There are two types of PTLD: early onset (occurring within two years after transplant) and late onset (occurring more than two years after transplant).
Recipients' EBV seronegative status is known to be an independent risk factor for developing early-onset PTLD after kidney transplant. It has traditionally been believed that late onset PTLD is not associated with EBV serostatus.
If I want to explore how many years after kidney transplant EBV serostatus affects the risk of PTLD (i.e., how many years after transplant before EBV serostatus no longer affects PTLD risk), what would be the best statistical method to do so? My hypothesis is that EBV serostatus negative remains an independent factor for PTLD even after 2 years post-transplant.
Below, I used a competing risk analysis (code stcrreg) and Nelson-Allen cumulative estimates, comparing risks between EBV serostatus negative and EBV serostatus negative recipients. Then, I move the time origin from time zero to at 1 year post-transplant, then origin at 2 years post-transplant, and so on and so forth. From this method, it appears that EBV serostatus is no longer a risk factor for PTLD at 4 years after transplant.
I don't think this is the right way to do it. There must be a more sophisticated way to answer my hypothesis. Please help!
Recipients' EBV seronegative status is known to be an independent risk factor for developing early-onset PTLD after kidney transplant. It has traditionally been believed that late onset PTLD is not associated with EBV serostatus.
If I want to explore how many years after kidney transplant EBV serostatus affects the risk of PTLD (i.e., how many years after transplant before EBV serostatus no longer affects PTLD risk), what would be the best statistical method to do so? My hypothesis is that EBV serostatus negative remains an independent factor for PTLD even after 2 years post-transplant.
Below, I used a competing risk analysis (code stcrreg) and Nelson-Allen cumulative estimates, comparing risks between EBV serostatus negative and EBV serostatus negative recipients. Then, I move the time origin from time zero to at 1 year post-transplant, then origin at 2 years post-transplant, and so on and so forth. From this method, it appears that EBV serostatus is no longer a risk factor for PTLD at 4 years after transplant.
I don't think this is the right way to do it. There must be a more sophisticated way to answer my hypothesis. Please help!
Code:
///start at time 0///
stset PTLD_ONSET, failure(PTLD_OUTCOME==1)
stcrreg GROUP if GROUP!=., compete (PTLD_OUTCOME == 2 3)
sts graph, cumhaz by (GROUP)
///start at 1 year////
stset PTLD_ONSET, failure(PTLD_OUTCOME==1) origin(time 365)
stcrreg GROUP if GROUP!=., compete (PTLD_OUTCOME == 2 3)
sts graph, cumhaz by (GROUP)
///start at 2 year////
stset PTLD_ONSET, failure(PTLD_OUTCOME==1) origin(time 730)
stcrreg GROUP if GROUP!=., compete (PTLD_OUTCOME == 2 3)
sts graph, cumhaz by (GROUP)
///start at 3 year////
stset PTLD_ONSET, failure(PTLD_OUTCOME==1) origin(time 1095)
stcrreg GROUP if GROUP!=., compete (PTLD_OUTCOME == 2 3)
sts graph, cumhaz by (GROUP)
///start at 4 year////
stset PTLD_ONSET, failure(PTLD_OUTCOME==1) origin (time 1460)
stcrreg GROUP if GROUP!=., compete (PTLD_OUTCOME == 2 3)
sts graph, cumhaz by (GROUP)
Comment