Survival Analysis - To Exclude or Not???

[Robert Bohler]

I am doing a survival analysis in Stata looking at one-year mortality outcomes after participation in a jail-based program. I have 350 unique individuals that attempted the program a total of 550 times and I am using a propensity-matched control group. Initially, I only examined the event (death within one year after release) after the last program attempt. However, I received feedback that I should be including all program attempts. This does not make sense to me as these program attempts would be clustered among certain individuals and, by definition, since they survived to attempt the program again, this would inflate my hazards ratio. Am I making the right decision on the study design to only examine the last program attempt? Or should I be using some form of survival analysis that accounts for multiple program attempts?

[Rich Goldstein]

your situation is not completely clear to me but it does seem that you are throwing away relevant information which will reduce your power and may introduce bias; note that whether you are using a semi-parametric model (stcox) or a parametric one (streg), you can use cluster adjusted standard errors (part of the "vce()" option); in some situations that I can imagine, a more formal use of the multiple occurrences might also be called for