salbm: Sensitivity Analysis for longitudinal binary missing data

In olssol/salbm: Sensitivity Analysis for Longitudinal Binary Missing Data

Description

Given data from one arm of a repeated measures clinical trial with binary outcomes, produces estimates of the expected value of the outcome at the final time-point for a range of sensitivity parameters.

Usage

salbm(data, trtname = "trt", trtlev = c(1,2), rf.ntree = 1000,
rf.seed = -1, rf.sampsize = floor(nrow(data)/4), rf.nodesize = 1,
nbootstraps = 1000, alphas )

Arguments

data	a dataframe in wide fromat with a column indicating treatment group and columns of outcome measures in time order.
trtname	The name of the treatment column in data.
trtlev	The levels that the treatment name takes.
rf.ntree	Number of trees to pass to the rfsrc function.
rf.seed	Seed to pass to rfsrc function.
rf.sampsize	Size of sample passed to rfsrc.
rf.nodesize	Minimum nodesize passed to rfsrc.
nbootstraps	Number of bootstraps to perform.
alphas	A vector of sensitivity parameters.

Details

The dataframe data is split into the two treatments arms using the column trtname. The remaining columns are coded as 0 or 1 for the binary outcome and should be in time order. Unobserved values are coded as NA.

Using the randomForestSRC package is used to estimate the joint distribution of the outcome values (including the non-observed outcome values). The sensitivity parameters are then used to tilt the outcome values to produce the underlying joint distribution of the underlying observed outcomes.

Value

salbm returns a list which includes the following:

Results1	the results from treatment 1.
Results2	the results from treatment 2.
ResultsD	the results from treatment 2 minus treatment 1.
bootstraps1	Bootstrap results from treatment 1
bootstraps2	Bootstrap results from treatment 2
bootstrapsD	Bootstrap results from treatment 2 minus treatment 1

See Also

The salbm_userDoc.pdf file in the Examples subdirectory.

Examples

data("salbmData")

Results <- salbm(
  data           =  salbmData,
  trtname        =  "trt",       # treatment indicator
  trtlev         =  c(1,2),      # treatment values 
    
  rf.ntree       =    25,        # num trees in the forest
  rf.seed        =  -172,        
  rf.sampsize    =    90,         
  rf.nodesize    =     3,

  nbootstraps    =     5,
  alphas         =  -1:1 )

olssol/salbm documentation built on March 19, 2020, 3:43 p.m.