Show stci output in hours

[Dorothea Ekoka Mbassi]

Hello,

I am performing some survival analyses. When performing stci, the output is given in milliseconds, thus rendering rather large (unreadable) values:

. stci, by(rand)

         Failure _d: censored
   Analysis time _t: (exitdatetime-origin)
             Origin: time firstdosedatetime
  Enter on or after: time firstdosedatetime

             | Number of
rand         |  subjects         50%      Std. err.    [95% conf. interval]
-------------+-------------------------------------------------------------
    Drug1 |        16    1.15e+08      168183.1      5.0e+07    2.7e+08
     Drug2 |        17    1.44e+08      4.80e+07      1.2e+08    2.0e+08
-------------+-------------------------------------------------------------
       Total |        33    1.19e+08      3.63e+07      1.2e+08    2.0e+08

.

How can I have the input given in hours? I could not find out how to apply the hours() command in this context.

Thank You so much,
Dorothea

[Andrew Musau]

I think you need to rescale the time variables.

stci, by(rand)

*RESCALED
preserve
foreach var of varlist t t0 _t _t0{
    replace `var'=`var'/(1000*60*60)
}
stci, by(rand)
restore

[Dorothea Ekoka Mbassi]

Thank You Andrew,
this looks useful but does not work yet, it says invalid syntax - Can You spot the error?
What varlist should it be, that is, from what variables does stci calculate its output?
The basis for my analysis was

stset exitdatetime, failure(censored) enter(firstdosedatetime) origin(firstdosedatetime)

Thank You

[Andrew Musau]

Your time variable is named "exitdatetime" and Stata creates the variables "_t" and "_t0". So without changing the stsettings,

stci, by(rand)

should give you your current results and following this,

preserve
foreach var of varlist exitdatetime _t _t0{
    replace `var'= `var'/(1000*60*60)
}
stci, by(rand)
restore

should give you the rescaled results. Otherwise provide a reproducible example.

[Andrew Musau]

There is a direct way that avoids the workaround in #2 - see highlighted.

stset exitdatetime, failure(censored) enter(firstdosedatetime) origin(firstdosedatetime) scale(`=1000*60*60')

The option is explained in

help stset

[Dorothea Ekoka Mbassi]

Both work wonderfully, You've been so helpful once again!