Flaw in CMS Logic Causes Cost

When the NPI Final Rule (and all of its after-final rules) created the National Plan and Provider Enumeration System, there were many unknowns: Which datapoints would be released for the industry to use? loomed large. But another issue has come home to roost.

Organizations (Type 2 providers under the rule) were permitted to have as many NPI numbers as they liked, and they could structure their assignment of NPIs any which way. For instance, one hospital might get separate NPI numbers for each of its business units, while another got and NPI for each of its physical locations, another for each of the cluster of corporations, while some clever hospitals got an NPI for each reimbursement channel. And then of course, some hospitals got just one.

No problem with that -- the various business optimization strategies are interesting to observe, and surely make sense in their various contexts. The problem is that there is no primary NPI number per hospital or health system. That is to say, there is no way to know from the NPPES records which if any of the NPI records is a parent, and which is a child. Oh, of course, an army of human analysts can pore over the records and find 37 hospital NPI records each identifying, say, Mayonaise Health System as its parent. But a computer finds that task a bit difficult, since it will find many variations in the records, e.g.,

• Mayonaise Hospital
• Mayonaise Health System
• Mayo Hospital
• Mayo Hospitals
• Mayonaise Hospitals
• Miracle Whip Health
• and on an on

Thus, it becomes essentially impossible to say how many hospitals there are, even though we are looking at the complete set of federal records on hospitals. Had there been a primary or master NPI required for each General acute care hospital -- regardless of how many business units and other NPIs are involved, it would be possible to perform much more significant research on hospital service areas, densities, availability of care, duplication of services, and much more. (We've just started putting state-by-state physician and hospital counts on our home page at CarePrecise.com, but for now, we are able only to show the total of all hospital records -- 29,946 at present -- which is far more than the roughly 5,000 actual hospitals to whom all those records belong.)

The coyness built into the NPPES was more or less deliberate. American hospitals are a contentious lot, engaging in constant competition, and they did not want any more known about them than absolutely necessary. Coy data costs everyone money, and adds opacity to the healthcare system. Still, with the HospitalCompare project and our subsequent mining of all of these data sources, much can be learned, and the reach of each hospital organization can ultimately be published. Stay tuned.