analyze_NEB: Analyze under NEB assumptions

In bblette1/psbinary: Assess effect modification by binary post-randomization variable

Description

Analyze data assuming no early treatment benefits

Usage

analyze_NEB(
  data,
  brange0 = c(0, 0),
  brange1 = c(0, 0),
  design = "full",
  weights = NULL,
  contrast = "logRR"
)

Arguments

data	Data frame containing the following variables Z: indicator of treatment Y: indicator of outcome Y_tau: indicator of early outcome S_star: intermediate biomarker value R: indicator of measurement of intermediate biomarker
brange0	Numeric (2 x 1) vector containing the specified lower and upper bounds of the range for sensitivity parameter β0
brange1	Numeric (2 x 1) vector containing the specified lower and upper bounds of the range for sensitivity parameter β1
design	String describing the study design / sampling scheme used. This allows for estimation of sampling weights. Options include "full", "cc" (case-cohort), and "other". When "other" is chosen the weights argument must also be specified
weights	Numeric (n x 1) vector containing pre-estimated sampling weights where n is the number of rows in data
contrast	Contrast function for estimand. Options include "logRR", "Difference", and "VE"

Value

Returns list consisting of 6 vectors corresponding to the ignorance intervals and EUIs of CEP(1, 0), CEP(0, 0), and the difference CEP(1, 0) - CEP(0, 0)

Examples

Z <- rbinom(500, 1, 0.5)
S_star <- rbinom(500, 1, 0.2)
R <- rep(1, 500)
Y_tau_0 <- rbinom(500, 1, 0.02)
Y_tau_1 <- Y_tau_0 + rbinom(500, 1, (1-Y_tau_0)*Z*0.02)
Y_tau <- Y_tau_0*(1-Z) + Y_tau_1*Z
Y <- Y_tau + (1 - Y_tau)*rbinom(500, 1, 0.1)
df <- data.frame(Z, S_star, R, Y_tau, Y)
analyze_NEB(df, c(-0.5, 0.5), design = "full", contrast = "VE")

bblette1/psbinary documentation built on June 18, 2021, 10:11 p.m.