Greg (version 2.0.2)

timeSplitter: A function for splitting a time according to time periods

Description

If we have a violation of the cox proprtional hazards assumption we need to split an individual's followup time into several. See vignette("timeSplitter", package = "Greg") for a detailed description.

Usage

timeSplitter(
  data,
  by,
  time_var,
  event_var,
  event_start_status,
  time_related_vars,
  time_offset
)

Value

data.frame with the split data. The starting time for each period is named Start_time and the ending time is called Stop_time. Note that the resulting event_var will now contain the time-splitted eventvar.

Arguments

data

The dataset that you want to split according to the time_var option.

by

The time period that you want to split the dataset by. The size of the variable must be in proportion to the the time_var. The by variable can also be a vector for each time split, useful if the effect has large varyations over time.

time_var

The name of the main time variable in the dataset. This variable must be a numeric variable.

event_var

The event variable

event_start_status

The start status of the event status, e.g. "Alive"

time_related_vars

A dataset often contains other variabels that you want to update during the split, most commonly these are age or calendar time.

time_offset

If you want to skip the initial years you can offset the entire dataset by setting this variable. See detailed description below.

The time_offset - details

Both time_var and other variables will be adjusted by the time_offset, e.g. if we the time scale is in years and we want to skip the first 4 years we set the time_offset = 4. In the outputted dataset the smallest time_var will be 0. Note: 0 will not be included as we generally want to look at those that survived the start date, e.g. if a patient dies on the 4-year mark we would not include him/her in our study.

Details

Important note: The time variables must have the same time unit. I.e. function can not dedu if all variables are in years or if one happens to be in days.

Examples

test_data <- data.frame(
  id = 1:4,
  time = c(4, 3.5, 1, 5),
  event = c("alive", "censored", "dead", "dead"),
  age = c(62.2, 55.3, 73.7, 46.3),
  date = as.Date(
    c("2003-01-01", 
      "2010-04-01", 
      "2013-09-20",
      "2002-02-23")),
  stringsAsFactors = TRUE
)
timeSplitter(test_data, .5, 
             time_var = "time",
             time_related_vars = c("age", "date"),
             event_var = "event")