Announcement

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

  • what to do when proportional assumption is not met in stata

    I have built a Cox multiple regression model. When I tested the proportional hazard assumption using estat phtest command, I realized the PH assumption is not met.
    Here is what I got:
    Code:
     estat phtest
    
          Test of proportional-hazards assumption
    
          Time:  Time
          ----------------------------------------------------------------
                      |                      chi2       df       Prob>chi2
          ------------+---------------------------------------------------
          global test |                     95.70       12         0.0000
          ----------------------------------------------------------------
    In this case, what is the next step? What to do when the proportional hazards assumption is not met?

  • #2
    Hello Eric,

    Your query is deeply rooted in the background knowledge related to time-to-event models. I strongly recommend you take a close look on this matter in a referential book on survival analysis. That said, among the strategies you may rely on if the PH assumption is violated, you shall include: 1) delving with the selection of variables so as to provide a fit model; 2) selecting the "culprit " variable as the strata; 3) extending the model and including time-dependent variables.

    Hopefully that helps!

    Best,

    Marcos
    Best regards,

    Marcos

    Comment


    • #3
      Cross-posted at http://stats.stackexchange.com/quest...t-met-in-stata

      Please note our cross-posting policy, which is that you should tell us about it. http://www.statalist.org/forums/help#crossposting

      Comment


      • #4
        Eric;
        you may be interested in this article: http://www.ncbi.nlm.nih.gov/pubmed/15757960
        Kind regards,
        Carlo
        (Stata 19.0)

        Comment


        • #5
          Thanks so much you all for the helpful comment and links.
          Dear Nick,
          Thanks for reminding me the cross-posting policy. Well noted.

          Comment

          Working...
          X