nphPower (version 1.1.0)

evalfup: Visualization of the Relationship between Follow-up and Sample Size

Description

evalfup function displays the graph showing the relationship between the follow-up time and the total sample size/event number required to achieve the the same power

Usage

evalfup(
  object,
  lower.time,
  upper.time,
  size,
  increment = 0.5,
  xlabel = "Follow-up Time",
  ylabel = "Total Sample Size/Event Number",
  title = "Relationship between Follow-up and \n Total Sample Size"
)

Value

a graph showing the relationship and a list of components:

approx.time

approximate follow-up time corresponding to specified sample size to reach the same target power

original

a list with elements of x and y. Vector x contains the follow-up duration and vector y contains the corresponding sample size

interp

a list containing the interpolated x and y included in original

Esize

a vector of events number corresponding to x in original

Arguments

object

returned object by function pwr2n.NPH

lower.time

a numeric value specifying the shortest duration time

upper.time

a numeric value specifying the longest duration time

size

an integer specifying the planned total sample size

increment

a numeric value specifying an increment number used for creating a sequence of duration times in plotting, Default: 0.5

xlabel

a text for labeling the x axis in the plot, Default: 'Follow-up Time'

ylabel

a text for labeling the y axis in the plot, Default: 'Total Sample Size'

title

a text for title in the plot: 'Relationship between Follow-up and Total Sample Size'

Details

The evalfun function helps to evaluate the relationship between sample size/event number and follow-up duration. It retrieves the trial design information from the object returned by pwr2n.NPH function. A sequence of follow-up times starting from lower.time and ending with upper.time are generated. The number of subjects and number of events required for achieving the specified power in object are calculated at each time point. An interpolation function approx from stats is applied to smooth the curves. In case of proportional hazards, the follow-up duration has little impact on the event number except for variations from numeric approximations, while in case of nonproportional hazards, the follow-up time imposes an important impact on both the total sample size and event number.

Examples

# The following code takes more than 5 seconds to run.
# \donttest{
# define design parameters
  t_enrl <- 12; t_fup <- 18; lmd0 <- log(2)/12
# define hazard ratio function
  f_hr_delay <- function(x){(x<=6)+(x>6)*0.75}
# define control hazard
  f_haz0 <- function(x){lmd0*x^0}
# perform sample size calculation using logrank test
# generate weight for test
  wlr <- gen.wgt(method="LR")
  snph1 <- pwr2n.NPH(entry = t_enrl, fup = t_fup, Wlist = wlr,
                    k = 100, ratio = 2, CtrlHaz = f_haz0, hazR = f_hr_delay)

# suppose the follow-up duration that are taken into consideration ranges
# from 12 to 24. The planned number of patients to recruit 2200.
# draw the graph
  efun <- evalfup(snph1,lower.time = 12, upper.time = 24, size = 2200,
               title = NULL)
# }