I was trying to do lasso inference for survival ,model.
The variables of interest are : AMIS, BMIS DRMIS
All of them are categorical variables
I want to assess the relationship between survival and AMIS with having BMIS and DRMIS in the model in all cases
Also , to assess the relationship between survival and BMIS with having DRMIS and AMIS in the model in all cases
Also , I want to assess the relationship between survival and DRMIS with having BMIS and AMIS in the model in all cases

I read in STATA manual for survival analysis ( [ST] Survival Analysis (stata.com)) , under "Example 9" , page "90-93"
that "Among others, Royston and Lambert (2011, sec. 4.5) show that you can obtain identical hazard ratios by fitting a Poisson model"
and to

So I did the following:

stset GTIME_KI, failure(GSTATUS_DTHCNS_KI)

generate time_exposed=_t-_t0
dspoisson _d i.AMIS i.BMIS i.DRMIS, controls( i.sex BMI_TCR COLD_ISCH_KI SERUM_CREAT END_CPRA AGE i.STEROIDS_MAINT i.induction i.DIAB i.dgf timeondialysis i.DGN_TCR i.ETHCAT KDPI i.finalcmv ) sqrtlasso missingok exposure(time_exposed)


My question is whether the above syntax is correct and would give me IRR which would be identical to HR?
Would this approach be a reasonable approach to get same results for performing lasso inference for survival analysis?

Thank you very much for disturbing and thanks for your support

Kind regards