We tend to have very complete data sets at the moment, so we are blessed, in a sense, in this country with very complete data sets from hospitals. In the discharge abstract database, for example, we have information on virtually every hospital encounter; it's a very high percentage. We have very high percentages for some of our other databases as well.
I think it's clear one can have less complete, less comprehensive databases that can be used to pinpoint issues. From a data and statistical point of view, I think you usually start these things in stepwise motions. You want to start collecting data in a certain sector. That would give you information you could then use to improve safety and reach conclusions.
If the question is if we started in hospitals, if we started in a certain place, would that be helpful, I think the answer would be yes. Certainly we have data collection systems for hospitals, for example. The technical means of doing that, certainly, is--