Particle Health, which uses an API platform to unify records for digital healthcare partners, recently released a white paper on the state of U.S. healthcare’s national network data exchanges based on the company’s experience gathering data from those networks. In an interview with Healthcare Innovation, Particle Health CEO Troy Bannister spoke about TEFCA, information blocking, individual access to data, and other issues.
HCI: Let’s start by talking about where Particle Health fits into the healthcare ecosystem.
Bannister: We are connecting to and aggregating and simplifying connectivity into massive networks across the United States, and making it really easy for third parties to connect to those networks.
HCI: So the third parties could by digital health companies that are trying to solve a chronic disease problem but can't get access to the data? Could it also be individuals themselves who want to aggregate their data from all these health system portals for their own purposes?
Bannister: Yes. However, I don't think you'll find a company like ours that is building anything direct to consumer. Companies like Particle need to spend our time and our focus on the infrastructure layer, not the user interface.
HCI: Your white paper details some of the progress made by the three large national health information networks — Carequality, Commonwell Health Alliance, and eHealth Exchange — as well as some ongoing challenges that they have. One of the things I found fascinating was that you were able to detail query success rates broken down by the state level. The top five were Michigan, Virginia, Alabama, Oregon and Washington. The lowest five were Vermont, Hawaii, New Hampshire, North Dakota and Montana. Is there an easy explanation for the disparities in those groups. Is it tied to HIE participation? Is it rural versus urban? Why would Alabama be near the top?
Bannister: I think there are a couple of reasons. The first thing we called out in the white paper is that this is data that we've aggregated as a synopsis of what we've seen with our queries, which are biased toward our customers, right? The more queries we do, and the more customers we have, the more statistically significant these points will be. The reasons why we think some states are good and some states are bad are kind of as follows: No. 1 is that there are some opt-in states and opt-out states for data sharing. And I think there are five opt-in states. Those five states are going to be towards the bottom.
The second piece is that places where there are unhealthy people will have more people participating within the healthcare ecosystem, so there will be more data in that state for us to find. We also see a big disparity between young people and older people. Older people have a lot more data we can go find than younger people.
Also, there are some states that are prioritizing indexing patient information, and they are good at matching and they are good at finding records connected to someone's demographics. There are other states that don't prioritize that. You know, the CRISP HIE in Maryland is a great organization doing a lot of indexing and management of this type of stuff. Manifest MedEx in California is another example. Then there are states that don't prioritize that and don't have it.
HCI: What kinds of data do you see coming back from queries and are there gaps?
Bannister: There are certainly gaps. We are seeing loose adherence to the USCDI requirements. We're starting to see insurance information come back in the returns. We're seeing payer type, your benefit plan group, your subscriber ID, your phone and address and type of plan. That is coming back now consistently. We get the majority of lab results. Almost any lab that's ordered through the EMR is making its way back. We're also seeing a lot more doctors’ notes come back — the unstructured contextual data, which is an equally if not sometimes more important aspect of this, depending on the customer or partner we are working with.
HCI: What kinds of things might be missing?
Bannister: A lot of specialty care notes. For example, we have a good number of cancer-oriented partners that we work with. They need very particular types of information in order to make decisions — tumor biopsy details, which are not always making their way back through the EHRs. So there are some nuanced, very specific but very, very important aspects.
HCI: Can we talk about some of the remaining challenges for those big three health information networks in sharing data?
Bannister: There are a few things that I'm very passionate about here. Hypothetically, if you have a million doctors query for patient records, you'll get a 90-plus percent response rate back. If you just switch out those doctors with individuals requesting their own records, you'll get close to a zero percent response rate. That is against the law. It is in direct violation of anti-information blocking. However, that is where we are today. The only difference is we're switching out the word treatment with the words individual request.
HCI: I guess we're waiting for a proposed rule from HHS that is supposed to give the rule teeth by saying what the penalties are. Something similar happened when they passed a law requiring price transparency from hospitals. Initially, a lot of them just ignored it.
Bannister: My sense is that we're going to follow a similar narrative here such that no one does it, no one does it, they issue a big fine, and then everybody freaks out and then they turn it on. It takes time.
HCI: Are there other challenges these networks face?
Bannister: I think there are some technical limitations that the networks have been trying to figure out. One is that this is a query-only framework across all three networks today. There is no way to be notified of a patient arriving or having new information created. There's no push notification rule mandated or pervasive yet across these networks. That is hugely limiting, right? If I'm a doctor, the only way to do this now is query for that patient record in some sort of regular interval, which is not allowed by the networks, so there is no way to be notified or to have regular updates on patient information other than tying it to an actual appointment of that patient coming in. So when we're talking about preventative care and population health and getting ahead of problems before they start, the networks don't allow it.
HCI: Aren’t these HINs working on event notification services of their own?
Bannister: They've already built it and piloted it. They can create amazing stuff, but unless it is mandated, it will either take forever to adopt or it will take significant work and pressure to make it happen. There's just not a lot of incentive from the organizations that have the power to make it happen.
HCI: So other than the individual access, are there changes to the “purposes of use” for data queries that are written into regulation that will lead to progress down the road?
Bannister: There are two buckets of purpose of use. There are the HIPAA-covered use cases — treatment payment and operations — and then there are non-HIPAA-covered use cases, which are individual access and things like clinical trials, public health, life insurance underwriting and, all these things. My belief is that the HIPAA-covered use cases are prioritized, and I know there's been a good amount of progress made on the operations purpose of use, specifically care coordination. Operations is a bit too broad of a term for everybody to agree on the policy and the technical framework for it, so it's broken down into smaller sub-purposes of use, if you will. I believe that one is making significant progress and should be turned on, hopefully this year.
HCI: Are these changes going to come through TEFCA and the RCE decision-making or is it through some broader rule-making at HHS?
Bannister: If you think about the hierarchy of all these organizations, you have the White House at the top, then you’ve got HHS, then ONC, the Sequoia Project and then Carequality. That is a lot of layers. Although it's slow, it is relatively effective. Because look at what we've been able to achieve with Carequality — it is truly working today better than we ever could have imagined it was going to. I think it requires an executive order from the President of the United States. Once those decisions are made, it can flow down through those different agencies and organizations. Obama signed the 21st Century Cures Act back in 2016, which included info blocking, and now in 2023 we're seeing info blocking start to matriculate. It takes a long time — half a decade — to even get it started. But it is starting, which is a massive milestone for the healthcare system.
HCI: HHS is planning to announce the first QHINs as part of TEFCA this month. Are there things that you expect to see in the first year from the QHINs and the RCE?
Bannister: I think the design of TEFCA and the nature of the U.S. government is to incentivize competition and use competition to accelerate innovation. I think that the number of applicants, which is in the twenties now, if I'm correct, is an awesome sign that there's going to be a lot of pressure for these different organizations to create really effective networks of data exchange that have a lot of value. I'm excited about that.
There are still some issues that are going to have to be figured out. One is that TEFCA is not valuable until a critical mass is there. I think there are going to be two or maybe three years of low-value, high-cost participation in TEFCA. That's just the nature of a network vision that we're building here, the chicken and the egg. The promise of TEFCA is being able to pull everybody's medical records across the United States into one place, but it will take a while to get there.
I think the only way that it does work is if you create such a carrot, such a value-add, that you are at a competitive disadvantage for not connecting to a QHIN.
HCI: Anything else from the report that I haven't asked about that you think is important for people to consider?
Bannister: The biggest message I want to get across is how ludicrous it is that you as an individual can't retrieve your own records, whereas the infrastructure to allow it is all built. We have looked up probably 30, 40, or 50 million records to this point. So why are we not letting individuals access their own records? It's mind-boggling to me.
I'm very happy that we get to do great work with some really awesome customers like Oak Street Health, One Medical and Omada Health, and we're not going to stop doing that. But that is not the dream, right? The dream is you have full agency over where your record goes. You can revoke access to an organization when you're done with them. You can grant access to an organization when you want to. You can share pieces of your data and not all of it. This is what we need. Look at what happened in FinTech when Plaid popped up. We got Mint and Robin Hood and Venmo. Where are the Mint, Robin Hood and Venmo of healthcare? We can't have that until this is turned on.
