qgcomp.cox.noboot: Quantile g-computation for survival outcomes under linearity/additivity

Description

This function performs quantile g-computation in a survival setting. The approach estimates the covariate-conditional hazard ratio for a joint change of 1 quantile in each exposure variable specified in expnms parameter

Usage

qgcomp.cox.noboot(
  f,
  data,
  expnms = NULL,
  q = 4,
  breaks = NULL,
  id = NULL,
  weights,
  cluster = NULL,
  alpha = 0.05,
  ...
)

Value

a qgcompfit object, which contains information about the effect measure of interest (psi) and associated variance (var.psi), as well as information on the model fit (fit) and information on the weights/standardized coefficients in the positive (pos.weights) and negative (neg.weights) directions.

Arguments

f: R style survival formula, which includes Surv in the outcome definition. E.g. Surv(time,event) ~ exposure. Offset terms can be included via Surv(time,event) ~ exposure + offset(z)
data: data frame
expnms: character vector of exposures of interest
q: NULL or number of quantiles used to create quantile indicator variables representing the exposure variables. If NULL, then gcomp proceeds with un-transformed version of exposures in the input datasets (useful if data are already transformed, or for performing standard g-computation)
breaks: (optional) NULL, or a list of (equal length) numeric vectors that characterize the minimum value of each category for which to break up the variables named in expnms. This is an alternative to using 'q' to define cutpoints.
id: (optional) NULL, or variable name indexing individual units of observation (only needed if analyzing data with multiple observations per id/cluster)
weights: "case weights" - passed to the "weight" argument of coxph
cluster: (from coxph function help) optional variable which clusters the observations, for the purposes of a robust variance. If present, it implies robust. This variable will normally be found in data.
alpha: alpha level for confidence limit calculation
...: arguments to coxph

Details

For survival outcomes (as specified using methods from the survival package), this yields a conditional log hazard ratio representing a change in the expected conditional hazard (conditional on covariates) from increasing every exposure by 1 quantile. In general, this quantity quantity is not equivalent to g-computation estimates. Hypothesis test statistics and 95% confidence intervals are based on using the delta estimate variance of a linear combination of random variables.

Examples

Run this code

set.seed(50)
N=200
dat <- data.frame(time=(tmg <- pmin(.1,rweibull(N, 10, 0.1))), 
                d=1.0*(tmg<0.1), x1=runif(N), x2=runif(N), z=runif(N))
expnms=paste0("x", 1:2)
f = survival::Surv(time, d)~x1 + x2
(fit1 <- survival::coxph(f, data = dat))
(obj <- qgcomp.cox.noboot(f, expnms = expnms, data = dat))
if (FALSE) {

# weighted analysis
dat$w = runif(N)
qdata = quantize(dat, expnms=expnms)
(obj2 <- qgcomp.cox.noboot(f, expnms = expnms, data = dat, weight=w))
obj2$fit
survival::coxph(f, data = qdata$data, weight=w)

# not run: bootstrapped version is much slower
(obj2 <- qgcomp.cox.boot(f, expnms = expnms, data = dat, B=200, MCsize=20000))
# checking whether missing data causes an issue
dat$z[1:10] <- NA
(objzmiss <- qgcomp.cox.noboot(f, expnms = expnms, data = dat))
(objzmiss_alt <- qgcomp.cox.noboot(f, expnms = expnms, data = dat[,c(expnms, "time", "d")]))
set.seed(110)
(objzmiss2 <- qgcomp.cox.boot(f, expnms = expnms, data = dat, B=3, MCsize=100))
dat$x1[1:10] <- NA
(objx1miss <- qgcomp.cox.noboot(f, expnms = expnms, data = dat))
set.seed(110)
(objx1miss2 <- qgcomp.cox.boot(f, expnms = expnms, data = dat, B=3, MCsize=100))
}