counterfactual.plot.discrete: Create a plot of the distribution of exposure under...

In pifpaf: Potential Impact Fraction and Population Attributable Fraction for Cross-Sectional Data

Description

Function that creates a plot of the distribution of exposure X under counterfactual scenario cft for discrete exposures.

Usage

counterfactual.plot.discrete(X, cft, weights = rep(1/nrow(as.matrix(X)),
  nrow(as.matrix(X))),
  title = "Exposure distribution under current and counterfactual scenarios",
  dnames = c("Current distribution", "Counterfactual distribution"),
  legendtitle = "Scenario", xlab = "Exposure", ylab = "Frequency",
  colors = c("deepskyblue", "tomato3"), x_axis_order = unique(X[, 1]))

Arguments

X	One dimensional vector with exposure levels.
cft	Counterfactual function of the exposure cft(X). **Optional**
weights	Normalized survey weights for the sample X.
title	String with plot title.
dnames	String vector indicating the names of the distributions for the legend.
legendtitle	String title for the legend of plot.
xlab	String label for the X-axis of the plot (corresponding to "a").
ylab	String label for the Y-axis of the plot (corresponding to "b").
colors	String vector with colors for plots.
x_axis_order	Order of names in xaxis for plot.

Value

cft_plot ggplot object plotting the shift from actual to counterfactual distribution.

Author(s)

Rodrigo Zepeda-Tello [e-mail]

Dalia Camacho-Garc<c3><ad>a-Forment<c3><ad> [e-mail]

References

Vander Hoorn, S., Ezzati, M., Rodgers, A., Lopez, A. D., & Murray, C. J. (2004). Estimating attributable burden of disease from exposure and hazard data. Comparative quantification of health risks: global and regional burden of disease attributable to selected major risk factors. Geneva: World Health Organization, 2129-40.

See Also

counterfactual.plot.continuous for plotting continuous counterfactuals, pif for Potential Impact Fraction estimation, pif.heatmap for sensitivity analysis of the counterfactual, pif.plot for a plot of potential impact fraction as a function of the relative risk's parameter theta.

Examples

#Example 1: Bivariate exposure
#--------------------------------------------------------
set.seed(2783569)
X   <- data.frame(Exposure = 
                   sample(c("Exposed","Unexposed"), 100, 
                   replace = TRUE, prob = c(0.3, 0.7)))
cft <- function(X){

     #Find which indivuals are exposed
     exposed      <- which(X[,"Exposure"] == "Exposed")
     
     #Change 1/3 of exposed to unexposed
     reduced                 <- sample(exposed, length(exposed)/3)
     X[reduced,"Exposure"]   <- "Unexposed"
     
     return(X)
}  
counterfactual.plot.discrete(X, cft)
  
#Example 2: Multivariate exposure
#--------------------------------------------------------
set.seed(2783569)
X   <- data.frame(
         Exposure = sample(c("Underweight","Normal","Overweight","Obese"), 
         1000, replace = TRUE, prob = c(0.05, 0.3, 0.25, 0.4)))
               
#Complex counterfactual of changing half of underweights to normal,
#1/2 of overweights to normal, 1/3 of obese to normal and 
#1/3 of obese to overweight
cft <- function(X){

     #Classify the individuals
     underweights    <- which(X[,"Exposure"] == "Underweight")
     overweights     <- which(X[,"Exposure"] == "Overweight")
     obese           <- which(X[,"Exposure"] == "Obese")
     
     #Sample 1/2 underweights and overweights and 2/3 of obese
     changed_under    <- sample(underweights, length(underweights)/2)
     changed_over     <- sample(overweights,  length(overweights)/2)
     changed_obese    <- sample(obese,        2*length(obese)/3)
     
     #Assign those obese that go to normal and those that go to overweight
     obese_to_normal  <- sample(changed_obese, length(changed_obese)/2)
     obese_to_over    <- which(!(changed_obese %in% obese_to_normal))
     
     #Change the individuals to normal and overweight
     X[changed_under,"Exposure"]   <- "Normal"
     X[changed_over,"Exposure"]    <- "Normal"
     X[obese_to_normal,"Exposure"] <- "Normal"
     X[obese_to_over,"Exposure"]   <- "Overweight"
     
     return(X)
}  

#Create plot of counterfactual distribution
counterfactual.plot.discrete(X, cft, x_axis_order = c("Underweight","Normal","Obese","Overweight")) 

pifpaf documentation built on May 1, 2019, 9:11 p.m.