Estimates of probablility of a hospitalization within short periods (10,7, 3 days) after a clinical event I have 6 months of observation during which "clinical events" can happen.  The endpoint during this period is a hospitalization (or censoring).  I want to develop estimates of the the likelihood of a hospitalization happening within 10, 7, and 3 days of the "clinical event" using survival analysis.  Is this possible?  Is it the wrong tool for this question?
Persons may have multiple "clinical events" but are followed only to their first hospitalization within the observation period. I have exact dates of clinical events and hospitalizations, and all subjects have the same 6-month window.
What i am trying to say is something like persons with "clinical event x" are likely to a hospitalization within 10 days, 7 days or 3 days.  
Any help is greatly appreciated and thanks in advance!
 A: This should be a comment. Hopefully it will spur on more discussion:
(1) It seems from the question that you have already dichotomized the outcome as 3 day hospitalization (yes/no), 7 day (yes/no) etc. and are thus less interested in time to event than the probability of these three outcomes. If that is the case three logistic regression models with these three binary outcomes makes sense, if it is not the case then the cox model makes sense. 
In hospital admission/readmission studies both logistic models and cox models have been used depending on how the results are to be presented. 
(2) You have an issue of clustering in that individual patients are may be at risk multiple times in the sample. Not sure how you wanted to address that. I don't know what the best way of doing this is. For logistic models one could choose the first clinical event for each subject, a random clinical event for each subject, or keep the multiple datapoints and adjust for clustering in the model (?Huber-White sandwich estimates?). 
