Draw an interactive responder threshold plot

Shows what a responder definition keeps and what it throws away. For two arms, the curves give the share of patients in each arm who improved by at least each amount, so the whole distribution of change stays in view, with the prespecified threshold marked. Beside them, the difference in responders is drawn across every possible threshold with its confidence band, which shows how much the conclusion depends on where the line is drawn. Under both, a table gives the responders in each arm, their difference, the number needed to treat, and the mean difference, which uses every patient.

Usage

ggresponder(
  formula,
  data,
  threshold,
  higher_is_better = TRUE,
  reference = NULL,
  xlab = "Improvement from baseline",
  level = 0.95,
  title = NULL,
  caption = NULL,
  family = "Lato"
)

Arguments

formula

A formula change ~ arm, where change is each patient's change from baseline and arm has two values.

data

A data frame holding both.

threshold

The prespecified improvement that makes a responder, in the units of change, such as a minimal important difference.

higher_is_better

Whether a rise in change is an improvement. When FALSE, as for pain, the change is turned around so that improvement is positive.

reference

The control arm. Defaults to the first level of arm.

xlab

Label of the improvement axis, with its unit.

level

Confidence level for the intervals.

title, caption

Title above the plot and note below it.

family

Font family. The package ships Lato and registers it on load.

Value

An object of class ggresponder, which prints as an interactive widget. Use graph_widget(), graph_plot() or graph_save() for the widget, a static ggplot or a file. The field responders holds the measures at the threshold and curve the difference at every threshold.

Details

In the widget, a slider or a drag on either panel moves the threshold, and every number and a sentence follow; a button returns to the prespecified threshold.

Responders have Wilson score intervals and their difference the hybrid score interval of Newcombe (1998). The number needed to treat is the reciprocal of the difference; when the interval of the difference includes zero, its interval runs from benefit through infinity to harm, as Altman (1998) describes. The mean difference has a Welch interval.

Examples

set.seed(3)
pain <- data.frame(arm = rep(c("Placebo", "Active"), each = 120),
                   change = c(rnorm(120, -1.3, 2), rnorm(120, -2.2, 2)))
ggresponder(change ~ arm, pain, threshold = 2, higher_is_better = FALSE,
            xlab = "Improvement in pain (points on a 0 to 10 scale)")