Transparent and machine-readable prices (and surprise medical billing)

Yesterday I posted Implementing the Hospital Price Transparency Executive Order; Making Health Savings Accounts Better. President Trump signed the Executive Order on Improving Price and Quality Transparency in American Healthcare to Put Patients First. CMS is currently requesting comments on its plan for implementing this order in this proposed rule. (Comment period ends 9/27/19)

Patients and Pricing

Section 3(a) “Informing Patients About Actual Prices” is where the executive order starts to get meaty. This is where price transparency requirements are articulated. 

This is the second time that the Trump administration has required hospitals to publish data about their prices. Previously, it was required that hospitals publish their chargemaster data online. Here is the FAQ from CMS that covers the details of that requirement

The first thing to note is how much broader this new mandate is. The specific language that matters is: “including charges and information based on negotiated rates”. For those that need some help translating industry parlance, that means that the rates of prices negotiated between health insurance agencies and hospitals should be made public. This is a huge victory for healthcare transparency and anyone in DC who was involved in making this happen has every right to be very proud of this. The patient community has needed this for so long, as hospitals have abused the fact that patients could not see how much more they were paying than insurance companies for the same procedures. Bravo! 

Here is the section in full, so that we can really get into the complex implications of the language:

Sec. 3Informing Patients About Actual Prices.  (a)  Within 60 days of the date of this order, the Secretary of Health and Human Services shall propose a regulation, consistent with applicable law, to require hospitals to publicly post standard charge information, including charges and information based on negotiated rates and for common or shoppable items and services, in an easy-to-understand, consumer-friendly, and machine-readable format using consensus-based data standards that will meaningfully inform patients’ decision making and allow patients to compare prices across hospitals.  The regulation should require the posting of standard charge information for services, supplies, or fees billed by the hospital or provided by employees of the hospital. The regulation should also require hospitals to regularly update the posted information and establish a monitoring mechanism for the Secretary to ensure compliance with the posting requirement, as needed.

Let’s get into some of the phrases: 

Post standard charge information

This is difficult because every hospital is free to implement their databases of “charges” in any way they see fit. This occurs even though industry standards exist, and even though hospitals frequently have EHR and billing systems in common. At CareSet, we spend a lot of time studying the different ways hospitals decide to bill Medicare for essentially the same work. There are huge differences between hospitals. So there will be some difficulty in nailing down what “standard charge information” means, and hospitals will use this grey area to try and pass off half-measures of data release as legitimate. It is critical that actual regulation narrow this down and make this concrete. 

Thinking X-Mas shopping by Hernán Piñera

photo by Hernan Pinera

For common or shoppable items or services 

What is “shoppable”? There are two distinct ways to look at this. If it is what hospitals encourage shopping for, it is essentially an empty list. No hospitals make it easy to shop today. If, on the other hand, you consider it from the “what has a patient tried to shop for” perspective, well then there is no healthcare service that I can think of that has not been shopped for by some patient. This includes emergency services, as people who are expecting emergencies frequently choose where to live based on the quality of EMS, paramedic and ER services. 

The reality is that some middle ground definition of “shoppable” needs to be developed. I suggest that a useful rubric might be “If patients typically wait more than two weeks, or could wait two weeks if needed to schedule the procedure, then it should be considered “shoppable”. Another guideline could be “if patients regularly have the procedure done in any city outside of their home city”. 

Both of these rules of thumb give a mechanism to ensure that patients who are already shopping for the best healthcare options are able to get the prices they need. If a patient is waiting for the availability of healthcare professionals they prefer, or if they are switching cities to get better care, then it is obvious they are already shopping and need price transparency around those healthcare services. 

Having said that, I am not convinced it is realistic to say that some services are “un-shoppable”, though it may be reasonable to say some are “not typically shopped”. However, in many cases, “un-shoppable” just means “there was a patient who tried to price compare this and was stone-walled at every turn”. 

Further, there are likely scenarios where services can be compared in a comprehensive way once pricing data becomes available. Let’s say, when considering all heart-related procedures, that a San Antonio hospital has a lower cost and higher quality than a Houston hospital. Some smart person could create an app that tells patients living between those cities, “if you have chest pains, drive to San Antonio”. So even though we might consider “emergency heart surgery” to be “un-shoppable” in standard models, there is a way to make it “shoppable”. 

Perhaps the right way to think about it might be “Everything is shoppable, if there are prices available. Therefore, there should be prices available for all the things”. 

More importantly, providing hospitals the opportunity to create areas of pricing secrets because “a patient cannot shop for that” will incentivize two things:

A. it will make providers put as many services out there as possible and;
B. it will result in hospitals trying to route patients into those “un-shoppable” buckets.

Easy-to-understand, consumer-friendly, and machine-readable format

Hospitals are never going to intentionally release something “easy-to-understand” which can be used to negotiate against them. They will, however, release something and label it “easy-to-understand”. The same is true of “consumer-friendly”. While these are good goals that can be achieved under this program, it is not the hospitals who will be producing these types of high-quality resources for consumers to facilitate shopping.

Instead, people like me and companies like CareSet might try to take the machine-readable data and turn it into something that is user-friendly and easy-to-understand. This will be followed by companies like eBay and Amazon (massive organizations who understand how to build optimal online shopping experiences), who will eventually facilitate not only the shopping but the purchase process. 

But that reality can only happen if the machine-readable files are released in a reliable manner, considering the constraints for building these types of shopping experiences. Specifically;

  • The files are updated when prices change. Within 24 hours is reasonable. 
  • The files use an open data format, either CSV or JSON, and the released data files are tested for successful compliance with those standards. Lots of data releases attempt to use these standards and then when you try and import the data, the process breaks because the standards were not actually honored. HHS should create a test suite to ensure that the CSV/JSON files open successfully using the top 10 costless programming languages, and ensure that hospitals are informed when they are failing. 
  • The data structures should specify which organization the data model is describing. This is critical because there is already a huge amount of, “Well our organization’s price for this is X, but the person who treated you here at this organization does not actually work for this organization and their price is 5X”. The first step is describing in detail where the prices apply. Do the prices apply to every hospital in the same city? The same system? Or just one hospital? The data is useless if it is ambiguous because every ambiguity will be used against the patient. 
  • Similarly, every hospital should require that any partner that charges its patients also have machine-readable price data available and should be required to incorporate those prices by reference. It should be possible, by looking at the machine-readable resource from a single hospital, and then downloading all of the pricing resources it refers to, to retroactively look up the prices for any services the hospital offers. This should not be limited to “shoppable” services, but cover any services offered. Anything short of this “retroactively verifiable price lookup” will be used against the patient by providing hidden corners of “not-published” prices and then routing patients into those corners. This is the only realistic way for hospitals to account for surprise medical billing (which is actually covered in Section 7 of the executive order).
  • HHS should seriously consider hosting an index document of the CSV/JSON price data endpoints. This was the approach taken with health plan data for the marketplaces and it worked pretty well. A person could look in one central place for the location of all the data offered by the organization and could find the deep links to the machine-readable data files all-at-once. HHS will have to develop a list like this to test for compliance (because of “…establish a monitoring mechanism for the Secretary to ensure compliance”), which is 90% of the work involved in centrally hosting. Given that such a resource would be FOIA-available once it existed, HHS should not ride the fence regarding this issue. 
Using consensus-based data standards

I cannot help but feel somewhat vindicated in our efforts to create such a standard already at As far as I know, this is the only hospital pricing data standard. We use the GitHub pull request and ticketing system to manage our consensus-building process. 

All of the “data standard issues” mentioned above that can be constrained by a data standard are already well-constrained in the data standard that we have released. With the possible exception of reflecting multiple pricing structures – and we are working on it!

These requirements are very exciting developments, and we enthusiastically support this new direction from the administration and from HHS.

Fred Trotter

Fred shapes our software development and data gathering strategies, which doesn't stop him from getting elbow-deep in the code on a regular basis. He is co-author of the first Health IT O’Reilly book Hacking Healthcare, and co-creator of the DIRECT protocol mandated in Meaningful Use. Fred’s technical commentary and data journalism work has been featured in several online and print journals including Wired, Forbes, U.S. News, NPR, Government Health IT, and Modern Healthcare.