A Tale of Rails and Ramps

Choo choo, all aboard for healthcare's utilities problem

Healthcare in the US has...some problems. It’s an industry where incentives are misaligned, where the patient/provider/payer triangle is heavily tilted against the patient, and where the debt-fueled arms race for expansion and growth forces an unparalleled upward trajectory of prices. When asked the biggest problem in sight, however, I focus on a different topic: Healthcare has a rails problem.

“Hold up,” you might say, “what in the world are rails? Are you talking about trains? Ruby?”

Rails is a concept primarily used in the banking world, defining the methods for payments accepted nationwide. In the US, there are at least 5 separate rails that have reached a level of ubiquity: cash, checks, credit, wire transfer, and ACH (the slow bank-to-bank one that you usually use to get your paychecks and/or pay bills). Each is utilized by hundreds of banks and tens of thousands of businesses, with complex supporting infrastructure in between to facilitate payments, reconcile accounts, and prevent fraud. At first glance, it’s quite curious that we’d have 5 parallel networks of sorts to perform the same simple function (move money from person A to person B in exchange for goods or services), but each comes with its own set of pros and cons (speed and cost being the most important).

ACH (bank transfer) is pretty sweet, but underutilized/less well known to consumers. Source

Internationally, there are many unique rails within each country, which are really interesting to compare/contrast in terms of differences of adoption (Dutch/Europeans don’t use credit cards like Americans, bank transfers are ridiculously easy, etc). Luckily, we see the emergence of cross-country payments rails, pushing the needle closer to frictionless for this core business transaction. There are also emergent new rails, such as cryptocurrency or real-time payments (RTP), that are rapidly expanding in acceptance on both the banking side and the business acceptance. Just this past month, we saw a major step towards that future:

Jeremy Allaire @jerallaire

1/ Breaking major news from US Treasury OCC, the largest US banking regulator (@USOCC), with new guidance allowing US banks to use public blockchains and dollar stablecoins as a settlement infrastructure in the US financial system.

occ.govFederally Chartered Banks and Thrifts May Participate in Independent Node Verification Networks and Use Stablecoins for Payment Activities …The Office of the Comptroller of the Currency (OCC) today published a letter clarifying national banks’ and federal savings associations’ authority to participate in independent node verification networks (INVN) and use stablecoins to conduct payment activities and other bank-permissible functions.

The payment rails exist in both the public and private spheres. Many of these names are well-known businesses, such as Visa or Mastercard, but others are fairly successful with less brand recognition, such as the Clearing House, Interlink, STAR, and Pulse. While some have a regional history, at this point, these networks are almost entirely a national affair. The federal government also takes an active role in facilitating payment rails, operating FedACH and Fedwire for ACH and wire transfers respectively, and planning to roll out FedNow for real-time payments.

Ubiquitous networks are powerful, even when the underlying technology is not. When one can assume that a given workflow can be fulfilled with little to no variation at a national scale, it can be treated as a commodity and built upon. The payment rails run on really ancient formats, but the fact that it can be used universally trumps the weird nuances and legacy standards. With payments networks fully ubiquitous (all banks, ranging from Citi, JP Morgan, or Wells Fargo down to the long tail of community banks and neobanks, participate in the payment rails), we see innovative companies like Stripe, Finix, or Moov acting as on-ramps, trading a simplified experience for some fees so that businesses that wish to utilize these commodities can do so with lower overhead and investment.

This is a trend that is not unique to banking, either. Other (but not all) heralded API companies follow a similar format of acting as on-ramps to legacy networks:

• Twilio: In their own “What does Twilio Do” page, they immediately hit on the format of acting as an easy on-ramp to the legacy (but ubiquitous) telephony networks - SMS, voice, Whatapp, etc - simplifying the experience and buffering the developer from having to interact with the complexity of legacy standards - (SIP, XMPP, RTP, GSM, etc)

• Mailgun (and SendGrid and Mailchimp and many others): Their products page has verbiage articulating how they act as an API on-ramp to the ubiquitous legacy technology of email - simplifying the experience and buffering the developer from having to deal with the pain of SMTP servers and MIME formats.

• Lob: Less well-known, but very well regarded, they act as an API on-ramp to the (very) legacy, ubiquitous postal network to allow the sending of mail nationwide, simplifying the experience and buffering the developer from having to deal with the complexity of the analog world (buying paper, printing, postage, sending).

  Hardy Farrow @farrowhardy1

  @garrytan @zachperret @truepill_rx @lob @easypost Lob is amazing. Can't imagine starting our business 10 years ago without it.

Compared with other types of API companies (aggregators and connectors), on-ramps aren’t sticky by default, as they’re using a network, not building one. The value prop of abstraction of a ubiquitous legacy network’s standards and intricacies is pretty easy to emulate on paper:

Sar Haribhakti @sarthakgh

It really is insane that Stripe is the dominant player in the US for online payments. There's no real network effects for customers (besides fraud detection & such). How have we not seen multiple competitors in the past decade?

Stripe offers a good template for how good API on-ramp companies build stickiness:

1. Amass profits as first-mover or (better) the simplest/least frictionful on-ramp (see red offerings in the screenshot below)

2. Use market presence and weight from 1 to build differentiated or at least supplementary services to lock in the customer base (see blue offerings), commonly analytics or insights only realizable via that market weight

3. Use brand and partnerships from 1 to start to build an adjacent unique function/network and upsell to captive audience locked in from 2 (see green offerings)

On-ramps allow for innovation. When ubiquitous functions are made easy and years of nuance and tribal knowledge reduced to well-documented open APIs, they become atomic units upon which all can build.

---

It’s easy to get tunnel vision on these on-ramp companies, though. Put simply, they’re exciting! As fundamental, necessary building blocks, they’re enabling new digital health organizations to build and grow faster than ever. However, before one can create the on-ramps, though, you need to build the highways.

Healthcare largely lacks infrastructure at the scale and pervasiveness of other industries, which is infuriating. The national business-to-business networks that should exist purpose-built for the most common healthcare transactions range from non-existent to completely lackluster.

There are many excuses to be had. Health is fundamentally more complex than payments, in that its data is significantly more structured. Payment has a lot of variations but is ultimately two accounts and an amount to be transferred, whereas healthcare has dozens of specialties with unique measurements, diagnostics, and interpretations relevant to care.

The incentives for data transfer and sharing are core to other industries - banks that can’t send and receive money with many institutions tend to fail, whereas healthcare organizations can focus locally or even internally and skate by, dealing with the occasional out-of-network patient.

Taking a step back and thinking about how enterprises interact in healthcare in an analog world (typically by fax today), you can get a picture of what should exist. Cross-organizational workflows don’t dramatically change with the introduction of new ubiquitous technology, at least to start. Checks have been largely replaced by ACH, but many of the core concepts and nuances are a carryover from previous rails.

Unfortunately for us, the bulk of these healthcare utilities either are tremendously lackluster, have been fragmented at the regional level, or simply do not exist. Thus, without further ado, we have the official…

2021 Healthcare Infrastructure Awards™

More like a participation trophy than anything else

1. Imperfect but Actually Kinda Useful Award

There are actually a fair number of healthcare workflows that are facilitated by ubiquitous networks (either private or public) at a national scale. They mostly take the form Bryne mentions in his excellent article - “A hodgepodge of legacy infrastructure and regulation colliding with private company middleware to make 1960s technology (barely) usable for 21st century”. Each member of this elite tier of mediocrity comes with their own rules, regulations, nuances, and caveats, but they’re the closest things we have to rails.

Eligibility and Claims Clearinghouses

Dan O'Neill @dp_oneill

@healthbjk I think we could probably describe Change, Availity and maybe a couple of others as ubiquitous in a practical sense. They generally have long-standing “interchange” agreements, so if one or the other is the last mile to a certain node, it’s still effectively ubiquitous.

• Positives: ☝️. It’s no surprise that where the money flows, ubiquity goes. These are the payment rails of healthcare, so the incentives are there to have full connectivity.

• Negatives: Like most ubiquitous networks, the technology is old and decrepit. X12, like its clinical EDI brother from another mother HL7, fundamentally sucks (and is unfortunately enshrined into law via HIPAA as the standard to be used for the foreseeable future).

• Competition: There are plenty of clearinghouses such as Availity, Optum, Change, and Waystar. The interchange has made broad payer integration a commodity, forcing the various clearinghouses into the difficult competition of constantly decreasing prices. As the pure-play clearinghouse has become less viable, we’ve seen consolidation in the space between various players, as well as differentiation by virtue of non-traditional offerings, such as telehealth (which is not exactly the first complementary offering I’d naturally think of alongside eligibility and claims).

• On-ramps: Pokitdok was the torchbearer in this area, offering eligibility, claims, and associated master data APIs backended to payers with X12. It actually acts less as an on-ramp and more as its own network, a modern clearinghouse directly connecting to insurance partners often. Its success in doing so led to an acquisition in 2018 by one of those legacy clearinghouses, Change Healthcare. Eligible and pVerify are the remaining independents of this mold. Some interesting discussion of those companies on Reddit (one of the few sources of semi-honest dialogue on the Internet, second only to Twitter) here.

E-prescribing

• Positives: Thanks to copious regulation and the fact that e-prescribing saves providers money / makes pharmacies money, there was massive adoption of Surescripts in the 2000s and early 2010s, so you can reliably count on a doctor being able to send your prescription to a pharmacy without printing it out or faxing 99% of the time.

• Negatives: It’s not perfect, though. Transfers between pharmacies still occur out of band if a patient changes their mind about where they want to pick something up. Providers don’t yet get updates about the dispense status sent back to them commonly. And upgrades to the network are slow (the 2017 updates to the core NCPDP SCRIPT standard have only just been adopted, despite regulation mandating such)

• Competition: It’s a single ubiquitous network (there’s no Mastercard to Surescripts’s Visa), possibly because of antitrust, but at least it works reliably. Truepill offers an alternative, in that providers can have their prescriptions fulfilled reliably, but it’s always by the Truepill pharmacy (not any pharmacy nationwide).

• On-ramps: Lackluster, as Surescripts has guarded against that to some degree to prevent competition (see antitrust case). Bravado Health comes up first in a quick “e-prescribing API” search and is clearly angling for reducing onboarding friction with their “Accelerate your development. No Surescripts certification required” pitch. DoseSpot was grandfathered in and is a common choice for new digital health organizations looking to simplify onboarding and not have to deal with NCPDP XML (although anecdotally they’re supposedly a bit lackluster in features and support). Change has an API listed (but not really documented). DrFirst gets some buzz there but I think their solution is more a portal/UI to jump to, not a developer-friendly API. GiftHEALTH is also doing some really interesting stuff to fill in the gaps and enable transfers of medications between pharmacies, while also aiming to reduce the administrative burden on pharmacists.

2. Pirate Hugh Grant Award

Ubiquitous but regionally fragmented solutions don’t piece together into a cohesive national utlity

While on paper, it seems like these healthcare problems may have been answered, this runner up tier is actually filled with partial solutions, segmented by geographic boundaries or the sharp divide of capitalistic competition.

Clinical Data Exchange

• Positives: The Carequality trust framework has woven together a number of regional and competitive players (Commonwell, eHealth Exchange, state HIEs, EHR specific networks) into a fairly cohesive national network for querying/pulling a patient’s record. Likewise, DirectTrust reaches 1.5 million or so providers (at least on paper), allowing the push of a patient’s record to a specific organization or doctor. Lastly, vendors like PatientPing, Collective Medical, and Audacious Inquiry (primarily through state HIEs) have made event notification for subscribed patients (understanding when a patient is admitted into another healthcare organization) fairly widespread.

• Negatives: “Wait, Brendan, why isn’t this in the top tier?” you might ask. While each individual network has wide deployment, the whole would be significantly greater than the sum of its parts. Event notification is fragmented geographically between PatientPing, Collective Medical, and Audacious Inquiry, meaning none can be fully counted on at a national scale. As a query only arrangement, Carequality is filled with extraneous traffic. DirectTrust suffers from data quality issues in regards to its organizational/provider directory (as entries are often out-of-date or invalid). A complete and comprehensive strategy at a minimum needs to be national in scope and ideally would combine or overlay the three functions. Proper and reliable event notification would reduce Carequality’s traffic problem, where the clean directory of Carequality would be hugely beneficial for DirectTrust when pushing data.

• Competition: For clinical data retrieval and push, Carequality and DirectTrust are well-established, but event notification is still a bit of a land grab, with no interchange or exchange between the three vendors. Beyond that, Patient Centered Data Home (PCDH) by SHIEC (a consortium of the state HIEs) has similar event notification and clinical data retrieval functions. Together, it adds up to an incomplete strategy.

• On-ramps: Various vendors, such as Redox (disclosure: the company I work for), Kno2, Particle Health, or Health Gorilla, offer easier onboarding and access to the Carequality framework. Health Information Service Providers are the certified on-ramps for DirectTrust, although not all have modernized with an API-driven approach.

Public Health Reporting

• Positives: As a result of Meaningful Use and other regulation, public health agencies have (for the most part) programs in place to receive syndromic surveillance, lab reports, and vaccinations. So there’s a nice illusion of ubiquity, at least.

• Negatives: Everything else. Each state has different processes for enrolling and connecting and often different transport standards. There’s a distinct lack of interoperability between them. Many do not pass the sniff test of “do they even actually work”. We saw massive failures in regards to lab reporting throughout the pandemic and we’re seeing similar issues with vaccination registries. I did a write-up on immunization registries, if you’re looking for a deeper dive.

• Competition: Ultimately, national problems need national solutions. The spiderweb patchwork of state registries with legacy technology is in sore need of modernization. At a minimum, this is an opportunity for the federal government to decide between CDC and HHS ownership of national lab reporting and to complete the Immunization Gateway project. But we should go further - if state registries are the bottlenecks, they need to be functional and funded. They also need modern provider onboarding (the Carequality trust agreements would be well-suited for this) and patient data access.

• On-ramps: APHL AIMS and Redox both offer national lab reporting, having aggregated the state and local ELR registries. Ironbridge and Redox do the same for vaccination reporting.

3. Gru’s Master Plan Award

The remainder of healthcare’s infrastructure use cases is still work-to-be-done - future networks that will hopefully be aggregated or forced into being by regulation.

• Patient Identity is so fundamentally core to all other problems, but we still have no overarching, national path to getting to a better place. India has shown (in a non-health specific way) the tremendous benefits that starting with identity can have on national rails - solve identity first and suddenly tons of friction and wasted effort for every point solution is alleviated. Most existing networks/rails delegate out matching to individual participants or utilize some probabilistic matching if they are more centralized. Private solutions like Experian Health, Verato, Clear, or even Redox (I work on this team, so if you’re interested, please reach out) really only help in limited capacities, sold and customized enterprise by enterprise. Ambitious proposals range all over the spectrum from centralized federal service à la India down to decentralized ledgers (hello, blockchain enthusiasts) but for any to be successful, a prescriptive path needs to be chosen. Otherwise, we can continue to pretend that credit companies are the best stewards of identity with knowledge-based questions like this:

Squeakycat and 3-5 knowledge based questions: a stalwart bastion protecting America against malevolent fraudsters nationwide

• Reference Lab Ordering and Resulting is still in network formation mode. Even if a large healthcare organization has built out their own clinical, pathology, or genetics lab, they often have niche tests that they do not have the right equipment for. Reference labs exist to service this need and provide convenience to the patient, in that the location to draw blood or other specimens may be closer to their home. Labcorp and Quest are the most prominent, so many health systems built point-to-point integrations with them, which has been the status quo solution. There has recently been an explosion of new reference labs, especially in the genetics space, so several players have attempted to aggregate them into a network and sell that access/connectivity to providers via API or simple portals, such as Rupa Health, ORDRS, Health Gorilla, Change Healthcare, or Redox.

• Prior authorizations (approval for procedures or admissions by a patient’s insurance company) are exceedingly manual in most cases, with the occasional point-to-point integration. We have seen regulatory progress as the CMS has mandated payers expose this through APIs, but the capability is not yet widely deployed nor exposed via an on-ramp aggregating those endpoints.

• Diagnostic image exchange (exchange radiology, cardiology, and other clinical imaging) is limited and balkanized in the US, with various vendors like LifeImage, Ambra Health, and Nuance having extensive proprietary networks. Carequality has advanced an implementation guide to unify the networks and other players, but there has been limited adoption.

• Study subject recruitment (enrolling enough patients with the diagnoses intended to be studied in clinical trials) is incredibly difficult in today’s system. 37 percent of sites selected for clinical trial studies under-enroll and 11 percent fail to enroll a single subject, leading to slower studies, wasted effort, and higher costs. Pharmaceutical companies and CROs lack a way to find patients with certain conditions, lab tests, or other criteria, which ultimately hurts providers and patients as innovative treatments are decelerated or blocked. The future here is unclear, as a network for this purpose would need to appropriately protect privacy (pharmaceutical companies are not HIPAA covered entities and have their own litany of regulation), but it’s definitely a gap today.

---

These awards are far from comprehensive, but they illustrate how digital health will continue to lag behind fintech. Ramps can’t exist before rails and building rails is hard work. Perhaps the incentives need to fundamentally be changed. There are plenty of examples of the healthcare industry in this country doing amazing, world-leading things - but unlike the banking industry, incentives have never been aligned to build these rails. 

It’s time to build. It’s time to build the networks we need, to put in that hard work of connecting hundreds, thousands, or more. It’s time for the government to build the regulation to incentivize, incubate and encourage the networks we need (finishing TEFCA would be a start). And for the networks that do exist, it’s time to build the on-ramps to accelerate innovation based on those commodities.

Thanks to Jamie for input, knowledge, and feedback on clearinghouses and to Colin, James, Nijay, and Bailey for editing!

Have a question? Did I miss something? Is there a particular topic you’d like to see expanded on? Leave a comment or reach out at brendan@redoxengine.com