interpret hazard ratios from a dummy variable correlated with time under risk

Question

My design is as follows:

• People become at risk upon entering University, and are censored when leaving university, or on event (childbirth) or by dec 2012.

• I have identified continuous university spells if various length (12-60 months).

• I have a dummy, “type”, taking values 0 if studies are taken in uni with characteristics A, anf 1 if uni has characteristics B.

• The length of the university spell differ by “type”: type=1 students tend to take shorter university spells, and here is the question: does this affect the interpretation of the results?

• "type" show a significant HR of 4.5

What does the HR mean here?

That type=1 students at any given time, compared to type=0 students, are quite alot more likely to become parents while in university?

Or does it merely reflect their shorter University spell, so that IF type=1 students become parents during University studies, the risk is higher as it is divided by a smaller exposure time?

Answer 1

That is the same: For as you compute a hazard rate, you compute a risk related to time. If you wanted to know the risk of becoming parent irrespective of the study time, you could compare proportions of students who became parents between types.