Plot variable over time

Function plotting a variable of interest over time (and by hospital, or other grouping variables).

Usage

plot_over_time(
  data,
  plot_var = NULL,
  time_var = "discharge_date_time",
  line_group = "hospital_num",
  color_group = NULL,
  facet_group = "hospital_num",
  time_int = "month",
  func = "mean",
  plot_cat = NULL,
  show_overall = TRUE,
  smooth_method = NULL,
  line_width = 1,
  ylimits = NULL,
  min_n = 0,
  colors = gemini_colors(1),
  base_size = 12,
  return_data = FALSE,
  ...
)

Arguments

data

(data.frame | data.table)
A table containing the relevant data to be plotted.

plot_var

(character)
The name of the outcome variable to plot.

time_var

(character)
The name of the variable specifying time. By default: "discharge_date_time".

line_group

(character)
Grouping variable representing individual lines. By default: "hospital_num" (unless "hospital_num" does not exist, in which case the function will check for "hospital_id" instead).

color_group

(character)
Grouping variable used for color coding.

facet_group

(character)
Grouping variable specifying facet plots.

time_int

(character)
Time interval used for plotting (i.e., x-axis intervals). Currently, the function can automatically calculate any of the following time intervals for any date-time variables:

"month" (default)
"quarter"
"year"
"fisc_year" for hospital fiscal year starting in April
"season"

For any other custom time intervals (e.g., weeks), users can calculate the desired time interval prior to running this function. As long as this custom time variable exists in the data input, users can then specify this variable as the time_int input. For example, users could derive a weekly time variable called "week" in their data table, and then specify time_int = "week". If time_var is equal to any of c("month", "quarter", "year", "fisc_year") and there is a user- provided variable of that same name in the data input, the function will default to the variable that exists in the user-provided input.

func

(character)
The summary function used to aggregate data by time & hospital. Has to be one of the following:

"mean" (default)
"median"
"%" or "prct"/"perc" (for categorical variables), users also need to specify plot_cat to specify which category to plot the percentage of (see below)
"n" or "count" to plot the count of rows per hospital/time period
"missing" or "na"

plot_cat

Required when func = "%" and/or when plotting categorical/factor variables. Users need to specify the level of any categorical variables to be plotted. For example, to plot the percentage of female encounters, plot_cat = "F". Multiple categories can be provided as a vector, e.g., to plot all non-male encounters: plot_cat = c("F", "O"). If no plot_cat is specified, the function will sort the unique entries of the specified plot_var and plot the percentage of rows matching the highest value (e.g., for logical variables, % TRUE will be plotted by default). plot_cat can also be used for numeric variables, e.g., plot_cat = 0 to plot encounters with Charlson Comorbidity Index = 0.

show_overall

(logical)
Flag indicating whether to plot thick line representing overall value across hospitals.

smooth_method

(character)
Character specifying which smoothing method to apply (if any). By default, this is set to NULL (i.e., no smoothing applied). If users specify a method (e.g., smooth_method = "glm"), individual data points will be shown as dots and the smoothed time trend will be shown as a line.

line_width

(numeric)
Width of individual lines. Summary line will be 2 * line_width.

ylimits

(numeric)
Numeric vector specifying limits for y-axis e.g. c(0, 100). To specify only the lower/upper limit, use NA (e.g., c(NA, 100) to fix upper limit only).

min_n

(numeric)
Minimum number of data points required for each hospital * time point combination. Data points with cell count < min_n will be suppressed.

colors

(character)
Character vector specifying line color(s).

base_size

(numeric)
Numeric input to determine the base font size for each subplot in pts (default = 12)

return_data

(logical)
Flag indicating whether to return a list of 2 data.tables with aggregated data ([1] by hospital and [2] overall). If FALSE (default), will return plot.

...

If a facet_group is specified: Additional arguments passed to lemon::facet_rep_wrap() (wrapper for ggplot2::facet_wrap), e.g., scales = "fixed" (default) vs. scales = "free", nrow/ncol etc.

Value

By default, returns ggplot figure. If return_data = TRUE, returns data.tables containing aggregated data.

Usage

Arguments

Value

See also