make_exclusions: Perform Sequential Exclusions

In stopsack/khsmisc: Miscellaneous functions for epidemiology research

Perform Sequential Exclusions

Description

This function performs sequential, user-defined filter steps on the input data set. It generates the filtered data and a tibble that can be directly passed on to exclusion_flowchart to plot a flowchart of exclusions.

Usage

make_exclusions(criteria, data)

Arguments

criteria	Tibble with filtering criteria. Must contain three variables: left String with description of data before applying the filter. right String with description of data after applying the filter. filter Filtering expression quoted using expr. The filter in the last row will not be executed, because the last row serves as a description of the final data set. See examples.
data	Tibble with data set on which the filtering criteria should be applied.

Value

A tibble. Each row is a filtering step. Variables:

• left: Labels for included subset that is "left" after the filter.

• right: Labels for excluded subset (which exclusion_flowchart plots to the right).

• included The data before applying the row's filter.

• excluded The data after applying the row's filter.

• n_left Number of observations before applying the row's filter.

• n_right Number of observations after applying the row's filter.

The last row, included, contains the data after applying all filters. Access this tibble using %>% pull(included) %>% last().

Example Output

Examples

# Example data set
data(cancer, package = "survival")
cancer <- cancer %>% tibble::as_tibble()
cancer

# Define exclusion criteria
criteria <- tibble::tribble(
  ~left,                   ~right,                ~filter,
  "All patients",          "Missing ECOG status", expr(!is.na(ph.ecog)),
  "Known ECOG",            "Exclude men",         expr(sex == 2),
  "Analytical population", "",                    expr(TRUE))

# Alternative, equivalent approach to defining the criteria
# Note the use of list() around expr(...)
criteria <- dplyr::bind_rows(
  tibble::tibble(
    left = "All patients",
    right = "Missing ECOG status",
    filter = list(expr(!is.na(ph.ecog)))),
  tibble::tibble(
    left = "Known ECOG",
    right = "Exclude men",
    filter = list(expr(sex == 2))),
  tibble::tibble(
    left = "Analytical population",
    right = "",
    filter = list(expr(TRUE))))

# Perform sequential exclusions
result <- make_exclusions(
  criteria = criteria,
  data = cancer)

# Show results
result

# Access study population after all exclusions
result %>%
  dplyr::pull(included) %>%
  dplyr::last()

# Plot flow chart of exclusions (might not display in the online reference)
result %>%
  exclusion_flowchart()

stopsack/khsmisc documentation built on Sept. 22, 2023, 12:26 p.m.