I am measuring the effect of a time event (Health Policy) on an outcome (Recent Use of Health Services). I have a baseline measure (use within 6 months at baseline), a pre-time event outcome measure (use within 6 months at pre-time event follow-up measure) and a post-time event outcome measure (use within 6 months of post-time even follow-up measure). The time-event is binary and is the main independent variable being tested as a main exposure variable. The outcome variable can either be the change from pre-post (ordinally: increase, decrease, no change) or as the post-time event outcome (binary) given I adjust for the pre-time event outcome (binary). Which method and framing to this question is optimal?