Hi there,

I ran the stpm2cr function on my HIV/AIDS data set. I considered AIDS, non-AIDS diseases and death unrelated to the AIDS or non-AIDS diseases as competing risks and occurrence of the first event is the failure status. The follow up time was 30 years. Because the form of CIF for AIDS was plateau after 21 years I added cure option to the AIDS equation. The independent variable is number of positive CD4 cells which is categorized.
I wrote the codes as follows:

. stpm2cr [AIDS: CD4g1 CD4g2, scale(hazard) df(3) cure] [nonAIDS: CD4g1 CD4g2, scale(hazard) df(2)] [death: CD4g1 CD4g2, scale(hazard) df(2)], events(event2) cause(1 2 3) cens(0) eform nolog

and stata kicks back this error:
initial values not feasible
r(1400);

I tried to add lininit option to the above code for solving the error as follow:
. stpm2cr [AIDS: CD4g1 CD4g2, scale(hazard) df(3) cure] [nonAIDS: CD4g1 CD4g2, scale(hazard) df(2)] [death:age, scale(hazard) df(2)], events(event2) cause(1 2 3) cens(0) eform nolog lininit

But again I faced this error:

( invalid name
) invalid name
r(7);
  1. I do not know what this error is related to? Is it syntactically?
  2. I tried to use other option of maximization including difficult, iteration, technique options but for all of them initial values not feasible.
  1. I changed the exit time in stset and set it to 20 years. In this case the command successfully implemented I got outputs but this is not my interest to reduce the follow up time because there are non-AIDS and death failures after 20 years.
Any suggestions that troubleshoot the problem would be appreciable.

Thanks,