simGG.simlinear {simPH}

Plot simulated linear, non-time interacted quantities of interest from Cox Proportional Hazards Models

Description

simGG.simlinear uses ggplot2 to plot the quantities of interest from simlinear objects, including relative hazards, first differences, hazard ratios, and hazard rates.

Usage

## S3 method for class 'simlinear'
simGG(
  obj,
  from = NULL,
  to = NULL,
  rug = TRUE,
  rug_position = "identity",
  xlab = NULL,
  ylab = NULL,
  title = NULL,
  method = "auto",
  spalette = "Set1",
  legend = "legend",
  leg.name = "",
  lcolour = "#2B8CBE",
  lsize = 1,
  pcolour = "#A6CEE3",
  psize = 1,
  alpha = 0.2,
  type = "ribbons",
  ...
)

Arguments

obj

a simlinear class object.

from

numeric time to start the plot from. Only relevant if qi = "Hazard Rate".

to

numeric time to plot to. Only relevant if qi = "Hazard Rate".

rug

logical indicating whether or not to include a rug plot showing the distribution of values in the sample used to estimate the coxph model. Only relevant when the quantity of interest is not "Hazard Rate".

rug_position

character string. The position adjustment to use for overlapping points in the rug plot. Use "jitter" to jitter the points.

xlab

a label for the plot's x-axis.

ylab

a label of the plot's y-axis. The default uses the value of qi.

title

the plot's main title.

method

what type of smoothing method to use to summarize the center of the simulation distribution.

spalette

colour palette for when there are multiple sets of comparisons to plot. Default palette is "Set1". See scale_colour_brewer.

legend

specifies what type of legend to include (if applicable). The default is legend = "legend". To hide the legend use legend = FALSE. See the discrete_scale for more details.

leg.name

name of the legend (if applicable).

lcolour

character string colour of the smoothing line. The default is hexadecimal colour lcolour = '#2B8CBE'. Only relevant if qi = "First Difference".

lsize

size of the smoothing line. Default is 1. See ggplot2.

pcolour

character string colour of the simulated points or ribbons (when there are not multiple sets of simulations). Default is hexadecimal colour pcolour = '#A6CEE3'.

psize

size of the plotted simulation points. Default is psize = 1. See ggplot2.

alpha

numeric. Alpha (e.g. transparency) for the points, lines, or ribbons. Default is alpha = 0.2. See ggplot2. Note, if type = "lines" or type = "points" then alpah sets the maximum value per line or point at the center of the distribution. Lines or points further from the center are more transparent the further they get from the middle.

type

character string. Specifies how to plot the simulations. Can be points, lines, or ribbons. If points then each simulation value will be plotted. If lines is chosen then each simulation is plotted using a different line. Note: any simulation with a value along its length that is outside of the specified central interval will be dropped. This is to create a smooth plot. If type = "ribbons" a plot will be created with shaded areas ('ribbons') for the minimum and maximum simulation values (i.e. the middle interval set with qi in coxsimSpline) as well as the central 50 percent of this area. It also plots a line for the median value of the full area, so values in method are ignored. One of the key advantages of using ribbons rather than points is that it creates plots with smaller file sizes.

...

Additional arguments. (Currently ignored.)

Details

Uses ggplot2 to plot the quantities of interest from simlinear objects, including relative hazards, first differences, hazard ratios, and hazard rates. If there are multiple strata, the quantities of interest will be plotted in a grid by strata. Note: A dotted line is created at y = 1 (0 for first difference), i.e. no effect, for time-varying hazard ratio graphs. No line is created for hazard rates.

Value

a gg ggplot class object

References

Gandrud, Christopher. 2015. simPH: An R Package for Illustrating Estimates from Cox Proportional Hazard Models Including for Interactive and Nonlinear Effects. Journal of Statistical Software. 65(3)1-20.

Licht, Amanda A. 2011. ”Change Comes with Time: Substantive Interpretation of Nonproportional Hazards in Event History Analysis.” Political Analysis 19: 227-43.

Keele, Luke. 2010. ”Proportionally Difficult: Testing for Nonproportional Hazards in Cox Models.” Political Analysis 18(2): 189-205.

Carpenter, Daniel P. 2002. ”Groups, the Media, Agency Waiting Costs, and FDA Drug Approval.” American Journal of Political Science 46(3): 490-505.

See Also

coxsimLinear, simGG.simtvc, and ggplot2

Examples

# Load survival package
library(survival)
# Load Carpenter (2002) data
data("CarpenterFdaData")

# Estimate basic model
M1 <- coxph(Surv(acttime, censor) ~ prevgenx + lethal +
            deathrt1 + acutediz + hosp01  + hhosleng +
            mandiz01 + femdiz01 + peddiz01 + orphdum +
            vandavg3 + wpnoavg3 + condavg3 + orderent +
            stafcder, data = CarpenterFdaData)

# Simulate and plot Hazard Ratios for stafcder variable
Sim1 <- coxsimLinear(M1, b = "stafcder",
                     Xj = c(1237, 1600),
                     Xl = c(1000, 1000),
                     qi = "Hazard Ratio",
                     spin = TRUE, ci = 0.99)
simGG(Sim1, method = 'lm', rug_position = 'jitter')
simGG(Sim1, rug_position = 'jitter')

## Not run: 
# Simulate and plot Hazard Rate for stafcder variable
Sim2 <- coxsimLinear(M1, b = "stafcder", nsim = 100,
                      qi = "Hazard Rate",
                      Xj = c(1237, 1600))
simGG(Sim2, type = 'points')
simGG(Sim2, type = 'lines')

## End(Not run)