[R] Survival data with time dependent covariate

[BrittD]

http://r.789695.n4.nabble.com/file/n4619765/survival_file.png 

Hi everyone,

This is what my data looks like, I haven't included the covariates, the
example would get too large.

         uid     id              date                   feverstart  
dumfever
130  75      346465    2011-04-11       <NA>           0
131  75      360287    2011-04-18       <NA>           0
132  75      373195    2011-04-25       <NA>           0
133  75      388540    2011-05-02       <NA>           0
134  76      4369        2010-11-02       2010-10-19 1
135  76      19789      2010-11-09       <NA>           0
136  76      33347      2010-11-16       <NA>           0
137  76      48872      2010-11-23       <NA>           0
138  76      62967      2010-11-30       <NA>           0
139  76      77395      2010-12-07       <NA>           0
140  76      92703      2010-12-14       <NA>           0
141  76     106219     2010-12-21       <NA>           0
142  76     119836     2010-12-28       <NA>           0
143  76     134060     2011-01-04       <NA>           0
144  76     151495     2011-01-13       2011-01-10 1
145  76     166549     2011-01-20       <NA>           0
146  76     182783     2011-01-27       <NA>           0
147  76     196972     2011-02-03       <NA>           0
148  76     213502     2011-02-10       <NA>           0
149  76     230432     2011-02-17       <NA>           0
150  76     245159     2011-02-24       <NA>           0

The uid variable is a number for a person.
The id variable is a number corresponding to a specific measurement on an
indivudual. 
The date is the date of the measurement.
Feverstart is the day the fever of a person started and dumfever is a
dummy-variable which tells use whether the person had a fever or not.

I need to build a dateset with a start and a stop variable for each day and
a variable that tells me whether the person got a fever that day, like this:

start                 stop               fever   
2011-04-11     2011-04-12   0 
2011-04-12     2011-04-13   0 

I already started to write a piece of code for this:

survival = as.numeric(NULL, NULL, 3)
for(i in 1:nrow(data2010)) {
start = data$date[i];   #first date
stop = data$date[i] + 86400;   #first date plus 1 day (day is in seconds)
sick = if(is.na(data[i, 7])) { 0 } else if (stop == data[i,7]){ 1 } else  0;
#if startfever is NA or date is not the same as
                                                                                                                  
#mentioned sick = 0, if date matches startfever
                                                                                                                  
#then sick = 1
	if(stop == data$date[i+1]) {      # if the stop matches the next date in
the data, go to the next line. 
		start = data$date[i+1] } else if {
 		start = stop }
}

My problem is, that if this loop is at the last measurement of a person, so
the last id of the uid, to get it to the first measurement of the next uid.
Maybe this whole way of programming is around the bush, than please tell me.

Greetings,

Britt

--
View this message in context: http://r.789695.n4.nabble.com/Survival-data-with-time-dependent-covariate-tp4619765.html
Sent from the R help mailing list archive at Nabble.com.