getGSDesign: Get Group Sequential Design

Description

Generate boundaries for a group sequential design using a one-sided test

Usage

getGSDesign(info.rates = NULL, looks = NULL, as.type = "asOF")

Value

Returns an 'rpact' TrialDesign object.

Arguments

info.rates: Analysis times expressed as rate of information accrual. Expects a vector with the last item representing the final analysis and equal to 1. For example, information rates for two-stage trial with interim analysis half way through is c(0.5, 1). One of two options for expressing analysis times. Either 'info.rates' or 'looks' must be specified.
looks: Analysis times expressed by number of patients accrued at each point. Expects a vector with the last item being equal to the maximum sample size. For example. looks for a three stage trial with maximum sample size of 300 and analysis planned every 100 patients is c(100, 100, 100). One of two options for expressing analysis times. Either 'info.rates' or 'looks' must be specified.
as.type: Time of alpha spending function to use. Options are as outlined by 'rpact'. Default is 'asOF' \(O'Brien-Fleming-type\).

Details

To generate confidence-based thresholds, we are interested in a one-sided test and alpha is set at 0.025. To generate the stopping thresholds, specify either looks or information rates, and the alpha spending function if difference from O-Brien-Fleming-type.

Examples

Run this code

# calculate critical values for a two-stage trial with an interim analysis half-way through
# Use Pocock-type alpha spending
design <- getGSDesign(info.rates = c(0.5, 1), as.type = 'asP')
critical.stagewise.alpha.levels <- design$stageLevels

# calculate values for a 6-stage trial with a maximum sample size of 1000
# interim analysis begins at 500 patients accrued and continues every 100 patients after
design <- getGSDesign(looks = seq(500, 1000, 100))