search

pharmaverse / ggsurvfit

http://www.danieldsjoberg.com/ggsurvfit/
Other
70 stars 19 forks source link

Changes in PARAM #104

Closed ddsjoberg closed 1 year ago

ddsjoberg commented 1 year ago

Previous comments from @SHAESEN2

  1. Check if the param/paramcd unique in the data used for estimation. If not, then the programmer forgot to make a subset.
  2. If param or paramcd was used as strata then label should default to Time
ddsjoberg commented 1 year ago

@SHAESEN2

  1. Check if the param/paramcd unique in the data used for estimation. If not, then the programmer forgot to make a subset.

in this case, I added all unique values to the label as a clear indication that either something is incorrect, or they feel like they've done something correctly and will then need to update the label themselves.

  1. If param or paramcd was used as strata then label should default to Time

PARAM = "The description of the analysis parameter (or time to event endpoint). Example: “Time to Progression”"

Do you have any examples where this is variable can accurately be used as a stratifying variable?

SHAESEN2 commented 1 year ago
  1. I feel it is better to warn the programmer upfront than when its too late (when creating the figure). Many programmers dont have the proper statistical background to understand/see the issue.
  2. Yes ADTTE typically contains different time to event parameters (eg time to hopsitalization, time to hospital discharge,...). We often use paramcd as strata to have an overall estimate for each parameter.