The government-mandated healthcare price transparency data is finally here and has immense potential to improve the performance of employer sponsored plans. However, traditional transparency tools are unlikely to be able to harness this potential. Given the complexity of the data and how it is delivered, employers will need new tools to help their employees get the information they need to access more efficient care.
Garner’s data science team has done an assessment of the new machine-readable files across the major insurers. We are excited to say that, while not perfect, the new healthcare transparency data is incredibly rich and useful. Particularly, these files finally make it clear the actual price of healthcare across all services, all providers and all health plans. The results are staggering: our initial calculations show an average of a 800% variation in the cost of common procedures within the same geography. This has been a powerful missing ingredient in past efforts to bend the cost curve of healthcare. However, to unlock the potential of this data, employers cannot rely on antiquated transparency programs. Instead, they must offer new tools to provide employees the information and incentives they need to see efficient providers.
At Garner, we have been planning for the release of this data for years. In particular, the new transparency data allows us to further refine our understanding of costs at the physician level, making our employee incentive accounts even more powerful. We are therefore excited about the healthcare transparency data as a way to not only shed light on the healthcare system’s inefficiencies, but to make this data actionable and create measurable cost savings as a result. Below we give our thoughts on the transparency data in more detail.
The enormous variation in cost among providers
Unlike the hospital transparency data a year ago, this health-plan-focused set of regulations have succeeded in bringing a sufficient quality and quantity of data to bear. For nearly all locations, providers, and procedures, we now know how much a healthcare service will cost . When looking at the data one is struck by how different the prices of common services are within the same geography. As one tiny example, below we show the cost of a colonoscopy across different locations in New York City. As you can see, the cost varies from as low as $300 to nearly $10,000, a 30x difference!
This type of variation in price is not a random example. As noted above, our initial calculations show that the average procedure has a 800% variation in price among providers in the same city. Put differently, if patients chose the most efficient provider for their care it would mean lower overall healthcare costs of 62%, equaling $7,542 per employee per year.
The data set is enormous and highly complex
The sheer size of the healthcare transparency data is staggering: when the data from all insurers is aggregated together we estimate the total size to be over 1,000 terabytes. For perspective, this is equivalent to over 1 trillion pages of text. Furthermore, many of the individual files are so large that they require custom cloud-based infrastructure builds to load into memory. As a result, it requires advanced data engineering work to load, aggregate and process the data.
You need access to claims data to unlock the potential of the transparency data
The healthcare transparency data makes it clear that there is no universal standard structure that governs how insurance companies and hospital systems negotiate on the price of medical services. Instead, there are a myriad of different billing, grouping and payment conventions that change the actual cost of a service. For example, take a very common outpatient procedure: a colonoscopy with a biopsy and related lab work. This one procedure gets billed many different ways. One contract may reimburse this together in a single bundle, whereas another might bundle the colonoscopy and biopsy together with the lab work paid separately and a third might bill all three services separately. Now let’s say one provider costs $2,000 for a colonoscopy whereas another costs $1,000. However, the first provider includes all biopsies and lab work while the second does not. Which is lower cost? To answer this question it is important to know the odds of a biopsy occurring during the colonoscopy. This requires historical healthcare claims. Therefore, performing any apples-to-apples comparisons using the healthcare transparency data requires not only lots of data science resources, but also access to this sort of claims data. Over the past few years, Garner has been able to collect over 50 billion individual claims, representing nearly 75% of all the claims data in the US, giving us a unique position to contextualize these insights.
History has shown that consumers will not “shop” for care using transparency data, no matter how simply it is presented to them
As we have written before, the idea of consumers “shopping” for cheaper procedures such as MRIs, labs and surgeries has been around for nearly 20 years and the data is clear that it does not work. Patients simply do not engage with search tools to find a facility or test location on their own. Instead, most people simply trust their doctor to refer them to a location for these services. For example, studies by the Journal of the American Medical Association (JAMA) and Harvard, only 3-4% of patients use digital search tools to find a lower cost location for a medical procedure. By contrast, over 70% of patients now use digital search tools to find a new doctor.
Furthermore, even when patients know enough to search for a cheaper location for a procedure, traditional plan design gives them very little incentive to do so. As shown below, 85% of employees in high deductible plans who have a major episode of care end up spending through their deductible. Thus, despite the desire for employees to have “skin in the game,” the employees who spend most of the healthcare dollars end up with very little incentive to lower costs.
In order for this transparency data to have any real impact on patient behavior, we believe employee-facing tools must influence patients at the time they are choosing a new doctor and be paired with incentives that actually moves the needle for consumer decision-making.
At Garner we have been awaiting this data and have structured our program to take advantage of it.
In particular, Garner’s approach to physician ranking does not rely on industry-standard episode groupers and instead goes “bottom-up” to measure each clinical decision an individual doctor makes. This process lets us very easily understand where each doctor sends their MRIs, labs, surgeries and other medical procedures and factor those costs into each doctor’s total cost of care. The new transparency data will therefore easily plug into our existing process and provide us greater accuracy when recommending doctors to our members.
On top of the benefits to our provider ranking methodology, Garner’s incentive accounts resolve the historical issues with the lack of employee engagement in transparency solutions. Many studies have shown that patients routinely use digital search tools to find doctors and want more information on doctor quality. Garner’s program takes advantage of this by working with employers to cover employee out-of-pocket expenses when they see a high quality, efficient in-network provider. These incentive accounts have enabled us to realize a 43% employee engagement rate each year with a 27% lower total cost of care per episode.
We are thrilled about the new healthcare transparency data and its potential to have a huge impact on employer-sponsored plans. If you’re interested in how Garner can help unlock this potential, please feel free to get in touch.
 It is important to note that many insurers are not yet in full compliance with the letter of the regulations. For example some have neglected to include key required data fields, others have only included physician bills and not facility bills and yet others have some providers missing. We are hopeful that these insurers will update their files with this information going forward.