Hi!
I have a dataset with approx. 30 000 patients. My outcome is metastasized colon cancer (mcc) and I have time to event from diagnosis of cancer to diagnosis of metastasized cancer (mcc_time). I have performed a Kaplan-Meier plot on non-imputed data, using sts list and stci to estimate median survival and estimated 5-year survival.
I have thereafter performed a multiple imputation and fitted a Cox regression on the MI-data. After that, I would like to predict the estimate median survival and estimated 5-year survival using the prediction from the Cox regression. I figure I would do that using the post estimation commands. However, they do not work when I use the imputed data.
Just to figure out the code, I tried doing it on the non-imputed data, but when doing that, I get this graph.
Something is obviously wrong, but I can't really figure out what. Any ideas? When I do the Kaplan Meier on the same data I get ordinary looking failure plots.
Grateful for any advice!
I have a dataset with approx. 30 000 patients. My outcome is metastasized colon cancer (mcc) and I have time to event from diagnosis of cancer to diagnosis of metastasized cancer (mcc_time). I have performed a Kaplan-Meier plot on non-imputed data, using sts list and stci to estimate median survival and estimated 5-year survival.
I have thereafter performed a multiple imputation and fitted a Cox regression on the MI-data. After that, I would like to predict the estimate median survival and estimated 5-year survival using the prediction from the Cox regression. I figure I would do that using the post estimation commands. However, they do not work when I use the imputed data.
Just to figure out the code, I tried doing it on the non-imputed data, but when doing that, I get this graph.
Something is obviously wrong, but I can't really figure out what. Any ideas? When I do the Kaplan Meier on the same data I get ordinary looking failure plots.
Code:
** Imputed data ** mi stset mcc_time, f(mcc) id(id_code) scale(365.25) mi estimate: stcox gender##age_categories c.(var1-var9) i.(var10-var20) stcurve, failure // does not work, not with the "mi estimate: stcurve, failure" either ** Non imputed data* stset mcc_time, f(mcc) id(id_code) scale(365.25) stcox gender##age_categories c.(var1-var9) i.(var10-var20) stcurve, failure
Grateful for any advice!
