Photo: istock.com/Atstock Productions
CareJourney President Aneesh Chopra is one of the foremost health care data and analytics experts and advocates. Prior to co-founding CareJourney, he served under former President Obama for three years as the first chief technology officer of the United States. Aneesh is also an author, graduate of Johns Hopkins and Harvard universities, and former candidate for Lieutenant Governor of Virginia.
Aneesh is extremely passionate about health care and the possibilities that new data sets and analytical techniques offer the industry. With open innovation and validation, he is confident the industry can establish effective clinical interventions that deliver superior care, better outcomes and lower costs through value-based care.
What were your responsibilities as chief technology officer for the United States?
My job was to advise President Obama on ways the country could better use technology, data and innovation to advance his priorities. My work began with economic recovery in 2009, and then I moved on to work on health care reform (including the Affordable Care Act), a clean energy economy, and up-skilling the workforce through math and science education. Additionally, we launched the Startup America initiative to support the growth of an important engine of our economy, as companies less than five years old account for the majority of net new jobs.
You’ve argued that an ambitious but achievable goal for health care cost inflation is gross domestic product (GDP) + 0 percent. How do we as a country and industry achieve that?
There are two choices. In option A, we put health care on a budget and force the system to figure it out. The blunt instrument in Washington would be Medicare for All on the left, or Medicaid block grants on the right. In both cases, the unit of intervention is basically some form of price control.
Option B would be to create a pathway for value-based care delivery that thoughtfully segments groups of people with specific interventions and outcomes. From a bottom-up perspective, it would mean redesigning the delivery system around capabilities, and it would mean considering how different population segments can be better severed—and serving them in that way.
I subscribe to the latter philosophy. I believe we can make the system better with more services and more capabilities—and the net result is lower cost. This may be in the form of more preventive services, more bundled care models or more primary care quarterbacking. Whatever it is, it’s more. That may result in a reduction of downstream services, but this is done in the best interest of the patient avoiding those services—the opposite of saying, “you need this service, but we’re going to cap the cost because we can’t afford it.”
Why has value-based care been slower to take off and slow to scale?
With the exception of a few Medicare-reimbursed models (examples being transitional care management or diabetes prevention), each risk-bearing entity needs to construct its own “recipe” for clinical intervention. My view is that we massively democratize these clinical interventions in a better public/private “handoff,” where models that show promise are tested and scaled via Centers for Medicare and Medicaid Services (CMS) authorities vested in the CMS Office of the Actuary.
One view is that this work is creating proprietary intellectual property that serves as part of the reason somebody should choose your plan. You have greater analytical capability, which leads to better, faster and less-expensive care. However, you may take an open innovation model view that welcomes academics, brings in vendors and creates a sort of GitHub for clinical models. That should result in significantly more people becoming informed about new clinical protocols and getting connected to service providers, which in turn should lead to better monitoring of interventions to make sure they deliver the expected results.
For actuaries working on these interventions and clinical models, do you believe there is still value to be learned from claims data?
My realistic view on this is “yes, and.” Have we squeezed all of the juice out of the claims database? I don’t believe so—there’s more to be done. I believe with standardization and the emergence of the Fast Healthcare Interoperability Resources (FHIR) application programming interface (API), we should be able to do a much better job of extracting the same kind of reliable signal from the electronic health record (EHR) infrastructure as we’ve relied on the claims infrastructure for the last couple of decades.
For a lot of reasons, I’m hopeful we’ll be introducing EHR data into the actuarial toolkit in a more reliable way. The resulting applications will result in real-time decision support for patients and doctors, which will ensure that the right folks get routed to the right treatment.
What is an API?
An API is a form of technology that operates like the valet key to a car. It enables the driver of the car to determine what capabilities it offers when it gives another party its key. So, say you’re going to a restaurant, and you give the valet a key that only starts the ignition. But if it’s a family member you trust, you might give them a key that opens up the glove compartment, the ignition and the trunk. Maybe you’re going to a friend’s party, and you want to limit the key to a one-time use. An API allows you to automate or digitize a contract between two parties, but then it standardizes the implementation.
APIs enable an open marketplace of applications that are powered by data sets held by an API server. Imagine wanting to connect with Walmart, Walgreens and CVS on pharmacy care coordination. Rather than building separate interfaces to each pharmacy, I could set up an API server that enables standardized application access to a set of data elements that each pharmacy wants (or is willing to standardize against). It would serve as a single door that allows a payer to communicate a common data set on mutually agreed terms and conditions. An API allows us to standardize the data that we expose but then to constrain and customize permission and usage rights through the key-making process.
What’s available today because of APIs? What’s coming next—for health care specifically?
Today, if any organization built a consumer-facing application and the consumer opted in, one is capable of getting access to a common clinical data set out of an EHR and aggregating that data wherever the consumer has given permission. In this example, we’ll call that trusted application the “Apple Health” database. We expect the number of available trusted applications to grow in the next two years to include clinical notes and other variables that are regulated under interoperability rules. However, today a consumer can only access medications, lab results, diagnosis and procedure codes, among other structured data.
Next, government-sponsored health plans will be required to publish claims and any collected clinical data via APIs by July 1, 2021, comparable to the mechanism consumers currently use via Apple Heath. My view is this will scale beyond government plans and into the commercial population. This should allow every American with an insurance product and phone to harness insight from their claims and some clinical history with their doctor in real time. This is opposed to relying on their health plan and their doctor to have previously figured out some sort of economic arrangement and technical mechanism to pull together that information.
Is there a challenge you would pose to the actuarial community?
If you believe we can deliver clinical interventions that allow us to expand benefits while also generating return on investment (ROI), can we as a community rapidly identify those interventions and those patient segments at scale? The country is looking at these two options, and it’s like the Sword of Damocles. If we can’t get the value-based care model working fast enough, at some point Washington is going to say “no more”, and we move to price-cutting and rationing.
The challenge is we need to move faster and scale what works more effectively. With modern techniques and the better health data infrastructure that we have today, we should be able to more reliably determine and screen interventions. Let’s not just offer a program to a generic member with diabetes. Let’s explore a broader set of chronic conditions, such as behavioral health, social isolation, chronic kidney disease and so on, and explicitly offer a richer benefit to the specific subgroup that merits an ROI. Can the actuarial community utilize modern data platforms and help identify the payer- and provider-agnostic recipes that maximize value in the delivery system?
Copyright © 2020 by the Society of Actuaries, Schaumburg, Illinois.