mets (version 1.3.7)

binregRatio: Percentage of years lost due to cause regression

Description

Estimates the percentage of the years lost that is due to a cause and how covariates affects this percentage by doing ICPW regression.

Usage

binregRatio(
  formula,
  data,
  cause = 1,
  time = NULL,
  beta = NULL,
  type = c("II", "I"),
  offset = NULL,
  weights = NULL,
  cens.weights = NULL,
  cens.model = ~+1,
  se = TRUE,
  kaplan.meier = TRUE,
  cens.code = 0,
  no.opt = FALSE,
  method = "nr",
  augmentation = NULL,
  outcome = c("cif", "rmtl"),
  model = c("logit", "exp", "lin"),
  Ydirect = NULL,
  ...
)

Arguments

formula

formula with outcome (see coxph)

data

data frame

cause

cause of interest (numeric variable)

time

time of interest

beta

starting values

type

"II" adds augmentation term, and "I" classical outcome IPCW regression

offset

offsets for partial likelihood

weights

for score equations

cens.weights

censoring weights

cens.model

only stratified cox model without covariates

se

to compute se's based on IPCW

kaplan.meier

uses Kaplan-Meier for IPCW in contrast to exp(-Baseline)

cens.code

gives censoring code

no.opt

to not optimize

method

for optimization

augmentation

to augment binomial regression

outcome

can do CIF regression "cif"=F(t|X), "rmtl"=E( t- min(T, t) | X)"

model

logit, exp or lin(ear)

Ydirect

use this Y instead of outcome constructed inside the program, should be a matrix with two column for numerator and denominator.

...

Additional arguments to lower level funtions

Author

Thomas Scheike

Details

Let the years lost be $$Y1= t- min(T ,) $$ and the years lost due to cause 1 $$Y2= I(epsilon==1) ( t- min(T ,t) $$ , then we model the ratio $$logit( E(Y2 | X)/E(Y1 | X)) = X^T \beta $$. Estimation is based on on binomial regresion IPCW response estimating equation: $$ X ( \Delta^{ipcw}(t) Y2 expit(X^T \beta) - Y1 ) = 0 $$ where $$\Delta^{ipcw}(t) = I((min(t,T)< C)/G_c(min(t,T)-)$$ is IPCW adjustment of the response $$Y(t)= I(T \leq t, \epsilon=1 )$$.

(type="I") sovlves this estimating equation using a stratified Kaplan-Meier for the censoring distribution. For (type="II") the default an additional censoring augmentation term $$X \int E(Y(t)| T>s)/G_c(s) d \hat M_c$$ is added.

The variance is based on the squared influence functions that are also returned as the iid component. naive.var is variance under known censoring model.

Censoring model may depend on strata (cens.model=~strata(gX)).

Examples

library(mets)
data(bmt); bmt$time <- bmt$time+runif(408)*0.001

rmst30 <- rmstIPCW(Event(time,cause!=0)~platelet+tcell+age,bmt,time=30,cause=1)
rmst301 <- rmstIPCW(Event(time,cause)~platelet+tcell+age,bmt,time=30,cause=1)
rmst302 <- rmstIPCW(Event(time,cause)~platelet+tcell+age,bmt,time=30,cause=2)

estimate(rmst30)
estimate(rmst301)
estimate(rmst302)

## percentage of total cumulative incidence due to cause 1
rmstratioI <- rmstRatio(Event(time,cause)~platelet+tcell+age,bmt,time=30,
                        cause=1,outcome="rmst")
summary(rmstratioI)

pp <- predict(rmstratioI,bmt)
ppb <- cbind(pp,bmt)

## percentage of total cumulative incidence due to cause 1
cifratio <- binregRatio(Event(time,cause)~platelet+tcell+age,bmt,time=30,cause=1)
summary(cifratio)
pp <- predict(cifratio,bmt)

rmstratioI <- binregRatio(Event(time,cause)~platelet+tcell+age,bmt,
                               time=30,cause=1,outcome="rmst")
summary(rmstratioI)

pp <- predict(rmstratioI,bmt)
ppb <- cbind(pp,bmt)