A very good reporter asked a very good question. I had told him that one of the major problems with the Dartmouth group’s studies of regional variation was that their metric of health care utilization was the average level of Medicare spending in each region. The problem is that quality within regions doesn’t depend on Medicare spending alone – it depends on total revenues from all sources, and total spending doesn’t correlate with Medicare spending. This hangs as a cloud over their studies of outcomes among regions (see “30% Solution – A Treacherous Prescription”).
The reported said, “but what about practice patterns? Dartmouth researchers found correlations between the number of days in hospitals for both Medicare beneficiaries and privately insured adults. Doesn’t that show that privately insured adults are getting treated similar to Medicare beneficiaries?”
Yes, but only within a particular hospital. Not within a region. As Dartmouth researchers showed, care is essentially the same for all patients (Medicare and others) within a hospital. But it is not the same for all patients between hospitals. In some hospitals, Medicare patients are mixed with more private patients and in some with more uninsured patients. All hospitals depend on total revenues when hiring nurses, creating systems, recruiting physicians, etc., and that’s what determines quality. No one payment source (Medicare or other) is representative of total revenues across a region.
This may be easier to understand in terms of race. Blacks and whites receive comparable care within particular hospitals, but the care that blacks receive across a region tends to be inferior because many are cared for in “predominantly black” hospitals, where care tends to be poor.
The lesson is that hospitals and regions are very different units of analysis, and observations in one should not be used to explain findings in the other. Unfortunately, that’s exactly what the Dartmouth group does.