smartest: Conduct statistical tests using a SMART data

In SMARTAR: Sequential Multiple Assignment Randomized Trial and Adaptive Randomization

Description

Return a message that contains the results of statistical tests to compare the values of adaptive treatment strategies defined in a SMART data. The statistical tests include (1) a global test (2) a series of pairwise tests.

Usage

smartest(
  data,
  family = c("gaussian", "binomial")[1],
  method = c("Gest", "IPW")[1],
  digits = NULL,
  common = FALSE,
  alpha = 0.05,
  adjust = NULL,
  ntest = NULL
)

Arguments

data	Input data frame of the SMART data used for analysis, which include the variables of stage-1 treatments (A1), intermediate outcome (O2), stage-2 treatment (A2) and final primary outcome (Y). If stage-1 treatment takes into account baseline information, baseline information (O1) also needs to be included.
family	A character string to specify the type of final primary outcome. The default is family=“gaussian”, which refers to the continuous primary outcome. If family=”binomial” then the primary outcome will be treated as binary variable.
method	Method used to estimate the value of adaptive treatment strategy. "Gest" for G-computation method and "IPW" for Inversed Probabiliy Weight method. Default is method="Gest".
digits	An integer indicating the number of decimal places for sequence-specific mean and variance. Default is digits=NULL.
common	If common=TRUE, the pooled variance across all the treatment sequences are used in estimation. Otherwise use the sequence-specific variance. The default is common=FALSE.
alpha	Significant level of confidence interval. The default is alpha=0.05.
adjust	A characteristic string to indicate whether the confidence intervals pairwise distance adjusted for multiple comparison. The default is adjust=NULL, which indicated no adjustment for multiple comparison. If adjust=”Bon”, the CIs are adjusted for the Bonferroni correction.
ntest	Number of pairwise tests adjusted for Bonferroni correction

Value

An objects of “Strategy” is return, which lists all the adaptive treatment strategy defined in the input data with an index number.

• ATS: Index of the treatment adaptive treatment strategy defined in the input dataset

• ds: the sequence of decision makings that define an adaptive treatment strategy

• N: number of subjects following an adaptive treatment strategy in the intut dataset

An objects of "Global.test" is return, which give the result of the global test.

• size: the total number of subjects in the input dataset

• nATS: the total number of adaptive treatment strategies defined in the input dataset

• df: the degrees of freedom of the global test, which is a chisquare test

• chisq: the chisquare test statistics for the global test

• Pvalue: the P-value of the global test

An object of "Pairwise.test"

• label: The labels of pairwise tests. The details of strategies are shown in $Strategy

• diff: Estimated pairwise distance between treatment and control adaptive treatment strategy

• lower.CI: Lower bound of confidence interval for pairwise distance

• upper.CI: Upper bound of confidence interval for pairwise distance

• Z: Test statistics of pairwise test

• P-value: P-value of pairwise test

References

Murphy, S. A. (2005),“An experimental design for the development of adaptive treatment strategies” Statistics in Medicine, 24, 1455-1481.

Ogbagaber S. B., Karp, J., and Wahed A.S. (2015), “Design of sequentially randomization trials for testing adaptive treatment strategies,” Statistics in Medicine, DOI: 10.1002/sim.6747.

Examples

smartest(data=codiacs,family="gaussian",method="Gest",
common=FALSE,alpha=0.05,adjust="Bon")

SMARTAR documentation built on July 31, 2020, 1:06 a.m.