No need to beat around the bush. Healthcare AI is succeeding without asking much of healthcare to change.

AI can make expertise and knowledge work dramatically cheaper, along with the coordination and execution required to put them to use. It changes what people and organizations can do, how they interact, and where scale still matters. In healthcare, that should reopen basic questions about the experience, work, institutions, intermediaries, and markets built around older constraints.

Instead, much of healthcare AI is being absorbed by existing arrangements. These products can create real value while healthcare remains expensive, difficult to access, and remarkably hostile to the people trying to use it. A generational reduction in the cost of knowledge work is making healthcare’s accumulated machinery cheaper to operate before we ask how much of it should still exist.

Efficiency is not structurally neutral. Making a bad structure cheaper to operate can make it more durable.

AI is early. Policy, consumer behavior, or further technical progress may still force deeper change. But the current trajectory already shows where attention is going and whose definition of value governs what gets built.

A perfectly rational place to stop

Companies looking for product-market fit usually start with the immediately apparent opportunity: work an identifiable buyer already pays for that AI can perform better.

Healthcare is full of work with a recognizable owner, an existing budget, and a measurable cost. A health system or payer looks for problems it recognizes, value it can capture, and changes it can absorb.

When a product finds traction, it proves the buyer, budget, and sales motion. Investors can underwrite the category, customers ask for variants, and more founders pursue the same terrain. The market gets better at finding more of what has already worked.

The cycle repeats.

Structural change rarely arrives as a customer request. It may ask whether the work, boundary, intermediary, or unit of purchase should exist at all. Its value may be spread across several parties, or it may require several institutions to change at once. An incumbent can justify a tool that improves its position; it has much less reason to fund one that weakens its revenue, control, or role.

And sometimes there is no deeper explanation. The immediate application is already enough to support an enormous company. Why divert the course?

Consider the provider-payer AI arms race. Providers use AI for coding and authorization requests; payers use it for utilization management and claims review. Each investment makes sense on its own. Taken together, they make both sides more capable inside the same adversarial arrangement.

No one has to be stupid, greedy, or unimaginative for this to happen. Rational actors can produce an unambitious whole.

One more turn

A company that hopes to change the structure later has to carry that intention from the jump. Its first buyers / users shape what it builds, how it measures success, and what it gets paid to do. Those choices compound, and the company can become steadily better at serving the arrangement it once hoped to change.

For that intention to become direction, the search might push past the immediate opportunity and into second-order implications. How behavior and institutions adapt, what becomes possible in combination, where constraints move, who gains or loses power. Tracing them might lead a company toward an aim larger than the sum of customer requests. (Note: not exclusive to AI.)

The boundaries of obvious shift with time. Ambient clinical documentation was once a non-obvious, second-order implication of AI. Now it is an established category, and the inquiry has to move again.

Consistent curiosity produces a lot of possible directions. Most go nowhere. A few get more interesting the longer you follow them, and staying with one long enough to find out whether the larger idea is real takes persistence. Occasionally, one survives the scrutiny and becomes the basis of a breakaway business.

Some questions that I come back to, non-exhaustive:

• As expertise, information, coordination, and execution become cheap, which work should disappear? Where does scarcity move next?

• Which functions still require institutional scale? What can individuals, independent providers, and other dispersed actors now do without joining or becoming a large institution?

• Where does responsibility end today? If a problem can be followed across organizations, is that still the right place for it to stop?

• Which advantages built on opacity, fragmentation, switching friction, or administrative scale weaken? What new competition—or new gatekeeping—forms?

• What defined problems of today can now be packaged as a larger superset, delivered and measured as one unit? Who should buy it, and who captures the value?

Patients currently learn the healthcare market one at a time. If that knowledge compounded and could be acted on collectively, fragmented demand might gain market power. Independent practices pushed toward consolidation for administrative scale could gain comparable capacity without joining a health system.

Intermediaries whose role depends on opacity or poor execution may disappear, while transactions too bespoke to clear today could support new brokers and clearinghouses. Always-on observation may surface more need than clinicians can absorb. Scarcity moves to prioritization, and whoever sets the escalation thresholds gains power.

Responsibility now often ends at institutional boundaries. An organization capable of operating across them could own the neglected work and be paid to see a problem through. Value scattered across several parties could be repackaged into a larger result with a natural buyer.

This is a crude, incomplete list. The point is to keep re-litigating assumptions against what has become possible, and to find shots worth taking, in service of a better system.

So far, we are mostly making the system we inherited better at being itself. We don’t have to squander the opportunity.

This is Part 2. Check out Part 1 here.

TL;DR

• LLMs are uniquely good at redaction and can significantly de-risk these approaches from any potential PHI handling that might occur.

• LLMs also can do redaction at small sizes, creating potential opportunities with client-side redaction.

• There are many novel workflows to explore now that could be potentially viable for PLG. There are still other limitations and headwinds with going bottoms up, but redaction LLMs can provide compliance for real-time end user workflows.

It’s redundant, but to be explicit: patient privacy is paramount, and HIPAA despite its flaws requires compliance at the risk of fines, data breaches, loss of reputation, and loss of business. The goal here is not to evangelize circumventing HIPAA and parade around PHI willy-nilly, but rather to surface swim lanes where one can find more straightforward paths to providing lift to overlooked end users of healthcare.

And these end users need lift, yesterday, and a lot of it. So much so that the utilization of non-compliant shadow IT is growing by the day. The OCR may be sporadic in its investigation cadence (which likely leads to this growing utilization), but still the risk of audit & fine stands. The threat of data breach remains persistent all the same. Fines and breaches can dramatically impact large Covered Entities; they’ll level any software new entrant.

There have been clinician-facing tools that simply request user consent to exclude PHI in inputs (DocsGPT by Doximity, Glass Health) - while this seems to drive adoption for now, one can see the ceiling in this method.

Perhaps burdens get heavy enough to instigate relevant regulatory change, but until then: if you’re building products that has the potential to interface with PHI, even if you take any of the previously mentioned approaches, you run the risk of handling PHI and the subsequent potential consequences. So I guess it’s all for nought and you’ve read all this for nothing.

Not so. Enter…

The LLM Healthcare Unlock - PHI Redaction

De-identification Context

PHI redaction and de-identification of patient data are certainly not new concepts. To provide a quick recap, there are two methods:

• Expert Determination: It’s in the name - an expert determines, using accepted statistical and scientific principles, that there is a very low risk that the information could be used to identify an individual, either alone or when combined with other available information. The expert must document the methods and analysis that support this determination.

• Safe Harbor: If by removing 18 types of identifiers throughout the entirety of the data in question, and one cannot use the resulting information (or combine it with other information) to identify an individual, then the data is considered de-identified.

Important context around de-identification and HIPAA:

• Once de-identified, the data is no longer subject to the Privacy Rule, and can be essentially “freely” exchanged, outside of the restrictions that HIPAA mandates.

• Only Covered Entities can “freely” de-identify PHI, and third parties that de-identify data on the behalf of Covered Entities must sign BAAs (as they must first willfully obtain PHI to de-identify it).

• If a cloud service provider encounters PHI, and so returns or destroys (e.g. de-identification) PHI, it is no longer considered a business associate.

• Re-identifying the data returns it to the protection of the Privacy Rule.

Putting de-identification into effect has high barriers to entry. An organization will require dedicated resourcing to deploy HIPAA-compliant, production grade redaction of their data. There’s often little reason to have this in-house. With breaches being costly (~$10M on average) and frequent, dedicating months to building (and more to maintenance) across data + engineering + compliance experts for something that isn’t revenue generative or a domain of expertise is a tough pill to swallow. If you turn to utilizing de-identification vendors, many offerings here are expensive and services-heavy, and turnarounds can range from days to weeks. It’s not unheard of for some vendors to go overboard in de-identification, rendering the data useless.

You could alternatively group patient profiles (aggregation) to de-identify and access aggregate-level insights and trends. This of course removes access to any individual-level granularity.

This has limited de-identification to use cases that make sense - commonly, de-identification via batch processing or patient profile aggregation enables macro-level research and insights for large provider organizations, policy makers, payers, pharma / CROs, life sciences and academia. AKA customers with large pockets whose data needs can tolerate slight delay. Building products for end users at Covered Entities (often smaller wallets), whose real time workflows need unimpeded, real time data flow for adoption to be viable, becomes essentially a no-go against the established path of building for enterprise and signing BAAs.

Meanwhile, end-to-end encryption can only enable safe data transfer; the entirety of the data is relatively useless for any additional processing while encrypted.

Now, back to everybody’s favorite new hype cycle, the LLM.

Democratization of Real-Time PHI Redaction

As you might know by now, LLMs are incredible at processing and analyzing inputs (text-based in particular) and converting said inputs into outputs of a desired format, in real time. This is directly applicable to de-identification with Safe Harbor - feed a properly trained LLM clinical data, and it can identify, redact, obfuscate PHI with incredible accuracy (definition below), while also enabling the remaining information to be processed unlike encryption.

To paint the picture: in a 2008 study, the poorest performing clinician annotator only achieved 63% accuracy; state-of-the-art models can achieve high 90s, with open source ClinicalBERT at 97.4% and OpenAI’s GPT-4 achieving over 99% off the shelf through strong prompting. If real world drift and bias can be monitored and accounted for, this seems to be production grade redaction at the ready. (Some additional thoughts by the ScienceIO team here).

One might observe that OpenAI’s BAA process has been reported to be challenging, and additional fine-tuning of open source models may be difficult in scope. Fortunately, there are already commercially available solutions, some of them even offering real-time APIs (Science.io’s Redact, Private AI, Gretel, potentially Mendel AI, more). Anthropic also recently announced HIPAA compliance and will sign BAAs.

This in and of itself isn’t groundbreaking; rather, it’s well known. However, it does have three implications.

1. As a reminder, while you’re handling PHI (including de-identifying it for further use), you are a business associate and require a BAA with your client. It’s inadvisable to go to market direct to clinician with willful neglect of HIPAA in hopes that the ephemeral presence of PHI prior to redaction is inconsequential. If audited, this will likely will not hold up in court, and it might get you banned by CISOs who deem the practice untrustworthy.

   Real-time redaction does however remove risk for the product developer. If clinicians were to accidentally or deliberately input PHI without a BAA in place, with intentional messaging and safeguards, real-time redaction removes the business associate relationship and the potential of audit & fine. More importantly, the risk of actual breach is greatly mitigated.

   For cloud-based tools that sit close to PHI workflows, this enables a more aggressive go-to-market.

2. If the redaction can occur on the Covered Entity’s servers / compute, subsequent information exchange can be compliant. Locally hosting ClinicalBERT requires not much more than 1 GB. A product that can leverage this while working with the limitations of hybrid or primarily on-prem architecture / local storage + compute can be downloaded and utilized by end users without violation of HIPAA. The barriers now lie on the prospect Covered Entity’s internal policies and utilization of endpoint management / access management systems or similar (Okta, Jamf, Windows Software Center, Citrix, Lookout, etc.). Though large health systems and payers will have stringent policies and systems, you could seek out smaller orgs with softer policies.

3. This might also unlock usage of some vendors who don’t yet offer BAAs or whose HIPAA-compliant tier is too costly. You will still likely want to obtain a BAA to CYA, or in the least be hyper-scrutinizing about how your de-identified data will be processed in the vendor of choice.

To summarize:

1. The Privacy Rule of HIPAA restricts exchange of data that contains PHI. Products & organizations that handle PHI must sign BAAs with Covered Entity clients.

2. If data with PHI is de-identified, the Privacy Rule no longer applies. Cloud service providers are no longer Business Associates and are safe from audits.

3. De-identification methods historically have been expensive and slow (often batch processing), and vary in accuracy.

4. LLMs enable highly accurate real-time redaction. Open source and commercial options are widely available today. Open source models can be locally installed for around 1 GB.

5. By increasing the universe of workflows that can meaningfully utilize PHI redaction, LLMs have increased the universe of products that can be securely deployed to end users at minimal risk to the product developer. The upper customer bound is with IT policies that restrict installation of local-hosted apps, which will be larger organizations - still, this is a new ceiling.

These points in sum, combined with the approaches from part 1, expose new possibilities for bottoms-up GTMs while staying compliant, compressing/circumventing BAA processes and maintaining patient privacy.

Reattribution

Nabla, one of the new entrant ambient clinical documentation companies (read more about this here), recently released this blog post regarding their approach to privacy and security. Particularly interesting to me was the detailed overview of their data flow. Nabla does the following:

1. Generates a transcript of conversation; this transcript is only stored locally (i.e. the physician’s computer)

2. Obfuscates PHI in transcript through pseudonymization; the pseudonym correspondence table is very temporarily stored.

3. Computing operations are performed on the redacted data to create a note

4. Re-attributes the note output with PHI through the temporarily stored pseudonym table; this output is only stored locally.

5. Deletes pseudonym table after each query.

Nabla’s data service is stateless, or at least almost so - save for the temporarily stored pseudonym correspondence table that is deleted upon each query, no inbound or outbound data is stored.

It seems to be the case that a couple organizations (Nabla, Athelas, many else) are utilizing this to go broadly direct-to-clinician (clinicians can utilize these products in live settings today on the free tier). I’m not yet perfectly convinced this is an infallible approach - as there are (brief) points in time where identifiers are processed and stored, this seems to operate in a gray area. It could be the case that moving redaction upstream to local servers would provide clearer compliance.

Bull case: the current approach drives fast adoption while avoiding breaches, leading to leverage and successful bottoms-up towards gaining market share. Bear case: I can see it draw scrutiny from CISOs with low risk tolerance, or competitors may file a complaint to the OCR.

I’ll be interested in seeing how the industry responds. To me, what this approach unlocks is preservation of context and output fidelity upon redaction.

However, assuming this approach can get buy-in, this further increases that universe of de-risked end user deployment. Real-time redaction (particularly on client side) de-risks workflows which have inputs with PHI. Real-time redaction combined with re-attribution + deletion also de-risks workflows which require PHI in outputs.

Utilizing this within one of the swim lanes from Part 1 creates intriguing PLG implications.

Which PHI-utilizing problem domains might now be viable for PLG?

We can break down this question with two factors: the input and the output of the workflow.

• Do the inputs of the workflow often possess PHI?

• Does the output of the workflow require PHI to sufficiently drive adoption?

In the previous era where de-identification via batch processing or aggregation were the only viable methods, the world of healthcare workflows looked like this.

In green are workflows that could be deployed to end users without signing BAAs up until recent. Below are examples.

• Batch processing / data aggregation enabled workflows

  • Clinical trial design

  • Population health insights and analytics (ex. Komodo Health)

  • Academic research

  • Drug development

  • Market research

  • Model development

• Real-time workflows that don’t require PHI in their input or output

  • General medical knowledge referencing (ex. UpToDate, Epocrates)

  • Clinician networking (ex. Doximity)

  • Navigation services

  • Referencing (ex. Medications - GoodRx); Price - Turquoise)

  • Back office services and practice management

The red colored nodes are restricted in adoption until BAAs are signed.

With the democratization of client-side real-time redaction by LLMs, the world of healthcare workflows starts to look like this.

Newly green are real-time workflows that utilize PHI in inputs, but do not require PHI in output. Through real-time client-side redaction and perhaps some feature adjustments, products that are solving for these types of workflows could utilize a product-led, direct-to-end-user motion if relevant.

Some examples:

• In-context information reference and analysis (ex. Atropos, Glass Health, AvoMD)

• Care plan generation (ex. Glass Health, Oler Health)

• Medication reconciliation

• Advanced CDS / symptom checking / event monitoring (ex. Bayesian Health)

• Non-PHI workflow triggering

With Nabla’s approach of de-identification + reattribution + deletion, combined with moving redaction to local servers, workflows internal to the Covered Entity that require PHI in outputs are viable as well. These are marked in yellow, and some examples below:

• Info summarization (ex. Hona)

• Documentation (ex. Nabla, Ambience, Abridge)

• Coding attribution and audit (ex. Fathom, Nym Health)

• Prior authorizations (ex. Develop Health, Latent Health)

• Internal task management

• Querying of patient information

The limitations here lie in the amount of compute that might need to be packaged with the client-side redaction app to ensure strong experience, and organizational policy and security systems. On HIPAA, it seems that this approach is compliant - please validate this with your legal advisor.

It’s critical to note that any workflows that require transaction of PHI with an external third-party will still run into HIPAA walls, and locally-installed applications will still need to be assessed by more stringent provider organizations.

What does this all mean?

In an industry marred with high costs, burnout, and poor experience, empowering the end user to be in the driver’s seat for product decisions might provide course correction. Furthermore, product-led growth motions can be a path forward for new entrants fighting uphill battles against incumbent organizations (more on these dynamics in my collaboration with Blake Madden and Paulius Mui).

If I had to guess, incumbents will likely continue to lean on their existing distribution networks - why wouldn’t they? To them, a product-led motion is unfamiliar and risky, and betting on ossification of user behavior and switching costs around their distributed products is a safer path. The added delta of cloud costs for redacting every query they process will be a tough pill to swallow as well.

Meanwhile, demand for improved workflows (for clinicians in particular) is increasing by the day. For instance, some physicians are paying for ambient documentation tools to the tune of $1500 per month out-of-pocket - producer surplus to be optimized.

To rehash some bits from my previous post - in creating defensibility against incumbents targeting your product vertical, you’ll probably want to target:

1. Strong UX focus and product execution - for AI products, this could look like AI-native, interlinked workflows such as downstream task triggering

2. Collaborative feature sets that create product network effects away from the legacy system

Not all workflows will be good fits. If the problem in question isn’t an end user (or end user team) problem, there’s less value in pursuing the motion. However, I think solutions for specific pain points can serve as excellent product-led wedges that can expand towards larger enterprise offerings. Finding overlooked problems that might have deeper product implications than they do on the surface can be key. With real-time redaction, one can strive beyond “product-led excitement” and gain enterprise leverage with actual utilization metrics.

There’s a ton of thread to be pulled at here. Some additional areas I’m curious about:

• Can the current approach of organizations like Nabla and Athelas survive the long run, or are there alternative approaches that provide the same effect?

• Where are the upper bounds for workflows that can be deployed client-side?

• Having been primarily focused on provider clinical workflow recently, I’m curious what realms within payers, pharmacies, labs, etc. are enabled for PLG (and actually are appropriate for it).

• With direct distribution to clinical end users being viable, is there any novel whitespace for patient-facing problems?

• What are best practices to running PLG motions within healthcare?

• One additional tangent - it’s interesting to imagine the possibilities for a “redaction rails” connector business. Such a business would sign BAAs with all relevant and willing CEs, and provide software vendors with de-identified data and deployment to customers while compressing cycles. I’d imagine a business like Redox (has done hard network building), Science.io (has CE customers utilizing Redact product) or Datavant (de-id network) could be well positioned here.

Share

Thanks for reading, and I’d love to hear your thoughts on any and all of this considering all the fuzzy / gray areas it covers. Feel free to reach out in the comments or at my email | Twitter | LinkedIn.

TL;DR:

• The promise of product-led distribution to end users in healthcare has been largely unfulfilled.

• This is primarily due to HIPAA’s restrictions around PHI exchange, driving Covered Entity organizations towards arduous review cycles.

• There are 5 currently established methods with which one could go direct-to-end-user. Product-Led Excitement; Non-PHI Workflows; Cash Pay Transactions; End User Represents Covered Entity; Patient Authorizations

Subscribe now

PLG Has Stalled In Healthcare

Over the past few years, with product-led-growth (PLG) proving its viability across other industries, there’s been optimistic musing about the potential of the motion within the healthcare space. To provide a quick definition: product-led growth (aka bottoms-up aka B2C2B aka Direct-To-Clinician/End User in the healthcare context) focuses on offering end users of enterprises a self-service version of the overall offering with the ultimate goal of driving enterprise sales - Notion, Slack, Datadog, Calendly, Airtable are all common examples.

Against some of the core problems that plague the industry, PLG seemed to offer a path forward:

• centering distribution around the end user mitigates the principal / agent problem

• historically maligned UX would be met by the iteration pace seen only in consumer-focus companies

• in overly saturated enterprise SaaS markets, high end user utilization metrics from a successful PLG approach drive leverage while avoiding pilot feature bloat common to B2B

• Lower CAC / efficient growth at scale compared to sales-led peers

With the stranglehold that incumbent distribution has on deployment of novel solutions (AI-enabled or not), tightly aligning with end users can also give new entrants opportunities to build brand, buy time, and stay closest to true problems. This is valuable in arenas where advantages are hard to come by and disappearing by the day.

The uptake of the motion companies has been steady for companies that face patients, scientists, developers, and employees. Organizations here have begun to successfully reach maturity or at least prove out the motion. Some examples (Jay Rughani’s post here for further reading):

• Maven or Noom’s initial distribution to patients followed by an employer focus.

• Benchling to scientists and academics

• Many of the healthcare tech stack for developers

• Bold for health plan members

However, for other end users in healthcare (and clinicians in particular), progress has been more stagnant, despite the evident advantages (further outlined in Morgan Cheatham’s great post on Direct-To-Clinician).

Title II of HIPAA (the Privacy Rule) has been the primary point of friction. For any organization creating products for Covered Entities (providers, pharmacists, payers, labs), if they process PHI, they will need to sign a BAA with the authorized executives of their Covered Entity clients. The enterprise stakeholder leapfrogging the end user in priority drastically reduces the potency of any product-led approach. Brendan Keeler said it best here:

So then, in what ways are end user distribution currently viable in healthcare? Where can you utilize such a motion, and how?

Not being a compliance expert, I dove head first into the murky waters of PHI regulation. I can’t say I came out unscathed. Strictly speaking, none of the below is legal advice (if it isn’t evident, I am not a lawyer) but perhaps interesting potential paths to explore in more depth.

This is a two-parter:

1. Paths that you might be able to take to go direct to Covered Entity end users

2. How LLMs might further unlock these paths (and make a new one) for new entrants

Paths to Healthcare PLG

Within the constraints set by the Privacy Rule lie a few paths forward:

1. Product-Led Excitement

2. Non-PHI Workflows

3. Cash Pay Transactions

4. End User Represents Covered Entity

5. Patient Authorizations

Product-Led Excitement

We start with the simplest path: providing end users access to varying layers of visibility and control of the product in non-production settings. Product videos, demo environments (perhaps populated with synthetic data) and similar collateral can serve as end user facing product marketing - do this well, and this “product-led excitement” might trickle up to the relevant stakeholders and move the needle. You might optimize this further by targeting communities of end users.

This approach has a safe but low ceiling. Without utilizing in live environments, users will have limited understanding of product capabilities, and no true adoption is occurring.

Non-PHI Workflows

To go past HIPAA’s restrictions, why not avoid PHI entirely? This has been the dominant way forward for companies taking on the motion, including many from the below market map from Bessemer.

If Covered Entities consist your target market, you have a few ways to directly reach your end user.

Non-PHI Primary Product: Product domains that avoid PHI entirely lean towards the more operational elements of Covered Entity organizations. Administrative (ex. Headway) and financial (Nitra) organizations can build entire product suites; if there’s a sensible wedge towards a self-service offering, it’s relatively smooth sailing here. Many developer-facing products for healthcare tech orgs will fall under this category as well.

PHI Adjacent Product: Naturally, the closer your product is to day-to-day PHI-handling workflow, the more HIPAA will come into consideration. It’s possible to still reside purely in the non-PHI domain (for ex. many reference tools above) - and maximizing value here will create end user stickiness - but deepening value for enterprise clients may mean crossing over into PHI-handling features. This is a sensible point to feature gate, and trigger enterprise sales conversations.

Butterfly Network serves as a great example.

We knew [enterprise health system’s] concerns around security and governance and integration, and so we built a software package for that. That then created the right dynamic, where we actually had a product to sell, which had higher recurring SaaS dollars attached to it. It was a win-win from that perspective: we could solve the clinicians problem of wanting to use it in the hospital, and then we’re driving increased subscription revenue and deepening our tentacles in terms of the integration into the organization and spanning across the organization to do more. - Bill Porter, Butterfly Network

PHI Agnostic Product: To draw the small distinction between this category and the previous: the core build workflow of a PHI agnostic product remains the same when handling PHI vs. not handling PHI; meanwhile, PHI adjacent products reside primarily outside of the PHI domain but likely require net-new product expansions to penetrate into PHI workflows.

An example here would be Awell - task automation can fully function to drive some live results for organizational end users without interfacing with PHI - and they did this to drive a bottoms-up sale.

Ultimately, I think the keys with this approach might be either:

• Finding non-PHI wedges that themselves possess high TAM

• Finding non-PHI wedges that might be lower TAM or conversion probability but can generate traction, combined with agility in expanding into PHI workflows that increase TAM

Cash Pay Transactions and DPC

Fun little quirk - HIPAA strictly applies to electronic transactions of specific information) between parties. An example: a provider organization becomes a Covered Entity upon electronically exchanging PHI with a health insurance company. These transactions are:

1. Health claims or equivalent encounter information.

2. Health care payment and remittance advice.

3. Coordination of benefits.

4. Health care claim status.

5. Enrollment and disenrollment in a health plan.

6. Eligibility for a health plan.

7. Health plan premium payments.

8. Referral certification and authorization.

9. First report of injury.

10. Health claims attachments

For providers that don’t exchange information electronically with insurance (i.e. practices that are strictly cash pay or strictly direct primary care), there exist ways to work around HIPAA restrictions, and BAAs are not required to be signed.

While interesting to note, it’s not yet a compelling area for building. The TAM is low (1762 total DPC practices in the US), and standard HIPAA restrictions will start to apply as soon as you enable third party electronic exchange for the above transactions (essentially all of which are mission critical interactions).

End User Represents Covered Entity

Legally binding BAAs must be signed by legal representatives of the Covered Entity - for ex., at larger provider organizations, CEOs, CISOs, Head of IT. To compress the BAA process, one can look to distribute to organizations where end users are themselves the legal representative, or very close to them. Essentially, sole practitioners and SMBs.

To these individuals, one could provide a simple flow with a standardized BAA, and enable immediate product utilization. Healthie has a version of this currently.

The most notable headwind in this motion is the currently shrinking market (in providers at least) - physician consolidation is steadily increasing, with roughly 7 of 10 physicians employed by hospitals or corporate entities. Moreover, the smaller complexity of problems in outpatient care compared to inpatient makes moving upmarket a formidable task that few have achieved.

However, as tensions rise between physicians and health systems, physician independence is increasingly championed. There’s a case to be made for intentionally targeting this overlooked market, making beautiful tools that alleviate specific burdens for these end users, and finding ways to seize outsized value. This could be tying into patient capture and accelerating it, or innovating within recent primary-care-centric alternative payment models (Making Care Primary; Primary Care First).

Patient Authorizations

Covered Entities can disclose PHI if the patient (or representative of patient) provides a written authorization. This agreement can be drawn up by the third party for signage by the patient. The Covered Entity is no longer liable for the PHI release, and third parties (naturally) can work with PHI-containing data.

In the healthcare tech wild, DTC flows and Meaningful Use Patient Access APIs have been the primary adoption methods of this path, but I’m not sure if there’s been meaningful use (ha!) on the side of CE-facing software, for understandable reasons. In an era of heightening privacy anxiety and suspicion of third party data use (WSJ’s account of Amazon Clinic’s utilization of this path here), there will need to be extra precautions taken to quell both patient and Covered Entity wariness. The slight uncertainty in obtaining authorization could lead to adoption friction, and organizations have instead opted for the tried and truth path of enterprise BAAs. Lastly, the size of universe of realtime workflow products for clinicians that are easy enough to build yet have reasonable TAM has been relatively limited (until now).

However, if you can present the patient authorization flow in thoughtful ways within the workflow of the end user, strive for security in data obtained, and the tool dramatically improves the end user’s working experience, I think this path has potential. It seems that Upheal is utilizing this at time of writing. It’s not hard to imagine a world where clinicians are evangelists of tools that relieve burdens, and patients can buy into experiences that help their clinicians deliver them better care.

A wrinkle to consider here is the obtaining of patient contact information to initiate the patient authorization process. Covered Entity end users cannot directly provide third party software with email, phone numbers, etc. as that is by definition a PHI exchange. As such, the end user will have to facilitate the connection between your product and the patient:

1. Provide a path for the patient to access your product and create an account

2. Provide a patient authorization form with your product as the recipient

Finding ways to streamline this workflow would be key. One interesting example from the more patient-facing side: Ciitizen tapped into trusted patient advocacy organizations as a large node for patients, and was able to obtain patient authorizations through this as well.

Patient Side Value

Further, patient authorization flows can be part of overall patient-facing experiences, which might drive growth loops. Although they don’t utilize their patient apps to unlock clinician side workflows (to my knowledge), Abridge and Elaborate are great examples of orgs that have patient-facing feature sets that can drive patient-clinician word-of-mouth; GoodRx is similar, though more patient-centered as a product. In particular if you’re building for Method 1 described above, creating value props for the patient might be critical to success.

Read Part 2 next week for how LLMs can de-risk these paths for the product organization, and unlock a potential new one.

Share

Thanks for reading, and I’d love to hear your thoughts on any and all of this considering all the fuzzy / gray areas it covers. Feel free to reach out in the comments or at my email | Twitter | LinkedIn.

Below, we cover:

• Context on what’s led up to the current AI zeitgeist

• Opportunity spaces for AI applications in clinical workflow where players are converging, and why

• Segmenting the current space and the competition dynamics between players

• Anticipating how new players can make space for themselves in the space

• A clinician’s perspective on these happenings, by Paulius Mui, MD

• A look at the recent financial and commercial activity in the space, by Blake Madden

This post is hefty - thanks for reading in advance :)

Subscribe now

Broad AI Context

In 2017, the Google Brain team’s landmark paper Attention Is All You Need introduced the Transformer architecture, which drastically improved performance and quality in building large language models (LLMs). What followed was a convergence in approach across ML / AI research, resulting in composability of ideas and compounding of improvements. 2020 saw the OpenAI team publish Scaling Laws for Neural Language Models which led to the emergence of the Scaling Hypothesis - TL;DR: pump model size, data, and compute and you’ll see increasingly sophisticated behavior in AI models. Then of course, a small “research preview” called ChatGPT surfaced the capabilities and potential of LLMs through a consumer-facing interface, and became the fastest growing consumer app of all time (100M MAUs in just two months!). Here’s a fantastic primer on most of the concepts behind the tech.

The resulting Cambrian explosion of AI innovation has occurred at a breakneck pace - so much so that, a few days ago, many notable figures across many disciplines signed an open letter to pause large scale AI experimentation for six months. Everything is happening everywhere, all at once.

Although the healthcare industry has been burned in years past by AI overpromise and the low use case / complexity ratio of task-specific models, the current moment feels like an inflection point for the industry as well, specifically due to the massive leap in performance. For one example: LLMs aren’t just passing the USMLE, they’re crushing it (GPT-4: 86.65%; Google’s Med-PaLM 2: 85% - both roughly the 90th percentile). Commonly stated concerns include hallucinations, equity & bias, and automation bias among others (interesting Chrissy Farr thread below).

My (Ben’s) take: the degree of attention and pace of innovation in the space will eventually render these concerns largely obsolete, although it is certainly likely that new ones will rise. For instance: inaccuracies in mathematical calculations were among the most commonly observed hallucinations - OpenAI’s introduction of third party plug-ins, and Wolfram-Alpha’s development of a plug-in has already dramatically reduced such mathematical hallucinations.

And the industry is responding accordingly. New healthcare-specific thoughts and papers (like this one!). Rapid fire Twitter musings by all who dare to comment. AI-focused roles from prompt engineers to the emergence of the Chief AI Officer in health systems - the latter certainly implies a sense of urgency in health system IT departments. The road to implementation & end user adoption at health systems is crowded with provider app vendors scraping by awaiting CISO legal / security review and integration - the mentioned trend might provide a zipline to the top for some chosen few.

Clinical Workflow in the Age of AI

The focus has predominantly landed on solving the increasingly burning problem of provider supply shortage - this is also where we’ll focus for the bulk of this post. Some more applications here by Eric Topol for your reference.

Here’s a quick(est) recap. Demand for care is slowly growing and will likely accelerate due to many factors including a growing, aging population and likelihood of higher patient acuity. This demand is far outpacing supply - the provider shortage is expected to be at ~57000 FTEs in 2025, and growing year by year.

As the shortage exacerbates, increasing pressure is put on providers to deliver more care with the same resources. Compounding this further are the burdens of administrative tasks placed on the provider - many related to the much-maligned UX of EMRs. This has resulted in high levels of burnout, driving physicians to retire or pivot roles/industries, and - you guessed it - increasing shortage.

In particular, the documentation and prior authorization spaces have received attention from builders, not only due to the severity of their burden on clinicians, but also their positions as key junctions connecting workflows in clinical operations, patient data, and reimbursement. The past ~decade saw a variety of approaches, ranging from RPA to integration-heavy sidecar apps to net new EMRs; you could count AI-enhanced products among these, but all such products required humans-in-the-loop to close the last mile of performance. (More on this later)

The recent leap in performance and quality of generative AI has resulted in renewed attention from many that are drawing the dots between the qualities of these tasks - time intensive, manual, delegable, content heavy - and the strengths of LLMs.

Who’s getting involved?

What’s interesting about what’s happening is the democratized access and capabilities for personas across varying degrees of complexity and healthcare involvement, including:

1. Care team end users using ChatGPT as a means to alleviate their current workflow pain points

2. Care delivery organizations buying AI solutions from vendors creating their own AI-enabled tools for the workflows of their in-house care teams (Curai, which has been AI-focused since inception; tech-enabled orgs - quote below from VP of Prod at Oak Street Health)

3. Net new AI-native SaaS organizations building for care delivery organizations (Abridge, Ambience, Nabla, Ethermed, Glass Health and Fairway from the latest YC batch, more orgs here)

4. Incumbent healthcare SaaS organizations transitioning to selling AI-enabled feature sets (Epic, Nuance, Augmedix, Doximity)

5. Industry-agnostic, large-scale orgs pouring $xxB in healthcare ventures (Microsoft, OpenAI, Google, etc.)

History and Current Dynamics

The documentation space (the most crowded one) has intriguing dynamics that may be translatable to other spaces. Let’s focus first on the healthcare-focused SaaS organizations, which would be categories 3 and 4.

Category 4 contains the more veteran players (namely Nuance, Augmedix; 3M’s M-Modal, also Suki) that have weathered the tumultuous path of healthcare industry distribution and its many challenges - long sales and implementation cycles, bespoke integrations, etc. - with services-heavy models. Serving as primary trailblazers in commercializing novel technologies (in this case, applying ASR + NLP / LLMs towards ambient documentation), these companies often augmented the premature capabilities of their models with humans-in-the-loop - typically cheaper talent such as offshore scribes, medical students.

These hybrid (human + AI) models proved superior to status quo; training and retraining human scribes took time, while dictation software and other AI-enabled competitors proved actually additive to clinician burden by requiring edits. Yet, hybrid models also offset the production cost and scalability benefits of AI, restricting distribution to larger organizations that can afford them.

The holy grail, however, remained full automation (in documentation - real-time, structured output that accounts for the full breadth of care settings) and the venture scale margins that come with it. With continual research elevating AI capabilities, new entrants (Category 3 - Abridge, Ambience) have surfaced who promise just that: near-to-full automation of administrative burdens and restoration of clinician time. A bearish perspective to take here would be to view valuable doctors and clinician end-users as the humans-in-the-loop (as opposed to the cheaper talent deployed by Category 4).

However, the most compelling among Category 3 do boast in-house high pedigree AI/ML talent and data partnerships that enable model fine-tuning (in pursuit of closing the last mile of quality). If we are to assume that the quality of models/products have reached (or will reach) sufficient enablement of full automation, Category 3 now faces the uphill battle of healthcare distribution - an arena where Category 4 orgs sit as advantaged incumbents.

Furthermore, many in Category 3 are beginning to strategize around second order implications of LLMs in healthcare. If the first wave of AI tools eliminates the negatives of clinician workflow, how does one then add positives and empower providers? What are the compelling ways one might compress downstream tasks or utilize real-time data generation towards care advancements? Pulling from the popular example of Github Copilot and its abilities to enhance developer productivity (rather remarkably, as the below chart shows), players here are positioning themselves towards this direction with words like “copilot”, “superpower” in their copy.

As there is yet to be an establishment of standardized benchmarks, it’s tricky to compare existing products beyond anecdotes and marketing language. This will be all the more critical as the space becomes saturated.

Heeding both the general capabilities & hype surrounding recent AI developments and the increasing threat of newer Category 3 organizations, Category 3 veterans have returned back into public eye in recent years, with particular types of activity:

1. Collaborating with Category 5 industry-agnostic tech behemoths through partnerships / acquisitions as a means for augmentation of AI capabilities and channel partnerships towards logo collection (Nuance acquired by Microsoft; Augmedix with Google, 3M collaborating with Amazon)

2. Announcements to release fully automated / generative AI solutions later this year (Nuance injecting GPT-4 into their ambient documentation product DAX Express; Augmedix piloting Augmedix Go, Suki announcement)

How will this play out?

Let’s take a quick step back. AI innovations are happening in almost every sector and function, including software development. Tools like the aforementioned Github Copilot to code creation enablers (Insane open source tool here - AutoGPT) are drastically driving down the marginal cost of building software. Applying this to the industry, small teams (or even one person) can build projects solving for previously labor-intensive healthcare problems in a fraction of the time.

It’s fair to cite history in being bearish towards ML use cases in healthcare reaching real life adoption, with many failing to bridge the implementation gap. The required time & costs of change management during deployment have battered many a challenger.

Still, here are examples of demos for medical coding automation built by one person - though likely currently far from suitable for enterprise at scale, is it that much of a stretch to think that something of that grade, for many AI-appropriate needs, will arise within the coming months given the pace of innovation?

The compounding ease with which one can inject AI into workflows and build applications means that the time window for startups (and Category 3 players, documentation or otherwise) to take advantage of some of the classic items in their favor - speed-to-market vs. Innovator’s Dilemma, etc., becomes compressed.

If one person can stand up a demo in less than a month, Category 4 incumbents for any workflow vertical should relatively quickly establish a new status quo of AI-enabled experience through deployment at scale across their sites without need to grind away at health system distribution and implementation, easily the heaviest bottleneck. Yet, point solutions in Category 4 themselves will be fearful of the foundational platforms and systems of record (namely Epic and its GPT-4 partnership) building such experiences themselves (though Epic’s deployments that aren’t among the ~200 Epic hosted will create additional bottlenecks).

And that doesn’t even begin to mention all the use cases that simply OpenAI + industry plug-ins eventually can and will sufficiently cover.

This begs the question: in this arena, what will the organizations that win look like? Do we chalk up the victory to incumbents and OpenAI and call it a day?

In the nearest term, well-known dynamics will persist. Many care delivery organizations (Category 2), in particular the larger ones with thin margins and strapped for resources, will be driven to address clinician shortage needs by partnering with vendors providing near-term ROI.

For even the small-to-medium care delivery orgs capable of experimenting with AI workflow injection, most will gravitate towards their primary business focuses in a capital-strapped environment, and perform risk-averse AI experiments at most as opposed to re-evaluate their operating model for AI-enabled efficiency improvements.

As such, the TAM of provider org prospects will remain stable, and such prospects will drive towards vendors (3, 4, maybe 5). The battle here will be between how quickly AI-Native Category 3 players can distribute and implement vs how quickly Category 4 can build a “good enough” solution and upsell existing relationships. Point to Category 4 incumbents.

A note on data moats

Currently, proprietary data is commonly cited as a moat, which provides a massive advantage towards incumbents. I’m not yet convinced that it’s unconquerable (someone please correct me) in the longest term. Below are some early observations:

• Outputs from a model can be used to train other’s (including competitors)

• Fine-tuning seems to cost less and require less data

• There’s at least some likelihood that future GPT-X models outperform specialized models

• Increasing data interoperability and data structuring ability in healthcare

Naturally, possession of such data provides accelerated speed-to-market for incumbents in the near term, which buys time for further advancements. Further, there’s still instances where proprietary data will be paramount (biotech for one). However, especially for use cases where a qualified individual (like clinicians) can verify and edit output (and such flow can even be a trust-building component), there might be diminishing marginal returns.

Differentiation

When Category 3 new builders might begin to take advantage is in the medium-term, with a fundamental and annoyingly cliche answer: “good product execution”. How good and fast are you at uniquely identifying the right problems for your customer, and subsequently the right things to build?

Some thoughts here on how AI-native organizations can start to gain market share:

• Agility in idea generation and constant delivery of innovative “AI-native” workflows allow the steady march towards overall differentiation. My (Ben’s) hypothesis: all incumbents will be able to relatively easily develop and deploy “point AI-feature sets”; however, AI-native organizations can build interlinked AI-enabled workflows towards a cohesive AI-enabled product ecosystem. This will be more difficult to replicate and provide an overall more effective experience than incumbents embedding AI into legacy applications. Examples:

  • Point feature sets

    • Real time patient data value unlock

    • AI-induced synthesis of clinical corpus

  • Interlinked workflows

    • AI-induced inter- and intra- product task automation from AI-generated output (Agents?)

• Workflows that work in harmony with existing healthcare systems and processes will be a no brainer. Layering this with workflows/interfaces that can transcend the traditional lock-in of healthcare systems enables flexibility in UX and hopefully sticky end user adoption. Many of Category 3’s gravitation towards browser-based or app-based workflows aligns with this concept.

• On the themes of end users and interlinked workflows, Category 3 players might find it interesting to optimize for product network effects to accelerate sticky end user distribution. An example might be collaborative feature sets that enable care teams to deliver better work more quickly when more users use the product together.

• Despite my earlier disparagement of data as a moat in this realm, constructing data network effects to continually extend data moat could be a compelling way to stay on top of the curve. This often means finding methods to continually elicit contribution from end users, which does seem to interlace well with RLHF (the most popular fine tuning method).

• There’s increasing optimism around product-led growth (PLG) for healthcare SaaS. It’s critical to note PHI concerns (great thread by Brendan Keeler here) for most use cases, but DocsGPT by Doximity and Glass Health achieving some virality across clinician circles show proof of concept - there’s something here. Further, open source efforts are surfacing quality LLMs that can be self-hosted and fine-tuned and enable circumvention of cross-organizational data transactions.

Of course, all this needs to be combined with effective change management especially for larger provider organizations.

From Rohan D’Souza (former CPO at Olive): “One idea might be to create experimentation models within systems to go through full implementations with stakeholder commitment. This will require budget to hold both customer and vendor/partner accountable with a nod to full scale deployment assuming the experiment succeeds. Without this, all we will hear is more AI amazingness in press releases with no actual deployments.”

Long term, constant product execution across the above thoughts will establish 1) brand equity and 2) true end user adoption that translates to an enterprise stickiness surpassing contract-induced “stickiness”. There’s some parallels here to Figma using a new paradigm (browser-based applications) to displace incumbent applications; I think this will be a compelling model for AI-enabled clinical workflow SaaS orgs to take after. Bessemer predicts that AI-Native companies can reach $1B ARR 50% faster than cloud counterparts - in healthcare SaaS, the above outlined thoughts might provide one such path.

Value Shifts and Parting Notes

Lastly, the eventual ubiquity of high-powered LLMs has us look to Clayton Christensen’s The Law of Conservation of Attractive Profits: “The law states that when modularity and commoditization cause attractive profits to disappear at one stage in the value chain, the opportunity to earn attractive profits with proprietary products will usually emerge at an adjacent stage.” If AI deployments achieve massive scale and many clinical workflow tools become commoditized, where then does value materialize in such a case?

Taking to the law, some early hypotheses to where value might accrue:

• Applications that enable data unlock for any care delivery organization that seeks in-house AI deployment - ScienceIO, Truveta

• “Verticalized” care delivery organizations that embed AI in their life blood across operational & clinical steps to reduce overhead + increase value, competing in selling care delivery as opposed to selling AI - Curai seems to be AI-native and is intriguing to observe from this lens.

• Enablement models (a la Passion Economy) that address administrative pain points to create and aggregate outsized capture of patient value - AI-enabled MSOs / ACOs?

If it isn’t evident through all of the above, I personally am tremendously excited about what’s happening and what’s to come. I’d love to chat if you are as well (or even if you’re a skeptic!) - feel free to drop me a note at btlee215@gmail.com. Special thanks to Rohan D’Souza, Samy Danesh, Samir Unni, Patrick van Nieuwenhuizen, Rik Renard, Ian Shakil, Alex Barbet, and many others for insights, perspectives, etc. on this post (not endorsements!). And of course, thanks in particular to Paulius and Blake :)

Share Quite A Few Claims

A Clinician’s Perspective

by Paulius Mui, MD

X = Primary Care

Solving for X

By Paulius Mui, MD

Prior to the release of ChatGPT, we published a primer on AI in primary care. The developments in AI since then continue to outpace everyone’s ability to keep up with them. For AI to truly disrupt clinical workflows, the advancements in tech need to be coupled with advancements in payment models. For now, AI tools are limited to tweaking components of the imperfect system. Nonetheless, we’re excited for what’s on the horizon.

Bespoke experiences are becoming the status quo. AI intake and triage tools will relieve access bottlenecks for low acuity, high volume needs in personalized fashion. This will redefine the “front door” experience for all stakeholders. Launch of services like Azure Health Bot lowers the barriers to train own AI models from trusted documents and deliver the kind of responses patients would expect from a doctor who knows them well. To get this right, optimizing reinforcement learning from human feedback will be an essential new clinical workflow for care teams. Scaling data annotation will be a huge unmet need and companies like Centaur Labs will become increasingly relevant. 

Patient engagement and data analysis will never look the same. Developments like HealthGPT showcase the potential of integrating personal health history with state of the art AI bots. As privacy and accuracy kinks get ironed out over time, insights into strategies for preventive health and chronic disease maintenance will stem from data across all domains of patient lives: financial, behavioral, and clinical. Primary care teams are uniquely positioned to partner with patients in managing such data assets because of the inherent nature of enduring relationships. Surfacing data intelligence will be a much easier lift without the need for the same depth of resources as it will be simply “conversational.”

There is much debate whether AI tools will facilitate health systems in entrenching their power or will it empower private practice clinicians to break away from incumbent tech to innovate? Standing up a tech forward practice from the ground up has never been simpler. When coupled with tools like Ease and expert full stack dev shops like Light-it, independent practices can deploy enterprise-grade tech at the community level. 

It’s evident that low-hanging fruit for AI to tackle is the booming clinical documentation market. “Communication is documentation” concept pioneered in virtual primary care will become the new standard in traditional care settings. These tools will shift the roles of clinicians from content creators to content editors. Will this make healthcare more human? Finding signal among documentation noise is already a real pain point with modern EMRs that may be exacerbated during the transition to a fully ambient documentation paradigm. We anticipate that winners in this space will be those who identify the most appropriate use cases for AI augmentation of the documentation experience based on deep clinical expertise. Leaning into the guidance of clinical advisors will be a key differentiator for AI startups because there is a large disconnect between clinicians and builders. Here are some of the companies whose work we’re following in this space.

Diving into Ambient Documentation Financials 

by Blake Madden

As Ben notes, ambient documentation solves a pressing need for health systems - whose relationships and contracts are extremely sticky.

Ambient documentation offers a sweet, sweet value prop for hospitals in 2023 when labor and margin pressures are top of mind:

• Quantifiable ROI to address clinician burnout (a $4.6B annual cost),

• A boost in clinician satisfaction - physicians love the product,

• An increase in patient engagement and medication adherence (hitting on consumerism trends and patient-facing communication/applications),

• Higher productivity (e.g., increase in visit volumes), and

• Optimized documentation for billing and EHR purposes.

For example, Augmedix (which is a public company and has data available) estimates that its solution saves doctors 2-3 hours a day.

Health system CFOs can do the math here. Between boosted physician productivity and burnout prevention/retention, seems like a pretty clear win-win-win for patient-provider-admin.

Just ask HCA Healthcare, the largest health system in the US that announced a partnership and investment in Augmedix today (4/20), providing a $12M capital infusion into the ambient player and giving the entire space even more investment credibility.

The pilot with HCA in two EDs will help Augmedix optimize the product for acute care settings. From the release:

Schlosser adds the pilot testing is an example of HCA Healthcare’s strategy of working closely with clinicians to design and integrate innovative technology into patient care. “This Augmedix proof-of-concept is not a complete solution, which is by design. We are capturing direct feedback from and ongoing collaboration with physicians and nurses to mold and develop the solution to meet their needs.” During the pilot, HCA Healthcare plans to gather data – including critical physician, nurse, and patient feedback – to determine the potential for use at additional HCA Healthcare hospitals.

So, it’s pretty easy to see why we’ve seen so much fundraising activity and new players enter the AI-enabled clinical workflow space. Some recent, non-exhaustive examples, total funding included:

As far as the financial and business case, investors should like and pay attention to the AI scribing space. It’s one of the few areas in healthcare with a more scalable offering and SaaS-esque margins to boot.

Over time, the investment thesis Augmedix is presenting as a public company is intact. Its gross margin since 2019 has expanded, implying scalability. Moving forward, when vetting these companies financially, take a look at gross margins as a proxy for automation and operating leverage, cash burn, and ARR growth as a proxy for clinician growth:

There’s a clear distinction here between players who provide hybrid offerings - with trained human medical scribes - versus pure-play, automated AI offerings.

Keep in mind that Augmedix operates the full spectrum of products here and its main ones, Notes and Live, still have a human component involved - its trained overseas medical scribe teams.

If new, fully automated products are actually worth a sniff (and without being a technical expert they seem to be trending in that direction), then gross margins would look rosier. I imagine that point is plastered in the decks of Ambience and Abridge, attracting the likes of Optum, a16z, OpenAI, Bessemer, Martin Ventures and other notables. The demo’s of new AI solutions are amazing. Like magic.

As these new automated offerings come online, it’ll be interesting to see whether they (1) truly are pure AI, automated and if so, (2) whether they cannibalize the current hybrid business models that have dominated the space recently (given that pricing is much lower for pure automated solutions).

As noted earlier in this piece, Augmedix thinks there’s still work to be done on the pure AI angle:

LLMs are particularly useful in providing rapid summaries and context from transcripts. However, use of LLMs alone today cannot produce medical notes that meet the [ accuracy ] standards or data structure requirements of our industry. A summary of the physician-patient encounter in the same sequence that the conversation unfolds and which contains some of the inconsistencies that invariably crop up in such conversations is not what the industry wants or needs. That is why we use other technologies together with LLMs to deliver highly accurate and structured medical notes and is a major reason why our note quality is considered so high by our customers. - Augmedix Q4 2022 earnings call, emphasis added

Kinda spicy take! It makes sense why the firm would make these comments given the highly specialized, trained medical scribes it employs in Bangladesh and other areas. To add to this, based on conversations I’ve had, in 2023, a pure AI solution will still require some form of human clean up. That gives the firms with the trained human medical scribe base the edge.

For now, everything I’ve seen points to AI as an augmenting tool rather than a full-fledged human replacement. Given the rate of change with AI lately though, I might be eating these words in like…a week.

Pure AI players will argue that the space is headed in this direction - that they’re more flexible, can compete on price, and differentiate by doubling down on AI-driven models. This angle makes sense - even Nuance and Augmedix are releasing and/or will release fully automated, software only products (DAX Express and Augmedix Go).

How Health Systems Vet Documentation Solutions

Given the human component involved, though, health systems are more likely to view existing hybrid solutions (Augmedix, Nuance) with a better sense of familiarity and comfortability. This dynamic is why Nuance and Augmedix dominate the enterprise space.

Leaders at health systems know how human medical scribes work. They don’t yet know how an AI-driven solution might fit in. It’s risky, and we all know how slow-moving these institutions are. Will it create compliance issues? Security issues? Is my hospital’s data at risk? All of these are likely addressable, but are additional hoops that AI-forward companies need to consider and deal with.

Here are some other considerations for documentation solutions from a health system perspective:

• Winning the contract is one thing; clinician change management is another. Physicians need to use these fantastic, time-saving tools - but many MDs are set in their ways and set aside time for documentation. Plus, lots of physicians prefer to work with human scribes. Perhaps there are ways to incentivize physicians to use new documentation tools, but adoption could prove a challenge.

• If the documentation is wrong or the code is wrong, then a human will need to spend time to fix it. Poor capturing could lead to a higher increase in denials from payors, which is a big issue.

• If these tools make clinicians 20%+ more efficient with their time, how will that truly translate in day to day clinical time? If physicians see even MORE patients in a day, could that still lead to productivity burnout? Health systems love the idea (and get sold on) of higher productivity and more visits in a day, but likely still need to tread carefully.

• Many health systems already view their employed medical scribes as a sunk cost - they have them, they trained them, and they might not want to get rid of them.

• In general, whether valid or not, generative AI creates a perceived security concern for hospital legal & compliance teams.

Inside the Health System bidding wars

As far as actual partnerships penned, the health system arms race is heating up. Like I noted, Nuance and Augmedix tend to be the two heavyweights when it comes to enterprise solutions, with some recent activity from Abridge:

• Nuance partnered with several large health systems recently including Cooper University Health Care, Providence, SCL Health, and WellSpan among others. Nuance also holds the ‘brand’ edge given its affiliation with Microsoft, OpenAI, and ChatGPT. Anecdotally, I hear it’s also the highest-priced vendor.

• Augmedix works with health systems including Sutter, Dignity, and UCSF and now HCA.

• Abridge recently won a partnership with The University of Kansas Health System

Health systems are looking for two primary things, which echoes HCA’s Sam Hazen’s commentary from my ViVE post - does the solution provide clear ROI to my organization? Show me the money. And…is this solution scalable across my entire organization? How does it integrate across different hospitals, specialties, and departments?

Pay attention to healthcare cloud dynamics and the major players trying to sell into health systems as well. They want to create compelling product offerings with the potential to cross-sell into health system clients.

• Azure is aligned with Microsoft-owned Nuance.

• Augmedix holds a close relationship with Google.

• Announced yesterday (yes, we have amazing timing), Amazon and AWS are now working with 3M’s M*Modal.

I’ve heard from industry personnel that Nuance and Microsoft are bidding to win health system contracts by cross-selling cloud capabilities. I imagine this pricing strategy grows more common in order to enter those sticky health system relationships.

Documentation is an initial first wedge product into health systems. From there, there are plenty of other products and solutions to expand into, and I’ll be excited to see how AI-driven offerings change the game in the crusty ‘old hospital landscape.

If those dimensions from Part 1 weren’t difficult enough to build for, there’s more!

To build an aggregate of relationships, the core approach has to consider longer-term scale: efficiently earning patient adoption & engagement, and aligning patient incentives with internal business outcomes. For NewCos, these have to be deeply considered in the early stages of building - retrofitting new structures will not only be significantly challenging but has the chance of non-trivially lowering the utility of your offering. 

Efficient patient trust acquisition 

Much has been said about the tension between healthcare distribution paths and their ability to engage patients. 

A quick recap: If you sell to incumbents (B2B2C), just one new contract results in a new mass of patients under management, but with that comes the principal-agent problem and its sub-problems: 1) customer acquisition does not equate to patient enrollment and actual utilization, and 2) deprioritization of the patient (as business decisions are driven by the dollar). It’s also well known that these customers themselves have become weary of solutions with low engagement. If the standard B2B route is the one taken, these must stay top-of-mind; otherwise, patient relationships will erode, and subsequently customer relationships. 

Philosophically, the strength of DTC / B2C2B approaches lies in their ability to circumvent these issues due to their truer alignment. Of course, this means that the challenges of patient enrollment are pushed more to the forefront, resulting in high CAC. There’s also no guarantee of continued engagement, only the hope of an easier path to it. 

Whichever GTM path you consider, patient adoption/acquisition and engagement stay paramount (of course). Standard marketing channels and other table stakes growth strategies will continue to be critical skillsets, but to maximize capture, you again can look to opportunities that provide outsized advantage in the acquisition of trust - the advantages being high average “trust size” creating higher conversion and engagement, and/or one-to-many nodes. Some of these might be:

• People patients (already) know: Essentially, the “embedded relationships” sub-bullet within the “duration” dimension. Working through individuals who are themselves incentivized to acquire, engage, and/or benefit patients introduces a non-linear trust aggregation pattern. Organizations like Spark Advisors working through independent brokers or DPC / ACO Reach companies like Taro Health and Pearl Health are investing in such relationships.

  • Perhaps there might be interesting plays here to utilize patient influencers, if handled with clinical scrutiny. Within WellTheory’s founding story, founder Ellen Rudolph shared her experiences with autoimmune disease on TikTok resulting in 1.6M views - serving as proof of unmet need and partly catalyzing the eventual founding. Generally, demand for content authenticity is rising, resulting in mediums like TikTok becoming an increasingly powerful organic growth channel. Enabling authentic content creates obvious utility in the short term (patient finds immediate solutions or empathy to their problems) while also accelerating duration through building a human connection. This seems to be a powerful technique towards defensible trust aggregation that’s largely underexplored in healthcare currently. Also a powerful technique towards increasing my weekly screen time 500%.

• Places patients already are: Essentially, the “proximity to routine non-healthcare interactions” sub-bullet within the “frequency” dimension. Meeting patients where they are, commonly through their homes/neighborhoods/ communities, reduces acquisition friction, while also revealing more about the health environment. Fabric Health and Live Chair Health are popular recent examples for their unique identifications of powerful community health locations. Recent rural care models such as Hopscotch Health strive to emphasize not only connecting with the rural communities they serve, but actively building community relationships.

• Patient-to-patient interactions: There’s similarity here to “People patients already know” - patients themselves can be one-to-many nodes for patient acquisition, and provide arguably the deepest baked-in trust. Charlie Health, Groups Recover Together and orgs/domains that naturally support patient groups (ex. Alcoholics Anonymous) lend particularly to patient referral processes as patients become evangelists. Nest Health provides care to the whole family, alleviating child health needs while also providing Medicaid VBC to parents.

  • What’s intriguing to me is seeking patient-to-patient interactions that may even drive patient adoption network effects. A potential option: online communities. Currently fairly underutilized by healthcare orgs, a simple search of Reddit for [insert condition here] can show demand and proof of concept. Half the time, I achieve more therapeutic benefit from ADHD memes than I do from actual therapy (kidding).
    In a digital age, it’s a no-brainer to say that patients spend far more time online than interacting with the healthcare system, and largely we’ve failed to harness that mindshare. Over time, online community assets provide higher LTV and lower CAC that hedges against healthcare services with lower LTV and higher CAC.
    In the immediate, I’d anticipate that this would be high potential for dominantly online populations (check out this great report on Gen Z healthcare by Springbank Collective and Able Partners), or conditions with high complexity or stigma. Roon is an interesting play here. Providing depth/quality of experience to common health problems with community could unlock true scale potential for virtual care.

• Lying tangential to community and also the point on “authentic content” is user-generated content. Tia again is a great example here, having doubled down on UGC for patient acquisition & engagement and creating positive brand associations (trust!) in doing so.

Patient-aligned incentives

To combat healthcare’s misaligned incentives, new organizations need to deeply consider building from the go with models that create positive financial outcome for patients and market pressure. This also should naturally scale better. If the patient is incentivized towards positive action, and such action results also in net gain for the business, there’s no tradeoff made to patient outcomes in the path to finding positive unit economics.

The most popular approach currently is providing patients with utility (direct or wraparound) while selling second-order trust implications (improved engagement, lowered cost of care, medication adherence, better outcomes, etc.) to payers. The obvious examples here are risk-based models - firsthand not only provides core SMI care and services, but also provide patients with easy access to benefits beyond “traditional” health (food, utilities, housing, legal, etc.) that improves trust volume.
The best and cleverest B2C2B models will have innovated here as well. In the case of Capsule, a patient pays nothing for convenient access to medication, while Capsule earns based on improved adherence.

Arbitrage models, though more uncommon, have exciting potential to help patients flex outsized leverage by surfacing previously obfuscated information (commonly price, quality of care, quality of experience). The most written about one recently is GoodRx. For those unfamiliar, GoodRx provides patients with access to cash pay pricing across all PBMs, providing patients with necessary medication at the same quality with cheaper prices while enabling price competition and downward market pressure. Some other domains I’m keeping an eye out here:

• The insurance enrollment domain has been ripe for this opportunity and has seen attention from recent companies in the ACA & MA brokerage and ICHRA domains. Given that the CMS Price Transparency Rules for Payers are starting to kick in, along with on-ramp / Patient Access API companies like Flexpa lowering barriers to data access, the attention in these domains may start to pick up.

• Cash pay marketplaces like Sesame and Mishe have emerged recently, and on paper disintermediate the patient-payer relationship (and provide price transparency). I’m looking forward to seeing how they address common questions such as: patients overindexing on price at the expense of quality; clinician experience (Mishe team thread here); expansion into more complex care.

• *Turquoise has been one of the more intriguing new companies recently - by increasing access to price transparency, they might unlock more companies to enable this form of arbitrage. They’ve also hinted at a consumer-facing motion - really curious to see how that plays out.

Generally, most arbitrage opportunities might require deep knowledge of some archaic section of the healthcare supply chain, then developing B2B motions with relevant (and likely archaic) players to gain an industry-holistic data perspective.

There’s also the pure incentivization of desired behavior. In its simplest form, this might look something along the lines of Oscar’s introduction of the step counter program (of which true outcomes are debated). More recently, Garner Health reduces patient out-of-pocket cost if the patient chooses a provider deemed quality under Garner’s analysis. These might be smaller engagement “hacks” as opposed to true business models, but perhaps they move the needle positively. 

Despite healthcare’s labyrinthine incentive and supply chain structures, there’s many more creative models to be unlocked with deep analysis. If the result is positive patient outcomes, lower costs, and a scalable business, is that not a worthwhile pursuit? 

Now what?

How quickly your organization can effectively amass the largest aggregate of patient trust, and average trust volume versus that of competitors, will determine early likelihood of success.

Even still, trust will be fragile. Delivering on what you promise the patients and continual improvements in experience & outcome (by increasing the trust volume you can capture) is what maintains trust, builds brand equity over time, and strengthens early defensibility. It’ll also be key to continue searching for opportunities that drive network effects and trust stickiness, in pursuit of the truest moat.

If defensible scale is reached, a natural benefit is pricing and contracting leverage. The greater the aggregate, the greater the leverage, and ideally this power is passed to patients - either through exertion on their behalf, or directly providing patients themselves with aggregate leverage.

For certain aggregators, particularly those which are care domain agnostic, there’s intriguing implications for steerage. As healthcare continues to unbundle, and new services arise while old ones expand - without building cross-service experience pipelines (referrals, care continuity, data, etc.) - fragmentation in the system continues to grow. The strategic answer from many larger orgs has been consolidation, with largely mixed results (wrote more about these trends with Brendan Keeler and Megh Gupta here!)

Orgs that can truly harness deep trust might earn the right to advise the best options for patients, while utilizing the above described leverage to drive said options toward patient-forward quality, cost, outcomes, etc. In the largest scale, this might result in such orgs owning the demand-side of the market - becoming the de-facto “front door”.

Special thanks to the below folks in particular for initial reads and critical feedback:

• Bobby Guelich - Elion (Check out the recent marketplace launch!)

• Rik Renard - Awell, CareOps

• Krish Maypole - Icon Health, Substack: Point of Care

• Rachel Menon - Buoy

• Seeking early defensibility of your healthcare business is more critical than ever, as investor expectations become higher and more scrutinizing, and differentiation becomes more difficult to earn.

• Early defensibility then can come from your engaged patient base - specifically, if you can aggregate the trust of many patients quickly, this buys you defensibility and more time towards differentiation.

• There are some domains to build in healthcare that present more advantages to patient trust aggregation than others. This article presents a view on how one might identify these advantages for new opportunities, business augmentation, or investment evaluation.

• One can view patient trust as a product of three dimensions: frequency of interactions, utility of interactions, and duration of interaction history. The higher the product of the three dimensions (aka volume), the higher the trust of the individual trust.

• To aggregate this trust scalably, you need to not only consider the trust volume potential of your strategic approach, but also how that might tie into effective patient acquisition, and patient-aligned incentives.

• Finding an elegant way to optimize across all of these criteria earns your business a competitive advantage, pricing/contracting leverage, and steerage potential.

Quickly stating the obvious: Virtual first-healthcare companies’ need for competitive advantage continually rises. The macro market landscape exacerbates pressures while competition and investor scrutiny stays persistent (or increases). Meanwhile, the necessity for a more patient-aligned system remains as pressing as ever. 

For patient-facing companies, the bar for execution on pure tech differentiation is high. With services both existing and emerging (via the digital health tech stack) to quickly enable table stakes experiences, and some even empowering net new, standing up a functional healthcare operation is easier than ever - for both you and your potential competitor. This means the opportunities for truest healthcare differentiation that comes from tech itself lie increasingly in uncharted waters, where value experiences growth on a non-linear scale. (Good examples here being companies with clinically proven outcome-improving tech itself like SWORD, or those with novel care delivery automation.) Difficult to identify, even more so to truly execute.  

In this arena, the proven path for some form of early defensibility will be to amass an aggregate of patients. Done successfully, and perhaps this might even buy some time for building towards that true tech differentiation. 

Naturally, building this aggregate isn’t as simple as it seems (and when is anything ever, in healthcare?). It’s not enough to have a rolodex of patients at the ready. To acquire sturdy defensibility and maximize value capture, one must build to aggregate patient trust - relationships that engage and retain. This will be the case for any patient-facing company, and especially for those looking towards risk-based models. And with runways tightening, it’s more imperative than ever to find quick paths to adoption and engagement.

As such, the objective will be to target opportunities that possess, among others: 1) high patient propensity towards trust, and 2) aggregation advantages. Below are my heady thoughts on how to approach that objective. For builders, hopefully this provides a way to find NewCo ideas, new business initiatives, new product lines, new acquisition channels that drive truer adoption. For investors, perhaps a qualitative way to examine investment opportunities. For healthtech aficionados, some fun thoughts to chew on. 

Let’s dig in.

Identifying opportunities for patient trust

Different parts of healthcare will pose different challenges in building patient trust. If capturing trust early might be a source of defensibility, the question becomes: where are the advantages?

A simplified way I view trust is as a product of three things:

• Frequency of interactions (“How often?”)

• Utility of interactions (“How much value?”)

• Duration of interaction history (“For how long?”)

The greater the volume of these three dimensions, the greater the trust. A cube of trust if you will. Sounds funny. 

Thus, to optimize the ability to capture a maximum volume, we can look at healthcare domains that possess advantages in the three.

Frequency

• Frequent healthcare utilization populations (ex. Chronic care, Medicare / Medicaid)

• Routine healthcare engagement (ex. Pharmacy / prescriptions, urgent care)

• Proximity to routine non-healthcare interactions (ex. Texting, community centers, barbershops / laundromats / convenience & retail)

A leading navigation company defines engagement (in its S-1) through “percentage of member families who have engaged at least once a year”, and that number sits around ~50% - healthcare’s engagement problem is egregious! This may be a symptom of the deeper issues of patient apathy and mistrust of the system, which will need to be attacked head on. If one aims to build habit-forming patient engagement in search of deeper trust, high frequency domains will provide more shots on goal than domains dominated by sporadic or episodic interactions. 

For many of the early wave of patient-facing orgs, frequent utilization populations & routine engagement domains made easy targets, partly for this reason. Patients here have higher propensity to need care frequently and on a regular cadence, so building for these domains allows for a non-trivial “alleviation” of the engagement problem at the root level.

Even with the above mentioned frequent utilization populations and routine engagement domains, often the engagement behavior is driven by necessity, not proactive choice (and even for proactive personas, the poor overall experience exerts negative inertia). It’s important to recognize that patients (read: humans) are generally averse to new behavior - new lifestyle choices, new apps, new services, new individuals to interact with.

Lowering the cognitive burden to engage drives engagement. This often can mean tapping into existing behavior patterns, or finding “proximity to routine non-healthcare interactions”. Going to where patients already are can in the least provide adoption/engagement lift, and perhaps even provide deeper trust opportunities - a great example being the NEJM Barbershop Study, recently made more famous by Nikhil Krishnan from Out-Of-Pocket. And to prove that I really practice what I preach, nothing says “lowering your cognitive burden” like making you read a super-long, extremely wordy blog post with like two pictures.

Utility

• Immediate patient value / convenience (ex. Care delivery, scheduling, health information & content, HSA & tax advantages, BNPL)

• Patient-favorable cost structure (ex. Discount coupons, CostPlusDrugs, HSA)

• High patient complexity / information asymmetry (ex. Insurance enrollment, specialist referrals, complex care populations) 

• Difficult access (ex. Specialty pharmacies, rural healthcare, inclusive care)

Utility to the patient can be obvious and/or deep.

Obvious utility domains provide immediate, conspicuous value to parts of healthcare top-of-mind to patients: one could 10x convenience and bring the experience to modern standards, improve patient’s financial outcome by reducing costs (or even earning them money in the case of HSAs), provide previously unheard-of clarity on quality, etc.. Some domains with obvious utility might even accomplish a blend of utility and frequency by tying into oft-frequented points in the patient journey (a good example being new age convenient pharmacies).

The key advantage with obvious utility domains is that they have an easier path to organic patient intent, for the average patient. The more apparent the value (and the more effective the communication of said value), the lower the cognitive burden. And of course, the lower the burden, the likelier the patient adoption & engagement and trust over time.

Meanwhile, deep utility domains tap into care journeys where the complexity or patient pain is high. Relieving said complexity / pain with empathy and quality will quickly ingrain deep trust. This may require meeting patients at eye-level and gaining an ethnological understanding of patient pain points, while providing all-around support to reduce burdens to receiving care. 

Duration

• Embedded relationships (capitalizing on previous patient relationships)

• Enabling relationships (accelerating the path towards a defensible human connection)

• Portability (ability to maximize patient LTV / future patient relationships)

Obviously it’s impossible to immediately build a long duration of trust. However, what we can do is optimize the ability to capture duration.

One such way is to look towards finding embedded relationships and enabling new ones. I’ve written on this concept in more depth here, but patients will lean (sometimes irrationally so) towards the comfort of human connection for their healthcare experience. Building through/for that connection creates a proxy for durational trust. Essentially, finding a pre-existing “cube” as opposed to starting from scratch and encountering the warts that come with doing so. If such relationships are not easily identifiable and acquirable, then providing relevant individuals with optimal ability to build such relationships is the alternate path.

Another way is portability. Healthcare’s reimbursement structures require the continual severance of relationships (care discontinuity) which equates to churn. Planning for portability, or working through wedges with high portability (or consistency of experience across all reimbursement forms) creates longer-term advantages by avoiding the need for this severance while also creating trust-building opportunity at each severance point. 

You might be able to assess your wedge/initiative/etc. by testing its strength (or lack there of) in any or all of these dimensions. How does the MVP offer advantage in these dimension especially against potential competitors? Further, does the initial approach have a path towards beefing up the dimensions where it may be currently weaker?

There’s strong parallels between this trust volume framework and LTV. To capture max trust volume (and increase LTV), you can imagine that there’s multiple ways to play with the three dimensions. If you provide high utility at high frequency, then patient trust might be high fairly quickly despite a low duration of the relationship. If the domain is more seasonal in frequency, then it may be beneficial to tap into high duration wedges (if they may exist) to capture an early, consistent trust advantage.

Each dimension can provide supplementary trust support that eases the path towards early engagement. Some might even increase your TAM, or your average revenue per patient - especially if you’re pursuing the path of increased trust → increased engagement → improved outcomes → increased savings.

There can also be overlap across dimensions. Embedded relationships allow for efficient acquisition and “hacks” a long duration of trust. Local communities and social work lower barriers to entry for building durational trust for vulnerable populations while providing access points toward high frequency. Building care / operating models that find elegant ways to weave through these overlaps could be differentiators.

I think Tia’s journey is a great example of strong execution on building on the trust framework over time:

• Early initiatives targeting frequency + "obvious utility" with a patient information app

• Deepening that early trust through building in-person clinics that rewire a traditionally cold and "untrustable" clinic experience for women into something that ethnologically understands their needs

• Steady aggregation leading to incumbent partnerships and steerage opportunities

• One could imagine that, early on, they could have partnered with high-network OB/GYNs and women's health specialists to tap into "duration" referrals and accelerate aggregation.

Lastly, maximizing this trust volume for a given patient eventually will mean that you will become the go-to relationship for that patient’s healthcare journey, over Google / Facebook / the growing presence of TikTok. Part 2 will explore what might be key considerations towards scaling collection of maximized trust.

Special thanks to the below folks in particular (and others!) for initial reads, thoughts and/or critical feedback:

• Bobby Guelich - Elion (Check out Elion’s recent marketplace launch!)

• Rik Renard - Awell, CareOps

• Krish Maypole - Icon Health, Substack: Point of Care

• Rachel Menon - Buoy

All jokes aside, we hope you enjoy it and look forward to any feedback or thoughts!

We may still be in the early innings of this budding era of healthcare innovation, but enough time has passed that we’re seeing the emergence of new trends. Of these, the trend of consolidation has been steadily gaining traction over the past year. Many newer health tech companies have reached a degree of technical and operational maturity on a national scale. Combine this with the unprecedented dry powder in the industry and expectations from large fundraising rounds, and it’s a recipe to drive big acquisitions.

Each merger and acquisition is fundamentally unique as you dive into the details, but given that’s no fun, this article attempts to build mental models to categorize, classify, and hastily draw circles around groups of consolidations.  

M&A Strategy

For D2C organizations, these acquisitions have been predominantly driven by rising CACs, resulting in a search for improved unit economics and overall TAM. Thematically, these organizations will seek expansion in depth/complexity of offering or supported verticals.

• Vertical Integration: D2C players will look to expand patient LTV by vertical integration. So far, the focus here has been to connect the virtual experience to the physical, either through in-home care (Ro + Workpath / Kit) or physical presence (Kindbody + Vios; Babylon + Higi). This trend should continue, as virtual players hit the ceiling of what patient care can be via primarily digital modalities and need some in-person touchpoints to continue to expand the care of their patients.

• Horizontal Expansion: Alternatively, these players can also launch/acquire new verticals, and in doing so increase cross-sell opportunities (LTV) and/or overall TAM. One could draw a parallel between the strategies of increasing margins of cross-selling new verticals to SaaS businesses targeting expansion revenue. The largest share of acquisitions of these type have been companies with tangential specialty conditions (ThirtyMadison + Nurx; Ro + Plenity / ModernFertility / Dadi), but there have also been instances of companies acquiring assets in an orthogonal vertical but with close product fit (K Health + Trusst). From another angle, the addition of new verticals creates a more holistic offering that is also compelling to employers, enabling these D2C companies to more easily activate a B2B channel. This type of expansion is reminiscent of the parallel consolidation we see in traditional healthcare markets, with large integrated delivery networks offering complete integrated care across many acquired specialties. 

With the increasing competition across the healthcare ecosystem, companies have also begun to expand to more complex reimbursement markets, and this has been reflected in merger activity as well. In combinations between D2C and Employer-focused companies (Calm + Ripple; Ginger + Headspace), you’ll ideally see accelerated distribution across markets and the combination of assets for a more diversified offering. Meanwhile, employer-focused organizations move towards government-funded markets (One Medical + Iora) for not only increased TAM but also the high-reward potential of value-based contracts. This brings unique stressors as these organizations need to build organizational and technical muscle to actually execute on that vision - enterprise sales experience instead of consumer oriented marketing, for instance, or tooling to monitor and provide care management proactively at the population level instead of simple encounter oriented experiences.

For a variety of reasons, value-based care (and tangential) organizations are the other domain of healthtech that has seen an increase in M&A volume in recent months. Here, organizations will target larger savings by increasing patient-side and/or provider-side efficiencies or extending their capabilities in care.

• Increase patient-side efficiencies: By increasing a patient’s ability to engage and utilize care, organizations can better patient health and reduce more costly downstream utilization. More holistic organizations will acquire point solutions that enable this - often in care management/navigation (Babylon + DayToDay; Aledade + Iris; Accolade + 2nd.md). Also common are M&As between navigation platforms and virtual care/telehealth orgs (Doctor on Demand + Grand Rounds; Accolade + PlushCare)

• Increase provider-side efficiencies: Risk-bearers also look for cost-controlling opportunities in the care delivery itself. Recent examples have been specialist consultations (OSH + Rubicon) for effective triage and decreasing downstream referrals or quality-focused provider networks (Transcarent + BridgeHealth).

• Extend care capabilities: Acquisition has enabled players to go deeper into care for key verticals and demographics, ideally resulting in not only increased savings, but cross-selling opportunities, much like the above D2C plays. As of late, the target areas have been deeper engagement for chronic conditions (Teladoc + Livongo) and vertically integrating physical presences / home health (Humana + Kindred), where chatter about adopting value-based care models is gaining steam.

Lastly, we’ve also seen acquisitions geared primarily towards marketshare expansion (UniteUs + NowPow, LetsGetChecked + BioIQ, Cedar + OODA, Collective Medical + Audacious Inquiry). As the private markets start to follow the downturn of the public markets, landgrab-type acquisitions might pick up as directly competitive companies with similar assets look to shelter together for warmth as the sweet waning rays of our decade long boom begin to wane. This represents the most traditional type of consolidation, removing competition by virtue of merger. In an era and administration of increased antitrust activity, though, marketshare expansion should be viewed more cautiously. 

Post-Merger Integration

Once consolidation occurs, the key to longer term viability will be navigating the challenges of integration, namely:

• Organizational dynamics: How tightly integrated will the two organizations be? If the idea is to keep the two entities structurally independent, challenges here might be more side-stepped. However, if the goal is to consolidate to achieve a more comprehensive offering, putting thoughtful structure in place will be the differentiator in whether the merger is smooth or jarring for the new organization, especially for consolidation between companies with different GTM / reimbursement strategies.

• Competitive dynamics: It can be the case that an acquired asset could result in odd internal competitive forces. A recent example here is the Ro + Workpath acquisition. Workpath, as a platform enabler of in-home health services, sells to virtual-first providers - some of whom might be directly competitive to Ro’s specialty condition business lines. Establishing the right set of incentive structures is critical to ensure alignment with the overall strategy and goals of the merged organization.

• Merging clinical care stacks: When multiple Covered Entities consolidate, it’ll be the case that the organizations will possess stark differences in their clinical care stack. These will need to be resolved to ensure a consistent approach to providing patient care, billing and organizational function. This is in addition to the more standard integration problems of merging tech stacks (see below). 

• Merging tech stacks: At the root of any consolidation between two product assets (as opposed to independent service lines) will be the challenge of technology integration: programming language, overall syntax, third-party vendors, etc. At worst, variations across the technology and stack used at these companies can turn mergers that on paper are pure winners into absolute Pyrrhic victories, leaving the product and engineering teams in full rebuild mode for months or years just to get back to parity. If the integration involves a full incorporation of a new asset into a set of existing assets (eg.. K Health converting Trusst into K Therapy, Teladoc acquiring Livongo, Doctor on Demand merging with Included Health and Grand Rounds), it’ll be critical to be thoughtful in designing across the new workflows and to ensure smooth handoffs. 

• GTM strategy: In order to successfully streamline CAC and increase LTV, there will need to be a comprehensive GTM strategy in place to ensure that marketing channels, conversion mechanisms and, if relevant, sales team are all aligned. Consolidation can bring immense cross-selling and cost saving opportunities if done right.

Looking forward

As the fallout from the public markets continues to drip into the private arena and well capitalized startups look for alternative avenues for growth, we anticipate that industry consolidation will continue. The M&A strategies we outlined above are not exhaustive, and organizations will surely iterate to optimize for success. As to whether this wave of consolidation will yield tomorrow’s behemoths is yet to be seen, but regardless of whatever might happen, the middle innings of healthtech are going to be real interesting. Let us know your thoughts in the comments!

Healthcare Examples

Companies utilizing this framework are already emerging across distinct, complex healthcare verticals. Four examples I find intriguing are Honor, Unite Us, Headway, and Aledade. 

Honor

Honor was among the cohort of mid-2010s companies that made waves trying to disrupt the home health industry through a Gig Economy model, along with HomeHero and HomeTeam. Difficulties stemming from changing regulatory environments, operational and supply acquisition challenges, and tying into healthcare incumbent contracts and workflows resulted in a tumultuous journey for all three. CEO Kyle Hill’s post-mortem after the cease of operations of HomeHero is now famous, while HomeTeam underwent an organizational overhaul and has rebranded as Vesta Healthcare.

Honor experienced similar road bumps as a DTC home care organization, before announcing with their Series C a pivot into a partnership-led strategy. Naming it the Honor Care Network, Honor leaned into establishing partnerships with independent home care agencies and aging services entities like senior living communities, after noticing the fragmentation and clear pain points of these independent operators. Marketed growth success and new large fundraising rounds indicate a more sustainable model. 

“The core opportunity Honor has really seen is that there’s a very high degree of fragmentation in the current home care market, and that does create a lot of challenges for clients to navigate, for [caregivers] to get vested, so we’ve had a very, you know, relatively simple mission — to create a national platform to address a lot of those issues that fragmentation has created.” - Nita Sommers, President

Framework Alignment: 

Business Model

• Partners with fragmented independent home care agencies, aging services entities (senior living communities), in exchange for a negotiated share of agency revenue. 

• Agencies stay independent, but agree to co-brand with Honor

Offerings

• Provides tech-enabled operational lift for back-office functions: billing,  payroll, legal and regulatory, compliance.

• Improves caregiver turnover rates (key agency pain point): caregiver recruitment, evaluation, onboarding, retention, scheduling, added staffing, training

• Improves client experience through augmenting with nationally-scaled data insights and technology: ex. Decreasing caregiver churn by improving on experience variables (such as travel distance)

• Enables agency growth through support in digital marketing, sales team training

Business / Industry Impact

• Enhanced delivery of localized experiences: “The locally-owned agency providers in Dallas, Los Angeles, and Albuquerque understood their communities better than we ever could. Not just cultural nuances, they knew the health and social support systems for older folks, who families turned to when looking for a caregiver, and the key institutions within the community who helped their patients and clients find quality home care. And they had spent years building real relationships and real trust, which is what this business is about.” - CEO Seth Sternberg

• Enhanced delivery of personalized experiences: “They ran local, home-grown businesses—all deeply embedded and respected in their community—and it resulted in a deep kind of personalization that technology alone could never replicate.”

• Industry-leading caregiver retention: 82% industry caregiver turnover vs 36% for Honor

• Accelerated scale for Honor: Within a year post-pivot, Honor tripled hours of providing client service, and doubled market coverage, while receiving inbound agency interest. 

• Criticism - Honor has received questions and public criticism about the occasionally aggressive revenue share it arranges with agencies and recent corporate turnover (which may not necessarily be due to the strategy shift).

Next Moves

• Honor intends to utilize the leverage of its scale to develop a national set of standards for home care and contracts with large scale insurance carriers / major health systems. 

• “It is currently not possible for a national insurance company or hospital chain to work with one single home care provider across markets to provide a consistent experience. We believe Honor can change that – with our technology and operational capabilities, it’s extremely do-able for us to provide a consistent, high-quality service across all markets we serve”  - Sommers

Unite Us

As veterans, co-founders Dan Brillman and Taylor Justice both experienced the pain points and inefficiencies of care transitions between healthcare and social services, and founded Unite Us in 2013, originally focusing on aiding veterans in utilizing VA services. 

With social determinants of health (SDOH) being an emerging market at the time, Unite Us went through early challenges and pivots. Initially, it provided a resource directory, but Brillman and Justice found that the lack of coordination and accountability among community based organizations (CBOs) resulted in patient dropoff and poor visibility into outcomes. Largely, this was due to CBOs being under-resourced and underfunded while having to utilize antiquated manual processes.

This realization fueled a pivot into a more deeply integrated solution. Brillman and Justice took a year to develop a framework around where CBOs aligned on SDOH outcome standards, then incorporated that framework into developing a tech-enabled CBO referral network that tracks those outcomes. Key to a sustainable network was building relationships with these CBOs and creating value for them. They then offered health systems access to outcomes data and access to this CBO network. Finding success through this pivot, Unite Us has now expanded to serve Medicaid, Medicare and commercial patients as well. 

“80 percent of someone's overall health happens in the community and not in a clinical setting. How do we then bring that into that person's overarching health picture? And so as we start to create those work streams and you create value for the health care system to reduce costs and engage with these community based organizations, we then had to create solutions in the back office for the CBO to submit invoices and so make claims back for those services. So we become this marketplace where health care, community based organizations are not only working together, but they're in a business relationship together as well. And the whole goal is to improve health, not just health care, but health.” - Taylor Justice, CBO/Co-Founder

Framework Alignment

Business Model

• Works with local CBOs to construct coordinated social care networks

• Charges health systems and payers one time fees for configuration and setup and recurring annual fees for network access, network maintenance, and third-party integrations.

Offerings

• Provides health systems with integrated software to securely coordinate with CBOs and track/analyze SDOH outcome data

• Builds and curates tech-enabled SDOH care coordination networks within communities by deploying localized implementation teams that onboard and train CBO service providers that understand local dynamics

• Provides these network service providers with software for screening / risk stratification, referral management, care plan management, intranetwork communication, bidirectional communication back into health systems, outcome data tracking, and longitudinal service data

• Provides onboarding, training, implementation, tech-enabled process efficiencies, and support to service providers

• Facilitates CBO reimbursement workflows for health systems

Business Impact

• Scaled patient distribution: Unite Us has coverage of more than 21 million patients as the largest social services network operator across 20+ states and active networks in 42 states 

• Enhanced local navigation: Unite Us networks saw a 93% reduction in average time to referral acceptance across all service types (utility payment assistance, food access, transportation, etc.) since 2014.

• Improved CBO partner efficiencies: “Some organizations are seeing an 85 percent drop, which is saving 15 hours per employee per week.” - Justice

• Success with larger payers and health systems: Improved 50% of KLAS-surveyed Unite Us customers said that they saw outcomes immediately, and 30% said that they saw outcomes within six months or less. This has led to partnerships with Aetna, Humana, and other major insurers 

Next Moves

• In the short-term, Unite Us sees data interoperability anad enhancing its own integration capabilities as a core initiative, likely seeking to further develop standards around SDOH data sharing. 

• Long term, Unite Us plans to utilize those standards to leverage tighter inclusion of CBOs and SDOH into value based care and the overall healthcare supply chain.

Headway

After struggling to find a therapist that accepts his insurance, CEO Andrew Adams learned that this was an indicator of a larger problem: the process for a therapist to accept insurance is extremely tenuous, leading to 70% of therapists not accepting new patients on insurance. Many practices (solo or group) do not have the bandwidth to navigate the bureaucracy of health insurance, an activity which often takes up entire billing and administration departments for larger hospital systems. Headway started under the thesis that making accepting insurance easier for therapists would increase therapist ability to take on patients, subsequently increasing patient access to therapy.

“Health insurance is built around a medical world dominated by billers and admins, but therapists are small practitioners and don’t have the bandwidth to handle that, so they don’t. So we thought if we could make it easier for them to, they would, and they have.” - CEO Andrew Adams

Framework Alignment

Business Model

• Provides software and services to patients and licensed, practicing therapists for free

• Takes commission from insurance carriers who pay for access to and patient engagement with a broader network of therapists

• (Headway is able to profit more from insurance commissions by automating away standard overhead needs)

Offerings

• Provides tech-enabled operational lift: insurance administration process automation (key therapist pain point), claims processing, payment processing, scheduling, billing

• Improves experience through therapist-centric EMR

• Enables practice growth through marketplace with localized, personalized demand-supply matching (variables like ZIP code, specialization, etc.) and digital marketing

Business Impact

• Improved patient access: As of November 2020, Headway was working with ~1800 therapists in NYC, of whom two-thirds were only accepting insurance through the company. (Number is now 3107 in multiple markets, and this makes it the largest group practice in the country). These therapists are serving “tens of thousands” of patients at an average savings of 81% per appointment.

• Improved patient retention: Headway advertises 90% retention and 93 NPS, both of which are industry-leading numbers.

Next Moves

• As an earlier company, Headway is exploring supply growth through primary care referrals and partnerships with enterprise companies as a benefit.

Aledade

Well established in healthcare, Dr. Farzad Mostashari served as the national coordinator for health IT in the U.S. Department of Health and Human Services during the Obama Administration. During the development of the Affordable Care Act, the concept of the accountable care organization (ACO), where primary care providers coordinate between each other under a value-based payment model began to crystallize. After his tenure as national coordinator, Dr. Mostashari spent time at the Brookings Institute formulating the ideas for what would eventually become Aledade, using the ACO as the backbone of the model. 

“We are different in that we have delivered results at scale across over a dozen different markets using our tech-enabled playbook. We don’t deliver care ourselves, we support those who do. We don’t buy practices. We empower them.” - Dr. Farzad Mostashari, CEO

Framework Alignment

Business Model Model Source 

• Unlike other primary care companies like One Medical, Oak Street, or Iora Health, Aledade doesn’t build brick-and-mortar clinics or provide care itself; it constructs ACOs out of independent primary care practices, then facilitates value-based care contracts between them and Medicare / commercial insurers and provides necessary product and services to ensure the success of these contracts. 

• Aledade shares the capitated payments with practices (where providers are paid a fixed amount by insurers for care, and can keep the medical cost savings). For every $1000 savings generated for health plans, $400-$750 is split between Aledade and physician partners.

Offerings

• Provides resources, expertise, process installations, legal/regulatory support, and scale needed for primary care practices to join ACOs and take on value-based risk.

• Integrates with EHR data, hospital discharge data, claims data, lab and pharmacy data

• Drives better care through insights on proactive outreach for algorithmically identified high-risk patients, point of care suggestions, follow-ups, and care coordination.

Business Impact

• Accelerated scale: In its pitch deck, Aledade reported 15x in growth of medical spend under management over the last 5 years. The company works with 7800 providers (with 84% DAU!) at 800 practices to provide care to 1.2 million patients. During the early stages of COVID-19, Aledade launched telehealth capabilities for 150 practices over a weekend.

• Business growth for Aledade and primary care practice partners: As of January 2021, Aleade’s practices have earned $115 million in capitation payments, while Aledade itself reached profitability in 2020. (In comparison, competitor One Medical sees continued losses, and saw net losses climb with increasing patient membership during certain periods.)

• Positive care outcomes: Patients at Aledade’s participating practices experienced 20% increase in primary care utilization, 14% decrease in ER utilization, 15% decrease in inpatient utilization, and 22% decrease in skilled nursing utilization

Future State

• Aledade’s focus with its most recent raise will be pursuing strategic partnerships with Medicare Advantage plans. Over 2020, it had already doubled patients served in MA contracts to 100,000, but this is still less than 10% of their patients under management. 

• It seeks to launch initiatives into overall population health: integrated telehealth, prevention of unplanned dialysis, reduction of racial disparities in hypertension control, etc. 

• Aledade also will aim to enable the long tail of primary care practices to join ACOs through it. 

Here are some more companies that I think fit this mold to some extent:

• AspenRx

• DocSpace

• Mindbody

• Nines

• Solv

• Strive Health

Special thanks to Nikhil Krishnan, Andrew Pajela, Chris Hogg, Alex Zhang, Neal Khosla, and Byron Edwards for helping workshop this. Candidly, this series was a whole mess before their feedback :)

Please feel free to leave comments or questions, and don’t hesitate to reach out to connect or collaborate!

btlee215@gmail.com | @_btlee_ | LinkedIn Profile

Current Implications

While on the surface, this framework may just seem like “Cameo / Shopify for Healthcare”, there are deeper implications that make this a particularly compelling strategy.

It’s well known that there are now two dominant approaches to building in patient-facing healthcare - B2B selling solutions to large payers and employers, and (to a much smaller degree) DTC. These approaches have propelled companies to varying degrees of success, but also encounter common struggles, especially as both approaches strive towards more complex healthcare. These problems can be readily addressed by the patient aggregation framework. 

Not only this, but I think constructing infrastructure around service providers will enable the collection of truer metrics around cost, quality, outcome, and experience, which will help facilitate the transition into value-based healthcare experiences for these verticals.

Most importantly, over time, this strategy will result in the aggregation of the long tail of service providers within the chosen healthcare domain, and an advantageous path to an aggregated patient base. With both of these forces aggregated and the added factor of a scaled tech/data infrastructure that can truly support the two, perhaps this creates the existential threat that can impose significant leverage against the healthcare establishment that the overall system desperately needs. 

Here are some of the problems patient-facing DTC and B2B companies face today: 

Problems:

Costly patient distribution:

Scaling patient distribution is expensive. DTC has finally succeeded in healthcare after years of failure by focusing on healthcare’s transactional domains and acquiring patients through heavy growth marketing efforts. However, as these organizations seek to extend patient LTV by reaching beyond transactional domains into deeper, more complex patient problems, they need to prove effectiveness of their growth strategies for these more complex issues.

Meanwhile, B2B is a tried-and-true GTM strategy in healthtech, but lengthy, extensively bureaucratic sales cycles and the threat of heavy product influence from high-leverage clients are well-known problems. With both the payer and employer channels becoming increasingly saturated, the competition has never been higher, and winning the market will depend on aligning with consumer preference and delivering true utilization - as Joe Connolly said in his recent piece, “instead of fighting for payer contracts, companies will fight for consumers”.

Low patient engagement

With out-of-pocket spend steadily increasing and more options available than ever (cash pay options, MA plans + individual exchanges, novel healthcare models, etc.), more patients are shouldering the burden of making healthcare decisions and becoming shoppers. New wave DTC companies zeroed in on very specific problems that are disruptive pain points for patients and create moments of high transaction intent, but low enough in complexity that the entirety of the patient journey could be virtually isolated, commoditized, then marketed. However, while this approach proved successful in earning high patient utilization in the less complex, transactional domains of healthcare, these companies will need to prove the sustainability of their high engagement as they move deeper into healthcare’s more complex problems. 

However, at large, patients are still at-best infrequent utilizers of healthcare, and creating strong engagement is a hurdle that even the best companies like Omada struggle with. Some main problems being:  

• Complexity: Choosing the optimal path of healthcare decisions requires a strong understanding of both health and the healthcare system, which most patients do not possess. To make matters worse, healthcare information lacks the transparency, usability, and interoperability needed for optimal decision making. This, combined with poorly-designed experiences result in deep patient frustration, which often leads to lack of engagement or outright neglect of important healthcare choices. For instance, 57% of Medicare enrollees do not review their coverage options, with many citing difficulty as a main turn-off. This in turn results in volatile engagement intent from patients, which is difficult to build around. 

• Patient Trust: Building relationships with patients is mission-critical for not only long-term engagement, but providing better care/experiences. This starts with trust in either the platform or the human contact service provider, which is both difficult to establish and challenging to scale.
  (Grabbing deep in my reference bag, I’m reminded of a particular episode of the 90’s hit show Doogie Howser, MD - When Doogie Comes Marching Home. During the episode, teen doctor Doogie Howser takes pride in having broken his previous record of patients seen, but also is challenged by the case of an elderly woman who’s increasingly becoming ill. By spending time with the patient, he learns a detail that helps him successfully diagnose and heal the patient.)

Localization

The more complex the healthcare problem space, the more localized its healthcare will be. 

For one, healthcare is embedded with localized workflows, regulation, and operational challenges that vary across communities. Additionally, if your company deals with care delivery, there may inevitably be a point where a physical presence is required, and this virtual-physical handoff is notoriously difficult to coordinate and accomplish successfully.

Meanwhile, building a physical presence (like a clinic) is of course extremely capital intensive. 

These challenges mean that the supply of service providers can’t be completely virtualized and commoditized - for instance, telemedicine companies like Teladoc may provide access to a broad net of providers, but service providers that can provide both localized market/navigation expertise and a physical presence to patients when needed are scarce, yet are also the individuals that can provide best-in-class experiences to patients. These problems impede the scaling and scope expansion of virtual-first platforms. 

How this strategy addresses these problems:

Compared to other healthtech strategies, the ideal execution of this framework provides the following advantages:

• Unlock one-to-many patient distribution: Similar to celebrities on Cameo (and the passion economy at large), service providers are nodes of patient aggregation through which organic distribution can occur. They may either already have penetration through existing relationships with patients or know best practices for developing new ones. Optimally, they are also self-incentivized to maintain and grow these relationships. A tangential example is GoodRx utilizing doctors to distribute to customers. From the Caseload: “Doctors don’t generate any value for GoodRx directly, but are critical as a marketing channel. A patient that hears about GoodRx from their doctor is not just aware, but getting a valuable endorsement from a trusted source.“ 

• *Also - Minimize enterprise-scale sale cycles: The bane of existence for many healthcare companies is the B2B sales cycle: pilot programs, relationship and implementation management, etc. Best-in-class organizations go live within 6 months which is an eternity, especially considering feedback and product iteration needs, and the graveyard of healthtech companies is littered with feature houses that suffered a death by thousand cuts from large enterprise requests.
  The monetization avenues for this strategy offer advantages to the standard B2B cycle. Selling directly to service providers will have greatly reduced sales cycles because of the lack of bureaucracy. Meanwhile, access to aggregated service provider networks is an easier sell than a particularly novel solution, especially as value-based care initiatives kick in.

• Capitalize on pre-existing utilization + engagement intent: Patients that have sought the services of these service providers are already inbound users with high intent for engagement.  For the remaining patients, service providers are incentivized to foster deeper engagement. Further, for domains with less frequent interactions / volatile patient intent, working through service providers that are patient aggregation nodes offers an ability to spread the risk of that volatility across many patients. 

• Achieve patient trust by proxy: Through association with service providers, a company has a faster path to building trust with the patient, which then can be leveraged into deeper engagement and relationships with value-add products/services that go beyond a provider’s abilities.

• Overcome localization challenges: Partnering with localized service providers allows access to their understanding of local market dynamics, their existing abilities in coordination and navigation at the lower level, and their referral networks to other local service providers (who can provide access to more patients). If the company provides care, the virtual-physical handoff hurdle is more easily overcome through the patient relationship with the physical presence of service providers. Service providers will also simply use the physical resourcing they already have, allowing you to be much more selective and strategic with capex spending and to pass on savings to patients. 

• Utilize existing complex healthcare problem expertise: By working with service providers’ existing knowledge, you’re afforded the ability to immediately provide services at current standards and also the time to incrementally improve the standards of experience.

At the end of the day there still will be a need to extend any existing engagement intent into a deeper relationship. There are keys to success here as well - a quick few thoughts:

Patient Experience

• When at a crossroads, the patient’s experience should take priority, and the experience of the service provider (and other stakeholders) should be curtailed. Enabling a tight patient-service provider relationship and incentivizing repeat patient engagement will account for this, as well as strengthening the patient’s ties to the company and brand.

• Investing in strong brand will enhance patient trust and become a competitive moat at scale. Companies utilizing this framework will be among those that unbundle healthcare from institutions, and unbundling companies need to systemize trust - AirBnB’s disruption of hotels is perhaps the best modern example. 

• Service provider accountability should be optimized through network design and/or patient choice clarity.

• Provide patient choice clarity and autonomy through patient-facing data transparency and interoperable data flows. I’m of the opinion that this is absolutely critical in becoming a true patient aggregator/advocate, as opposed to a PBM-like deadweight middleman. 

Provider Experience

• To start, create authentic, lasting partnerships by driving conspicuous business value for service providers, with the aim to handle business complexities while enabling service providers to do what they do best: growth & lead generation, process optimization, operational cost reduction, patient (client) retention, service support.

• The most business-critical will be to facilitate reimbursement workflows between service providers and larger healthcare incumbents. Not only is this a key value proposition for service providers, but absorbing part of the reimbursement can be the primary business model for the framework.

• Referral and coordination workflows with other service providers will often be a core value proposition, but it will also enable supply growth loops and potential new vertical expansion opportunities. 

• Defining and incorporating metrics on cost, quality, outcome, and experience will enable provider accountability. This will require building data infrastructure and capturing relevant data from the onset. 

• Identifying vertical-specific granularities early on will empowering service providers’ ability to provide specialized, personalized experiences to patients and power enhanced supply-demand matching.

Looking Forward

As healthtech as a whole begins to focus on more complex problems, this framework will become more ubiquitous. Here are some of the verticals where I see potential for the framework (curious to hear where you might see opportunities!):

• Direct Contracting

• Skilled Nursing Facilities and Post-Acute Care (Now-defunct Call9 is an interesting case here)

• Mid-High Acuity Behavioral Health

• Dentistry

• Autism Spectrum Disorder

• Medicare

• Rural Healthcare

• Specialized outpatient clinics

I also think there’s implications for compelling potential new business initiatives for existing companies as well - for one, it may be intriguing for Cityblock Health to explore expanding reach through providing care coordination software, operations, and logistics to clinics in heavy Dual-Eligible population regions.

Assuming this strategy becomes continually utilized, there’s growing business need for:

• Patient triage and funneling

• Supply growth optimization

• Back office APIs - compliance, contracting, payroll, licensing, etc.

• Healthcare-specific workflow APIs - case management, optimized patient engagement for service providers (ex. Capable Health)

• Contracting with incumbents and reimbursement workflows as a service

• Supplementary clinical resourcing as a service (ex. Axle Health)

• Supplementary operational resourcing as a service (ex. Wheel for care navigators)

Source: Julie Yoo, a16z

At the end of the day, the framework starts with providing operational efficiencies as a means to acquire relationships with service providers and by proxy their patient populations, but the most critical thing to be focused on will be the patient’s overall experience. With a proper tech infrastructure laid down and patient aggregation in hand, one should act with urgency to capitalize - can you deliver on better care, better experience, better cost, better outcomes? Are you aligning, incentivizing, and enabling service providers to do this?  

It’s common to become jaded in healthcare but I think there’s reason now to be less so. Regulation trends, data interoperability, understanding of systematic problems in healthcare, industry interest from key talent, and other factors are all trending in the right direction and will result in the development of new business models that can truly better American healthcare. Companies that follow this framework will be steps in the right direction towards that future state.

Special thanks to Nikhil Krishnan, Andrew Pajela, Chris Hogg, Alex Zhang, Neal Khosla, and Byron Edwards for helping workshop this. Candidly, this series was a whole mess before their feedback :)

Please feel free to leave comments or questions, and don’t hesitate to reach out to connect or collaborate!

btlee215@gmail.com | @_btlee_ | LinkedIn Profile

Part 1 (this post):

• Growth strategies employed by passion economy companies

• How these strategies have parallels in healthcare, leading to a framework for patient aggregation

Part 2:

• How the application of this framework may be advantageous over other healthcare strategies

• Potential impact and opportunities

Part 3:

• Case studies of companies using this framework

Opinions are my own. Let’s dive in!

In 2008, Kevin Kelly (Wired magazine’s founding editor) wrote the evergreen 1,000 True Fans, which has since become the philosophical backbone for the passion economy. Its core themes are: 

1. For success, individual creators need to obtain and maintain direct relationships with ~1,000 true fans.

2. “The big corporations, the intermediates, the commercial producers, are all under-equipped and ill-suited to connect with these thousand true fans. They are institutionally unable to find and deliver niche audiences and consumers”

The key word here is “relationships”. By providing platforms of tools and services for “1,000 fan” writers, video game streamers, small business owners, and creators, businesses like Substack (you’re here now!), Twitch, Shopify enable creators to capitalize on relationships with their new and existing consumers. What’s more is that these relationships are not abstracted away (as would be the case for successful Gig Economy businesses like Uber) - rather, they’re celebrated as core components to the business. 

Source: Li Jin, a16z

But why did Substack, Twitch, Shopify, etc. focus on these relationships? The powerful element of relationships is their ability to accelerate growth and consumer distribution. Let’s look at one of my recent favorite examples, Cameo.

You most likely have heard of it by now, but if you haven’t, Cameo has been equally lauded (TIME's 50 Genius Companies of 2018) and dismissed (“a respirator for C-list celebrities gasping for their last bit of relevancy and revenue”) because of its business. Essentially, Cameo offers a marketplace for consumers to purchase personalized video messages from celebrities, ranging from a TikTok-famous monkey to Floyd Mayweather. 

One of the most undeniable components of Cameo’s business is its growth strategy. Lenny Rachitsky broke it down best recently: instead of recruiting thousands of users, “Cameo found that if they recruit celebrities, the celebrities themselves bring the fans to the marketplace”. 

Source: Lenny Rachitsky

(With Gig Economy models, this type of growth cannot occur. Drivers do not build/maintain relationships with riders because they provide a commoditized service and can be easily replaced with other drivers. Therefore, Uber must recruit both driver and rider, as they cannot take advantage of driver-rider relationships for growth.)

Cameo’s focus on tapping into the creator’s relationship and ability to connect with the consumer exponentially improves their customer acquisition ability. One-to-one becomes one-to-many. 

In sum, passion economy businesses achieve scale by:

1. Identifying domains with “1,000 fan” creators who cultivate relationships with consumers 

2. Providing creators with tools and services which empower their business and encourage consumer connection

3. Capitalizing on creator-consumer relationships to recruit consumers 

Healthcare Parallels

Let’s apply these learnings. How do these passion economy strategies relate to healthcare?

Just as Cameo and Shopify zeroed in on under-resourced creators / small-business owners with relationship-building ability, healthcare companies have the opportunity to zero in on similar individuals, who I’ll refer to as “service providers”. Essentially, this will be all individuals or small scale organizations who are involved in a given person’s care experience - healthcare practitioners, healthcare support, care agencies, insurance agents, social workers, etc. 

Healthcare is commonly viewed as having three main players: patients, payers, and providers.  However, “providers” here more often than not looks to the large revenue, incumbent health systems, and less so the individuals that comprise the system. Any individual/organization outside this realm commands even less mindshare. 

Service providers however are the ones with the most frequent and direct contact with patients, and thus are in the most advantageous position to develop relationships, increase engagement, and influence care decisions. As patients increasingly become (or are forced to be) decision makers in their own health, it’s not unlikely that they will turn to these service providers for deeper assistance. In other words, service providers will become nodes of patient aggregation. This is especially the case in more complex healthcare problem spaces. 

Every domain of service providers also is rife with pain points to be solved. Broadly speaking, service providers are:

• Under-resourced: Burnout is a well-documented problem for service providers. They also often need additional resourcing to be able to make or implement optimal decisions for patients. An example: during her time at Clover, Uno Health CEO Anna de Paula Hanika noticed that the task of enrolling individuals in cost-savings programs often fell on either clinical or insurance staff, who often lacked the time and/or necessary expertise to provide such assistance. 

• Misaligned: Healthcare’s poor incentive structures can create conflict between a patient’s best interest and a service provider’s business priorities. 

• Inefficient: Service providers have specific capabilities, but patients have a wide variability of needs. This results in a supply-demand matching problem, which, if unsolved, leads to inefficient coordination and additional costs for both patient and service provider. 

• Lacking in scale: By nature of functioning in smaller settings, service providers do not have access to the benefits that come with scale - not only resourcing, but pricing leverage, scaled data insights, etc.

There are (at least) two ways to monetize this avenue as well:

1. Sell directly to service providers

2. Sell access to aggregated service providers (and their patient relationships) to payers or incumbent provider systems.

By utilizing service providers to engage with patients (through initially solving service provider patient points), healthcare organizations can unlock the same one-to-many distribution to consumers (aka patients) that passion economy companies experience. Applying the passion economy strategy to healthcare results in a framework for patient aggregation:

Healthcare Patient Aggregation Framework

1. Identify the domains of healthcare where: 

   • The core service provided requires (or is greatly enhanced by) a longitudinal relationship with the patient

   • There exists a long tail of fragmented service providers

   • This long tail of service providers is underutilized and/or under-resourced

2. Empower service providers with tech infrastructure, process improvement, data insights, services, support so that each can:

   • Grow their own business 

   • Lower operating costs

   • Provide patients with top-of-the-line experience / care

   • Nurture patient relationship and engagement 

3. Through service providers, recruit and develop long-standing relationships with their patients

4. Continually elevate patient experiences and outcomes with data/tech infrastructure

5. Utilize engaged patient population to exert scaled leverage on incumbent systems

Aggregation is the name of the game in all fragmented industries, and especially in healthcare. As digital health pioneer Chris Hogg penned in his recent piece, “Healthcare organizations in the U.S. strive to obtain scale and mass in order to wield power in pricing and contracting. This level of scale and mass comes from the aggregation of patients, physicians, facilities and buyers (employers).” It creates more than just larger revenues - done correctly, aggregation creates influence.

Of course, this is no secret, and the incumbent system knows this as well as anyone. Their misuse of this leverage combined with inadequate/misaligned business models, inefficient infrastructure/technology, and inability to innovate leads to what happens everyday - higher prices passed onto patients without marked improvement to the quality of services. With no meaningful way to aggregate, patients cannot collectively bargain against the system. The failures of the system more often than not are forced upon them, while anything better than the status quo experience relies on goodwill from healthcare incumbents.

Yet, patients are obviously the true end-user in healthcare, and patients, in aggregate, will sway the decisions of incumbents. In an arena of Goliaths, how do you create a David-like advocate for patients? Can you aggregate patients for good and exert patient-favoring leverage? How do you do this as fast as possible and build a successful, sustainable business?

This framework provides an answer to those questions. 

Click here for Part 2 or Part 3.

Special thanks to Nikhil Krishnan, Andrew Pajela, Chris Hogg, Alex Zhang, Neal Khosla, and Byron Edwards for helping workshop this. Candidly, this series was a whole mess before their feedback :)

Please feel free to leave comments or questions, and don’t hesitate to reach out to connect or collaborate!

btlee215@gmail.com | @_btlee_ | LinkedIn Profile

This post is Part 3 of a series. Click for the Overview, Part 1, and Part 2.

• Expand your core product to new markets

• Expand your core product itself

Expand your core product to new markets

At this point, ideally your product has three qualities: product/market fit in your first area of focus, end-to-end with embedded modularized tech and processes, and a formed relationship with your patients. Now is the time to expand towards new strategic focuses.

Your modularized components and relationship with patients will facilitate these expansion efforts, but of course it’s important to anticipate where these elements are applicable and where new considerations are necessary. Blind spots are inevitable, but you can mitigate their total volume strategically.

Notably, Oscar prioritizes new markets through “Network Scorecards”, through which they’ve evaluated 350+ markets based on developed metrics considering access to care, quality, population health capabilities of health systems, etc. Meanwhile, while launching Rory (Ro’s women’s health vertical), Dr. Melynda Barnes noted that a key learning was regarding awareness of women’s health issues — while Roman was able to learn from 20 years of DTC Viagra marketing, Rory’s launch felt the challenges of women’s health issues not being mainstream. As such, they had to build processes around education and awareness for relevant topics.

Source

If you’ve constrained geography, these are the paths for geographical expansion:

• Pure geographical expansion (ex. relocating to new city)

• New customer types (ex. aggregated demand, product distributors, self-insured employers, MA markets, etc.)

• Strategic business development

If you’ve constrained categories, these are the paths for categorical expansion:

• Needs of existing patient base

• New patient types

Strategic partnerships can provide both categorical and geographical expansion. If the partnership company’s values and product aligns thoughtfully with yours, it’s a powerful way to expand your reach.

Capsule

• Geographical expansion: After three years purely in NYC, Capsule has begun its national expansion, starting with Boston, Chicago, and the Twin Cities, and citing the cities’ health systems’ innovative histories and COVID as the main reasons.

• “To support this unconventional model, we have an unconventional growth plan. Rather than open one market at a time, we are working to lay the foundation — the people, the process and the technology — to enable us to expand in a different way, and in continued partnership with all of the stakeholders in health care.” — Kinariwala

• Strategic partnerships: Capsule hasn’t been very public about its partnerships, but in December 2019, a partnership with Oscar was announced.

Cedar

• Strategic business development: Cedar engages in innovation development partnerships — notably health systems ChristianaCare and Novant Health in creating a pre-visit product.

• New customer types: Originally starting with provider groups, Cedar began expanding to AMCs, national staffing agencies, oncology groups, and HCOs.

Maven

• New customer types: After starting DTC, Maven was one of the pioneers of selling to self-insured employers, and has since also sold to health plans.

• Existing patient needs: Maven’s launch of Maven Campus was inspired by the organic demand and niche needs of college women.

• Strategic business development: Most recently, Maven partnered with MassHealth to provide free COVID-19 telehealth assessments in Massachusetts. Early on, Maven converted consumer company Snap into an evangelist customer, allowing for further success with Fortune 500s. Maven also piloted Maven Campus at NYU, Columbia, and Fordham.

• Strategic partnerships: Maven’s partnership with the health benefits navigation platform Castlight Health allows easy Maven integration for Castlight customers.

Oscar

• Geographical expansion: Oscar will be expanding to 12 new markets in 2020 to a total of 26, as a part of its plan to “expand at a rate of 4–5 cities every year, in individual, small group, and other market segments.”

• New customer types: Medicare Advantage, Small Business

• Strategic business development: Oscar’s network construction often relies on partnering with marquee health systems in the market, like Cleveland Clinic and Mount Sinai, or insurance companies, like Humana and AXA.

• Strategic partnerships: Oscar’s forays into new customers were augmented by partnerships with Uno Health (for Medicare Advantage) and Cigna (for small business owners).

Ro

• Existing patient needs: Zero was launched as a result of “thousands of members asking for our help on how to attack the root cause of their ED.”

• New patient types: Rory provides Ro with access to the 50% of the population not covered by Roman.

• Strategic partnerships: Ro partnered with Gelesis to distribute Plenity, entering into weight management.

Trialspark

• New patient types: Trialspark continually offers new conditions — currently with capacities in dermatology, pulmonology, gastroenterology, and endocrinology. Trialspark’s partnership with Pfizer is focused on inflammation and immunology, extending beyond existing programs (at the time) in skin conditions (vitiligo, atopic dermatitis, psoriasis and alopecia).

• Strategic partnerships: Trialspark partnered with 23andMe to enhance patient recruitment and enable genotyping-based studies, and with Limbix to extend into digital therapeutics. Most recently, Trialspark, with Sam Altman (YC, OpenAI) and Dr. Mark Fishman (Harvard, Novartis), launched Project Covalence, a project focused on COVID-19 project acceleration.

• Strategic business development: Pfizer, Novartis

Expand your core product itself

We’ve established that deepening your relationship with your patient through a longitudinal experience is critical. Naturally, this can be accomplished through impacting your patients through more parts of their patient journey. You need to identify net new offerings for your patients, upstream or downstream from where you currently operate in your patients’ care journey. If your direction is thoughtful, you’ll have the added benefit of unlocking new patient personas, expanding your broader reach.

This requires moving beyond your current core competencies. While new market expansion meant modifying your modularized components, here you will have to build essentially from scratch, while continuously assessing fit with the existing product and still designing for scale. Strategic partnerships can also be used to provide lift.

Capsule

• Capsule has focused on national scale as its key priority for its next 18–36 months, and has been generally vague about any other priorities. However, Eric Kinariwala has hinted at plays in medication adherence and a patient front door experience on their patient-centric efforts.

• “That needs to be the go-to, high-trust brand that you go to get anything with your healthcare whether that’s your prescriptions, whether that’s seeing a doctor online, whether that’s booking an appointment, whether that’s buying non-prescription products.” — Kinariwala

Cedar

• Cedar is expanding from the post-visit experience (patient billing) to activities that occur pre-visit, including check-in tools and price estimates for visits, the latter enabled by a new partnership with Waystar. The company envisions eventually branching out beyond financial workflows as well.

Maven

• Maven’s moves downstream have included forays into return-to-work care with Maven Milk, a nationwide breastmilk delivery service, and the acquisition of Bright Parenting to complement its forthcoming Parenting / Pediatrics product. Meanwhile, Maven Wallet accommodates the expenses/reimbursement experience and minimize claim rejections

• Maven Planning, Maven’s latest launch, sees Maven moving upstream to innovate the preconception experience.

Oscar

• Among its many initiatives, Oscar has continually invested into extending downstream into elements of care delivery. Virtual Primary Care includes virtual visits with primary care providers, but also a wide variety of other services for no cost to the patient — Tier 1 Prescriptions, durable medical equipment (DMEs), diagnostic imaging, vitals monitoring kits and in-home lab draws, all free and delivered to a patient’s home. In addition, Oscar announced a $3 Prescription List for commonly prescribed drugs.

• Oscar’s partnership with Capsule will bring same-day delivery to eligible markets, while its Eligible integration enables point-of-visit cost transparency.

• Oscar Center, launched in 2016 through a partnership with Mount Sinai, is a test lab for new provider products, but also could be a play for an expanded brick-and-mortar presence, as Nisarg Patel notes.

Ro

• Ro’s $200M Series C will go into building its Collaborative Care Center, enabling downstream patient experiences in remote patient monitoring, chronic disease management, urgent care, and at-home testing through integrated devices. Additionally, Reitano envisions creating a more holistic care experience through building for additional downstream health providers, such as nutritionists, therapists, and social workers.

• Ro’s partnership with Ribbon enables Ro to direct its patients to an offline, in-person care experience.

• “We built our diagnostics ingestion and management infrastructure with these advances in mind so that as they enter the market, we can seamlessly integrate connected devices and sensor data into our platform.” — Todd Levy, Ro CTO

Trialspark

• Trialspark envisions moving into patient outcomes in the long-term — “After a drug’s been approved, it’s almost just as important to understand the outcome these interventions make. We think we’ll play a non-trivial role in the value-based reimbursements discussion — how do we understand, after a drug has been approved, what outcome it’s making for a patient? There’s opportunity for us to contribute the evidence that’s needed to understand that.” — Liu, 38:57

Looking forward

Building in healthcare now is eaier than ever. The examples set by these organizations offer digital health legitimacy and unprecedented clarity in strategy. API enablers are emerging that allow newer organizations to quickly plug into and work with the healthcare data ecosystem — Particle Health and Redox for patient data, Ribbon Health for provider/insurance/cost data, Eligible for billing information, Commure for healthcare platform development, to name some of the best. The incubator and venture studio models (Redesign Health, AlleyCorp, Oxeon Partners) are seeing success as well, providing shared talent and ideation while distributing risk.

While how to build becomes an easier question to answer, what to build remains a persistent question. To start, the entire patient journey hosts a wealth of problems. I’m compelled by the approaches of Spring Health and Ginger in mental healthcare, Tia for women’s health, and Ride Health for patient logistics and transportation, largely due to their similarities with the approaches discussed in this series. Additionally, I think the continued rise of companies in this mold holds intriguing implications down the line in the problem spaces of patient routing, deeper automation of CBT, outcomes quantification/attribution, microinsurance, and value-based healthcare pricing/payment.

Still, there are so many trails left to blaze. The examples set by these companies establishes a clear theme — healthcare needs more boldness. As Malay Gandhi says, “the most exciting health tech companies will ignore the parts of healthcare that have become obviously anachronistic, actively discarding convention”. Rip out the assumptions, have a vision of what healthcare should be, then strive to be an entity to be reckoned with in that future.

Source

This post is Part 2 of a series. Click for the Overview, Part 1, and Part 3.

• Constrain your focus

• Own the full value chain

• Optimize for scale in tech / processes

• Empower a relationship between your product and your patients

Constrain your focus

In his famous series How to Kickstart and Scale a Marketplace Business, Lenny Rachitsky researched 17 of the most notable marketplace companies. He found that:

• “with the exception of one company, every single marketplace that I interviewed constrained their initial marketplace”.

• “The research points to two ways to constrain a marketplace: (1) by geography, and (2) by category.”

It turns out that this is the case for patient-facing healthcare as well. The connection makes sense — all healthcare products that deliver some form of care are marketplaces, connecting the demand of patients to the supply of care. As such, there’s two general routes for constrained focus: own everything about a given experience and expand geographically, or own one particular patient experience then expand to new categories.

The early focus becomes a superpower — you can fill gaps in the patient journey and create a personal, in-depth patient experience for your specific patient category, or optimize operational efficiency in your select geography, in ways that the incumbent experience never could. Healthcare isn’t one size fits all, rather quite the opposite.

Capsule

• Capsule initially served New York City in 2016, minus Staten Island, and took +4 years to announce launches in new markets.

• “We’ve spent, you know, five years in New York building the operational and technological capabilities, expertise to be able to scale the business really rapidly until we were able to turn that on pretty quickly.” — Kinariwala

Maven

• Though providing access to a variety of women’s health care providers from early on, Maven dialed in on the category of the fertility experience — prenatal and postpartum care — and identified on-demand access to information and care at this point as the core focus.

Oscar

• After launching in New York in 2014 and briefly expanding, Oscar notably constrained focus further and halved its New York footprint, citing curation of its provider network as core to its strategy. Although it faced initial criticism, it paid off a year later in profits and in patient experience.

• “In a system where every network is broad, where insurance largely gets sold through big employers, where there’s a big incentive to keep everybody in the network, where there isn’t the kind of competitive pressure on the value chain, there isn’t that much of a reason to develop a better correlation between cost and outcomes, or even visibility into how certain parts of the system perform.” — Schlosser

Ro

• Ro started out with a vision for men’s health, but focused down to ED, citing reasons of personal experience, stigma-created opportunity, and ED’s ties to deeper health issues.

• “The problem was we didn’t know where to start. We knew we couldn’t tackle all of Men’s Health from Day 1. We have all started companies before and know that maniacal focus is vital to success. We would never make it to point C if we didn’t first go from point A to point B. The day after my being told I needed medicine again, Saman, Rob, and I were jamming. I brought up the idea of starting our men’s health company with ED.” — Reitano.

Trialspark

• Trialspark originally started as purely a clinical trial recruitment product.

• The company creates geographical focus by utilizing de-identified patient data sets to identify geographic markets that are “high-potential ‘hot spots’ of potential trial participants” — this was the approach since its launch with New York sites for ankylosing spondylitis partnering with Novartis. Sites are established based on logistical complexity, potential patient identification, and sponsor requirements, among others.

Source

Cedar

• **The marketplace analogy applies less to companies like Cedar as it deals less with care and more purely on the patient experience, but the importance of constrained focus remains.

• “When we first started out, like most companies, you’re extremely narrowly focused on solving the one very micro problem. So for us, it was figure out how we can basically replace paper statements with a much better process after a patient has had a visit. Today, if you look at what Cedar does, we still do that. That’s the core of our business. But we’ve now expanded our vision.” — Arel Lidow, Co-Founder / President

Own the full value chain

“There are all kinds of middlemen in that process. And what the middlemen do for the most part is remove the competition in the health care value chain that would go towards building something that’s a compelling, seamless user experience. They remove the cost-containment pressures as well. That’s one big reason why health costs in the US have risen so much.” — Schlosser

With your focus constrained, the key priority is to own all the problems concerning the area of focus. In essence, all the problems that prevent the patient from accomplishing their end goal, which is why they came to you in the first place — from where your patient meets you, to when you deliver value to your patient. Patients need healthcare to be convenient, affordable, outcome-driven, and personal, and your end product isn’t valuable until it creates such an experience.

This means owning the entire value chain (within reason). You can’t build the experience you want and your parents need plugging into incumbent systems. Through this, you disintermediate middlemen, take ownership of the experience, and use holistic data insights & product design to improve the existing points in the patient journey AND gaps between points. This is how you can actively target embedded complexities, costs, and inefficiencies/poor outcomes for the patient. This is the foundation that enables you to scale.

• Complexity: Ingest navigational complexity of the existing experience for patients, surface something actionable.

• Cost: Operational efficiency (reduced SG&A) + improvements in unit economics over time mean you can pass on cost savings to patients. Also, the more

• Inefficiency/Outcome: End-to-end data ownership empowers the promise of digital health — enabling the right action for the right person at the right time.

“We wanted to be the company that takes on the complexity, so our members never have to. We wanted to not only offer the best experience in the present but innovate on the future.” — Reitano

Of course, full-stack ownership is no small feat. The larger the scope of ownership, the larger the need for capital. Clarity of vision becomes all the more crucial as your company starts to seek investors to fill that need for capital.

To mitigate that challenge, the companies of this post have all started with an end-to-end vision, but took a gradual approach, initially only building the very core of their value prop to their patient, and either outsourcing or backlogging the remaining parts of the end-to-end experience.

Capsule

• “If you think about what you need to get started in this business, you need to build a pharmacy. You need to sign a lease, and you need to construct your pharmacy, and you need to buy drugs, and you need to have inventory, you need to hire people, you need to build a brand. So, I had raised the first round concurrently with me really saying, “This is what I’m going to spend the next decade or more of my life doing. So, we had done that in May of ’15. We spent a year building out the rest of the team, the brand, the technology, the regulatory permits, everything. Then we launched the business in May of ’16.” — Kinariwala

• “We’ve built a completely proprietary software stack that powers our experience. And that has been a really key differentiator in delivering what is the only true end-to-end modernmobile first ecommerce experience in pharmacy. And nobody else has that where you can end-to-end transact and manage your medications from your phone.” — Kinariwala

Cedar

• “But Cedar Pay is just the beginning. To solve the real problems of healthcare delivery costs and pricing transparency, we need to arm patients pre-procedure with the information and self-service they need to avoid the sticker-shock when the medical bill arrives many weeks post-procedure. And Cedar is doing just that, rolling out a full patient engagement platform that will not only make the post-procedure payment process work smoothly, but engage the patient pre-procedure so she understands what she will owe the healthcare provider and can manage her information-exchange digitally.” — Scott Kupor, a16z GP / Cedar Board

Source

• “We have to build a secure, scalable data pipeline from a myriad of complex data sources. We have to continuously synchronize patient data from multiple clients’ EHRs (Electronic Health Record systems), accounting for varying EHR vendors and business processes.” — Hassan Sultan, Cedar CTO

Maven

• “We were really lucky because Zach [Zaro, former CTO] called very early on the importance of the HIPAA compliance, privacy and data ownership issues. We actually built almost everything in-house, even our analytics platform where we track how members are using the app. … We built MPractice, which is our provider app, which has an EMR, … and we built the patient software which is the video appointments and scheduling. After we built the core telemedicine components, that was when we started building a personalized pregnancy program to drive healthy outcomes.” — Ryder (31:58, slightly edited for length and clarity)

Oscar

• Oscar, initially focused purely on patient-facing products (consumer technology, plans, network contracts, customer support infrastructure) prior to open enrollment in 2014 (year after launch), and contracted insurance operations to external vendors. Over time, the company brought in operations in-house and built the tech infrastructure to support and optimize those operations, which now encompass the full-stack of health insurance.

• In 2014: “It’s an interesting thing — we’re an insurance company, and we pay the claims, right? But there’s a next step, which is where we don’t just help you manage your health care but we guide you to care. And then there’s a third step, where maybe we provide some of the medical care ourselves. Maybe we set up our own MRI.” — Schlosser

Source

Ro

• “When we first started out, for example, we spent $300K of our $400K pre-seed round on building a pharmacy. A lot of people thought we were crazy to try and do that given the level of complexity — we bet the company we could”. — Reitano

• “When we first started, we dreamt of the experience we wanted to deliver and worked backwards. This meant rebuilding the infrastructure from scratch. We built our own physician EMR, our own pharmacy software, patient application, and even our own pharmacy fulfillment center, which now power both Roman and Zero.” — Reitano (2018)

TrialSpark

• Trialspark scaled up to being full end-to-end (being the trial site, CRO layer, and tech stack all in one) two years ago, four years after launch. [Liu — 19:36]

• “In healthcare, there are often local optimizations that exist in the market that benefit incumbents, but subsequently create a macro-misalignment of incentives. Full-stack solutions can begin to align these incentives, creating drastically more efficient solutions. However, these solutions often take longer to implement so as a founder, it’s important to create a roadmap for your team to take a more ambitious, full-stack approach,” — Liu

• “Because we own and operate the full-stack, we’re able to reimagine things like biostatistics. It’s one data layer, and you can imagine everything in a clinical trial is pre-specified. If all the data is structured in the right way, we can run interim analyses in a more scalable way with software.“ — Liu — 8:44

Optimize for scale at every point in tech / processes

A known conundrum in healthcare entrepreneurship is its lack of scalability. Human interaction is core to the healthcare experience,but human capital doesn’t scale. Patients, conditions, processes, data — all have a ton of variety and too few standardized best practices. The more revenue that a business with negative unit economics generates, the more money it loses.

To scale, you have to proactively build with scale in mind, and reactively adapt new learnings for scale. As you build for your area of focus, either identify the modular components of your tech / standardizable components of your processes, or design them into the core infrastructure of your end-to-end offering. What are the tech components, processes, resources, business relationships that are translatable towards net new efforts?

Then, as Alex Zhang, former Head of Product at Cedar, puts it,

“The ideal outcome when confronted with a unique process is to converge it towards a standardized best practice.”

It will inevitably be the case that there’s a line to be drawn between the limits of a quality, focused product and where a human experience should take over, whether it be to manage large enterprise clients, or to provide the human element of care, or anything in between. Automate manual processes, and where human interaction is necessary for patients, augment that interaction to its fullest extent (but more on that later).

Capsule

• “So business is highly regulated and it is complicated to roll out the business geographically the way we are doing. Because we spent five years building this underlying technology platform, it’s lending us to accelerate the pace at which we do that because there’s a scalable system and technology platform under that.” — Kinariwala

Maven

• “Though it was harder to raise funding in the early days when women’s health wasn’t a category, being one of the only companies in the space also gave us the opportunity to really focus on co-developing solutions with our early customers and building a foundation to serve the market at scale.” — Ryder

Oscar

• “Perhaps most critically, we empower our business colleagues to perform semi-technical operations and control their own destiny. Flexibility, modularity and self-service are now the central themes driving our efforts. Key to these efforts is our own self-service configuration platform, which we call Automat.” — Sara Wajnberg, Oscar CPO

Ro

• Ro utilizes its onboarding flow process, physician infrastructure, and pharmacy for every new condition it treats.

• “While it might have looked like we were building a DTC ED company, we were building the core infrastructure to treat not just one, but hundreds of conditions via telemedicine.” — Reitano

Trialspark

• “The operations of our industry are complex, filled with manual workflows and processes. We keep an eye out for automation in everything we do.” — Linhao Zhang, Trialspark CTO

Empower a relationship between your product and your patients

By optimizing your stack for scale, you increase your ability to reach more patients. But how do you deepen your ability to provide value to each individual patient? As mentioned earlier, all patient-facing healthcare converges to an engagement problem, and this is done through relationships.

It’s critical to recognize the role your product plays in your patient’s healthcare journey. Patients don’t view “products” in abstract, they are enduring emotional duress on top of their physical problems, and will view interactions with healthcare entities through an emotional lens. By earning their trust at each interaction, you earn continued engagement, and foster a deeper relationship that creates a personal experience.

Beyond standard engagement best practices, this is now what will be table stakes:

• Branding / communication: The voice and perspective of your company needs to be one that resonates with empathy for the patient’s problems. Your messaging needs to be catered to the patient, meeting them where they are, personalized to their preferences.

• Provider enablement: Every human interaction your patient has while using your service colors their view of the experience. So, these people are extensions of your product, and it becomes imperative for you to ensure that they can deliver the best experience possible. This is done through ensuring provider quality (through vetting or network design), absorbing administrative hurdles, clinical support, and/or financial incentives.

• Long-term view: To quote Jay Parkinson, “telehealth is a feature, not a company.” This extends beyond telehealth. Your company isn’t the care or service you provide, but the overall experience. Building a relationship with your patients means being thoughtful about building an experience that considers long-term impact on patients, even when it creates short-term negatives. Doing right by your patients pays off.

Capsule

• Brand: One of the three pillars Capsule was built around was creating an “emotional resonate brand”. As such, it emphasizes tracking brand resonance, consumer engagement via retention, and customer love; its third hire was a consumer brand expert; it has deployed unconventional marketing strategies such as taxi cab advertisements.

Source

• Providers: “Doctors across New York have adopted the tools we’ve built enabling us to provide real-time feedback loops to deliver better care than has ever been possible before.” — Kinariwala

• Long-term: “Over the past year, our team has made thousands of calls to insurance companies to save our customers time. We’ve found and applied coupons that add up to millions of dollars in medication cost savings.” — Dr. Sonia Patel, Capsule Chief Pharmacist

• On taxi-cab ads: “The out-of-home is super fun for us, and it’s really fun for the team. It makes our business come to life, and it makes it really tangible. It also happens to have the nice primary benefit of being highly effective for our business in creating credibility, awareness, and trust.” Kinariwala

Cedar

• Brand / Communication: Cedar has invested heavily into its design and data science teams, who have conducted research projects with IDEO and FBI negotiators to develop a deeper understanding of different patient archetypes and the best ways to engage each one.

• Providers: “The technology we deliver to providers gives them the ability to better communicate with patients and clearly see all the steps along the patient payment cycle. It was surprising to us when we started the company two years ago that a lot of hospital and health system leaders did not know their out-of-pocket collection rates or their times-to-collect. Moreover, they couldn’t break down this information by insurance, specialty, or location. Cedar offers all these functionalities, letting organizations drill down into the details and gauge performance.”- Otto

• Long-term: Among its core features, Cedar offers customizable payment plans and data-driven discounting, and developed tools for claim denial and enabling Medicaid enrollment.

“For instance, Cedar communicates bill reminders via text message for relevant patients. In the short term, we could be tempted to increase the frequency of bill reminders via text messages because it leads to higher engagement, which in turn leads to higher collection and resolution rates. But this has a negative long-term effect: it creates high unsubscribe rates and an understandably negative patient experience; not having the option to text a patient has a negative impact on collecting future bills.” — Yohann Smadja, Cedar Head of Data Science

Maven

• Brand: “Because healthcare is so personal, building a strong reputation of compassion, confidence, convenience, and quality ended up being the foundation for Maven’s success.” — Ryder

• Providers: Maven’s extensive vetting process, (which was the first thing Maven built) accepts only 35% of providers. Maven providers are equipped through Maven’s M Practice app for administrative tasks.

• Long-term: “Telemedicine was only used for a quick prescription or answer, without building a longitudinal patient-provider relationship. Maven’s platform is designed around this relationship.” — Ryder

• “I also think that one of the most important things about building a new brand, but particularly in healthcare, is trust. And so how do you build trust with a new healthcare product? And I think having a very personalized network for women that’s built for women by women is one of the first ways to do that, so actually about 98% of our providers are women.” — Ryder

Oscar

• Brand / communications: Oscar’s Concierge system was built for building patient trust at all touchpoints, and is empowered through in-house tools.

• Providers: Oscar’s network construction process heavily leans on determining provider quality metrics and curating in-network providers accordingly. Oscar’s Clinical Dashboard “synthesizes a member’s health history, medical encounters, lab results, active and past prescriptions, hospital admits and discharge alerts, allergies, and more into a readable profile of a member’s current state of health.”

• Long-term: Oscar has continually delivered in this regard, from providing fitness trackers for free and paying members for meeting step counts in 2014 to its recently announced free, unlimited Virtual Primary Care.

• “We remind ourselves that a member’s trust is a delicate thing — earned with great difficulty and lost at the first sign of complexity or the impression that we don’t have their best interests at heart.” — Wajnberg

Ro

• Brand: Ro’s launches into new verticals corresponded with the creation of new brands (Zero — addiction; Rory — women’s health), citing the need to “ speak to this community in a very different way than we speak to our current members.“

• Providers: Ro provides clinical support and administrative lift (software maintenance, payment processing, background checks, provider credentialing, and malpractice insurance procurement.)

• Long-term: Ro publicly takes pride in its cancelling workflow, and provides discounts for in-person PCP visits, along with unlimited communication, personalized treatment plans, and access to blood tests

“We built an amazing cancel flow. That’s right, Ro has the easiest way to delay or cancel a shipment in healthcare and we’re damn proud of it.” — Reitano

• “How online visits, treatment plans, even their advertisements are written needs to make patients feel like Ro really understands them. When Ro can convey that, that’s how they build trust and eliminate the feeling of something being stigmatized.” — Dr. Melynda Barnes, Rory Clinical Director

Trialspark

• Brand / communication: For its personalized patient outreach campaigns, Trialspark has utilized omni-channel communication approaches, digital / social media advertising, and partnerships, and employed creative designers for resonant branding.

• Providers: Trialspark’s “site-in-a-box” model operates on creating standardized simplicity for its physicians, managing everything from payments and legal to logistics and data collection/analysis while also providing the necessary training and equipment.

• “TrialSpark is uniquely positioned to be a consistent, guiding voice for patients throughout their clinical trial journey. From the very first time we meet a patient — perhaps before they are even looking for a trial — all the way through to FDA submission. We ask a lot from our patients: They are people that are struggling with conditions and symptoms that may not even have cures on the market yet. We’re asking them to volunteer months and years of their time and energy making visits to our sites. We’re asking them to share intimate details about their medical histories. In some cases, we’re asking patients to leave their current treatment options and switch onto a new, investigative one. It’s our duty and our privilege to partner with our patients and physicians to drive the future of medicine forward.” — Frank Wu, Trialspark SPM

A brief note

Easier said than done, but it’s so important to find proper product/market fit in this first area of focus prior to expanding to any new category. New products you build at any point prior to this will only be additional things to juggle on an unstable foundation, continually deteriorating your patient’s experience until they leave and take their (or their aggregator’s) money with them.

Click here for Part 3

This post is Part 1 of a series. Click for the Overview, Part 2, and Part 3.

• Identify a broken patient experience

• Isolate the core opportunity

• Align monetization and stakeholder incentives to maximize patient experience

• Think beyond the status quo experience

Identify a broken patient experience

There’s a difference between finding a problem, and finding the right problem. More often than not, successful founders personally witness a broken patient experience, and/or have the depth of healthcare exposure to be able to identify patient-facing bottlenecks.

**Note: Throughout these series, I will be providing examples from these organizations. These examples are ordered in alphabetical order, not necessarily priority.

Capsule

• “I had this moment like, ‘What is going on? How could the experience of getting something so simple as a Z-pak be so terrible and broken that I wasn’t even able to get it? How is there a pharmacy on every street corner in America, and is this happening to the millions and tens of millions of Americans who go to the pharmacy way more often than I do?’ So, I just started digging in.” — CEO Eric Kinariwala

Cedar

• “My fiancé had a terrible billing experience after a short ER visit in Manhattan hospital. The two-hour visit resulted in three separate bills that were very difficult to understand. One of the bills ended up going into collections. I thought, if a first-tier hospital in Manhattan has such an antiquated billing technology, the entire industry must be broken.” — CEO Florian Otto

Maven

• “About six years ago, most of my friends were starting to have children, or, like myself, thinking about having children, and I was witnessing their experiences with a broken healthcare system. Even though healthcare is a primarily female market it wasn’t built to completely support women, especially women in the workforce.” — CEO Kate Ryder

Oscar

• “Having a baby is one of the only positive reasons to interact with the healthcare system, but our experience left us confused about our options and costs. There wasn’t an easy way to interact digitally with providers. There was no good way to understand what the costs would be for a procedure. And, generally the insurer didn’t seem to be particularly pleasant to talk to.” — CEO Mario Schlosser

Ro

• CEO Zachariah Reitano’s personal experiences with erectile dysfunction and its ties to deeper health issues (provided in-depth here) inspired him to obsess about healthcare.

• “With Ro, I found something that I feel so fulfilled obsessing about. I get so excited going down the healthcare rabbit hole, reading some thesis or economic report from 1975 about physician supply, trying to understand the space better to serve the products we build.” — Reitano

Trialspark

• “In my graduate work, my lab discovered a few candidate drugs. When we talked to pharma companies about exploring a clinical trial, we were exposed to how expensive and time consuming that process was. It was eye-opening to learn how they review hundreds of drug candidates a year, and that a new clinical trial could cost hundreds of millions of dollars across all phases.” — CEO Benjamine Liu

• “I think the industry is recognizing that we need to engage more patients in clinical trials, and one way to do it is to allow patients to participate with their doctors.” — Liu, 5:50

Isolate the core opportunity

Take the broken experience, and drill down to find the corresponding underlying problem in the healthcare system. This involves understanding the status quo of healthcare within that domain. What are the assumptions upholding the current experience? Where are the bottlenecks along the value chain?

Where is it complex, expensive, ineffective, and impersonal? Why are things the way that they are? Finding answers to these and similar questions unveil wrinkles of competitive advantage.

Regulation changes also create pockets of new opportunity. The resulting gold rush can sometimes lead to the adoption of poor products that meet the new criteria, but also introduce new problems. When riding these tides, it’s crucial to build with a patient-centric mindset, both for the sake of the patient and for longevity.

Source: Julie Yoo, GP @ a16z — Healthcare: The Great Unlock

Capsule

• Key opportunity: Resources dedicated to inefficient physical pharmacy locations could be best used towards improving the patient experience.

• Additional insight: The pharmacy is the most frequent interaction in the healthcare ecosystem, but these interactions are underutilized.

• “Why do they spend so much money on rent, and what happens if we actually didn’t have thousands and tens of thousands of stores, and we took all of the money that you invested in rent, and we put that back in hiring better people who can care for you better?” — Kinariwala

Cedar

• Key opportunity: The complexity of patient billing creates friction for patients trying to resolve their bills, although healthcare organizations are increasingly reliant on patient payments.

• “Despite the growing importance of this revenue stream, providers fall short in effectively capturing patient monies. There is a common misconception that people don’t want to pay their healthcare bills and avoid the task. Research indicates that most patients are willing to pay — yet only 30 to 40 percent actually do. What this means is that healthcare organizations have more of a process problem rather than a patient-willingness issue.” — Otto

Maven

• Key opportunity: Despite women controlling about 80% of healthcare decisions, women’s health itself is underserved, with a severe lack of products and brands built for women by women.

• Additional insight: Poor maternity leave policies are detrimental to workforce productivity for self-insured employers, an underexplored customer group at the time of Maven’s launch.

• “Why, for example, aren’t there easier ways for women to navigate fertility problems, get postpartum care or access birth control? Why is seeing a good doctor so difficult and inconvenient that getting a prescription for something like a urinary tract infection requires putting your entire life (and your family’s life) on hold? How is it even possible that the United States has the worst maternal mortality rate in the developed world?” — Ryder

Oscar

• Key opportunity: User engagement tied with virtual care is underutilized in healthcare, but will unlock lower unit economics while maintaining or even improving user satisfaction and care quality.

• Additional insight: Health insurers have great data visibility into the healthcare ecosystem, but haven’t taken advantage of this.

• Key regulation: The passing of the ACA opening health insurance exchanges created the ability for insurers to go direct-to-consumer, as well as catalyzing Oscar’s early growth by increasing the market in downstate New York from 20,000 to one million.

• “As health care becomes increasingly fragmented and crowded with point solutions that fail to gain traction, value in health care will accrue to personalized, engaging member experiences.”

• “As the insurance company, we’re in a unique situation to essentially say that the more cases our telemedicine doctors can resolve, the more we can lower total cost of care and increase the satisfaction for the member.” — Schlosser

Ro

• Key opportunity: The convergence of emerging telemedicine regulation, patent cliffs, and increasing deductibles creates an opportunity for direct-to-consumer healthcare, with high pain point conditions as a clear entry point.

• Additional insight: Physicians and care providers are seeking alternative income sources and flexibility as HIPAA-compliant tools that can provide those benefits become easier to build.

• “The insight that we had is that there are certain health conditions that prevent people from thinking about their long term health. We wanted to start with ED because we started with a present pain point that’s indicative of someone’s underlying health, that builds a bond and creates trust with a physician, but the relationship is founded on something that we know we will have the opportunity to add additional value to their life over time.” — Reitano

Trialspark

• Key opportunity: Finding and enabling eligible patients to participate in clinical trials expands the total pool of patient participants.

• Additional insight: Variability and inefficiency in clinical trial operations exacerbates costs and prevents participation from willing physicians despite significant shortage.

• “If you don’t live near an academic center or an existing trial site, you often will need to travel far distances to access clinical trials and leave your doctor to participate. Our model changes this, enabling patients to participate in clinical trials at their local doctor’s office.” — Liu

Align monetization and stakeholder incentives to maximize patient experience

This isn’t just the nice thing to do, it’s the smart thing to do.

People demand results. In order to make sure healthcare is outcome-driven, your company should be outcome-driven. By structuring your business model so that you’re paid for the value you create, you put your money where your mouth is, and champion the patient experience as a result. Everybody needs checks-and-balances.

Meanwhile, for other stakeholders, participation in improving the patient experience was previously not conducive to their best interest. Create the system where working for the patient is intuitive, not full of friction, for everyone involved. This then aligns with your mission and your bottom line.

Alignment creates an easier sell and hones your organizational focus. Patients are going to vote for the best experience with their feet and their wallets. Enterprise clients see an oversaturation of solutions, most of which deliver mediocre or unclear results while racking up fees that don’t match the created value. Since you have to improve the experience to succeed, it’s clear to buy what you’re selling — an easy pill to swallow, if you will. It’s worth noting here of course that, as the market grows increasingly saturated, the demand for better outcomes and smarter incentives alignment grows as well.

With the customer acquired, your focus turns to retention. In DTC healthcare, you seek a maximized LTV / CAC ratio, whereas for B2B2C, you optimize for efficiency of delivering results for clients. With incentives aligned for the patient, the differentiating retention strategy in all patient-facing healthcare converges to patient engagement. By finding new ways to elevate the patient experience (through convenience, cost, outcome, or personalization), you improve engagement, and continue to retain and/or reduce operational costs as a result.

B2B2C

Capsule

• Free delivery for patients, Capsule generates revenue per prescription filled primarily through insurers and collection of copays.

• Kinariwala quotes that only “fifty percent of prescriptions go filled”, mainly due to high patient friction. Capsule reduces this friction to achieve higher fill rates and improved recurring revenue per patient, balancing out its heavier investments into marketing / branding and delivery logistics. It’s also possible that Capsule charges premium rates to insurers by demonstrating improved patient adherence.

• Capsule provides software to doctors, hospital systems, insurers, conventional pharmacies, and manufacturers, delivering information on patient adherence, inventory prediction, etc. that improves clinical and financial outcomes.

Cedar

• Cedar reduces frictions, increases flexibility, and adds transparency to improve a patient’s historically terrible billing experience. For providers, this increases collection volume while decreasing collection time, with the added benefit of an improved patient satisfaction rate. Cedar is paid a percentage of bills collected through the platform.

• “The Cedar business model is based on aligned incentives. In short: we get paid when our client partners see results. Our performance-based approach helps ensure that we continue to deliver the most innovative solutions to providers and an outstanding experience to patients.” — Otto

Trialspark

• Promising faster, better clinical trials, Trialspark doesn’t “get paid by the sponsor unless the data gets checked by the monitors.” [Liu, 12:12]

• To accomplish this, Trialspark addresses bottlenecks throughout the trial process such as patient recruitment, which accounts for up to 40% of clinical trial expenditure. By taking the clinical trial to where eligible patients are and providing a smoother experience, Trialspark is able to dramatically reduce trial attrition compared to its incumbent CRO competitors.

• “There are also a few misaligned financial incentives: the CRO business model is a cost-plus model, so the more variability there is at the site level, the more billable hours they can charge to the sponsor.” — Liu

B2C

Maven

• Maven’s B2C product is offered at less than common copays. Early on, Maven appealed to mission-driven providers to accommodate for the lower compensation rates they provided due to being an early-stage company.

• As one of the early players in the self-insured employer market, Maven provided better fertility experiences and alleviated one of employers’ top benefits expenses in pregnancy and maternity leave. Here it was a double-whammy: reduced medical expenses and faster return-to-work, in addition to improved employee productivity & satisfaction.

• “We cover everything from helping women transition back to work in a healthy and productive way (to help drive retention) to helping employers reduce maternity-related costs, which are often the #1 or #2 healthcare cost for companies.” — Ryder

Oscar

• “The other important characteristic was that the economic interests of insurers and patients are actually aligned — they’re both better off if costs come down. If an insurer is able to help keep its members healthy, to the point where they need less care or even less-expensive care, both insurers and patients win.” — Schlosser

• “If we prove we can pay claims faster, maintain a selective network and show good NPS scores, the providers want to work with us.” — Schlosser

Ro

• “Ro makes money in two ways: 1. Providing software and administrative services to providers. 2. Fulfilling patients’ prescriptions if they choose the Ro Pharmacy Network.” — Reitano

• Ro’s patients pay for their prescription with cash, and $15 to physician consultation and ongoing care, which goes purely to the physician (minus processing fees).

• Reitano wrote up an extensive dive into Ro’s DTC metrics. At large, with cash pay prescription fulfillments being its significant revenue driver, Ro’s incentives are aligned with the patient’s best interest. By focusing on continually adding value for the patient, retention metrics improve, justifying CAC and maximizing LTV / CAC. This results in a “virtuous retention cycle”, well-drawn by Reitano below.

Source: https://medium.com/ro-co/dtc-metrics-explained-29ff99ff5657

Think beyond the status quo experience

Finally, you need to be bold in how you set your vision. Some of these companies are still riskier, relatively unproven models, but they’re swinging for the fences trying to change healthcare as opposed to approaching it incrementally. Reimagine healthcare with new standards in a way that makes sense, creating an escape from the pointlessly convoluted maze of American healthcare.

Capsule

Capsule seeks to be “ the hub of healthcare”, as the “single most trusted, simplest place for consumers to engage with their health on their phone.”

Cedar

“In the next 5 years, we aim to be the leading comprehensive healthcare consumer engagement platform for providers, engaging with over half of the US patient population to improve their lives.” [Source]

Maven

“For Maven, success in 5 years means enabling women everywhere to get the information they need, affordably, to help them make good healthcare decisions for themselves and for their families.” — Ryder

Oscar

“We founded Oscar to catalyze in American healthcare the same positive market forces that have long existed across every other major consumer market: the need for a clear and differentiated value proposition, competition, and choice.” — Schlosser

Ro

“Our ultimate vision is to be every patient’s first call. So whenever they have a question or a symptom, they reach out to Ro first.” — Reitano

Trialspark

Reverse Eroom’s Law by running trials “50% faster“ than the previous year [Liu, 13:11]

Click here for Part 2

Note: in this series, by “patient-facing healthcare”, I am referring to all elements of healthcare where patient interaction is significant, inclusive of both direct-to-consumer models and B2B2C models. Excluded in this definition are solutions that are purely B2B intended on solving problems for just organizations or care providers, with patients as a lesser priority.

Opinions expressed here are my own.

Intro:

Building anything in healthcare is hard. Just ask Nikhil Krishnan (Why It’s Almost Impossible to Build a Health Care Startup That Works). Or Halle Tecco, Rock Health’s Co-Founder, on the common pitfalls in digital health. Or Jay Parkinson on his many stabs at dragging primary care to the present. Or Kyle Hill building in home care with HomeHero. Or Jamie Campbell on the complexities of medical data portability. Or …

Building in patient-facing healthcare is even harder. At least when you’re building products centered around healthcare’s incumbents, there have been success stories (EHRs like Epic / Cerner / Athena, Veeva, Flatiron, etc.). Why is this the case? Among a million reasons, some of the most critical ones are:

• Incentives: At the macro scale, decisions and money primarily flows through larger incumbent players like payers and providers, with patients commanding little leverage. With essentially no incentive to build strictly for the patient, considering the patient experience isn’t mission-critical, it’s charity. Good luck convincing thin-margin incumbents to spend a proactive dime for patient-centered innovation.

• Building: Building and maintaining software that’s consistently compliant with regulation while overcoming the healthcare data bottleneck presents massive hurdles prior to even developing a core product. (Side note:check out my other series where I dive into healthcare’s data problems) Then comes the problems with actually creating a solution: finding the right balance between tech and human interaction, accommodating healthcare’s complexities, optimizing for scale, etc.

• Opacity: The conspicuous difficulty of this arena deters inflow of talent and new perspectives. This deterrence reduced the amount of precedent for the critical problems of building a business (figuring out product, pricing, selling, etc.). This, combined with the long feedback loops, meant that new entrants had less data points to work with compared to other industries. Essentially, throwing darts blindfolded.

For patients then, healthcare is complex, expensive, ineffective, and impersonal, but nobody wants to solve their problems. That’s messed up.

However, times have changed. Recent economic and regulatory trends in healthcare point to a crucial development: patients are steadily becoming both where the money and the data reside.

Now, it’s no longer just good will, it’s opportunity. Incentives are slowly aligning in healthcare so that, finally, success in healthcare necessitates actually solving problems for the patient. The historic neglect of prioritization of the patient means that delivering an improved patient experience can creates disruptive levels of differentiation. It’s almost exciting. (What a weird thing to say in healthcare!)

So then let’s take that earlier statement, and flip it on its head. If healthcare is complex, expensive, ineffective, and impersonal for the patient, how can you create an experience in healthcare that’s convenient, affordable, outcome-driven, and personal?

Luckily, there’s precedent to follow. I dove into the paths of high trajectory patient-facing healthcare companies, namely Capsule, Cedar, Maven, Oscar, Ro, and Trialspark. Though operating on very different pieces of the healthcare pie, these organizations’ paths to success have a lot of overlap. This overlap distills into a three step playbook:

1. Vision —Identify a broken patient experience that speaks to deeper complexities and misaligned incentives in healthcare, then connect this to a bolder vision of healthcare’s future.

2. Build — Succeed in building a full-stack experience that strives for scalability while emphasizing a relationship with the patient.

3. Scale — Utilize your scale-optimized components and established brand to reach the full potential of your product, then extend beyond it towards your greater vision.

Building is your problem to solve, and incentives are slowly aligning. Here’s my stab at breaking open the black box of building in healthcare and reducing the opacity. Let’s break this down.

Thanks to Nikhil Krishnan, Alex Zhang, and Rachel Lin for providing thoughtful feedback.

If you have any questions or thoughts, or would like to connect, please don’t hesitate to reach out via LinkedIn, Twitter, or e-mail at btlee215@gmail.com!

Sign up now so you don’t miss the first issue.

Subscribe now

In the meantime, tell your friends!

This is Part 1 of a series. Here’s the Overview, Part 2, Part 3, Part 4 and Part 5.

The current healthcare landscape.

Let’s level-set on the healthcare landscape. The three major stakeholders that are consistently influencing the dynamics of healthcare are patients, providers, and payers. Overall, the power dynamics of healthcare is shifting to the patient, and providers and payers are feeling more pressure than ever to accommodate patient demands.

Patients:

• Patient access to care is rapidly diminishing as the pandemic makes known issues in healthcare more pressing than ever. The impact of social determinants of healthcare is a growing threat, as lower-income communities are hit harder. This population is growing — jobless claims have spiked, reducing income levels of millions of families. Meanwhile, healthcare providers were already in shortage, but this is exacerbated by COVID-19. Additionally, operating in close proximity to each other, providers are even more in danger from contaminating each other.

• The percentage of persons under 65 enrolled in a high-deductible health plan (graph below) has steadily increased. Meanwhile, the increasing costs of American healthcare are eventually passed down to the consumer as 17.8% of premiums go to operating costs. These factors combined (along with a large uninsured population) means consumers are increasingly paying more out-of-pocket (OOP) for their health.

• The historic lack of transparency and competition between health systems have uncoupled quality of care from its cost.

• The above, combined with the fragmentation of the healthcare experience has meant that the patient is playing darts blindfolded, hoping to hit the bulls-eye of a healthcare experience that meets their needs. This could mean coordinating their care (potentially between multiple specialties), understanding the complexities around insurance and billing, all the while managing disruption in their lives and the difficulties of their healthcare needs. As the system gets more complicated, the individual patient finds it less valuable (while paying the same or more).

• The patient population is becoming increasingly digital-native. 60% of consumers expect their healthcare digital experience to mirror that of retail, and expectations for healthcare to be personalized and accessible are growing.

• Patients pay with their feet towards the healthcare experience with the best price, quality, and convenience. (check out this video with Ro CEO Zach Reitano for more detail)

Providers and health systems:

• Tremendous costs from administrative waste ($265.5B), pricing failure ($240.5B), failure of care delivery ($165.7B), overtreatment / low-value care ($101.2B), fraud and abuse ($83.9B), failure of care coordination ($78.2B), among other things, cripple health systems and also payers. 20–25% of healthcare spending is wasteful — as seen in the chart below, even despite the high spend, the US leads its peers in preventable deaths.

Source: Mary Meeker’s 2019 Report

• Physician burnout from administrative bureaucracy and high stress, though recently decreasing, is a prominent threat to healthcare and directly affects cost and patient outcomes. Clinicians spend 5.9 hours of their 11.4 hour work-days on EHR and administration related tasks.

• Instead of innovating to solve inefficiencies, health systems uses the unscalable approach of throwing bodies at the problem (see graph), exacerbating costs year-over-year.

• Meanwhile, median operating margins for hospitals are razor-thin [source] at 1.7% in 2018, meaning healthcare executives have to overturn every stone to unearth every last pebble of profit.

• The increase in OOP costs for patients has led to increased leverage for the patient population — patients have the potential to affect 61% of all healthcare spending. Though a consumerized experience was previously just a strong desire in the healthcare industry, it has now become table stakes for health systems [#2 for healthcare execs]

• PR / moral nightmares, such as surprise billing and harmful, aggressive collections tactics, sour the reputation of hospitals in the consumers’ eyes.

• The Affordable Care Act created guidelines around the first value-based care (VBC) model, which ties payment of care to value to the patient as opposed to volume of services. Since then, the adoption of VBC models and people covered under such contracts is steadily increasing, which means that health systems revenues will be increasingly tied to patient outcomes. This has in turn increased provider attention and scrutiny in population health, coordination of care, and improved understanding of quality.

Payers:

• Functioning as an aggregator of consumers’ purchasing power and mediating patient-provider interaction, payers have historically been the ones with the most influence on the healthcare system. Unfortunately, as Sidney Primas notes, payers have also been incentivized to increase the cost of medical care over time for profit.

• However, tides are shifting in healthcare — just like with health systems, patients are gaining leverage and are increasingly demanding a consumerized insurance experience. The trend of increasing enrollments in Medicare Advantage adds to this effect. There’s a lot of ground to make up here, as health insurance companies ranked in the bottom five of industries in customer satisfaction, lower than airlines.

• New insurance startups are riding the tides of these trends to threaten incumbents. Clover (see graph from Mary Meeker’s 2019 Report below), gained market share through personalizing healthcare plans to the needs of members. Oscar utilized a deeply personalized and engaging patient experience en route to a $3B valuation. (As Nisarg Patel cites in his insightful dive into Oscar, “41% of [Oscar’s] members are monthly active users. Compare that with telemedicine market leader Teladoc, where only 2–3% of its members use the service annually”)

• Payers also suffer from the costs of inefficiency mentioned above for health systems (spending almost $250B of excess a year on administrative issues such as eligibility processes, coordinating reimbursement, and handling claims). To optimize margins, they are driven towards investing in methods to deliver improvements to enhanced patient experience/outcome and reduction of administrative and medical cost.

• In particular, VBC has proven more cost effective in providing better patient outcomes compared to fee-for-service plans, according to a Humana report. This has led to a need for tighter collaboration between payers and providers, and formalized partnerships between the two stakeholders (such as accountable care organizations (ACOs), 50–50 joint ventures) are gaining in number.

• Payers are recently turning their eyes towards startups as they now “need to differentiate in a stagnant commercial market and compete,” note Venrock partners Bob Kocher and Bryan Roberts.

In sum:

• If they can even access care in the first place, patients are using larger shares of their income to pay for a healthcare experience that is unnavigable and broken, so they need the right tools to better navigate their care and have improved. outcomes. However, this also means patients have increased leverage.

• Health systems need to meet patient demands and improve outcomes to optimize margins as costs related to burnout, waste, and inefficient processes pile up to staggering amounts.

• Formerly at the top of the healthcare pyramid, payers now face an existential threat as patients command more leverage and the attention of new healthcare entrants/competitors. They too need to innovate through personalized patient experiences, tech-enabled services, and process efficiencies to differentiate in a much more competitive market.

• Value-based care initiatives are driving payer-provider partnerships in which each focus on respective strengths in order to create a differentiated consumer experience. This in turn requires tighter collaboration between the two stakeholders.

Next, let’s learn about the emergence of digital health as an attempt to resolve these problems. Click here for Part 2.

This is Part 5 of a series. Here’s the Overview, Part 1, Part 2, Part 3, and Part 4.

Where are we now?

On March 9, 2020, the Department of Health and Human Services finalized two rules that will start to loosen the stranglehold of the data bottleneck. In a high-level summary, the rules:

• Define and prevent information blocking practices (as an extension of the Cures Act)

• Establish requirements around a FHIR-based Patient Access API that supports a patient’s ability to access and control their electronic health data from providers for free, through third-party applications

• Require health plans in Medicare Advantage, Medicaid, CHIP, and on federal Exchanges to share claims, encounter, and clinical data electronically with patients and their EHR, with an implementation deadline of 1/1/21

• Require payers regulated by the Center for Medicare & Medicaid Services (CMS) to implement a FHIR-based Provider Directory API, with an implementation deadline of 1/1/21

• Require all Medicare and Medicaid participating hospitals to send notifications of patient events (admission, discharge, transfer) to other providers

• Require payer-to-payer data exchange at the request of the patient, starting 1/1/22

• Require EHR vendors to enable an export of all patient records when a provider switches between vendors

There are some concerns to note. For now, these rules aren’t enforced through monetary penalty. When such penalties are established, it remains to be seen whether systems will see them as a cost of doing business. Additionally, the deadlines have been described as “extremely aggressive” and difficult to realistically and properly implement. Epic, on brand with its history of being uncooperative towards interoperability, rallied together almost 60 health systems against the rules.

However, despite these concerns, it is still difficult to overstate the potential impact of these regulations when combined with recent industry trends. We are most certainly at a definitive inflection point in healthcare. To recap:

• The patient is becoming king in the patient-provider-payer dynamic.

• The digital health industry is growing as companies seek to either solve problems for the patient or empower providers and/or payers in creating a better patient experience.

• The COVID-19 pandemic has dramatically increased attention, pressure, and demand on the healthcare system to improve its inefficiencies.

• The problems associated with healthcare data (lack of interoperability and standardization, negative influence of legacy players) have been core contributors to recent American healthcare inefficiencies and the bottleneck to effective digital health solutions.

By requiring compliance from providers, payers, and vendors to play along, the new rules start the breakdown of pre-existing barriers that will enable healthcare organizations to first bring healthcare into the 21st century.

An opportunity in healthcare

Concurrently, the time is now extremely ripe for a healthcare API infrastructure solution that facilitates interoperability and connection of healthcare’s multi-sided marketplace. In a recent a16z podcast episode (a must listen), general partner Julie Yoo outlined the most compelling implications of the above rules, the most notable being the potential emergence of interoperability infrastructure players in healthcare.

Josh Nussbaum’s post on API startups lists the key factors that need to be at play for the success of an API startup, among which are (paraphrased):

• Industry large enough to support a large number of API calls

• Recent changes that increase need for API’s data and/or functionality

• Core to a “new economy”

• API’s data and/or functionality is difficult to manage

• Creates net positive value

• Creates network effects that add to business defensibility

Sound familiar?

In the paraphrased words of Balaji Srinivasan (a16z general partner, former Coinbase CTO), a function that takes simple inputs with high backend complexity is a billion-dollar function. API startups that provide developer-first products for industry bottlenecks have the capacity for tremendous growth — for instance, take payment processor API startup Stripe’s growth over little more than a year.

Perhaps the most potent analogy to this scenario is Plaid’s impact on the financial services industry. Much like the healthcare industry, the financial services industry suffered from data interoperability issues that impeded fintech innovation and consumers lacking robust access to their financial data. By building the developer-friendly API infrastructure that enables connections between the consumer, banks, and third parties while also providing operational flexibility for financial institutions, Plaid became the backbone behind the modern consumer financial experience, powering companies such as Venmo, Coinbase, Acorns, and Betterment.

Source: Mary Meeker 2019

Similarly, the cream that rises to the crop of healthcare interoperability players will bridge the gap between healthcare’s present and its tech-empowered future. First, these solutions will gain market share by aiding panicked providers and payers meet rule deadlines and connecting them to digital health solutions that expand their capacity to combat COVID-19. Among the most notable companies currently developing API solutions to these problems are the following:

• Ribbon — Provider data, care & insurance navigation, cost & quality (with Julie Yoo on their board)

• Particle — EHRs (care coordination, reimbursement, operations, access)

• Redox — EHRs (care coordination, patient engagement, telehealth, devices)

• Noyo — Health insurance

• Eligible — Revenue cycle management and billing

• Truepill — Pharmacy

Of these, the most compelling are likely the ones laser-focused on provider data and patient access (due to the pressure exerted by the new rules), such as Ribbon, Particle, Redox.
*Not an API startup, but PatientPing, which provides real-time care coordination notifications, is also noteworthy here in light of the new rulings.

Source: Julie Yoo of a16z

Once the most urgent needs are met, these interoperability players and others will be key in ushering in the new age of healthcare as new digital health companies take advantage of the more freely flowing data pipelines to build much-improved, patient-centric healthcare experiences. This will potentially include, but certainly will not be limited to:

• Deeper incorporation of current healthtech solutions into healthcare value chain

• Deeper utilization of consumer health data in care delivery

• Patient autonomy over medical records (and an experience around sharing them)

• Holistic care coordination (of all dimensions of care — acute, ambulatory, behavioral, home, etc.)

• Optimal matching of provider to patient

• Consumerized healthcare engagement experience, involving frictionless access provider/payer shopping experience based on price and quality transparency

• Real-time health insights and care management for patient (especially in chronic care)

• Unlocking insights from compiling real-world evidence

• Deeper personalization of coverage and care payment

• Disruption of EHRs

All such solutions will depend on the infrastructure created by these and other API companies to deliver on their value proposition. Each instance of a potential API call represents potential revenue for these infrastructure players in becoming billion-dollar companies.

In sum:

• The HHS released new rulings that require stakeholders to promote interoperability, through API implementation and unimpeded data exchange.

• This, combined with recent trends, points to the current moment being an inflection point in healthcare.

• Solutions to the new rules’ requirements, such as API interoperability infrastructure, will lay the groundwork for the new era of tech-empowered healthcare. Companies providing these solutions can follow the precedent of startups in other industries (Stripe, Plaid, etc.) in becoming unicorns.

Series takeaway:

Past and current trends, the scope of healthcare’s data problem, and new pressures from regulations and a global crisis set the scene for a company to follow precedent in becoming a unicorn for healthcare data, empowering the future of care.

This is Part 4 of a series. Here’s the Overview, Part 1, Part 2, Part 3, and Part 5.

How does healthcare’s data problems impact digital health?

As patient experience and outcome becomes increasingly crucial to healthcare, more and more healthtech startups are entering the foray. There’s two schools of thought for the survival of these startups. While Kevin O’Leary believes “if you’re trying to disrupt the healthcare system, you are going to need to partner with incumbents effectively in order to stand any chance”, Neal Khosla’s take is that companies need to become “existential threats” that “[scare] the shit out of the healthcare establishment.”

Essentially, to be truly disruptive, these startups have to demonstrate that they can create a viable business through solving a core problem in patient experience, through two options: partner with providers and/or payers and align their incentives & behaviors for the patient’s interest, or replace these incumbents to directly interact with the patient themselves.

In pursuing one of these two options, healthtech startups will inevitably encounter the complexities of navigating healthcare data. Here, too, exists two options: integrate with existing data sources, or build your own data infrastructure.

For healthtech companies taking the partnership route, integrating with legacy data infrastructure will be a bitter pill to swallow. When your clients are change-averse health systems or payers with paper thin-margins, the ability to provide immediate ROI is the line between life and death.
However, integrations are expensive and time-consuming. Conservatively, it will take an industry-leading integration team (involving full-time account managers, engineers, etc.) 1–2 months per new enterprise healthcare client, and that number can quickly expand based on variables such as type of EHR, implementation of change management processes, etc. (for larger, slower organizations, this could easily extend to +6 months). Take this process and multiply it by the amount of sales cycles a start-up could go through, and it quickly becomes a multi-million dollar scaling problem for the start-up over time.

Talk to employees within the start-up, and you’ll see this materialize into multiple, tangible pain points. Partnerships and account management will seek faster turnaround time and reduced costs on client go-lives, but implementations and integrations are slowed with every new data quirk. Product wants lowest scope efforts for highest impact, yet any build involving healthcare data requires high scope for an incremental amount of impact. Engineering dreads wrangling clunky legacy data and interacting with ancient code bases lacking detailed, friendly documentation. The organization as a whole feels a drain as these data problems start to shift focus away from delivering value to end users.

Alternatively, if a company seeks to circumvent integrating with cumbersome systems, it will have to own the experience itself and build its own data infrastructure. This is the approach for companies like Oscar, Flatiron, Ro, which have redesigned claims systems, EHRs, patient relationship management platforms, etc. to be able to ingest and utilize data for their specific use cases and deliver a modern healthcare experience.

Obviously, building modern versions of these systems is a laborious undertaking that requires significant time and capital — it’s the reason that only the most well-capitalized companies have found any degree of success. Yet, beyond this, maintaining the usability of the data itself persists as a massive business problem. Oscar, for instance, makes 2,200 data edits per day to its own provider directory to maintain accuracy and effectiveness.

Finally, as care continues to unbundle from the hospital, these new healthtech companies are developing new ways to access, pay for, and experience your care, creating new sources for healthcare data. However, it’s critical that these solutions create a synergistic ecosystem that elevate the patient experience, not exacerbate fragmentation. All participants need to be able to access and contribute to a patient’s longitudinal data history with minimal friction. This in turn will dramatically improve their value proposition, addressable market, and utility for the patient.

In healthcare’s current data landscape, this is simply not accomplishable. Aside from hindering healthtech companies of today, data interoperability and quality issues slow healthtech from bringing in the future of care.

Ultimately, all healthtech startups, regardless of their approach, are individually reinventing the wheel every time they interact with healthcare data, unable to maximize their utility for the patient. This is healthcare’s bottleneck.

In sum:

• To be successful, digital health companies have to solve for an improved patient experience through two paths: partner with incumbent players and integrate existing data sources, or replace incumbents and build their own infrastructure.

• Integrating existing sources is expensive, time-consuming, and difficult. Variables in the process mean that this has to be repeated for each new sales cycle, becoming a critical scaling issue decreasing business health and increasing employee backlogs.

• Both of these paths cause each healthtech start-up to reinvent the wheel, creating a bottleneck.

• Building new data infrastructure is impossible in and of itself, but managing that data’s usability significantly adds complexity.

• As healthcare data increases in quantity and variety, the opportunity to create a healthcare ecosystem that works for every stakeholder is larger than ever. Healthcare’s data bottleneck actively impedes this from happening, driving fragmentation instead of synergy.

What’s happening right now that points to the alleviation of this bottleneck? Click here for Part 5.

This is Part 3 of a series. Here’s the Overview, Part 1, Part 2, Part 4 and Part 5.

With the digital health industry seeing such steady growth as a result of tailwinds from industry trends and increased demand from a global pandemic, a core set of questions arises: can the industry maintain the growth and meet the demand? What challenges lie in the way?

The Problem with Healthcare Data

The answer to these questions lies in the titular bottleneck. Let’s take a second to understand the context surrounding healthcare data and how it creates a bottleneck for the digital health industry.

Largely speaking, healthcare data lacks interoperability and usability. For the sake of simplification, I’ll observe healthcare data in two large buckets: health data and administrative data. Health data will refer to all data that’s relevant to a patient’s health, and administrative data refer to all data that’s relevant to the administration of care, such as provider data, insurance coverage data, claims data, etc.

Health Data

Prior to 2009, a vast majority of hospitals still operated through outdated processes, leaning on fax, paper, and landlines. The HITECH Act (2009) incentivized the adoption and utilization of health information technology (to the tune of $38B from the CMS by 2018), resulting in an explosion in adoption of EHRs (electronic health records) from 8% across hospitals and health systems in 2008 to 96% in 2017. The problem? These EHRs are unsatisfactory to say the least.

The largest EHR players at the time of the HITECH Act (namely Epic, Cerner) started out in the pre-World Wide Web-era building software for particular hospital record keeping functions, such as billing, and tacked on features over time to accommodate the growing list of demands from client health organizations. They eventually became one-size-fits-all solutions for all of health systems’ software needs, despite initially being designed just for records.

To paint the picture of the antiquated nature of these solutions, many of the main EHR players use MUMPS, an outdated programming language invented in 1966 (Epic, MEDITECH, GE Healthcare, the list goes on). Not only did the obsolete software and design practices lead to a burnout-causing, unusable product, the costs surrounding EHR contracts combined with implementation and change management overhead are sky-high. To only make matters worse, the dollars and hours attributed to these contracts mean that healthcare executives are emotionally committed to their systems’ EHRs. Many health systems are stuck with their EHR for the foreseeable future.

As enormous as these downfalls are, they may not even be the worst part — that honor goes to the lack of data interoperability. EHRs store massive amounts of information, and are the primary repositories for patient health data, but aren’t designed towards using or sharing the data in a meaningful way. Due to this, less than half of hospitals electronically integrate data from other hospitals outside their system, and many hospitals still resort to printing and faxing to share data, despite having spent billions of dollars implementing EHRs.

The above flow diagram (from another great read, The Fragmentation of Health Data by Datavant) displays only some of the steps where healthcare data is created and exchanged in a patient’s interaction with the healthcare system (notably, the complexities of provider-payer data exchanges are not included). For many health systems, EHRs are the owners of most of these steps and significantly involved in others. And, if you recall, they are poorly designed one-size-fits-all products that are not conducive to sharing or analysis.

At each of these points, multiple data points are generated by multiple entities, perhaps incorrectly or duplicatively, then exchanged. Thus, as each transaction increases potential for additional complexity / fragmentation, the usability of health data decreases.

This has materially impacted patients, consumers, and physicians. In fact, in the time of COVID-19, it’s a matter of life and death. CMS Administrator Seema Verma said, “We’re seeing this with the situation with coronavirus and patients coming into a hospital and providers are stressed with dealing with sick patients and they have no time to gather patient information to have a complete medical record.”

In a great breakdown of the impact of medical record fragmentation, Ilana Yurkiewicz, Stanford physician and medical journalist, elaborates, “When I meet a new patient, I have to gather slips of these histories from various sources — electronic records, paper documentation, outside faxes, notes in wallets, family members — to piece together a meaningful narrative. … When a patient with a complex medical history … arrives under my care, it’s like opening a book to page 200 and being asked to write page 201. That can be challenging enough. But on top of that, maybe the middle is mysteriously ripped out, pages 75 to 95 are shuffled, and several chapters don’t even seem to be part of the same story.”

Meanwhile, access to medical records is nearly impossible for a consumer. To cite a high-profile scenario, former Vice President Joe Biden said, “I was stunned when my son for a year was battling stage 4 glioblastoma. I couldn’t get his records.”

Part of the reason EHR companies viewed data as proprietary (as opposed to belonging to the patient), and thus were prohibitive to other EHRs and healthcare software companies in accessing patient data, causing a deeply fragmented data landscape.

Particularly, Epic (the market leader) has taken a “walled garden stance” as it seeks to be the only EHR on the market. It’s been historically uncooperative in data interoperability, charging fees for data exchanges with non-Epic providers, impeding data access, and lobbying against interoperability rules.

Though Epic has responded to interoperability criticisms through a variety of ways, these efforts are half-hearted. For instance, gaining access to AppOrchard, an app store-type platform for housing third-party tools on Epic, has been known to be an expensive, time-consuming, unnecessarily complicated process that requires playing by Epic’s rules.

Some notable efforts have arisen to combat these issues, listed below:

• The 21st Century Cures Act (2016) — policy that, among other items, creates clarity around interoperability and prevents information blocking

• Fast Health Interoperability Resources (FHIR) — the most recent and popular data API standard created by HL7, structured in programming-friendly languages

• The CommonWell Alliance is a non-profit collaboration between EHR companies, led by Cerner seeking to define standards and best practices around data interoperability. Notably, Epic is not a member, going so far as to publicly identify the Alliance as a “weapon” against market share.

• Another collaboration, the CARIN Alliance, includes industry giants such as Apple, Microsoft, Google Humana, Walgreens, Blue Shield of California, Salesforce, etc., and its members have pressed for expedience in interoperability.

Though these efforts have been steps in the right direction, they have not yet resulted in definitive progress.

Concurrently, healthcare data is increasing in volume and variety: the quantity is expected to be 15 times greater in 2020 than in 2013, and consumers are increasingly using digital health tools to complement or even supplement their healthcare. With the proliferation of consumerized genomic data, wearables, and other forms, healthcare data is an expanding gold mine with severe obstacles around mining.

Source: Rock Health

Administrative Data (Provider and Claims Data)

Healthcare administrative activity is reliant on back-and-forth between providers and payers, in data-heavy processes such as claims (submission, inquiry, payment), prior authorization, eligibility verification, contract management, referrals and credentialing. However, as mentioned before, administrative inefficiencies consist the largest portion of healthcare’s exorbitant costs. According to a16z general partner Julie Yoo, healthcare revenue cycle/billing is “roughly 8X less productive than other sectors when it comes to the number of employees affiliated with a certain level of revenue.” This, too, is largely due to the core issues of healthcare data: interoperability and usability.

VBC-initiatives are driving the growth of payer-provider partnerships. These partnerships heavily rely on data-sharing to enable their pursuits of improved care coordination, better interventions and better analytics around population health measures. This is reflected in the astonishing volume of administrative data transactions — according to a 2017 CAQH Index Report, “an estimated 925 million manual transactions and nearly 13 billion electronic transactions were conducted by medical plans”, which was a 38% increase over the prior year. Another CAQH Whitepaper found that “a typical practice … must manage 8,400 data points” of provider data for contract management. Utilizing claims and provider data is a critical function in healthcare.

Sadly, healthcare claims have problems similar to health records. For one, many claims processes are still excruciatingly manual. A McKinsey report finds that automation of payer data activities could achieve cost savings between 15–40%, but, in order to automate, a solution needs to integrate with the EHR and payer claim data sources, which we know is an incredibly tall order. An example of these problems is prior authorizations. 83% of physician survey respondents said prior authorizations are “very” or “extremely” burdensome, one reason being that 84% of medical necessity attachments are manually exchanged, often containing the wrong amount or wrong type of information.

Even when automated, electronic claims systems were designed without consumer needs in mind, and are difficult to manage. As Mario Schlosser (Oscar CEO) puts it, “the biggest purpose for these data files is for claims to get paid. This data wasn’t built for driving better clinical outcomes, attaching better clinical payloads to the data transmissions, for having more real-time insights and things like that. It’s a payments system first and foremost, and that’s it.” Strikingly, 835 files (electronic claim payment information file) could have a 30–40% error rate.

This all leads to 1 in 5 claims being mishandled, a problem which compounds over multiple complex processing steps that occur between the provider and payer.

On the other hand, provider data is woefully inaccurate and difficult to maintain. 52.2% of provider directories have at least one inaccuracy and the data changes frequently: 2–3% every month. To put this into context, this consistently inaccurate data worsens provider communications, causes delayed provider payments, impedes patient access, and all of the administrative processes shown in the below diagram.

Most critically, poor provider data creates severe challenges in care coordination and fluid payer-provider partnerships. Though health systems’ networks are growing through partnerships and mergers, poor provider data quality (for ex. lack of granularity into specialties) limits optimal patient-to-provider matching and provider-to-provider referrals, leading to more back-and-forth and inappropriate care. 75% of specialists received a clinically inappropriate referral in the last 12 months, while 62% of referring physicians say they lack reliable information for in-network specialists.

In sum:

• All healthcare data suffers from two key problems — interoperability and usefulness.

• The administration of healthcare is highly complex with multiple touchpoints and data transactions between many stakeholders, including patient, provider, and payer. As fall off can occur at each step, the problems of inaccurate data (patient, provider, claims) and inefficient data sharing processes snowball into severe administrative waste, a multi-billion dollar problem

• Current solutions are poor fits to solve for these problems. Manual processes are unscalable, and incumbent software solutions do little to provide operational lift.

• Health systems widely adopted expensive EHRs that are cumbersome products designed around logging information rather than to create meaningful use of healthcare data.

• EHRs are historically uncooperative with each other, with other software, and even internally. This creates multiple, disparate data silos throughout the healthcare landscape. Efforts to encourage interoperability have been so far unimpactful.

• Claims systems and provider directories suffer from similar problems, making provider-payer transactions difficult and slowing the progress of value-based care.

• Physicians are paying for this with burnout, and patients have worse outcomes.

• Health data is continually increasing in quantity and variety, but the healthcare industry is unable to take advantage because of these issues.

These issues clearly are detrimental to the healthcare system — how do they impede digital health?

Click here for Part 4.

This is Part 2 of a series. Here’s the Overview, Part 1, Part 3, Part 4 and Part 5.

The growth of digital health

To combat the ever-growing list of increasingly urgent problems and conflicting incentives, key stakeholders in healthcare have turned to digital health solutions. Here’s a quick, by-no-means-exhaustive list of the categories of digital health solutions that have emerged.

• Tech-enabled health insurance — Oscar, Clover

• Real world evidence — Flatiron, Verana, Evidation

• Full-stack pharmaceuticals — Ro

• Pharmacy — Capsule

• Vertical-specific digital care providers — Omada, Maven, Cricket, AbleTo

• Tech-enabled primary care — OneMedical, Iora

• Digital therapeutics — Noom, Proteus, Pear

• Care coordination — Quartet, PatientPing

• Patient access — ZocDoc

• Population health — Komodo

• Patient financial engagement — Cedar

• Healthcare process automation — Olive

• Clinical trials — TrialSpark

Source: CBInsights

All start-ups operating in digital health will live or die based on their abilities to completely own a segment of the patient experience, align the incentives of the three stakeholders, or both.

Consequently, the digital health industry has amplified. Total venture funding in the industry grew >7x over the 2010s, growing by an average of 32% year-over-year from $1.1B in 2011 to $8.2B at its relative peak in 2018.

Source: Rock Health

With this explosion has come the interest of Big Tech. Among some notable recent activity:

• Amazon partnered with JPMorgan and Berkshire Hathaway in creating a healthcare venture, Haven (here’s a great Nisarg Patel post that dives into Haven’s potential strategy).

• Apple has been prolific in its healthcare approaches, from partnering with larger organizations (teaming up with Johnson & Johnson and Evidation Health for a heart study) to start-ups (consumer-focused diabetes product, OneDrop), and tackling health records. Tim Cook commented that Apple’s greatest contribution to mankind will be in health.

• Beyond its ties to Verily, Google is steadily growing its Google Health division, and its acquisition of FitBit made waves last November.

COVID-19 and its impact

Healthcare’s challenges with innovation have often been attributed to the inertia of the system and the complexity behind creating new habits. With COVID-19’s recent emergence creating urgency for digital health development, this may soon change. The systemic shock provided by the pandemic is certain to fundamentally alter how everyone perceives and approaches healthcare, especially around the definition of basic care, how we access it, and how we pay for it. The longer this pandemic extends, the more changes stakeholders will have to accommodate, leading to stickier habits in healthcare. Simply put, this isn’t a temporary delta, it’s a permanent step function.

As coronavirus raises Medicare spending from $38.5 billion to $115.4 billion over the next year, health systems and payers are scrambling to expand their capacity, ramping up their usage of digital health solutions — particularly those that enable telemedicine, telecommute, and virtual care management. Payers, especially those serving higher risk populations, could collapse under the steep incline in costs if the pandemic is prolonged.

Relief efforts will be crucial in alleviating the impact of this sudden spike in demand (telehealth services experiencing a 312% increase in demand from New York patients alone!). Through the CARES Act, hospitals will receive +$100B, $200 million is specifically going to aid providers offering telehealth services, and $275 million will be allocated to expand capacity and services for rural hospitals. Big Tech is responding with major contributions and partnerships of their own.

With the crisis elevating global panic, anxiety, and stress, the need for impactful behavioral health solutions has risen as well. Funding of behavioral health startups in Q1 2020 alone ($462M) has already exceeded half of the full 2019 year amount ($750M). Yet, similar to other domains of healthcare, the supply of behavioral health care has been unable to meet the demand spike and, as National Alliance of Mental Illness CMO Ken Duckworth notes, US mental health infrastructure is a “patchwork quilt”. Both providing immediate access and re-architecting this quilt are critical to managing the fallout of COVID-19.

As is the case with mental healthcare, the many gaps of the healthcare system command severe attention. A prolonged pandemic will mean the healthcare system will need to rely on solutions that can efficiently bridge these gaps, perhaps including companies that can provide:

• Management of healthcare supply chain and logistics

• Scalable staffing/onboarding of healthcare employees

• Accelerated clinical trial operations

• Drug development infrastructure

• Navigation of regulatory complexity

These developments have fueled digital health to record funding levels in Q1 2020. Nevertheless, it remains to be seen whether this trend will beat out the larger-scale macroeconomic trends of the pandemic’s impact.

In sum:

• The increasing demands of the three stakeholders of healthcare have lead to the explosive growth of the digital health industry and the subsequent interest of venture funding and big tech.

• The COVID-19 crisis has created pressure to expedite development and deployment of digital health. Resultant changes to healthcare will have a lasting impact on stakeholder expectations.

With the digital health industry seeing such steady growth as a result of tailwinds from industry trends and increased demand from a global pandemic, a core set of questions arises: can the industry maintain the growth and meet the demand? What challenges lie in the way?

Click here for Part 3.