Sts graph on Longitudinal data for time varying covariates: Extended KM? or not?

[Sakshi Rajatbhai Tewari]

Hello,
i was wondering...if i have a longitudinal database by visit, with covariates populated at each visit per person..and some variables are modifiable like smoking, drinking, hash use etc. If i were to stset this longitudinal data..and run

sts graph, by(modifiable var)

would stata create kaplan meiers where at each event time, the risk sets that are created account for changing group memberships? Is that something i could use as a better alternative to the "classic kaplan meier"; that only uses a single time point of baseline/when person starts being at risk for event?
Thanks in advance!