Preloader Icon

The Informonster Podcast

Episode 20: Working Towards Price Transparency in Healthcare

June 30, 2021

On this episode of the Informonster Podcast, Charlie Harp talks about the purpose of the CMS Hospital Price Transparency Initiative, how price transparency would be implemented in our current system, as well as the difficulties we would have to face in order to achieve it.


View Transcript

Follow Us

Have a question or topic idea?

Get our News and Updates

Get notified about new podcast episodes, upcoming events and webinars, and more!


I’m Charlie Harp, and this is the Informonster Podcast. Today on the Informonster Podcast, we’re going to talk about the CMS Hospital Price Transparency Initiative from a practical informatics perspective.

Now, if you’re unfamiliar with what I’m talking about, the Hospital Price Transparency Initiative, effective January 1st, 2021, is designed to help Americans know the cost of a hospital item or service before receiving it. To support this, each hospital operating in the United States will be required to provide clear, accessible pricing information online about the items and services they provide, uh, in two ways, essentially: The first way is they need to provide a comprehensive, machine-readable data file, like a spreadsheet, with all the items and services listed. In this list, they’re required to share the following information for each item: The gross charges or the non-discounted rate as reflected in the hospital Chargemaster, the discounted cash price for the item, that’s what the hospital would charge individuals who pay cash or cash equivalent, payer-specific negotiated charges for the item, which is a rate that the hospital has negotiated with a specific third party, um, like an insurance company, for that item or service, a de-identified minimum negotiated rate for the item, which is the lowest price they’ve negotiated with any third party payer without identifying anybody, and the de-identified maximum negotiated rate, which is the highest price they’ve negotiated with any payer, um, and without identifying the payer.

The second thing they have to provide is something called “shoppable services,” where they have to provide at least 300 shoppable services that a consumer can schedule in advance in a consumer friendly format, whatever that means. Now, the idea behind price transparency is that by providing this information, it’ll be easier for consumers to shop and compare prices across hospitals, and estimate the cost of care before going to the hospital or getting a service.

Now, this is all tied up in the general idea that competition will naturally drive down the price of a product or service. And this principle has not always worked in healthcare in part because the consumer or patient doesn’t always play an active role in choosing a product or service. They also aren’t always the primary payer for that product of service, and the price and costs of those products and services are often buried underneath the complexity of healthcare. And that combined with the negotiated rates, depending upon the payer and everything else, can make it quite challenging, actually, to determine something actually costs. And that’s been the case for a long time.

Now I’m not going to get into the controversies around pricing, and, you know, providers, and insurance companies, and all the other stuff that’s going on. What I’m really going to talk about are the pragmatic challenges relating to price transparency as things stand today and how that could impact our ability to achieve the objective that I think the authors of the price transparency legislature intended because that’s kinda what we talk about on the Informonster Podcast. That kind of stuff. While the idea of price transparency on the surface seems pretty straightforward, there are a couple of issues that complicate achieving this objective, um, and make it non-trivial. The first issue is a pretty basic idea that, in order to accomplish the objectives of price transparency, the consumer or agent has to be able to compare the price of an item or service from one hospital to that item or service at another hospital. Now the big scary word in that sentence is the word “compare.” This is because, in order to compare the prices of two items, I have to determine whether or not they’re actually the same thing. And that brings us to the second issue, which is semantics.

Now, as somebody who works in the business of comparing things, uh, when it comes to normalization or interoperability, I like to think that I can speak intelligently about what it takes to facilitate comparative analysis. Now, in order to talk about the challenge, let’s start by talking about how we compare things in healthcare for things like semantic interoperability and normalization. How do we do that today? Well, with clinical normalization, which is where I spend a lot of my time, we often are comparing things like a lab result term coming from a source to a lab result term in a target terminology, like LOINC. We also do this kind of thing with medications in RxNorm, diagnoses and SNOMED, procedures in CPT, you name it. Now to facilitate this type of comparison, when you’re doing mapping or normalization, we examine the textual representation of the new term on the left and semantically break it into its defining characteristics. So, we break it down into its parts, and we kind of normalize the parts. We also do this with all the terms in the target terminology on the right side. Now from this point, comparative analysis is just a matter of comparing the characteristics, the parts, of the new term on the left with the characteristic, or parts, of the potential matches on the right, and the one that has the highest degree of overlap, the most similar, is the best fit. Pretty straightforward. Now this is possible because most of the clinical terms we use in healthcare are semantically rich, and when I say semantically rich, I mean that the words we use to represent the term typically define the term semantically. Let’s say I receive a medication term, “Warfarin, 10 milligram oral tablet.” Now, since I know it’s a medication, I can semantically break it down into its parts and identify that the ingredient is warfarin, the intended route is oral, the dose form is tablet, and the strength is 10 milligrams. The term itself, this representation, is semantically rich. It told me what it was, and it’s rich enough for me to look at RxNorm, for example, and find a 10 milligram warfarin tablet. (Which I don’t think there is one, so forgive me for that.) But it’ll allow me to find an equivalent in the target because their semantic parts are the same. And once I do that, I can map it, and I can feel good about the fact that I mapped it because they were semantically equivalent.

Now, unfortunately, when it comes to the Chargemaster files in healthcare, the terms associated with each price item are usually not semantically rich. Um, they are different kind of animal entirely. You might be wondering, “What happens when I receive a term that’s not semantically rich? Why would someone create such a thing, and what happens when you try to use it?” Now, in order to delve into this concept, I’m going to provide two examples and I will warn you one of them is a tiny bit nerdy. If you’re a fan of “Star Trek: the Next Generation,” you may have at some point in your journey watched the episode called “Darmok.” Now this is the second episode in the fifth season, and in this episode, the USS Enterprise is on a mission to attempt to establish communication between the Federation and an alien race called the Tamarians, and this is after several attempts over the last century had failed. The two parties meet and try to communicate, but, just like with previous attempts, neither party can comprehend the other. And at some point the Tamarian Captain, who’s frustrated, essentially kidnaps captain Picard, and takes him to the surface of a nearby planet to try to work things out. Now, I’m not going to describe the entire episode. It’s only a 15 minute podcast. Um, you’ll have to go watch it for yourself. But throughout the episode, the two captains continue to try and fail to communicate with one another, and get very frustrated. And one of the phrases the Tamarion captain keeps repeating is, like, “Darmok and Jalad at Tanagra.”

Now, despite the fact that the universal translator, (which would be pretty cool if we had one of those, outside of Google, I suppose,) that the universal translator is working, Picard can’t make sense of anything that he’s saying. The words are coming out, but it doesn’t make any sense. “Darmok and Jalad at Tanagra” means nothing. Now, spoiler alert, as the episode transpires, we find out that the reason the Federation can’t understand the Tamarians is because Tamarians don’t communicate concepts semantically. They communicate information through allegory and metaphors based on their shared history. So the words they say are really just placeholders or pointers that reference something that the other party is expected to know.

Now, you’ve probably done this and referred to it as an “inside joke.” You know what I mean: Where you turn to your friend and say, “Hey, this is just like that time at Applebee’s when we were discussing Indiana Jones.” Now, when you say that they knew exactly what you were talking about, but anyone that did not have the shared context would be clueless. And as it turns out in the episode, “Darmok and Jalad at Tinagra” was the Tamarions’ way of saying they would form a bond of kinship through shared hardship, like the historical figures in Tamarian history, Darmok and Jalad, at Tanagra. Now Picard eventually responds with a similar allegorical response from earth history, and they both smile, hold hands, and walk off into the sunset together. Now, even though the idea of an entire race that communicates in this manner is farfetched, and that Picard’s solution of responding using what would be equally confusing nonsense, it’s a fun episode and it provided me with a very nerdy example of semantically sparsing contextually-drived information. So I had to go for it. I’m sure you understand. You’re welcome nerds.

Now for a more accessible example of what I’m talking about, especially if you’re not a science fiction nerd, let’s talk about the concept of a value meal. So let’s say I sent you a McDonald’s and I say, “I’d like you to tell me if the ‘value meal number one’ at McDonald’s is as good a deal compared to the ‘value meal number one’ at Burger King.” Now you might struggle a little bit with determining if those items that you purchased are equivalent, unless you were to go to both McDonald’s and Burger King. This is because the term “value meal number one” doesn’t tell you anything about the meal itself. It’s semantically sparse; contextually derived. The McDonald’s “value meal number one” could be a double quarter pounder with cheese, a large fry, and a large drink. The “value meal number one” at Burger King could be chicken nuggets, apple slices, milk and tiramisu. In this case, the term “value meal number one” is semantically sparse and contextually defined. So regardless of which example you prefer, Star Trek or fast food, the bottom line is these types of terms are challenging to compare without the detailed knowledge of what they represent. This is exactly the problem we have when it comes to price transparency because, like the value meal, the Chargemaster was never intended to be a semantically rich terminology. It’s meant to be a bucket of things that have a price associated with them. Now, many Chargemaster files have been created as hospitals evolved. It’s not like a terminology, where somebody, a subject matter expert, necessarily sat down and is thinking about, “Well, how do I define this thing so that it fits into the universe, and do it in a semantically understandable way?” They’re putting things into a list, um, and they’re assigning a price. So there’s going to be duplicates. There’s going to be roll-ups. There’s going to be all kinds of variability that you typically wouldn’t see in a semantically rich terminology. Furthermore, since the actual things that make up that Chargemaster item come from a dictionary and a financial system somewhere else, usually, not only is the term not semantically rich, it’s typically kind of a mess. And when I say that, I mean that, you know, Chargemaster items aren’t necessarily meant to be shown to a doctor or a patient. Um, they’re meant to be used for wiring things behind the scenes. And so they also typically have limits to how big they can be. So you could be dealing with a 40 character term, and to be able to fit what they want to call it in there, they might truncate it, they might abbreviate it, um, they might have things that only somebody in that facility would understand, like that time at Applebee’s, when we were talking about Indiana Jones.

Now speaking as somebody who’s seen his fair share of Chargemaster files, often with a request to semantically analyze them, I can tell you firsthand that it’s not my idea of a good time. In fact, it reminds me of Darmok and Jalad at Tanagra. So, what this means is that I can produce a file with Chargemaster items and prices, and someone else can produce a file with Chargemaster items and prices, and even if the Chargemaster text is exactly the same on the two line items, I can almost guarantee, unless it’s a “single, medium, latex glove,” that any comparison you might try to draw would have a fair amount of uncertainty between the two.

Now, I’m not saying that price transparency is a bad idea. I’m just saying that I think we need to manage our expectations as to what we can reasonably accomplish by trying to compare Chargemasters from different institutions as if they were semantically rich. I say this because the worst case scenario is that the consumer, seeing two items that look like they’re the same thing, might make a choice, not realizing that the item that was more expensive was the “value meal” that included everything, and the item they thought was the more economical item was actually only the tiramisu. the result would accomplish the opposite of the intended purpose of price transparency.

For price transparency to actually work, we would need to find a way to create a common understanding, like a universal “value meal” taxonomy. If we could establish some kind of common pricing item in healthcare, we would enable those providing care and those participating in care with a common basis to assess value on an item or service level. Now, this idea is non-trivial, and it’s not necessarily something you could use existing standards for, like SNOMED, RxNorm, or LOINC. Um, it also depends on how far you want to go. I mean, does a consumer want to know how much a single glove costs? Probably not. At the end of the day, the objective of price transparency is to allow the consumer, when facing hard healthcare related financial choices to make good decisions. It can also allow providers to determine if their pricing is out of alignment with the rest of the providers in their region. But despite our best intentions, none of this can occur if we can’t semantically compare one item to another. It just won’t work. I think as we continue to collaborate with the notion of value in healthcare, we’ll need to iterate, evolve, and develop models and approaches that help payers, providers, and patients get a better handle on our spending and the return on the investments of those healthcare dollars.

So that’s kinda my, my thoughts on, you know, pragmatically evaluating and doing this comparison of these files. And, you know, we gotta start somewhere, and the idea of transparency is a good one. So I have to imagine that at least we’ve taken a few steps on what should be a fruitful journey. Now, speaking of return on investment, if you’ve listened this long to this podcast, and you see me at the upcoming hymns in August at booth 5654, if you use the phrase “Darmok and Jalad at Tanagra,” you’re going to receive a special swag item, so I hope to see you there. So anyways, thank you all for listening to this edition of the Informonster Podcast. I hope you found it interesting, and maybe even a smidge entertaining. And I look forward to any feedback, thoughts, or ideas you’d like to share. Uh, you can find me at Thanks.