Create time variables in long-format data as set-up time-varying covariate analysis

[spencer keene]

Hello,

I have the following data in long-format (i.e. panel data).

clear
input int idno double duration
4 .09308692676249145
4 .17522245037645448
4 .28473648186173856
4 .38329911019849416
4 .48186173853524983
4  .5913757700205339
4  .8569472963723477
5 .12594113620807665
5   .243668720054757
5 .32854209445585214
5  .4353182751540041
5   .731006160164271
end

and I would like to create _t0 and _t variables for time-varying cox regression analysis. Similar to the below:

clear
input int idno double(duration _t0 _t)
4 .09308692676249145                  0 .09308692676249145
4 .17522245037645448 .09308692676249145 .17522245037645448
4 .28473648186173856 .17522245037645448 .28473648186173856
4 .38329911019849416 .28473648186173856 .38329911019849416
4 .48186173853524983 .38329911019849416 .48186173853524983
4  .5913757700205339 .48186173853524983  .5913757700205339
4  .8569472963723477  .5913757700205339  .8569472963723477
5 .12594113620807665                  0 .12594113620807665
5   .243668720054757 .12594113620807665   .243668720054757
5 .32854209445585214   .243668720054757 .32854209445585214
5  .4353182751540041 .32854209445585214  .4353182751540041
5   .731006160164271  .4353182751540041   .731006160164271
end

Any help would be much appreciated.

[Øyvind Snilsberg]

bysort idno (duration): gen double t0 = cond(_n>1, duration[_n-1], 0)
gen double t = duration

[spencer keene]

Thank you, Øyvind !