Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • cumulative incidence AND competing risks

    Hi all.

    I am doing a survival analysis and trying to figure out the cumulative incidence at different time points. But I have no idea about how to produce it. The command I use is as follows:

    stset time, failure(outcome=1)

    stcrreg drug age male ef nyha, compet(outcome=2)

    predict cif0, basecif

    predict xb, xb

    gen cif1 = 1-(1-cif0)^exp(xb)

    sort time

    list time outcome xb cif0 cif1 if time ==1 or time ==2 or time ==3

    Could anyone help check this command, please? Thank you very much!



    Li

  • #2
    Take a look a the unofficial -stcompet- command (Stata Journal):
    Code:
    search stcompet

    Comment


    • #3
      I've also written a wrapper for stcompet which reports the CIF at requested time points, analogously to sts list. You can install it from my website using:
      Code:
      net install stcomlist, from(http://www.bukharin.net/stata)

      Comment


      • #4
        Unfortunately, your code will not do what you want. (1) The proportional hazard calculation ("cif1 = 1-(1-cif0)^exp(xb)) applies only to the sub-hazard CIF, which is uninteresting, not to the actual CIF (see the manual). (2) Even if the calculation did apply to the actual, CIF, it wouldn't be of much use: the predicted CIF for an individual is at his/her analysis time "t", whether that was the time of the primary outcome, ing risk, or censoring.

        I haven't tried Phil's wrapper, but stcurve after stcrreg will plot the CIF at designated covariate values for all times in the data set. You can use the undocumented outfile() option to save the results.
        Steve Samuels
        Statistical Consulting
        [email protected]

        Stata 14.2

        Comment

        Working...
        X