Create a Lexis object
Create an object of class Lexis to represent follow-up in
multiple states on multiple time scales.
Lexis( entry, exit, duration, entry.status = 0, exit.status = 0, id, data,
merge=TRUE, states, notes=TRUE, tol=.Machine$double.eps^0.5, keep.dropped=FALSE )entry |
a named list of entry times. Each element of the list is
a numeric variable representing the entry time on the named time
scale. All time scales must have the same units (e.g. years).
The names of the timescales must be different from any column name in
|
exit |
a named list of exit times. |
duration |
a numeric vector giving the duration of follow-up. |
entry.status |
a vector or a factor giving the status at entry |
exit.status |
a vector or factor giving status at exit. Any change in status during follow-up is assumed to take place exactly at the exit time. |
id |
a vector giving a unique identity value for each person
represented in the Lexis object. Defaults to |
data |
an optional data frame, list, or environment containing
the variables. If not found in |
merge |
a logical flag. If |
states |
A vector of labels for the states. If given, the state
variables |
notes |
Logical. Should notes on entry states and time be given. |
tol |
Numerical tolerance for follow-up time. Rows with duration less than this value are automatically dropped. |
keep.dropped |
Logical. Should dropped rows from |
The analysis of long-term population-based follow-up studies typically
requires multiple time scales to be taken into account, such as
age, calender time, or time since an event. A Lexis object is
a data frame with additional attributes that allows these multiple time
dimensions of follow-up to be managed.
Separate variables for current end exit state allows representation of multistate data.
Lexis objects are named after the German demographer Wilhelm Lexis (1837-1914), who is credited with the invention of the "Lexis diagram" for representing population dynamics simultaneously by several timescales.
The Lexis function can create a minimal Lexis object
with only those variables required to define the follow-up history in
each row. Additional variables can be merged into the Lexis
object using the merge method for Lexis objects. The
latter is the default.
There are also merge, subset and transform methods for
Lexis objects. They work as the corresponding methods for data-frames
but ensures that the result is a Lexis object.
An object of class Lexis. This is represented as a data frame
with a column for each time scale (with names equal to the union of
the names of entry and exit), and additional columns with the
following names:
lex.id |
Identification of the persons. |
lex.dur |
Duration of follow-up. |
lex.Cst |
Entry status ( |
lex.Xst |
Exit status (e |
If merge=TRUE (the default) then the Lexis object will also contain
all variables from the data argument.
Only two of the three arguments entry, exit and
duration need to be given. If the third parameter is missing,
it is imputed.
If only either exit or duration are supplied it is assumed that
entry is 0. This is only meaningful (and therefore checked) if there
is only one timescale.
If any of entry.status or exit.status are of mode character,
they will both be converted to factors.
If entry.status is not given, then its class is automatically
set to that of exit.status. If exit.status is a
character or factor, the value of entry.status is set to the
first level. This may be highly undesirable, and therefore noted. For
example, if exit.status is character the first level will be
the first in the alphabetical ordering; slightly unfortunate if values
are c("Well","Diseased"). If exit.status is logical, the
value of entry.status set to FALSE. If
exit.status is numeric, the value of entry.status set to
0.
If entry.status or exit.status are factors or character,
the corresponding state variables in the returned Lexis object,
lex.Cst and lex.Xst will be (unordered) factors with
identical set of levels, namely the union of the levels of
entry.status and exit.status.
Martyn Plummer with contributions from Bendix Carstensen
# A small bogus cohort
xcoh <- structure( list( id = c("A", "B", "C"),
birth = c("14/07/1952", "01/04/1954", "10/06/1987"),
entry = c("04/08/1965", "08/09/1972", "23/12/1991"),
exit = c("27/06/1997", "23/05/1995", "24/07/1998"),
fail = c(1, 0, 1) ),
.Names = c("id", "birth", "entry", "exit", "fail"),
row.names = c("1", "2", "3"),
class = "data.frame" )
# Convert the character dates into numerical variables (fractional years)
xcoh <- cal.yr( xcoh, format="%d/%m/%Y", wh=2:4 )
# See how it looks
xcoh
str( xcoh )
# Define a Lexis object with timescales calendar time and age
Lcoh <- Lexis( entry = list( per=entry ),
exit = list( per=exit,
age=exit-birth ),
exit.status = fail,
data = xcoh )
Lcoh
# Using character states may have undesired effects:
xcoh$Fail <- c("Dead","Well","Dead")
Lexis( entry = list( per=entry ),
exit = list( per=exit, age=exit-birth ),
exit.status = Fail,
data = xcoh )
# ...unless you order the levels correctly
( xcoh$Fail <- factor( xcoh$Fail, levels=c("Well","Dead") ) )
Lexis( entry = list( per=entry ),
exit = list( per=exit, age=exit-birth ),
exit.status = Fail,
data = xcoh )Please choose more modern alternatives, such as Google Chrome or Mozilla Firefox.