Comment by jcrawfordor
2 years ago
I'm not so confident from my experience in healthcare IT, admittedly shorter than yours. One of the issues seems very telling, the fact that one of the listed rates from the insurer matched the portion paid to the radiologist but not the total. Problems with matching up different line items are really ubiquitous in healthcare systems because, despite the standardization that theoretically exists, there's huge variation in how different accounting and analytics systems represent line items. It seems extremely plausible to me that all the numbers in these MRFs are real numbers, but aren't the numbers they're supposed to be. Probably the reports were generated by people using BI tools that didn't have the expertise in the underlying data to understand what values they should actually be reporting on, so it ends up being a hodgepodge of different dollar amounts that are often not the actual reimbursement amount but instead a subtotal or breakdown line items used for analytics.
Sure, insurance companies ultimately know what they paid, but consider that these MRFs are almost certainly not being prepared by the people responsible for that knowledge. They were probably tasked to one or two data analysts who quickly banged them out in whatever BI/reporting tool they use and did nothing to verify correctness. It's not like they had an accountant audit these if they weren't absolutely required to (they weren't). Most healthcare analytics tools are complete junk drawers of data from numerous systems and getting these MRFs right was probably never a priority for anyone. Just a total "I have no idea if these are right but they sure are numbers" exercise.
I mean, how many times have you seen some Tableau report that's all screwed up because some of the MRNs aren't actually MRNs (even though someone named the model field community_mrn) but file numbers from the scheduling system, and now you've got duplicate patients? BI systems just breed that kind of problem unless you are extremely careful about managing them, and since they're "not systems of record" (these are scare quotes) few people are.
>> Problems with matching up different line items are really ubiquitous in healthcare systems because, despite the standardization that theoretically exists, there's huge variation in how different accounting and analytics systems represent line items.
Maybe the problem IS the line items. You know the labor cost of tracking the fact a person was given Tylenol is way more than the cost of the pill? Just give people the incidentals and stop billing for them. Half your overhead might vanish.
BTW, yes every medication need to go on their chart. But it does not need to go through the entire finance system and to insurance.
It's a good point and ties in to the broader issue of opaqueness of reporting, especially now that we've empowered self-serve report generation. We can democratize data systems... but documenting them is a whole different level of effort.
> Sure, insurance companies ultimately know what they paid
It came to mind reading the above that a more accurate/useful perspective might be "The insurance companies' system knows what they are paid," but those system may comprise multiple software systems, none of which have data in compatible formats.
Ergo, even though the insurance company "knows" operationally (it can generate a number on request), it might be unable to generate a list of all numbers (effectively: every path through the system).
But that's why mandates work in insurance: if CMS pushes hard enough, eventually the insurers will develop the functionality.