most US hospitals are tiny outfits in small cities and rural areas.
I believe that those tiny little hospitals are those between 5,000 and 10,000 (or beyond) - and note that I didn't use them in my calc. For instance the US World and News Report has a search engine for hospitals that has more than 5,000 without including the tiny little community hospitals.
Also note that if IMS wants to have an accurate estimate of sales that they need to have a polling population that is representative of larger population - e.g. I'd bet that pretty much across the board the smaller hospitals use substantially different procedures and different drugs than the bigger hospitals. Maybe it isn't completely equally spread - but still I'd expect at least 1/4 of their polling to be done in community hospitals (the last 5,000). So again my stat is conservative given the data I have.
Just to reiterate - it isn't that I am, per se, claiming that this happened to Recothrom. But I am claiming that the IMS data for February is extremely low fidelity. Something that bears watching, but not much more - and will be massively trumped if Zymo puts out accounts data in the format that they have previously talked about their marketing campaign - e.g. accounts in won in the top 100, accounts won in the next 200, ... . Once they have hit enough accounts then, perhaps, the IMS data becomes more meaningful (still subject to errors due to lag - which errors can be big in a ramp up). But right now I would suggest it is only a marker of something that should be watched.
Market Data 
Markets 
