table1 (version 1.1)

table1: Generate an HTML table of descriptive statistics.

Description

There are two interfaces, the default, which typically takes a list of data.frames for x, and the formula interface. The formula interface is less flexible, but simpler to use and designed to handle the most common use cases. It is important to use factors appropriately for categorical variables (i.e. have the levels labeled properly and in the desired order). The contents of the table can be customized by providing user-defined `renderer' functions. Customization of the table appearance is deliberately not attempted, as this is best accomplished with CSS. To facilitate this, some tags (such as row labels) are given specific classes for easy CSS selection.

Usage

table1(x, ...)
# S3 method for default
table1(x, labels, groupspan = NULL, rowlabelhead = "",
  transpose = FALSE, topclass = "Rtable1", footnote = NULL,
  render = render.default, ...)
# S3 method for formula
table1(x, data, overall = "Overall", rowlabelhead = "",
  transpose = FALSE, droplevels = TRUE, topclass = "Rtable1",
  footnote = NULL, render = render.default, ...)

Arguments

x

An object, typically a formula or list of data.frames.

...

Further arguments, passed to render.

labels

A list containing labels for variables, strata and groups (see Details).

groupspan

A vector of integers specifying the number of strata to group together.

rowlabelhead

A heading for the first column of the table, which contains the row labels.

transpose

Logical. Should the table be transposed (i.e. strata as rows and variables as columns)?

topclass

A class attribute for the outermost (i.e. <table>) tag.

footnote

A character string to be added as a footnote to the table. The default NULL causes the footnote to be omitted.

render

A function to render the table cells (see Details). immediately displayed? Otherwise an HTML fragment is printed to stdout.

data

For the formula interface, a data.frame from which the variables in x should be taken.

overall

A label for the "Overall" column. Specify NULL or FALSE to omit the column altogether.

droplevels

Should empty factor levels be dropped?

Value

An object of class "table1".

Methods (by class)

• default: The default interface, where x is a data.frame.

• formula: The formula interface.

Details

For the default version, is is expected that x is a named list of data.frames, one for each stratum, with names corresponding to strata labels.

Examples

# NOT RUN {
dat <- expand.grid(id=1:10, sex=c("Male", "Female"), treat=c("Treated", "Placebo"))
dat$age <- runif(nrow(dat), 10, 50)
dat$age[3] <- NA  # Add a missing value
dat$wt <- exp(rnorm(nrow(dat), log(70), 0.2))

label(dat$sex) <- "Sex"
label(dat$age) <- "Age"
label(dat$treat) <- "Treatment Group"
label(dat$wt) <- "Weight"

units(dat$age) <- "years"
units(dat$wt) <- "kg"

# One level of stratification
table1(~ sex + age + wt | treat, data=dat)

# Two levels of stratification (nesting)
table1(~ age + wt | treat*sex, data=dat)

# Switch the order or nesting
table1(~ age + wt | sex*treat, data=dat)

# No stratification
table1(~ treat + sex + age + wt, data=dat)

# Something more complicated

dat$dose <- ifelse(dat$treat=="Placebo", "Placebo",
                   sample(c("5 mg", "10 mg"), nrow(dat), replace=TRUE))
dat$dose <- factor(dat$dose, levels=c("Placebo", "5 mg", "10 mg"))

strata <- c(split(dat, dat$dose),
            list("All treated"=subset(dat, treat=="Treated")),
            list(Overall=dat))

labels <- list(
    variables=list(sex=render.varlabel(dat$sex),
                   age=render.varlabel(dat$age),
                   wt=render.varlabel(dat$wt)),
    groups=list("", "Treated", ""))

my.render.cont <- function(x) {
    with(stats.default(x), 
        sprintf("%0.2f (%0.1f)", MEAN, SD))
}

table1(strata, labels, groupspan=c(1, 3, 1), render.continuous=my.render.cont)

# Transposed table
table1(~ age + wt | treat, data=dat, transpose=TRUE)

# }