st commands - nested case-control study help

[shravya kakulamarri]

I am having some trouble properly setting up my data for a nested case-control study using the st commands. I am trying to determine possible predictors of intramedullary nail breakage. These will serve as my cases. They will be matched to controls on the following parameters: gender, bone affected, and age. My dataset also has surgery date and follow-up date, and I have used these two dates to calculate days between surgery to follow-up, which I will also be matching on (days_btwn_surgery_fu). I will be comparing and running my analyses on the remaining variables in my dataset. Additionally, my data is set up in a way that it has multiple rows per patient if they had followed up at the hospital numerous times.


These are the commands I have been using:

. stset days_btwn_surgery_fu, failure(broken_nails_cases) id (CaseID)

. sttocc CaseID, match (Gender bone age days_btwn_surgery_fu)

Ideally, I’d be setting a seed so I can replicate the match.

My question is whether or not I have set up st command properly. I have read up on snapspan and stjoin, which are related commands that can clean up my data, but I am unsure whether or not I need to use them in this context.

[Raoul Reulen]

You shouldnt have to match on analysis time as sttocc aready does this. In fact, it ensures that the analysis time in the control is the same or greater than that of the corresponding case. I do not know whether you have stset the data correctly without more info.