/Copenhagen
#> tzcode source: internal
#> 
#> attached base packages:
#> [1] splines   stats     graphics  grDevices utils     datasets  methods  
#> [8] base     
#> 
#> other attached packages:
#> [1] timereg_2.0.7  survival_3.8-3 mets_1.3.7    
#> 
#> loaded via a namespace (and not attached):
#>  [1] cli_3.6.5           knitr_1.50          rlang_1.1.6        
#>  [4] xfun_0.53           KernSmooth_2.23-26  jsonlite_2.0.0     
#>  [7] future.apply_1.20.0 listenv_0.9.1       lava_1.8.1         
#> [10] htmltools_0.5.8.1   sass_0.4.10         rmarkdown_2.29     
#> [13] grid_4.5.1          evaluate_1.0.5      jquerylib_0.1.4    
#> [16] fastmap_1.2.0       numDeriv_2016.8-1.1 yaml_2.3.10        
#> [19] mvtnorm_1.3-3       lifecycle_1.0.4     compiler_4.5.1     
#> [22] codetools_0.2-20    ucminf_1.2.2        Rcpp_1.1.0         
#> [25] future_1.67.0       lattice_0.22-7      digest_0.6.37      
#> [28] R6_2.6.1            parallelly_1.45.1   parallel_4.5.1     
#> [31] Matrix_1.7-4        bslib_0.9.0         tools_4.5.1        
#> [34] globals_0.18.0      cachem_1.1.0

Recurrent events

Klaus Holst & Thomas Scheike

2025-08-30

Overview