Introducing the digital health tech stack
We go deep on digital health developers' buy/build/partner decisions
So you’re starting a digital health company. If you’re providing care to patients as part of the experience, you’ll have a lot of decisions ahead of you about where to invest vital development resources. That’ll bring you to your tech stack, ranging from your EHR to your API for ordering lab tests.
To help guide that process, my colleague Marissa Schlueter and I have spent the past few months going deep on the tech stack companies. And we’ve asked founders as well as product /operations leaders how they’re making all important build vs. buy decisions.
We’re doing that because funds we respect like a16z and Greycroft are betting big that the future of digital health isn’t just the care experiences. It’s the software toolset, i.e. the “building blocks,” that makes all of that happen in a more efficient and more scalable way. What remains to be seen is whether these tools that are custom built for digital health can migrate over to the enterprise. The big question on everyone’s minds here is around the Total Addressable Market or TAM. But the positive here is that it may be enough, at least for a while, to sell into other startups. As we’re seeing time and time again, digital health is selling into digital health and becoming unicorns in the process.
This post features a cross-section of perspectives, some of which you may agree with and others you may not. Frankly, it’s hard to generalize with this stuff as companies will have very different needs depending on their care model, patient population, and so on. Also, I’m sure I have some gaps here so please reach out if you’ve got an addition or edit to make!
So with that preamble behind us, let’s talk through the layers of a typical digital health company’s tech stack. This post is really intended to be more of an open source project than anything else, so please let us know what we’re missing and we’ll do an update in a few months.
Peeling back the layers
So the consensus amongst the experts we talked to was this: Digital health companies will make build or buy decisions based on their reimbursement model.
Those starting up a tech-enabled services company or even a “pay-vider”, which is both a payer or provider combined, will have a very different set of choices to make than a D2C company. There’re also nuances when it comes to the payment model, whether it’s value-based and is some kind of risk-bearing entity.
Let’s start by focusing on the tech-enabled service space at large. This list below and the inspiration behind it comes via the collective brains of a mix of folks we’ve talked to over the past few months with an extra special shout out to Josh Tauber, COO of a stealth virtual care startup; Joe Tischler, co-founder of Well Health and formerly of SteadyMD; Brendan Keeler, a senior Product Manager at Zus; Jon Lensing, Co-Founder & CEO of OpenLoop; Julie Yoo at a16z (special thx for the Airtable!); Erica Jain, Co-Founder & CEO of Healthie; Sebastian Coates, Co-Founder & CEO of Tellescope; Dhruv Vasishtha, VP of Product at First Hand; and Corey van der Wal, Co-Founder & CPO of Frame Fertility.
The elephant in the room is, of course, the electronic medical record – and that’s also an important initial decision. Startups these days tend to gravitate between DrChrono, AthenaHealth or Elation – although we’re seeing Canvas Medical gain traction with early-stage businesses.
Depending on the EHR, there may be a need to supplement with external vendors. And outside of the EHR, there’s a lot of other decisions to make and many companies will opt to outsource or leverage tools that aren’t specialized for the healthcare vertical.
This list of ‘building blocks’ is not exhaustive but should be a good start.
To be clear, there are TONS of these types of vendors selling into traditional, less tech-savvy brick-and-mortar providers (e.g., independent practices, hospitals, health systems, etc.) but a much smaller – although growing – list caters to digital health businesses specifically. Our intention with the following list is to highlight those serving digital health and tech-enabled healthcare service businesses to some degree.
EHR – EHRs like athenaHealth (the incumbent), Elation, DrChrono, Canvas Medical, Healthie, Akute Health can be used for a wide swath of use cases, while others are more specialty-specific (e.g., Modernizing Medicine for specialties like dermatology, gastroenterology, orthopedics, urology, etc. while others like Osmind are tailored to mental health). There’s also the option to build your own. Many companies are doing this custom-fit to their specific need – e.g., psychiatry with PHQ9 templates integrated. Vendors like Avon Health offer no-code/low-code platforms and APIs to enable custom EHR builds. It’s important to not only choose the right EMR but use it in the right way. Tischler thinks for companies providing longitudinal care, using an EMR like athena that’s established out the gate is a step up because of the robust reports on the ever-changing “clinical, administrative and regulatory requirements” that come out of the box. Note: For those taking Medicare and/or Medicaid payments, Tischler recommends reading the CEHRT requirements first.
Practice Management – For fee for service, this requires a true practice management system. The choice of vendor very much depends here if the company is DTC/B2B2C/B2C2B/value-based.
Plug-and-Play Provider Staffing – An important consideration here is a vendor’s ability to dynamically and flexibly scale provider supply up/down to meet real-time patient demand. Wheel, SteadyMD, OpenLoop etc. are all working with digital health startups in different ways. While most focus on MDs, there are some focused on NPs (Nurse-1-1), subclinical staff like coaches, social workers, etc. (Pillar), and pharmacists (RxLive). There are also some pared-down Wheel alternatives such as MD Integrations.
Legal Infrastructure – If a company will be delivering any type of healthcare to patients, they will require a full legal set up. Companies need to own and operate PC (professional corporations) that are wholly owned by a licensed practitioner in that state in order to deliver care within that state. If setting up care to be delivered in all 50 states, Jon Lensing from OpenLoop tells us this will take an average of 18 months and around $750k. Some companies offer this as a service and place clinicians underneath those PC groups to deliver the care.
Provider Licensing, Credentialing, and Data Management – This is hard because of 50 states with different licensure and credentialing orgs. Medallion, Verifiable, CertifyOS, Symplr and others are in this space. And there’s niches within that, like licensing just for the Substance Use Disorder (SUD) space. Other vendors like Ribbon Health are building specifically for the provider data management piece.
Content & Marketing – This is important for both patient acquisition and ongoing engagement and includes everything from website design/development tools, content management systems (Wordpress, HubSpot, Contentful, etc.), newsletter/email marketing platforms (MailChimp), patient-specific marketing tools, and more. Notably, if companies are going to break into a higher data/tech touch environment to usher in a true tech-enabled healthcare ecosystem, then this revolves around highly curated, applicable content to patients. The question becomes “how do you make a patient more educated, thereby making them more self-sufficient with technology?”...and this will result in more articles, resources, how-tos, etc. that companies will want to buy off-the-shelf. Some of these off-the-shelf content developers & creator economies are beginning to emerge – the ones we’ve seen are early-stage/in stealth and focused on specific specialties/patient populations.
Scheduling – This can be challenging with EHRs that are built for brick-and-mortar, and there’s no great resource out there yet for digital health. The problem lies within state-to-state matching to the correct specialty. Some decent/average workarounds include Calendly and Acuity. It sounds like this is a big item on DH tech stack developers’ future roadmaps.
Patient Registration & Intake – Most EHRs like Elation and Canvas have a patient intake portal. PhaseZero and HealthNote are trying to make some headway here.
Revenue Cycle Management – Once your MDs are licensed to practice medicine in the specific states the company is operating in and you have credentialing in place, you have a network. Then there’s the payer enrollment process (credentialing, contracting on behalf of the payers) and then the operational function once the company is in network. The middle piece in payer enrollment, meaning the contracting and credentialing, is where the least number of companies are. Serif Health and Cerebrae help companies traverse the payer contracting piece. For eligibility and prior authorizations Eligible, and Availity are big players. For RCM, the list includes established companies like R1 RCM and Change Healthcare as well as newer, API-accessible players like Candid Health and Apero and specialty-specific vendors like Nirvana Health (for mental health).
Administrative Task & Workflow Management – there aren’t many companies here but Dock Health is a good example.
Security, Privacy, & Compliance – Startups are running into SIG questionnaires, SOC2, HITRUST requirements earlier and earlier. “Invest in security framework toolkits early on to avoid losing deals & wasting product engineering time on unnecessary redesigns,” notes Tischler. From a privacy perspective, there’s a lot to say here but it’s very important to understand HIPAA, etc. Vendors in this space include Vanta, Aptible, Virtru, Medstack, Skyflow, and Accountable. A note here: If you’re managing and sharing PHI, there may be more options available outside of those focused on the health industry. Fraud and abuse software is also a big, wide open need.
A lot of companies use the same analytics tools like Tableau, that aren’t specific to health. It’s worth noting here that lots of companies are using non healthcare specific tooling, including analytics and data pipelines. a16z has a good list of best-in-class enterprise tools here.
Customer Data Platform/CRM – Something simple might be fine to start with but definitely do look for one early on because it’s not worth losing deals due to lack of organization. Examples that are health-specific include Freshpaint, PhaseZero, and Healthnote.
Population Health Tools – Pop health analytics tools are needed to help providers perform risk stratification across large populations and configure and execute effective care plans for distinct groups. Why are these tools needed? Because EMRs have historically been bad at surfacing any sort of relevant or actionable insights. Back in the day, analytics were limited to SSRS and Crystal Reports, and any data visualization had to be assembled manually in Excel. In fact, most pop health tools (e.g., Lightbeam) are data repositories loaded with proprietary scripts – white-labeled Tableau with prebuilt workbooks – and are artifacts from those earlier years. If you’re starting out today, you’re hopefully analyzing more relevant datasets and need to design feedback loops into your product to maximize engagement and retention. Interestingly, though, we haven’t seen many great vendors here focused on the digital health segment. Innovacer is one to watch in the space.
An Analytics Datastore – That requires mashing together data from EHR, apps, website, and other tools.
Analytics Visualization – Visualizing the analytics data store, usually after manipulation with a tool like dbt. Looker, Mode, Redash, and Metabase are all popular as well.
Reverse ETL/ELT – Take insights from analytics datastore back into operational tools. Hightouch and Census are the two main players but others like Ursa are beginning to emerge.
Care Team Communication & Coordination – This is critical, especially in virtual environments with decentralized care teams. Central comms systems should not only facilitate seamless, secure communications but also intelligently fill in gaps and prompt care team members when they’re needed and for what (“smart hand-offs”). This is essential not just for operational purposes, but also for the patient experience. We continually hear that one of the biggest pain points for patients is having to “tell their story over and over and over again.” Vendors like Tellescope, HealthHive, and Welkin are working on this.
eConsults – Primary care-focused digital health providers looking for occasional oversight or a second opinion from specialists may opt to partner or integrate with a number of the existing eConsult vendors. There’s also companies focusing on making better referrals to specialists, including PicassoMD.
Patient Communications (which includes both clinical and non-clinical communications) – Some of these are built into certain EHRs while others are available a la carte. Synchronous, HIPAA-compliant audio/video conferencing can be done via Zoom, Vsee, Doxy.me, Vidyo, etc. For asynchronous communications (e-mail, web-based chat, SMS, etc.), “patient portals”/patient-facing apps could be available via the EHR but could also be homegrown. Some example vendors include: twilio, Sendbird, Messagebird, Welkin, and Sinch. Pair Team is also an option but is more tech-enabled services-oriented. Luma Health is used by health systems and clinics. Note: There are also important regulations to understand as part of this process, like the 2015 FCC TCPA Ruling, which is little known outside of the industry but fueled the adoption of patient engagement platforms nationwide. Tischler notes that this can be a “game changer” when it comes to retention when used appropriately to communicate with patients about their health.
Peer-to-Peer & Community Communication – Moderating staff physicians to answer questions, organizing groups of patients with similar conditions. Reddit, Circle, Facebook are non-health focused options.
Interoperability – Either via EHR connectivity to Commonwell, Carequality, state HIE or an on-ramp like Particle, HealthGorilla, Redox, Zus, etc. There are also API-first players that pull payer, rather than provider data, including newer players on the block like Flexpa. Others like OneRecord pull both medical records and claims data.
Patient-Generated Health Data (PGHD) – As the name implies, PGHD is collected and managed by patients, but it’s increasingly being incorporated into clinical workflows, often through EHR integration. These data sources are also becoming a bigger focus now that CMS has expanded RPM/RTM CPT codes. Vendors like Validic, Human API, Thryve Health, Terra, and others offer APIs to facilitate PGHD integration and SDKs to build with it.
Clinical Documentation – Even in a virtual world, there’s a pressing need for physicians and care teams to operate more efficiently at the POC. In the brick-and-mortar realm, clinical documentation has been a hot target for waste and inefficiency reduction for several years now, and we’re just starting to see these digital innovations cross over into virtual care environments (e.g., AI scribes/voice assistants developed by vendors like Nuance, DeepScribe, Suki, Ambiance, iScribe).
Care Pathways & Clinical Workflow – In the same vein, operational efficiencies can be generated by digitizing and [partially] automating clinical care protocols. This is true in both brick-and-mortar and virtual care environments. Companies like Tellescope and SourceHealth are early but taking a stab at this.
Ordering & Fulfillment – This may also include a services wrapper to go alongside the management of the script itself.
Prescription drugs - e-prescribing can be done either via EHR or DoseSpot, or via pharmacy-as-a-service companies like TruePill or Medly. Custom branded e-rx direct to patients’ households is becoming commonplace now (e.g., SmartScripts).
Labs - via EHR or HealthGorilla, Change Healthcare, ELLKAY, TruePill, or one of a handful of newer players like Ash Wellness, ORDRS, ixlayer, Rupa Health, Getlabs, LEAA, Axle Health, etc. Though mobile phlebotomy has historically suffered from logistical and labor constraints, companies like Getlabs hope to overcome these barriers and scale by adopting an Uber-like model.
DME - DME rollups are a real thing and would be neat to one day see the "SteadyMD of DME.” Impilo Health’s API infrastructure helps digital health companies with fulfillment & distribution logistics as well as device data management aspects of RPM programs. Others, like Equipt, manage the logistical aspects of getting prescription DME into patients’ homes.
Referrals & Care Transitions – Sometimes this is done via EHR but, sadly, the vast majority of referrals are still made via fax, DirectTrust, etc. Kno2 is an option for fax. Pair Team offers end-to-end referral, imaging and lab coordination services.
So what do you build and what do you buy?
Well, a lot depends on how you think about differentiating yourself. If you pride yourselves on something and you don’t see it as a commodity, don’t rely on a third-party vendor.
For Vasishtha, who runs the product team at mental health provider Firsthand Cares, his framework for thinking about it is quite simple. Tools that involve interacting with patients need to be insourced as that’s often the differentiator and the core value proposition. Another useful framework is the Reforge matrix, which he uses for a gut check. The prompt he uses with his team when determining the build or buy decision: is it something the market hasn’t already solved where they believe it’s “directionally feasible from a tech or data perspective [and] we have unique assets to leverage?”
Other experts caution against relying too much on vendors. “There's folks in the space who want to outsource everything and then folks who want to do it as strategically appropriate,” said Nate Lacktman, a telemedicine lawyer at Foley & Lardner. Lacktman believes that companies shouldn’t rely on third-parties so heavily that they become little more than a good looking website.
And what if I’m D2C?
It’s worth a reminder that D2C health companies will have different infrastructure needs. Those companies are often more focused on patient acquisition and retention, while providers that bill insurance will put more resources into the clinician, the care model, and the tooling needed to capture, analyze, and report on outcomes.
In Vasishtha’s view, D2C companies might often focus on the marketing aspects of their stack while those that bill payers might look more at risk stratification, interoperability and care outcomes. Tischler, who previously ran product at SteadyMD, suggests D2C companies use tools to continuously track blended customer acquisition cost along with lifetime value across all sources; software for testing and challenging assumptions around the product as it scales; using tools to monitor campaigns, and so on.
In terms of the metrics, Tischler doubled down on companies sourcing the best tools to really grasp the nuances for how their business really works and accurately determining such calculations as LTV: CAC. “I’ve seen a great deal of metric gamification and misrepresentation in virtual health around terms like ‘eligible’ versus ‘registered’ versus ‘enrolled’ versus ‘onboarded’ versus ‘active onboarded’ versus ‘churn.’ “Investors and customers could be “easily misled,” he said.
For D2C, one of the big challenges is when the space gets too crowded with very little differentiation in how consumers perceive the product. “If you aren’t offering a unique service, you will eventually saturate your target cohorts and competition will drive up customer acquisition cost. How bad can this get? See Talkspace,” said Tischler.
His parting feedback? “Fight to create continuous value to retain customers or you won’t survive. If you are already at razor thin margins at the outset and your COGS cannot be re-engineered, you have a huge problem.”
So where are the gaps when it comes to the tech stack?
Vasishtha didn’t find many great options for the CRM. He also hasn’t seen much software that’s truly geared to modern Medicaid providers or behavioral health, which are both areas that his company fits into. There may indeed be more specialization with vendors going forward, at least in pockets of the digital health stack.
“For Medicaid / BH focused software, for example when looking for ETL data management platforms, Ursa was the only company that could connect us with another Medicaid behavioral health reference customer,” he said.
Tischler also has a wishlist. He hasn’t found a great HIPAA-compliant CDP, which ingests activity from all systems and tools into one view and governs the data elements that gets shared with each system.
We’ve heard from plenty of other founders that they desperately need a service that can help them better predict their physician staffing needs as they’ll occasionally over-hire in certain states and then realize they don’t have sufficient demand. This is a problem most keenly felt by those who largely in-house their providers. Getting this wrong can be a big expense and it can slow down the growth of the business as providers won’t necessarily stick around if their services aren’t needed.
Lacktman, who consults with dozens of digital health companies, has a couple ideas based on his interactions with clients. In his view, pretty much anything that the doctor must order from the outside hasn’t yet been solved for. Pharmacy is more seamless now, but there’s still labs, imaging and so on. We’ve heard recently from entrepreneurs that “diagnostics-as-a-service” is a big opportunity for the right team. There isn’t yet a digital front door for Quest and Labcorp. With pharmacy, a physician might ask where a patient wants to pick up their prescription, whether that be a nearby pharmacy or their home. With labs, that flexibility is rarely offered. That might happen with a network-based approach as well as some new innovation to bring testing to the home or even with mobile vans. Lacktman is excited about mobile phlebotomists or “click and mortar'' models for tests that are required to be in person.
The other idea is in payments/e-commerce. For instance, Lacktman is still on the hunt for a real-time benefits checker that can determine what’s covered, what isn’t covered, the price of the co-pay, and so on.
Another gap that’s pervasive is the data interoperability piece. For Alessandra Henderson, CEO of Elektra Health, a company that supports women experiencing menopause, that’s the “pie in the sky” dream that she’s looking for from tech stack vendors. For her company, the eventual goal is to more easily receive and share a member’s prior health information with their past, present and future care team. She’s far from alone in wanting to give users that full view of their health within seconds. It’s a problem the industry has been trying to crack for decades.
Finally, a bigger problem for the category is really getting down to the specifics of the use case. Corey van der Wal, chief product officer and co-founder of Frame Fertility puts it this way: “The challenge for the digital health startup looking for tools is finding the needle in the haystack that actually meets your criteria,” he said. “Sometimes there simply isn’t one.” And for van der Wal, who previously worked as a product leader at athenahealth, that’s an ongoing challenge as companies will need to serve a large enough audience without diluting their value by trying to solve for too many.
Breaking it down by user type
Now that we’ve discussed the specifics of the tech stack and some potential gaps, another way to think about this is to put yourself in the shoes of a specific user. As many of us have learned the hard way, one of the reasons that digital health is so hard is it’s a two-sided network, not unlike ride sharing.
As Tischler puts it: “The clinician is the product to a patient, in the same way that a driver is the product to a rider. The provider and patient experiences are inextricably linked together. If a provider isn’t happy, they can easily power down or switch onto a different platform. If providers aren’t available, you’re going to lose prospects and patients to someone else.”
Arguably, the patient has long been a focus for companies in the space as it’s how they differentiate from incumbents. What’s coming next as digital health companies scale is a much greater level of attention paid to the provider, including the doctor and nurse. Think about how much Lyft and Uber spent annually on retaining its drivers. “I believe it’s inevitable that we’ll see referral / bonus systems in place for virtual health, too, as demand for virtual clinician availability increases,” said Tischler.
There’s other stakeholders too. Dr. Rebecca Mitchell, VP of Product at Teladoc, breaks down into the following groups:
1. Patient/primary user
2. The clinician (internal)
3. The clinician (external)
4. The client
Let’s start with the patient
It can be really challenging to solve for a consistent experience with a largely outsourced stack. How do you create coherence in the user experience between say digital and device monitoring components of a program and what do the clinicians do?
“This is everything from tone, to how the actual tools for interacting with the clinicians fit into the larger experience, to ensuring care plan/clinical consistency, to how you treat user profile/health history elements that a patient directly edits versus medical record elements by the clinician,” Mitchell explained. “This is harder than it sounds, especially if you don't directly employ the clinicians.”
There’s also a problem that needs to be solved around payment. What happens if insurance rejects a claim? What’s the right method to collect co-pays? Depending on the code that gets billed, patients might also be on the hook for unexpected costs that no one is taking the time to educate them on. A good example of this, according to Mitchell, are the remote monitoring companies leveraging CPT codes. “There are a lot of rules for who and how often those codes can be submitted, she said, ”and I haven't really seen any examples of this being coordinated across different providers who might try to bill with this code.”
There also needs to be processes for capturing patient health information from disparate sources, whether that’s devices, diagnostic tests and so on. And founders must also consider how to manage the patient’s care outside of the product, if they need to see a specialist for instance.
And that’s just the start! But in Mitchell’s view, all of these considerations will determine how to think about the tech stack, as well as the all-important “build or buy” decision.
The clinicians (internal and external)
For tech-enabled services companies, it is critical to hire and retain high quality clinicians who are good at providing care virtually. That could be one of the biggest “differentiators,” according to Mitchell. And that’s only going to get more challenging for companies given how Covid has accelerated current provider shortages. “Do not underinvest in the experience of your internal clinicians,” she stresses.
So internal docs and nurses will want clinical applications to allow them to do the fundamentals of medicine first and foremost. That involves reviewing and synthesizing information to make a treatment plan, order or prescribe, and make referrals for patients consistent with their health plan coverage. This clinical team will also require tools for billing and RCM; tools for quality management; for communicating with the patients as well as for internal clinical ops management.
Mitchell notes that other teams like client support or user support will open up new responsibilities to stay compliant. So don’t forget that!
Some companies will also choose to bring in clinicians via ‘on-demand’ staffing services like Wheel, OpenLoop and SteadyMD. Digital health providers will need to think about those users too & tools to support them could vary based on the clinical use case/patient population/density of the user-base in a specific region.
Depending on who the client is, providing virtual care might unlock a lot of different opportunities for reporting. That includes both virtual care usage, referring based on preferred networks and clinical quality, notes Mitchell. That presents big opportunities to add value but Mitchell warns that spending too many cycles here might eat up an entire product team’s roadmap.
A parting thought
As I asked founders how they were navigating these uncharted waters, I got a mix of responses. But I think van der Wal from Frame Fertility summed it up best by boiling it down to two questions:
1) Is this my core capability?
2) Is the effort versus impact worth it?
If it’s a core capability, he said, he would want to build it himself to be able to iterate at his own pace based on his company’s requirements.
“Ideally, a tool saves me time or money,” he explained. “If it doesn’t, then I have to weigh the effort it will take to implement versus the impact it will have. If the potential impact isn’t clearly greater than the effort, the tie goes to building it because building it enables more control.”
As a reminder, we are always looking for your thoughts and feedback. So give us a shout if you have perspectives on this topic at @chrissyfarr and @MSchlueterMoore on Twitter.