Hello, my last post didnt have any luck so i'm reframing and posting again with more useful info..
Data example below:
i stset it using:
I read online that to account for multiple failures i should not use the id(id) option.
my questions: 1) is my stsetting method right?
2) because i didnt use id(id) while stseting, how do i account for the correlation by id? could cluster(id) as an option suffice when i run stcox?
3) for time varying cov like current smoking/drinking etc, do i need to specify tvc? or something else? [ i have not included the time varying cov in the data example]
Data example below:
- we have ID, HPV straintype and visit number..followed by the HPV result at that visit and a clearance outcome..(which is 1 at first occurence of a negative result after a positive result). clear1enter is the enter variable and clear1futime is the time to clear1 or followup which i am using as the exit variable..
- clearance is a strain specific outcome..so can have multiple outcomes of clearance per id.
- The clearance var are only populated in certain windows as window is from the eligibility for clearance (first positive result) until clearance.
Code:
* Example generated by -dataex-. For more info, type help dataex clear input double(id strain_type visit) float(Result clear1 clear1enter clear1futime) 10042 66 52 0 . . . 10042 66 53 0 . . . 10042 66 54 0 . . . 10042 66 55 1 0 . 0 10042 66 56 1 0 0 189 10042 66 57 1 0 189 371 10042 66 58 0 1 371 553 10042 66 59 1 . . . 10042 66 60 1 . . . 10042 66 61 1 . . . 10057 16 52 1 0 . 0 10057 16 53 1 0 0 161 10057 16 54 1 0 161 343 10057 16 55 1 0 343 532 10057 16 56 1 0 532 714 10057 16 57 1 0 714 896 10057 16 58 1 0 896 1078 10057 16 59 1 0 1078 1260 10057 16 60 1 0 1260 1442 10057 16 61 1 0 1442 1638 10057 16 102 1 0 1638 2912 10057 16 103 1 0 2912 3276 10057 52 52 0 . . . 10057 52 53 1 0 . 0 10057 52 54 0 1 0 182 10057 52 55 0 . . . 10057 52 56 0 . . . 10057 52 57 0 . . . 10057 52 58 0 . . . 10057 52 59 0 . . . 10057 52 60 0 . . . 10057 52 61 0 . . . 10057 52 102 0 . . . 10057 52 103 0 . . . 10069 16 52 0 . . . 10069 16 53 0 . . . 10069 16 54 0 . . . 10069 16 55 1 0 . 0 10069 16 56 0 1 0 182 10069 16 57 1 . . . 10069 16 58 1 . . . 10069 16 59 0 . . . 10069 16 60 0 . . . 10069 16 61 1 . . . 10069 59 52 1 0 . 0 10069 59 53 1 0 0 168 10069 59 54 1 0 168 350 10069 59 55 1 0 350 532 10069 59 56 1 0 532 714 10069 59 57 1 0 714 896 10069 59 58 1 0 896 1099 10069 59 59 1 0 1099 1260 10069 59 60 1 0 1260 1442 10069 59 61 1 0 1442 1624 10081 51 52 0 . . . 10081 51 53 1 0 . 0 10081 51 54 0 1 0 189 10081 51 55 0 . . . 10081 51 56 1 . . . 10081 51 57 0 . . . 10081 51 58 0 . . . 10081 51 59 1 . . . 10081 51 60 0 . . . 10081 51 61 0 . . . 10102 18 52 0 . . . 10102 18 53 0 . . . 10102 18 54 0 . . . 10102 18 55 0 . . . 10102 18 56 1 0 . 0 10102 18 57 1 0 0 182 10102 18 58 0 1 182 388 10102 18 59 0 . . . 10102 18 60 0 . . . 10102 18 61 0 . . . 10120 45 54 1 0 . 0 10120 45 55 1 0 0 161 10120 45 56 1 0 161 343 10120 45 57 1 0 343 532 10120 45 58 1 0 532 731 10120 45 59 1 0 731 899 10120 45 60 1 0 899 1032 10120 45 61 1 0 1032 1242 10120 45 102 0 1 1242 2481 10120 45 103 0 . . . 10136 58 52 1 0 . 0 10136 58 53 1 0 0 165 10136 58 54 1 0 165 333 10136 58 55 1 0 333 508 10136 58 56 0 1 508 690 10136 58 57 0 . . . 10136 58 58 1 . . . 10136 58 59 0 . . . 10136 58 60 0 . . . 10136 58 61 1 . . . 10339 56 54 0 . . . 10339 56 55 0 . . . 10339 56 57 0 . . . 10339 56 58 0 . . . 10339 56 59 1 0 . 0 10339 56 60 1 0 0 182 end
Code:
stset clear1futime, failure(clear1==1) scale(365.25) enter(clear1enter) exit(clear1futime)
my questions: 1) is my stsetting method right?
2) because i didnt use id(id) while stseting, how do i account for the correlation by id? could cluster(id) as an option suffice when i run stcox?
3) for time varying cov like current smoking/drinking etc, do i need to specify tvc? or something else? [ i have not included the time varying cov in the data example]
Code:
stcox gender, cluster(id)
