People have been talking for decades about decentralizing much of healthcare away from massive Big Bricks on the Hill — for lots of good, sound reasons revolving around the efficacy of convenient primary care and home care, and the improvement in communications. It hasn’t happened much because these are all “should” reasons, not “have you by the throat” reasons.
The HYBTT reason is arriving: In the approaching liquefaction of healthcare, provider organizations (today’s institutions or others that will arise to compete with them) increasingly will be identifying particular populations in their service area whose primary care (or chronic conditions) they can manage under risk contracts. They will be doing this so that they can give these people earlier, smarter, resource rich care, and profit from driving the cost of care down and the effectiveness up. To serve these populations, providers will locate wherever is most convenient to that population, because especially in chronic and primary care, convenience is clinical. It makes a big difference if the people you are serving can get care downstairs, down the hall, or down the block, instead of across town, down the freeway, at the end of three bus transfers. So we will see a lot of “forward stationed” clinics in workplaces, union halls, schools, convalescent homes, neighborhoods, wherever the risk-contracted population hangs out.
This drives the second strategic observation: Geographic datamining. Providers are not doing this yet very much, but when they come to be at risk for the health costs of populations they will be all over it like your favorite metaphor. It is now becoming trivially easy to mine your records and discover where your patients come from — not just to the zipcode level, or even the block, but to the level of the individual address (HIPAA compliant, as dots on a map).
In a 




