Pulls the top five percent of U.S. medical spenders and calculates proportions of race.

analyze.r

library(Hmisc)
library(survey)

# Import data from SAS transport file
data <- sasxport.get('data/h171.ssp')

# Build survey object using person weights
full_population_survey <- svydesign(ids=~1, data=data, weights=~perwt14f)

# Find cutoff for top 5 percent of spenders
top_five_percentile_cutoff <- svyquantile(~totexp14, full_population_survey, c(.95 ))[1]

# Get subset of survey that's just the top 5 percent of spenders
fivepercent <- subset(full_population_survey, totexp14 >= top_five_percentile_cutoff)

# Build a table tracking the proportion each race comprises of the top 5 percent
race_fivepercent <- prop.table(svytable(~racethx, design = fivepercent))

# Build a table tracking the proportion each race comprises of the total survey
race_full_population <- prop.table(svytable(~racethx, design = full_population_survey))

# Get percent change between five percent population and total population
(race_fivepercent - race_full_population) / race_full_population

# Returns this:
#           1           2           3           4           5 
# -0.48050311  0.20804408 -0.08487131 -0.46280463 -0.19622867