principal stratifictation sensitivity analysis using the HHS method.

Share:

Description

Performs a principal stratifictation sensitivity analysis using the method described in Hudgens, Hoering, and Self (2003).

Usage

1
2
3
4
5
sensitivityHHS(z, s, y, bound = c("upper", "lower"), selection,
               groupings, empty.principal.stratum, ci = 0.95,
               ci.method = c("bootstrap", "analytic"), na.rm = FALSE,
               N.boot = 100, oneSidedTest = FALSE, twoSidedTest = TRUE,
               isSlaveMode=FALSE)

Arguments

z

vector; contains the grouping values (e.g., treatment assignment) for each record.

s

vector; indicates whether a record is selected.

y

outcome vector. Can be NA for unselected records.

bound

vector selecting which bound should be calculated, upper and/or lower. Partial string matching is performed.

selection

The value of s indicating selection.

groupings

Vector of two elements c(g0,g1), first element g0 being the value of z which delineates the first group, the last element g1 being the value of z which delineates the second group.

empty.principal.stratum

vector of two elements c(s0,s1); describes the s values that select the empty principal stratum. If empty.principal.stratum=c(s0,s1), then stratum defined by S(g0)==s0 and S(g1)==s1 is the empty stratum. In this example s0 and s1 refer to the two possible values of s. (Note: method only works if s0 != s1).

ci

numeric vector; confidence interval level, defaults to 0.95

ci.method

character; method by which the confidence interval and variance are calculated. Can be “analytic” or “bootstrap”. Defaults to c("analytic","bootstrap"). Currently only works for “bootstrap”.

na.rm

logical; indicates whether records that are invalid due to NA values should be removed from the data set.

N.boot

integer. Number of bootstrap repetitions that will be run when ci.method includes “bootstrap”.

oneSidedTest

logical. Return a one sided confidence interval for ACE. Defaults to FALSE

twoSidedTest

logical. Return a two sided confidence interval for ACE. Defaults to TRUE

isSlaveMode

logical. Internal Use only. Used in recursion.

Details

Performs a sensitivity analysis estimating the average causal effect among those who would have been selected regardless of treatment assignment (ACE). The method assumes no interference (i.e., potential outcomes of all subjects are unaffected by treatment assignment of other subjects), ignorable (i.e., random) treatment assignment, and monotonicity (i.e., one of the principal strata is empty). ACE is still not identified after making these assumptions, so this method computes the lower and upper bounds of the estimated ACE. These bounds correspond to the values one would get if using sensitivityGBH and specifying the sensitivity parameter beta as -Inf or Inf.

Value

an object of class sensitivity2d.

ACE

ACE=E(Y(g1)-Y(g0)|S(g1)==S(g0)==selection). Vector of the estimated ACE values at the specified bounds.

ACE.ci

vector; confidence interval of ACE determined by quantiles of bootstrap if ci.method includes “bootstrap”. Otherwise calculated using analytic variance with large sample normal approximation (NOT YET WORKING).

ACE.var

vector; estimated variance of ACE.

y0

vector of unique y values in the first group.

Fas0

matrix of estimated empirical distribution function values for y0 in the first group in the always selected principal stratum at the bounds. Pr(Y(g0) <= y0|S(g0)=S(g1)=selection)

y1

vector of unique y values in the second group.

Fas1

matrix of estimated empirical distribution function values for y1 in the second group in the always selected principal stratum at the bounds. Pr(Y(g1) <= y1|S(g0)=S(g1)=selection)

Author(s)

Bryan E. Shepherd
Department of Biostatistics
Vanderbilt University

Charles Dupont
Department of Biostatistics
Vanderbilt University

References

Hudgens MG, Hoering A, and Self SG (2003), "On the Analysis of Viral Load Endpoints in HIV Vaccien Trials," Statistics in Medicine 22, 2281-2298.

See Also

sensitivityGBH, sensitivityJR, sensitivitySGL

Examples

1
2
3
4
5
6
7
8
data(vaccine.trial)
est.bounds<-with(vaccine.trial,
                 sensitivityHHS(z=treatment, s=hiv.outcome, y=logVL,
                     selection="infected", groupings=c("placebo","vaccine"),
                     empty.principal.stratum=c("not infected","infected"),
                     N.boot=100)
                )
est.bounds