I found this interesting:
From the Abstract:
Outcomes are related to different types of treatment intensity, however: patients assigned to hospitals with high levels of inpatient spending are more likely to survive to one year, while those assigned to hospitals with high levels of outpatient spending are less likely to do so. This adverse effect of outpatient spending is predominately driven by spending at skilled nursing facilities (SNF) following hospitalization. These results offer a new type of quality measure for hospitals based on utilization of SNFs. We find that patients quasi-randomized to hospitals with high rates of SNF discharge have poorer outcomes, as well as higher downstream spending once conditioning on initial hospital spending.
I have not read the entire paper yet. My intuitions come from having worked in both acute hospital and SNF in a wide variety of geographic areas, and having participated in d/c planning. I suspect there are certain patient demographic factors that work favorably for the better performing hospitals that d/c direct to home, vs those that d/c to SNF in ways that are very hard to control from an aggregate, statistical level.
I am very interested in how the authors controlled for medical condition complexity and social/family factors. These are critical components on how D/C plans are determined in the real world.