a matrix or data frame with rows corresponding to subjects and columns corresponding to variables. Note that for a data frame or matrix containing multiple time-to-event data (e.g., time to recurrence, time to death, and time to last follow-up), one column is required for each specific event.

subset of rows of original matrix or data frame to place in event chart. Logical arguments may be used here (e.g., treatment.arm == 'a', if the data frame, data, has been attached to the search directory; otherwise, data$treatment.arm == "a").

subset of columns of original matrix or data frame to place in event chart; if working with a data frame, a vector of data frame variable names may be used for subsetting purposes (e.g., c('randdate', 'event1').

column(s) or data frame variable name(s) with which to sort the chart's output. The default is NA, thereby resulting in a chart sorted by original row number.

logical flag (which takes effect only if the argument sort.by is utilized). If TRUE (default), sorting is done in ascending order; if FALSE, descending order.

logical flag (which takes effect only if the argument sort.by is utilized). If TRUE (default), NA values are considered as last values in ordering.

logical flag (which takes effect only if the argument sort.by is utilized). If FALSE, sorting data (via sort.by specified variables or columns) will be performed prior to row subsetting (via subset.r); if TRUE (default), row subsetting of original data will be done before sorting.

variable name or column number of original matrix or data frame with which to scale y-axis. Default is NA, which will result in equally spaced lines on y-axis (based on original data or sorted data if requested by sort.by). Otherwise, location of lines on y-axis will be dictated by specified variable or column. Examples of specified variables may be date of an event or a physiological covariate. Any observation which has a missing value for the y.var variable will not appear on the graph.

type of variable specified in y.var (which will only take effect if argument y.var is utilized). If "d", specifed variable is a date (either numeric julian date or an S-Plus dates object); if "n", specifed variable is numeric (e.g., systolic blood pressure level) although not a julian date.

logical flag (which takes effect only if the argument y.var is utilized). Due to potential ties in y.var variable, y.jitter (when TRUE) will jitter the data to allow discrimination between observations at the possible cost of producing slightly inaccurate dates or covariate values; if FALSE (the default), no jittering will be performed. The y.jitter algorithm assumes a uniform distribution of observations across the range of y.var. The algorithm is as follows:

size.jitter <- ( diff(range(y.var)) / (2 * (length(y.var) - 1)) ) * y.jitter.factor

The default of y.jitter.factor is 1. The entire product is then used as an argument into runif: y.var <- y.var + runif(length(y.var), -size.jitter, size.jitter)

an argument used with the y.jitter function to scale the range of added noise. Default is 1.

logical flag. If TRUE, subset observations are listed on y-axis from 1 to length(subset.r); if FALSE (default), subset observations are listed on y-axis in original form. As an example, if subset.r = 301:340 and y.renum ==TRUE, y-axis will be shown as 1 through 40. However, if y.renum ==FALSE, y-axis will be shown as 301 through 340. The above examples assume the following argument, NA.rm, is set to FALSE.

logical flag. If TRUE, subset observations which have NA for each variable specified in subset.c will not have an entry on the y-axis. Also, if the following argument, x.reference, is specified, observations with missing x.reference values will also not have an entry on the y-axis. If FALSE (default), user can identify those observations which do have NA for every variable specified in subset.c (or, if x.reference is specified, also those observations which are missing only the x.reference value); this can easily be done by examining the resulting y-axis and recognizing the observations without any plotting symbols.

column of original matrix or data frame with which to reference the x-axis. That is, if specified, all columns specified in subset.c will be substracted by x.reference. An example may be to see the timing of events before and after treatment or to see time-to-event after entry into study. The event times will be aligned using the x.reference argument as the reference point.

the “now” date which will be used for top of y-axis when creating the Goldman eventchart (see reference below). Default is max(data[, subset.c], na.rm =TRUE).

logical flag. A feature utilized by the Goldman Eventchart. When x.reference is specified as the start of follow-up and y.var = x.reference, then the Goldman chart can be created. This argument, if TRUE, will cause the plot region to be square, and will draw a line with a slope of -1 from the top of the y-axis to the right end of the x-axis. Essentially, it denotes end of current follow-up period for looking at the time-to-event data. Default is FALSE.

line type of now.line.

line width of now.line.

color of now.line.

graph option, pty='m' is the default; use pty='s' for the square looking Goldman's event chart.

date of origin to consider if dates are in julian, SAS , or S-Plus dates object format; default is January 1, 1960 (which is the default origin used by both S-Plus and SAS). Utilized when either y.julian = FALSE or x.julian = FALSE.

title for event chart. Default is 'Event Chart'.

column or data frame variable name used for y-axis labels. For example, if c('pt.no') is specified, patient ID (stored in pt.no) will be seen on y-axis labels instead of sequence specified by subset.r. This argument takes precedence over both y.axis = 'auto' and y.axis = 'custom' (see below). NOTE: Program will issue warning if this argument is specified and if is.na(y.var) == FALSE; y.idlabels will not be used in this situation. Also, attempting to plot too many patients on a single event chart will cause undesirable plotting of y.idlabels.

character string specifying whether program will control labelling of y-axis (with argument "auto"), or if user will control labelling (with argument "custom"). If "custom" is chosen, user must specify location and text of labels using y.axis.custom.at and y.axis.custom.labels arguments, respectively, listed below. This argument will not be utilized if y.idlabels is specified.

user-specified vector of y-axis label locations. Must be used when y.axis = "custom"; will not be used otherwise.

user-specified vector of y-axis labels. Must be used when y.axis = "custom"; will not be used otherwise.

logical flag (which will only be considered if y.axis == "auto" and (!is.na(y.var) & y.var.type== "d"). If FALSE (default), will convert julian numeric dates or S-Plus dates objects into “mm/dd/yy” format for the y-axis labels. If TRUE, dates will be printed in julian (numeric) format.

two-dimensional vector representing the number of units that the user wants to increase ylim on bottom and top of y-axis, respectively. Default c(0,0). This argument will not take effect if the Goldman chart is utilized.

single label to be used for entire y-axis. Default will be the variable name or column number of y.idlabels (if non-missing) and blank otherwise.

logical flag. If TRUE (default), lower and upper limits of x-axis will be based on all observations (rows) in matrix or data frame. If FALSE, lower and upper limits will be based only on those observations specified by subset.r (either before or after sorting depending on specification of sort.by and value of sort.after.subset).

character string specifying whether program will control labelling of x-axis (with argument "auto"), or if user will control labelling (with argument "custom"). If "custom" is chosen, user must specify location and text of labels using x.axis.custom.at and x.axis.custom.labels arguments, respectively, listed below.

user-specified vector of x-axis label locations. Must be used when x.axis == "custom"; will not be used otherwise.

user-specified vector of x-axis labels. Must be used when x.axis == "custom"; will not be used otherwise.

logical flag (which will only be considered if x.axis == "auto"). If FALSE (default), will convert julian dates or S-plus dates objects into “mm/dd/yy” format for the x-axis labels. If TRUE, dates will be printed in julian (numeric) format. NOTE: This argument should remain TRUE if x.reference is specified.

two-dimensional vector representing the number of time units (usually in days) that the user wants to increase xlim on left-hand side and right-hand side of x-axis, respectively. Default is c(0,0). This argument will not take effect if the Goldman chart is utilized.

a factor whose reciprocal is multiplied to original units of the x-axis. For example, if the original data frame is in units of days, x.scale = 365 will result in units of years (notwithstanding leap years). Default is 1.

single label to be used for entire x-axis. Default will be “On Study Date” if x.julian = FALSE and “Time on Study” if x.julian = TRUE.

column or data frame variable name for plotting unique lines by unique values of vector (e.g., specify c('arm') to plot unique lines by treatment arm). Can take at most one column or variable name. Default is NA which produces identical lines for each patient.

vector of line types corresponding to ascending order of line.by values. If line.by is specified, the vector should be the length of the number of unique values of line.by. If line.by is NA, only line.lty[1] will be used. The default is 1.

vector of line widths corresponding to ascending order of line.by values. If line.by is specified, the vector should be the length of the number of unique values of line.by. If line.by is NA, only line.lwd[1] will be used. The default is 1.

vector of line colors corresponding to ascending order of line.by values. If line.by is specified, the vector should be the length of the number of unique values of line.by. If line.by is NA, only line.col[1] will be used. The default is 1.

a 2xk matrix with k=number of pairs of additional line segments to add. For example, if it is of interest to draw additional line segments connecting events one and two, two and three, and four and five, (possibly with different colors), an appropriate line.add argument would be matrix(c('first.event','second.event','second.event','third.event', 'fourth.event','fifth.event'), 2, 3). One line segment would be drawn between first.event and second.event, a second line segment would be drawn between second.event and third.event, and a third line segment would be drawn between fourth.event and fifth.event. Different line types, widths and colors can be specified (in arguments listed just below).

The convention use of subset.c and line.add must match (i.e., column name must be used for both or column number must be used for both).

If line.add != NA, length of line.add.lty, line.add.lwd, and line.add.col must be the same as number of pairs of additional line segments to add.

NOTE: The drawing of the original default line may be suppressed (with line.col = 0), and line.add can be used to do all the line plotting for the event chart.

a kx1 vector corresponding to the columns of line.add; specifies the line types for the k line segments.

a kx1 vector corresponding to the columns of line.add; specifies the line widths for the k line segments.

a kx1 vector corresponding to the columns of line.add; specifies the line colors for the k line segments.

vector of pch values for points representing each event. If similar events are listed in multiple columns (e.g., regular visits or a recurrent event), repeated pch values may be listed in the vector (e.g., c(2,4,rep(183,3))). If length(point.pch) < length(subset.c), point.pch will be repeated until lengths are equal; a warning message will verify this condition.

vector of size of points representing each event. If length(point.cex) < length(subset.c), point.cex will be repeated until lengths are equal; a warning message will verify this condition.

vector of colors of points representing each event. If length(point.col) < length(subset.c), point.col will be repeated until lengths are equal; a warning message will verify this condition.

a single number (may be non-integer), which is the base multiplier for the value of the cex of the plotted points, when interest lies in a variable size allowed for certain points, as a function of the quantity of the variable(s) in the dataset specified in the point.cex.mult.var argument; multiplied by original point.cex value and then the value of interest (for an individual) from the point.cex.mult.var argument; used only when non-NA arguments are provided to point.cex.mult.var; default is 1. .

vector of variables to be used in determining what point.cex.mult is multiplied by for determining size of plotted points from (possibly a subset of) subset.c variables, when interest lies in a variable size allowed for certain points, as a function of the level of some variable(s) in the dataset; default is NA.

vector of variables in the dataset to ignore for purposes of using point.cex.mult; for example, for some variables there may be interest in allowing a variable size allowed for the plotting of the points, whereas other variables (e.g., dropout time), there may be no interest in such manipulation; the vector should be the same size as the number of variables specified in subset.c, with NA entries where variable point size is of interest and the variable name (or location in subset.c) specified when the variable point size is not of interest; in this latter case, the associated argument in point.cex is instead used as the point cex; used only when non-NA arguments are provided to point.cex.mult.var; default is NA

logical flag; if TRUE, a legend will be plotted. Location of legend will be based on specification of legend.location along with values of other arguments listed below. Default is FALSE (i.e., no legend plotting).

will be used only if legend.plot = TRUE. If "o" (default), a one-page legend will precede the output of the chart. The user will need to hit enter in order for the event chart to be displayed. This feature is possible due to the dev.ask option. If "i", an internal legend will be placed in the plot region based on legend.point.at. If "l", a legend will be placed in the plot region using the locator option. Legend will map points to events (via column names, by default) and, if line.by is specified, lines to groups (based on levels of line.by).

title for the legend; default is title to be used for main plot. Only used when legend.location = "o".

size of text for legend title. Only used when legend.location = "o".

line location of legend title dictated by mtext function with outer = FALSE option; default is 1.0. Only used when legend.location = "o".

location of upper left and lower right corners of legend area to be utilized for describing events via points and text.

vector of pch values for points representing each event in the legend. Default is point.pch.

text to be used for describing events; the default is setup for a data frame, as it will print the names of the columns specified by subset.c.

size of text for points and event descriptions. Default is 2.5 which is setup for legend.location = "o". A much smaller cex is recommended (possibly 0.75) for use with legend.location = "i" or legend.location = "l".

option to put a box around the legend(s); default is to have no box (legend.bty = "n"). Option legend.bty = "o" will produce a legend box.

if line.by was specified (with legend.location = "o" or legend.location = "i"), this argument will dictate the location of the upper left and lower right corners of legend area to be utilized for describing the different line.by values (e.g., treatment.arm). The default is setup for legend.location = "o".

text to be used for describing line.by values; the default are the names of the unique non-missing line.by values as produced from the table function.

vector of line widths corresponding to line.by values.

number used for locator argument when legend.locator = "l". If 1 (default), user is to locate only the top left corner of the legend box. If 2, user is to locate both the top left corner and the lower right corner. This will be done twice when line.by is specified (once for points and once for lines).

additional par arguments for use in main plot.