Skip to contents

Prognostic Variables Assisted Treatment Effect Detection

Source: R/pated.R
pated.Rd

pated is a wrapper function of jointCovariance for testing treatment effect in randomized clinical trials. It assumes that prognostic variables are fully randomized. This assumption can help enhancing statistical power of conventional approaches in detecting the treatment effect. Specifically, the sensitivity of the conventional models specified in ... are improved by pated.

Usage

pated(
  ...,
  data,
  nboot = 0,
  compute_cov = TRUE,
  seed = NULL,
  transform = "identity"
)

Arguments

...

model specifications built by glm_(), coxph_(), logrank_(), gee_(), mmrm_(), km_(), or quantile_(). The first specification is the primary outcome whose treatment effect is being tested; the rest are prognostic covariates used to tighten the SE.

data

either a single data frame (when all models are fitted on the same dataset) or a list of data frames (one entry per data_index). Each data frame must have a pid column carrying subject identifiers; records with the same pid across different data frames refer to the same subject.

nboot

non-zero integer if bootstrap is adopted. By default 0.

compute_cov

logic. If TRUE, empirical covariance matrix is computed using bootstrap estimate and returned. Bootstrap estimate will be abandoned. If FALSE, bootstrap estimate will be returned and no empirical covariance matrix is computed.

seed

random seed when generate bootstrap data.

transform

character. Now only supports "identity".

Value

a data frame of testing results.

Examples

## More examples can be found in the vignettes.
library(survival)
library(mvtnorm)

genData <- function(seed = NULL){
  set.seed(seed)
  n <- 300
  sigma <- matrix(c(1, .6, .6, 1), 2)
  x <- rmvnorm(n, sigma = sigma)
  z1 <- rbinom(n, 1, .6)
  z2 <- rnorm(n)
  trt <- rbinom(n, 1, .5)

  bet <- c(-.2, .2)
  y <- -.5 + x %*% bet + z1 * .3 - z2 * .1 + .1 * trt - .1 * rnorm(n)
  death <- rbinom(n, 1, .8)
  data.frame(
    y = as.numeric(y), trt = trt,
    z1 = z1, z2 = z2,
    x1 = x[, 1], x2 = x[, 2],
    death, pid = paste0('s-', seq_len(n))
  )
}

dat <- genData(seed = 31415926)

## `data_index` defaults to 1 in every spec constructor and a single
## data.frame is auto-wrapped into a list, so neither needs spelling out
## when all models are fitted on the same dataset.
fit <-
  pated(
    coxph_(Surv(time = y, event = death) ~ trt),
    glm_(z1 ~ trt, family = 'binomial'),
    glm_(z2 ~ trt, family = 'gaussian'),
    glm_(x1 ~ trt, family = 'gaussian'),
    glm_(x2 ~ trt, family = 'gaussian'),
    data = dat
  )

fit
#>                            term family     estimate     stderr       pvalue
#> 1 Surv(time = y, event = death)  PATED -0.288684461 0.08056904 0.0003395805
#> 2 Surv(time = y, event = death)  coxph -0.201743178 0.13062378 0.1224769862
#> 3                            z1    glm -0.269021294 0.23351160 0.2492930398
#> 4                            z2    glm  0.110597580 0.11184325 0.3227304829
#> 5                            x2    glm  0.039205611 0.11165593 0.7254917405
#> 6                            x1    glm  0.007879826 0.11228370 0.9440521271
#>       method       corr
#> 1      PATED         NA
#> 2   Standard  1.0000000
#> 3 Prognostic -0.4589612
#> 4 Prognostic  0.4075322
#> 5 Prognostic -0.3087809
#> 6 Prognostic  0.2148699