plotRCS (version 0.1.4)

rcsplot: Plot restricted cubic splines curves

Description

Drawing of restricted cubic spline (RCS) curves form a linear regression model, a logistic regression model or a Cox proportional hazards regression model.

Usage

rcsplot(
  data,
  outcome = NULL,
  time = NULL,
  exposure = NULL,
  covariates = NULL,
  positive = NULL,
  group = NULL,
  knots = c(0.05, 0.35, 0.65, 0.95),
  knots.line = FALSE,
  ref.value = "k1",
  ref.line = TRUE,
  conf.int = TRUE,
  conf.level = 0.95,
  conf.type = c("shape", "line"),
  pvalue = TRUE,
  pvalue.digits = 3,
  pvalue.position = c(0.02, 0.98),
  pvalue.label.overall = "P for overall",
  pvalue.label.nonlinear = "P for nonlinear",
  fontsize = 12,
  fontfamily = "serif",
  linesize = 0.25,
  linecolor = "#0072B5FF",
  alpha = 0.1,
  xbreaks = NULL,
  ybreaks = NULL,
  xlab = "",
  ylab = "",
  explain = TRUE,
  ...
)

Value

A ggplot2 object with class 'rcsplot' containing the attributes of 'title' and 'note'.

Arguments

data

a data frame contain the columns of outcome, time, exposure, covariates, and group.

outcome

the name of outcome variable in the data.

time

the name of time variable in the data, for Cox regressions.

exposure

the name of exposure variable in the data.

covariates

the names of covariate variables in the data.

positive

in which positive of outcome variable to make the comparison. By default, positive is automatically defined. If outcome is a factor variable, then positive is defined as the highest level. If outcome is a numerical variable, then positive is defined as the largest value.

group

the name of group variable in the data.

knots

location of knots, detail see knot function.

knots.line

logical indicating whether or not to show the vertical lines for the knots, default FALSE.

ref.value

referrence value for the RCS curve, 'min' means using the minimum value of esposure as a reference, 'median' uses the median, 'mean' uses the mean, 'k1' uses the first knot. 'k2' uses the second knot, 'k3' uses the third 'knot', and so on. In addition, you can directly set the numerical vector as the reference value.

ref.line

logical indicating whether or not to show the referrence line, default TRUE.

conf.int

logical indicating whether or not to draw confidence interval. Defaults to TRUE.

conf.level

the confidence level to use for the confidence interval if conf.int = TRUE. Must be strictly greater than 0 and less than 1. Defaults to 0.95, which corresponds to a 95 percent confidence interval.

conf.type

confidence interval type of 'shape' (default) or 'line'.

pvalue

logical indicating whether or not to show P values, include P for overall association and P for nonlinear, default TRUE.

pvalue.digits

digits for P values, default 3.

pvalue.position

position for P value, numeric vector of length two (x-axis and y-axis).

pvalue.label.overall

label for P value of overall.

pvalue.label.nonlinear

label for P value of nonlinear.

fontsize

font size, default 12.

fontfamily

font family, default 'serif' (Times New Roman).

linesize

line size, default 0.25.

linecolor

line color, default '#0072B5FF'.

alpha

alpha for the shape of confidence interval, default 0.1.

xbreaks

breaks of x-axis.

ybreaks

breaks of y-axis.

xlab

label of x-axis.

ylab

label of y-axis.

explain

logical indicating whether or not to explain the figure, default TRUE.

...

further arguments.

See Also

rcs, knot

Examples

# View data
head(cancer)

# RCS curves for a linear regression model
rcsplot(data = cancer,
        outcome = "size",
        exposure = "age",
        covariates = c("sex", "race", "metastasis"))

# RCS curves for a logistic regression model
rcsplot(data = cancer,
        outcome = "status",
        exposure = "age",
        covariates = c("sex", "race", "size", "metastasis"))

# RCS curves for a Cox regression model
rcsplot(data = cancer,
        outcome = "status",
        time = "time",
        exposure = "age",
        covariates = c("sex", "race", "size", "metastasis"))

# Unadjusted covariates
rcsplot(data = cancer,
        outcome = "status",
        time = "time",
        exposure = "age")

# By group
rcsplot(data = cancer,
        outcome = "status",
        time = "time",
        exposure = "age",
        covariates = c("sex", "race", "size", "metastasis"),
        group = "sex")

# Set 5 knots from 'kont' function
rcsplot(data = cancer,
        outcome = "status",
        time = "time",
        exposure = "age",
        covariates = c("sex", "race", "size", "metastasis"),
        knots = knot(5))

# Set the second knot as the referrence value
rcsplot(data = cancer,
        outcome = "status",
        time = "time",
        exposure = "age",
        covariates = c("sex", "race", "size", "metastasis"),
        knots = knot(5),
        ref.value = "k2")