Hello
I have been asked to make curves for: time to progression and
progression free survival (PFS) based on a table that has:
1) Date patient entered on the study (DOS)
2) Last know a live (LKA)
3) Date of progression (TTP)
4) Date of death (DOD)

I tried to find exact definition of TTP and PFS but no luck.

I appreciate very much if anyone can help or refer me to a site or
example with real data.

ahlin777@gmail.com wrote in news:1166565117.306355.30140@
73g2000cwn.googlegroups.com:

Hello
I have been asked to make curves for: time to progression and
progression free survival (PFS) based on a table that has:
1) Date patient entered on the study (DOS)
2) Last know a live (LKA)
3) Date of progression (TTP)
4) Date of death (DOD)

I tried to find exact definition of TTP and PFS but no luck.

I appreciate very much if anyone can help or refer me to a site or
example with real data.

Time to progression ... if TTP is real the date of progression: (the name
of the variable suggests that the subtraction may have already occurred.
Please check.)
= TTP-DOS
If you were making a curve "they" may want the cumulative number of
progression events divided by the number at risk. It is also common in
survival analysis to look at the cumulative hazard, which will no be just
the sum of events divided by the number at time=0.

When calculating progression free survival you want the survival estimate
to _not_ include any persons who have progressed. Once they have
progressed they are censored, i.e removed from the risk set.

Y(0) = N
Y(t) = Number at risk at event
= Y(t-1) - number of events - number of censorings prior to t
n(t) = number events at time t

PFS(t) = iterated product at each event time of (1- n(t)/Y(t) )

There are varying ways to account for the time of exposure contributed by
the censored cases. You could multiply the censorongs by 1/2 if you
thought some of their time should have been counted. You would be saying
that there could have been an event at any time between t and t-1 had
they not been censored. In practice if makes little difference.

Klein and Moeschberger's text was used by the R-project.
Their datasets are here:
http://www.biostat.mcw.edu/homepgs/klein/menu.html

The R-project's version of the data is here:
http://www.maths.lth.se/help/R/.R/library/KMsurv/html/00Index.html

Gramsch and Therneau. I think data may be in their survival package:
http://mayoresearch.mayo.edu/mayo/research/biostat/upload/surv6.tar.gz

--
David "live long and prosper" Winsemius

When calculating progression free survival you want the survival estimate
to _not_ include any persons who have progressed. Once they have
progressed they are censored, i.e removed from the risk set.

When calculating progression free survival you want the survival estimate
to _not_ include any persons who have progressed. Once they have
progressed they are censored, i.e removed from the risk set.


This is not how I interpret progression-free survival. For PFS, you
would like to know the time until sometime experiences the event of
Progression, and hence those who progress should not be censored but
should instead be counted as failures. Usually, people who die are also
considered failures since they "progressed" to death.

Using this definition, time-to-progression (TTP) and progression-free
survival (PFS) are the same thing. See
http://jjco.oxfordjournals.org/cgi/content/full/32/1/19 for some
discussion (and a definition) of PFS.

-J

When calculating progression free survival you want the survival
estimate to _not_ include any persons who have progressed. Once they
have progressed they are censored, i.e removed from the risk set.


This is not how I interpret progression-free survival. For PFS, you
would like to know the time until sometime experiences the event of
Progression, and hence those who progress should not be censored but
should instead be counted as failures. Usually, people who die are
also considered failures since they "progressed" to death.

Using this definition, time-to-progression (TTP) and progression-free
survival (PFS) are the same thing. See
http://jjco.oxfordjournals.org/cgi/content/full/32/1/19 for some
discussion (and a definition) of PFS.