We invite you to sign-up for the upcoming virtual XPC IdeaFest on Nov 1 for hands-on problem solving about what “good care” means.

A patient asks her primary care doctor to recommend a “good cardiologist who will listen.” A hospital system allocates scarce resources to a fall-prevention campaign. A clinic director shortens visit slots to allow for better access (and margin). A startup creates an automated onboarding process to increase scalability.

In each scenario, someone has explicitly or implicitly answered the same unspoken question: what is good care? The answer to this question determines how we train clinicians, what processes we build, and how we allocate time and money. But in a system with limited resources and countless stakeholders, how do we align on this definition?

In this article, we explore:

• Why stakeholders in healthcare each have unique insights and blind spots on the question of good care

• Key tensions and tradeoffs when thinking through good care

• Case examples to illustrate how the same care may represent a success or failure to different stakeholders

Why stakeholders can mean well but think differently

Ask five New Yorkers what makes a good pizza and you are likely to get (at least) five different answers. Healthcare is no different. Not only is healthcare subject to the same subjective preferences of any service industry, the number and diversity of stakeholders involved in healthcare is staggering. Each of these stakeholders has their own values that inform what they believe good looks like and what tradeoffs should be made.

Imagine we were determining how to measure “good care.” We would need to first agree on a few basics. Before even getting to the meat of what “good” means, we run into diverging stakeholder opinions (Figure 1). What timeline do we measure over? Who is the focus of our measurements - individuals, panels, communities, countries? Patients, payors, and health tech companies are likely to think in terms of 1- 3 year intervals for reasons of human bias, payor churn, and funding runways. Policymakers or primary care clinicians may think in terms of decades or even generations. Patients and clinicians are more likely to try to prioritize individual outcomes, putting a premium on patient autonomy and individual optimization. Health systems, payors, and policymakers think about optimizing outcomes at a population level. As for the definition of “good,” each stakeholder’s unique vantage point comes with special insights as well as blind spots.

The insights and blind spots of healthcare stakeholders

Healthcare’s stakeholders can roughly be divided into three groups: people, organizations, and industry. Each stakeholder views good care through a different lens, affording both unique insights and potential blind spots.

People

Patients

The majority of stakeholders provide or administer healthcare. Patients experience it. This is a unique vantage point in healthcare, and it has often been undervalued. Patients are not a monolith, but when they are asked about what makes good care, three broad priorities consistently emerge: interpersonal connection; access to visits and assistance between visits; and structural elements such as affordability, wait times, and even parking.

For decades, the medical community treated the patient experience as a “nice to have,” separate from the technical elements that were deemed to determine good care. This was a mistake. Patients’ experience of care directly impacts the results of their care. For example, patients consistently say that good care makes them feel respected and is rooted in a strong, caring relationship. This is not just a preference, it has real consequences: patients with higher trust in their clinicians show more beneficial health behaviors, have more confidence they can impact their health, show higher self-efficacy, and report higher quality of life. Patients also prize care that is individualized to their circumstances over care they may feel is more routine or impersonal. Again, this preference impacts outcomes: patients whose care plans reflect their unique context are more likely to improve, an effect seen in outcomes ranging from lung function to depression symptoms.

While relationships and individualized care matter deeply, other factors sometimes outweigh them. The popularity of the urgent care model as well as direct-to-consumer telehealth platforms such as Hims and Hers, Ro, and Wisp show that for some situations ease of access and transparent pricing matter more than a long-term relationship.

Patients do have blind spots. Most patients do not have the clinical knowledge to judge the technical elements of their care (Was that the right diagnosis? The first-line treatment?). Patients do care about clinical quality, but they must judge through proxies such as how trustworthy they found their clinician, the name brand of an institution, or overall service quality. While patients may be cost-sensitive for their own bills, they generally do not consider the allocation of limited resources across a population when thinking about the quality of their own care. For example, a patient whose insurance covers an extra hospital night may define good care as “staying to be sure” without having to consider who else may need the hospital bed or the societal cost of an extra night that does not have a clear medical indication.

Patients, particularly those with complex needs, often rely on family members or friends who coordinate or deliver informal care. These caregivers experience the healthcare system slightly differently from patients themselves. For caregivers, navigation, communication, and structural ease become even more critical to good care. Caregivers are often the ones scheduling appointments, managing medications, and handling paperwork for a patient– and this work is on top of the caregivers’ own career, parenting, or personal health obligations.

Clinicians

Clinicians spend years mastering their clinical skills and decades embedded within complex healthcare systems. It is not surprising, then, that their perspective on good care often emphasizes clinical and technical skill, effective care processes, and organizational support. Their training gives them unmatched expertise in defining the clinical and technical components of good care. Indeed, professional organizations spend countless hours creating and updating specialty-specific practice guidelines and recommendations to define good care. Yet clinicians, too, have blind spots.

Most clinicians care for one patient at a time, and they understand that part of their value is adapting the science and the guidelines to the unique person in front of them. This creates complexity when trying to identify good care at a population level, as there are always exceptions to every guideline or best practice. Additionally, clinicians are generally not trained or reimbursed to think about the outcomes of their patients as a panel. Although value-based care and quality metrics now encourage population-level thinking, a clinician’s predominantly visit-based calendar and reimbursement structure anchors them in a 1:1 practice. A clinician’s focus on understanding the needs and goals of one patient at a time - as well as a need to maintain rapport and prioritize among many issues - often creates tension and resentment when clinicians are told to meet population-level goals to meet standards of “good care.”

For example, a doctor may believe in screening mammography, but when told to try to get her screening rates from 80 to 85% she may feel dispirited and disempowered, citing patient choice, the unique demographics of her panel, socioeconomic barriers like transportation she cannot control, and insufficient time to provide all guideline-directed care. Asking clinicians to consider cost at population level in their care decisions for an individual meets with even more resistance. On the one hand, clinicians generally agree that they should adhere to guidelines that discourage marginally beneficial care. However, 85% of clinicians reject the idea of denying beneficial but costly services so that resources can be redirected to patients that need them more.

Finally, physicians are at risk of forgetting that the outcomes that they care about physiologically, such as Hemoglobin A1c targets, are often meaningless to patients unless they are tied-in to goals that the patient values, such as function or independence. Clinicians may discount the importance of providing clear information to patients and families. They may overvalue the importance of what happens in the exam room and operating suite and undervalue the importance of support, education, and coaching between appointments.

Increasingly, care is delivered not by solo clinicians but by interdisciplinary teams. These teams broaden the definition of good care by increasing the value of seamless care coordination, communication, teamwork, accountability, and whole-person care that addresses mental, physical, and socio-economic factors. A team can often support patient self-management, proactive check-ins, and panel-level management in a way that is not feasible for a solo clinician. Different members of the team may have different definitions of “good care.” Specialists often define quality through precision and mastery of their domain, often specific to an organ system; primary care clinicians, in contrast, think about whole-mind-body, preventative care that is integrated into a patient’s life and context.

Organizations

Educational institutions

Universities and professional schools play a central role in shaping how future health professionals think about “good care,” yet they often present a limited view. Most programs disproportionately emphasize a single element of the biopsychosocial model, reflecting what they feel to be the needs of their cohort. Medical and nursing schools focus on the biological and technical aspects of disease; psychology and therapy programs focus on mental health; social work programs highlight socioeconomic determinants of health. Each discipline develops expertise, but in isolation, leaving them with gaps in understanding of how to provide holistic care.

Interprofessional education may also take a back seat, as the goal is to achieve depth within a discipline rather than a broad understanding across fields. Students may have brief exposure to team-based care, but meaningful, sustained opportunities to train alongside other disciplines are rare. As a result, graduates can enter the workforce unaccustomed to the teamwork and communication that other stakeholders see as integral to good care.

Finally, educational institutions still train students to manage individual patients, one encounter at a time, rather than to manage the health of an entire population. While some schools incorporate lectures on population health management or the economics of care delivery, the opportunity to see or apply these ideas in practice are rare.

Advocacy Organizations

Advocacy organizations represent the interests of specific patient communities, often defined by a particular disease, condition, or identity (e.g. National Brain Tumor Society). Their primary imperative is to amplify the patient voice and advance the needs of their constituency. For them, “good care” is defined by the lived experience of the patient - it must be deeply patient-centered, responsive to individual goals, and ensure access to the most effective treatments. This perspective extends beyond clinical encounters to encompass social support, education, and empowerment as integral parts of care.

This unwavering focus on a particular population’s needs can put advocacy groups at odds with other stakeholders. Their calls for rapid access to new therapies or specialized resources may conflict with the cost-containment priorities of payors, the standardization imperatives of health systems, or the evidence thresholds of policymakers. Moreover, their success in raising awareness and attracting funding can create a competitive environment for finite resources, where compelling narratives and organizational visibility determine which conditions receive attention.

At the same time, advocacy organizations play a crucial integrative role in the healthcare ecosystem. They connect individual stories to system-level change and translate patient experiences into policy, research agendas, and public awareness. When aligned with other stakeholders, advocacy groups can help bridge the gap between what the system delivers and what patients truly need.

Health systems

Health systems are quite heterogenous, but they typically consist of an affiliation of healthcare organizations that include at least one acute care hospital and one primary care group. (Many other definitions exist.) Whether non-profit or for-profit, health systems are medium to large businesses with a survival imperative: grow the patient base, operate efficiently at scale, and balance the budget. These pressures shape how they define “good care.”

For a health system trying to sustain a market foothold, good care must drive growth. Access, in particular for new patients, takes on high importance for its impact on both patient experience and growth. But easy access can come at the cost of time with clinicians and continuity with a trusted provider, eroding the trusting relationships that clinicians and patients prize. Health systems do invest in service-quality initiatives such as AIDET to boost patient reviews and retention, yet the need to staff flexibly across departments and shifts often fragments the patient experience into many fleeting encounters.

Scale brings another bias: good care must be safe and effective across large, complex systems. Systems aim to minimize errors and produce repeatable outcomes across large populations; health care systems must also hire, train and then sustain the performance of large numbers of staff. This creates a bias that good care must be embedded in strong processes rather than relying solely on good people. While a focus on system and process is critical for safety, it can clash with patients’ and clinicians’ desire for individualized goals and care. A health system’s push to reduce variability among processes can impinge on individualization and connection by creating too many clicks, too many restrictions on care pathways, and too many distractions from the patient-clinician interaction.

Because of the sheer volume of patient interactions, health systems define and track outcomes at the population level. Population-level data help keep operations focused, but the need to rely on high level metrics can flatten the nuance of care. It is easier to track length of stay than it is to track whether the care was appropriate for the patient’s context or goals. What is tracked gets focus; what cannot be tracked at times gets forgotten.

Finally, financial realities drive a bias towards reimbursable care, which often means more care, especially while fee-for-service remains dominant. Although researchers and policymakers advocate for primary care to be the quarterback for most patients, health systems have incentives to route patients to more (and more highly-reimbursed) clinicians. Activities like care coordination or patient advocacy, which many stakeholders view as essential to good care, receive little emphasis because they generate little revenue. Just look at the calendar of a typical physician in a health system; the vast majority of their time is earmarked for back-to-back visits, with little (if any) time dedicated to coordination or advocacy activities. Alternative payment models attempt to tie payments to quality and might shift some of these ingrained incentives; however, the underlying architecture of care delivery remains fee for service.

Private Payors

Payors are generally medium to large businesses managing very large patient populations. However, in sharp contrast to the health systems whose revenues grow with utilization, the payor’s business imperative is to control healthcare costs. For them, good care must above all be cost-effective. This aligns with policymakers concerned about the burden of rising healthcare spending on society, but it also introduces distinct biases.

Payors’ emphasis on cost-effectiveness is often at odds with other stakeholders’ values. Narrow networks, theoretically designed to steer patients towards high-value clinicians, encroach on patient choice. Utilization management mechanisms such as prior authorization or step-therapy, intended to prevent unnecessary or low-value services, can cause delays or create administrative burdens that frustrate both clinicians and patients.

Payors also prioritize interventions that yield near-term savings. Members rarely stay with the same private insurer for long, on average as little as 2 years and typically less than ten. This churn means that investments whose payoff is delayed by decades, such as intensive childhood obesity programs, may be harder to justify when the financial benefits may accrue to a competitor.

Private payors’ focus on quality and performance is heavily shaped by federal standards. Most commercial insurers mirror the metrics and incentives set by Medicare and Medicaid through programs such as HEDIS (Healthcare Effectiveness Data and Information Set) measures or CMS Star Ratings. While private payors retain flexibility and may innovate beyond public program requirements, many default to equating “good care” with these population-level benchmarks for preventive services, chronic-disease management, and readmission rates.

Policymakers

A diverse group of organizations contribute to healthcare policy in the United States, ranging from Congress; federal agencies such CMS, the FDA, the CDC; state governments; professional associations; lobbyists; and independent task forces. Each operates with its own set of incentives and biases, biases which may shift as leadership or political winds change. Historically, agencies like CMS, the CDC, and national task forces were viewed as well-positioned to balance considerations of scientific evidence and fair resource allocation across large populations and across longer time horizons.

However, biases exist. Elected and politically appointed policymakers may favor policies that reflect the ideology of their base and interventions that will create impact during their terms, often just two to four years. Policymaking is also an exercise in compromise: competing interests force negotiated solutions that can result in complex, sometimes contradictory regulations. Because policymakers rarely implement these directives themselves, they may underestimate or tolerate administrative complexity that places a burden on clinicians or health systems. Finally, when thinking across such large populations, policymakers often prioritize perceived public good over individual choice as most essential for good care.

Industry

Health tech companies

Health technology companies can bring fresh perspectives to a healthcare industry that is historically slow to change. Startups that serve patients must be highly attuned to patient preferences in order to attract and retain users, and many are born from a single pain point identified by patients, clinicians, or payors.

However, start up culture and the reliance on venture capital create their own biases. Startups favor fast, disruptive innovation over incremental change, which means the care or products they deliver sometimes outpace the evidence for safety and efficacy (see the book Careless People). To avoid complexity, startups often start with a focus on a very limited number of problems or health conditions and may go direct to consumer; while the episodic care they provide may be useful, this often leads to overall fragmentation of a patient’s healthcare experience and erosion of their longitudinal care relationships.

Venture capitalists expect exponential growth and short-term profitability, adding to pressure on health tech companies to cater to customers’ needs, particularly those needs that are faster to solve. Providing quick access or offering care that may stretch or sidestep standard practices, such as extensive hormone testing or unproven longevity treatments, is much quicker to deliver than coordinated, relationship-based care. Scalability itself becomes a core definition of good care: services must be tech-enabled, repeatable, and scalable at low marginal cost. This growth imperative can undervalue the slower, less-scalable human elements of care such as relationships, nuanced clinical judgment, and shared decision-making.

Pharma and devices

Pharma and device companies drive important innovation in clinical care. Through extensive research and development, they bring to market products that can improve or even revolutionize the care patients receive. However, their business model shapes how they define good care. By nature, they bring a bias that good care requires long-term medications and devices more so than processes, relationship-based care, risk-benefit discussion, or lifestyle interventions that can at times be equally or more effective.

The patent and regulatory environment rewards novelty, creating strong incentives for pharma and device companies to define good care as the newest drug or technology. This can lead to “me-too” drugs or iterations that provide only marginal incremental benefit at exponentially higher prices.

Another way to think about the multitude of perspectives on “good care” across healthcare stakeholders is to frame them on a simple 2x2 matrix (Figure 2). Patients, caregivers, and clinicians focus on individual relationships and what happens in the room between people. Policymakers, payors, and educational institutions, think in terms of systems, rules, and populations. In between are groups like startups and health systems that try to bridge the two, scaling what works while keeping care personal. Each group brings something essential, but also carries blind spots. Seeing them together helps explain why the system often feels disjointed and where the opportunities lie to connect the human side of care with the structures that sustain it.

Some key tradeoffs

Any definition of good care will require tradeoffs. The combination of stakeholder opinions and actions has shaped the current state of these tradeoffs, whether or not we as a society have explicitly made a choice. We will explore a few of these tradeoffs; many others exist.

Present versus future outcomes

Modern healthcare has evolved with an implicit bias toward short term outcomes. Long term outcomes are harder to track and harder to link to a specific intervention. For centuries, medicine was organized around episodic, acute care; recently, advances in medical knowledge, lengthened lifespan, and changes in our environment have led to a new epidemic of chronic disease. Yet many stakeholders still favor the near-term: humans, both patients and clinicians, tend to devalue delayed rewards; payors and venture capitalists want fast return on investment; politicians want impact during their term. We may claim to value proactive, forward-looking care, but our actions - how we reimburse, invest, and measure performance - suggest we think good care is care with a short time to impact.

What would happen if we committed to a new definition of good care, one where proactive thinking and long-term outcomes were treated as foundational? First, we would need to rebalance investment toward prevention; right now, primary care and public health, both central to prevention, are underfunded compared with procedural specialties. Our public health infrastructure lags behind much of the developed world. Next we should fund and scale interventions that keep people healthy or reverse early disease. These tools are less often medications or medical interventions and more often lifestyle changes. Some well-studied tools exist, such as motivational interviewing, SMART goals, and brief action planning. These historically have been high-touch and time intensive for clinicians, with minimal reimbursement, but new technology promises the possibility of retaining the high-touch and personalized nature of these interventions with less time requirement for the clinician. Just as importantly, we would need to find ways to reward both patients and care teams for choosing lifestyle interventions; while pharma has incentive to market blood pressure medications, who gets rewarded for the more time-intensive process of education and supporting a patient who is trying to follow the DASH diet while on SNAP?

The longevity movement and the success of wearables have shown that there is consumer interest in more extensive biological monitoring; if we shifted the idea of good care towards more long-term goals, we would need more focus on how to make these data actionable in an evidence-based way.

Patient-defined outcomes vs. expert-defined outcomes

Today’s health system largely equates good care with expert-defined metrics: blood pressure control, Hemoglobin A1c control, colorectal cancer screenings, length of stay, readmission rates. While these measures have some overlap with the things patients care about, they are often several steps removed from what matters most to patients. In some cases, they may even conflict with a patient’s goals or beliefs.

However, patient-defined outcomes have historically been far harder to measure and track; they are inherently personal, endlessly variable in content and timeframe, ever-evolving, and resistant to neat, structured data fields. In addition, in more paternalistic eras most clinicians felt that patients should not or could not be trusted to know what mattered most. Thus, definitions of good care have long had a bias toward expert-defined, population goals rather than patient-defined, individualized goals.

What would happen if we changed our definition of good care to center more on patient goals? There have been some attempts to create this shift. The Larry A. Green Center has developed a Patient Centered Primary Measure to track whether a patient feels they have a strong relationship with their primary care team and whether their goals are being met. Patient confidence measures have been shown to correlate with outcomes, and gauge whether practices help patients gain the skills and agency to control their health. The Geriatric 5Ms help highlight priorities such as mobility, mentation, and “what matters most” that are often highly important to patients but at times overlooked in disease-specific metrics. Artificial intelligence may allow for unobtrusive ways to track whether patient goals are being identified, revisited, and met.

However, if we were to truly refocus good care on understanding and helping patients meet their goals, we need further change. We need payment models like direct primary care or APCM that reward deep discussion, goal-setting and frequent check-ins. We must address socioeconomic barriers that often are the true barriers to goals. Finally, we need better technological support to document goals, not just diagnoses; we need streamlined ways to identify, document and track evolving patient priorities and shared-decision making alongside condition lists and medication data.

Continuity vs. access

For much of the 20th century, continuity of care was a given. Your primary care physician would see you in the clinic and follow you in the hospital; seeing multiple specialists was uncommon; urgent care centers did not exist. But a number of factors have shifted the balance away from continuity and relationships and towards a focus on access at all costs: a shortage of primary care; the vast explosion of medical knowledge and subspecialization; the advent of hospitalists, clinician work-life balance expectations; urgent care and direct to consumer episodic telehealth services; consumer expectations. There are times that patients need access above all else, but continuity with the same physician has been shown to actually reduce mortality. What would we need to do to recenter our definition of good care around relationships and continuity?

First, we would need to measure continuity reliably. Few systems track whether patients consistently see the same clinicians or what percentage of their care is with their core team. Next, we would need to align relationship-based care. In the fee-for-service model, it makes little business sense to leave slots open for same-day or semi-urgent appointments for one’s own patients; if those slots go unused, that is revenue lost. Payment models must instead reward the availability and flexibility that enable continuity. Finally, we need team-based and technological solutions that maintain connection and provide access without fragmenting care. Dr. Ilana Yurkiewicz explores these themes of data and care discontinuity in her book Fragmented.

Thought experiments to make these ideas more tangible

Think through the following cases. What tradeoffs are at play? How would different stakeholders evaluate this care, and why?

1. Mr. Thomas is a 78 year old retired security guard with severe congestive heart failure (CHF). He is on oxygen around the clock and gets breathless speaking more than a few sentences. Three months ago he was intubated and successfully extubated during a CHF exacerbation, and while the experience was harrowing he feels these heroic efforts afforded him what he values most: time with his wife and children. This experience leads him to repeatedly tell his primary care doctor, who he trusts deeply, that he wants full resuscitation and only would choose comfort care if there is no hope of quality of life. During the last three months of his life, he is admitted to the ICU for bipap (a form of non-invasive ventilation) on 5 separate occasions. On the final admission, he requires intubation to support his breathing. His hospital team is unable to successfully extubate him after several weeks, and his family chooses to transition him to comfort care. He dies peacefully.

2. A local free clinic is able to support uninsured patients with diabetes with reduced cost medications using part of their donation income. Insulin in vial form is significantly cheaper than insulin pens. Although most people prefer the pens for their portability and ease of use, the clinic makes a policy that only patients who are unable to use vials due to physical limitations (poor dexterity, etc) qualify for free pens.

3. A patient goes to a clinic that offers screening full body MRI scans every 2 years. She pays out of pocket. After seeing her results, she is motivated to make some lifestyle changes when she sees that she already has visible plaque built up in her arteries. She is also found to have a 1.1cm thyroid nodule. She undergoes a biopsy, which is negative but results in a small hematoma and requires repeat monitoring by ultrasound. Her next scan shows a 6mm pulmonary nodule; she undergoes a repeat CT scan for follow up in 6 months (insurance will not pay for another MRI) and the nodule is gone.

Figuring out how to best leverage healthcare workers across all levels of training is hard. That’s why it’s helpful to learn from labor transformation efforts in other industries. The team at Anthropic is looking at how specific tasks and associated skills will be impacted through automation or augmentation rather than trying to conceptualize how entire occupations will change (Figure 1). Large companies have been created to measure “skills intelligence.” Some notable examples include Workera (general purpose skills intelligence) and DX (developer productivity). Similarly, incumbent healthcare workforce development platforms like HealthStream and Amplifier service health system stakeholders but have yet to achieve meaningful reach within the complex realities of primary care.

Evaluating clinician performance, especially in chronic care, is hard for multiple reasons. For example, what is considered an appropriate referral for new onset atrial fibrillation in a middle-aged woman will depend on the local context (e.g. private practice vs. hospital affiliations) and the payment model of the practice (e.g. FFS vs. capitation). Additionally, condition specific skills and competencies (e.g. hypertension management) have long feedback loops that are difficult to surface in fragmented episodes of care and have fuzzy boundaries of onset and resolution. Bringing visibility to clinician performance has been a passion of mine. This year, that vision has matured from an idea into a business with paying customers. Here’s what resonates with them:

1. Identifying the skills index of their organization and individual clinicians (Figure 2)

2. Upskilling each clinician based on alignment of their educational gaps with organizational goals (Figure 3)

3. Doing all of this without additional administrative friction while meeting regulatory requirements for chart reviews and continuing education

4. Scaling business operations on a foundation of clinical excellence

We are often asked (particularly by non-clinicians): why not intervene with performance insights at the point of care? First, this is incredibly hard to do well. Deploying tools that dictate behavior in real time is fraught with unintended consequences like automation bias and de-skilling. With the advent of autopilot, the aviation industry has left us with decades of learnings about the impact of technology on high-stakes human performance. Second, the space between the clinician and their patient is already congested with continuous interruptions like care gap alerts, best practice advisory (BPA) notifications, and proliferation of other clinical decision support (CDS) widgets. Third, CDS AI will likely face regulation first. As their rapid, and sometimes cavalier, adoption accelerates, patient harm stories are inevitable - and staying in the game will demand serious capital.

With AI making its way into clinical workflows, understanding how to evaluate clinician performance will get even messier. We’re already seeing early studies on how models like AMIE are outperforming PCPs across a multitude of domains, including empathy and medical decision making. Paradoxically, some evidence suggests that AI tools perform better in isolation than with human involvement. This signals that our ability to wield them is fairly primitive. To change this phenomenon, a new set of AI competencies are emerging for integration into medical education. Through this work, it will be more apparent why humans are not the weakest link in healthcare and that many tech startups are automating the wrong part of the care experience.

Healthcare has fallen behind other industries in developing meaningful insights into workforce performance. Over $100M has been invested in chart review companies in recent years, only to see their technical prowess converge on compliance, revenue cycle management, and monetizing work for pharma. We’re overdue for a different way of leveraging technology to make a difference in healthcare operations. For this reason, we regularly convene a clinical skills intelligence (CSI) working group for healthcare leaders (e.g. Chief Medical Officers, Quality Officers, Medical Directors). Similar to how experts in aviation safety come together to share best practices, our sessions focus on problem solving around operationalizing good care. We trade notes on:

• Quality improvement initiatives (how to identify educational gaps, verify competencies, and develop new clinical protocols)

• Clinical team management (candidate vetting and retention strategies and personalizing upskilling interventions)

• Evaluating AI technology (decision frameworks for implementing pilots, governance and legal implications)

Three principles guide our work: 1) everyone deserves good care; 2) good care comes from quality clinicians; and 3) every clinician can improve. Developing a more capable workforce is a tall challenge and as Larry Bauer puts it “stop trying to change the system alone.” We want to meet others committed to empowering good care. Encourage your clinical leaders to apply for CSI membership (free to XPC customers).

P.S. If you are an engineer, and would like to work on the problems we’re tackling, reach out to paulius@xprimarycare.com

P.P.S. We hope to see you at our upcoming virtual IdeaFest on Nov 1. It’s going to be an interactive 4-hour unconference to collectively tackle the question: what does good care look like? across all healthcare stakeholders. You’ll get to network with many brilliant minds shaping care delivery. There is no cost to participate, but space is limited. Sign-up here.

Right now, an incredible community of innovators is tackling healthcare's toughest challenges. If you're one of these builders, we want to connect with you. Apply here to join Jay and me for an immersive five-day retreat in the Colorado mountains this June. Spots are limited to 10 guests for this unique opportunity to dive deep, share knowledge, and connect with peers and mentors.

Primary care physicians have range. We intuit the problems worth solving. The first road block for builders is navigating distribution challenges. Much has been written about this, and there are case studies and playbooks to pour over. What isn’t captured anywhere is how to trail blaze with taste. I asked Jay if that is even possible in primary care – and if so, what corner of the world will it come from?

Next are musings on “taste” by Jay Parkinson

Taste is something I've grappled with for decades, a bit unknowingly. I've always been a visual person. I often think in images. Photography is something I've been able to use to express my taste. I appreciate balance from symmetry and simplicity and light and the unexpected look on the subject’s face and all the things associated with an image you'll hopefully remember forever.

And then I've come to realize there's some overlap with the way I think about elegance and simplicity in how I think about experiences. We've all been to a restaurant that just knocks your socks off. Everything is on point— from the second you walk in, until the moment you sit down, to the visit to the restroom to wash your hands, to the food, to the farewell. You walk out knowing that was one for the books. TMI, but I think restrooms in restaurants are the window into the soul of the place, similar to our retinal vasculature. It could easily be the weakest link, but if every detail in the restroom is on point, you can tell the designer and visionary for the place is obsessed with the entirety of the experience.

There's something else about an experience like this. It feels unique. It looks unique. The food tastes like no other. And that's the magic. You can replace a restaurant with any type of service— coffee shop, buying a car, buying tires, you name it. Any experience can be magically memorable if the designer cares enough about designing the touch points and hiring the right people to make the vision come alive.

The Homogenization of Experience

Essentially, the designers of these experiences are creating art with time and place. We can all envision the Mona Lisa or Starry Night or Andy Warhol's Campbell's Soup. Experience designers are doing the same thing as these artists— branding your brain with a memory.

And then I listened to an Ezra Klein interview with Kyle Chayka, a staff writer at The New Yorker, who wrote a book called Filterworld. Kyle talks about traveling the world, in need of his daily coffee:

"It was just this strange, uncanny experience I was having that as I traveled around the world doing these freelance magazine assignments, I would land in a particular city, whether that was Tokyo or L.A. or Berlin or Beijing, and I always managed to find a particular style of coffee shop. It was this minimalist box of a cafe with white subway tiles on the walls and midcentury Scandinavian furniture and handmade ceramic mugs with nice cappuccinos in them. And this was not the work of a parent company. It wasn't a Starbucks. It wasn't a global chain. Instead, it was all of these completely independent coffee shops, baristas, entrepreneurs, who had decided to mold themselves into the same aesthetic. So I started wondering what connected them all together. And my conclusion was that digital platforms, whether Instagram or Yelp or Google Maps, were feeding this series of cafes to me as a recommendation. Like, my taste was being evaluated, and then I was getting recommended the same minimalist cafe over and over again, and I was being driven to these places by the platforms themselves."

The internet seems to be putting us in a box (I guess the word is more "persona") and using algorithms to amplify what that persona is expected to like. The almighty algorithm is turning us into good little consumers— it's almost communist, for capitalism's sake.

Applying Taste to Primary Care

This got me thinking about primary care. Has the algorithm or the groupthink done that to us, the innovators of healthcare? What are the subway tiles of primary care? Or the midcentury Scandinavian furniture? And, of course, you'll find both of those in One Medical offices, proving the coffeeshop-ification of primary care.

For some reason, and I can't really tell you why, I've been obsessed with chasing anything exceptional that gets burned into my brain as a memory. I guess that's why I lived in NYC for 16 years, because New York was the densest concentration of these kinds of things. I didn't really think about it or know what I was doing until Kyle Chayka defined taste in this way:

Taste is a process of collection. Like you're grabbing on to the little voices and artists and touchstones that make you who you are and give you your sense of self. You're drawn to something without knowing why.

I've built some pretty cool primary care experiences over the last 18 years. A few have been novel experiences nobody had ever seen before. The things that tie them all together is this innate sense of art meets business. Business is financial talent. Art is burning exceptionally positive memories into brains. It's making people feel a way they'll never forget.

In a lot of ways, I was the Creative Director of the primary care experiences my teams were building. But what does that even mean?

What does a Creative Director for Primary Care do?

They have an intuitive sense of what patients want

And what do they want? They want to communicate, easily. Because all healthcare is is communication. Weirdly enough, they don’t always want to be their healthiest, they just want to be heard and validated. Again, that’s communication.

They have an intuitive sense of what clinicians want

They want freedom to be their best. They want an active, evolving toolkit that increasingly makes all aspects of their job as easy as possible. It's literally that simple.

They steal from others

Creative primary care steals from exceptional experiences typically found outside of healthcare. It's the way Apple helps you fix a broken MacBook— first with self-helpy online content, then, stepping it up to real time online chat, and, if necessary, in-person at the Genius Bar. Notice how that playbook works great for fixing broken health too. It's the way ChatGPT has a fascinating spoken conversation with you about anything, for as long as you want. It's the way you walk into your favorite coffee shop and every employee knows your name and they treat you like you're part of their family. It's the way Meta knows a thousand times more information about you than a traditional PCP. It's a process of collection. It's paying attention to what's great in the world and constantly asking yourself "how does this apply to delivering exceptional primary care?"

They have an intuitive sense of knowing when something feels iconic, long before the data proves it

Think about Virgil Abloh, Rick Rubin, Steve Jobs, etc. They used certain kinds of cherry-picked data to make iconic things, via their gut. People like that have an innate conviction based on their intuitive understanding of humans and what they like. And they put their vision through their own personal filter to create category-smashing things. Lately, Product has been hijacked by data. Product leaders gather data and use that data to build or improve user experiences, disregarding taste and vision. That strategy is the best way to build incremental improvements. And when you think about it, in the early days of a company, Product's role is to launch, find product market fit, and drive value. That's super creative. And as the company matures and is forced to drive revenue, it's Product's role to drive enshittification. Primary care doesn't need enshittification, it needs a fundamentally new product.

They know how to create experiences that feel good when folks use it

Let's face it, medical care is never fun. How many times have folks ever said, "goodness gracious, going to my doctor was just so easy and pleasant!" It's hard to define taste, just as it's hard to define a feel-good experience. It's not "From a scale of 0 to 10, how likely would you recommend blah, blah, blah?" That feeling I’m talking about is etched into your mind. It's kind of like eating a steak at Peter Luger's in Brooklyn— it ruins all steaks you'll ever have the rest of your life because they'll never live up. That's how we want our patients (and clinicians!) to feel when they use creative primary care.

They gather the right talents

You need a Creative Director to define the vision. Remember, they have an intuitive understanding of what's iconic. You need the artist to design the visual identity. You need the operational wizard to design and iterate on the workflows. You need the hospitality expert to design how the brand talks and how that makes users (both patients and clinicians) feel. You need the clinician, who intuitively understands how clinical policies make patients feel, to define what care gets delivered for whom. You need the Product designer to interpret the mission into a UX that's immediately obvious. You need the writer who knows how to communicate the vision. And you need a business development talent that intuitively understands how to acquire populations and individuals.

They choose their target

I can’t think of an experience designed for every type of person imaginable that instills joy in us all. Great experiences pick who they want to impress, design the experience around that persona, and deliver. The problem with primary care is we take all-comers. It’s so general, it works for nobody. Give us your sick and we’ll do our thang. That’s why I’m excited about primary care being built around the unique experiences of people of size; LGBTQIA+ folks, and latinos. You even see this in the Hotel Trade primary care centers in NYC caring for people who work in the hotel industry. Choosing who these services want to care for gives that population a home they’ll never leave. People love brands that get them— it’s our innate tribalism. The brand doesn’t have to be high end. It just has to have a target and an opinion.

And finally, they tie this all together

Almost like how a movie director's job is to help you escape to a whole new world, a creative director's job is to ensure every moving part functions in a way that makes users say "wow, that whole thing, whatever it was, ain't never seen it before...that was delightful."

The Future of Tasteful Primary Care

But I get it. I've been around the block long enough. Pies in skies and stuff. What does this even mean? Does having taste look like One Medical? Or Biograph? Do they have "taste?" If so, is taste just a veneer on appointments and patient portals and tests?

I believe primary care can absolutely have taste. It emerges when we stop trying to simply improve the existing system and instead reimagine it entirely. Tasteful primary care will come from the intersection of technology, hospitality, and human-centered design. It will feel as different from today's standard primary care as an exceptional restaurant feels from a fast food chain.

The future of primary care with taste isn't just about prettier offices or more convenient apps—it's about creating healthcare experiences that people actually want to engage with, that clinicians find fulfilling to deliver, and that solve problems in ways that feel almost magical. It's about building something that makes people say, "I never knew healthcare could feel like this."

If you are building in healthcare and want to dig deeper into topics like this with other kindred spirits, you should apply to join our next retreat. Know someone who’d be a good fit? Feel free to share this opportunity with them.

How we learn to practice medicine has not changed in decades. It’s never been more exciting to reimagine how medical education and clinical training can catch up with modern practice. Here’s what drives my thinking about what’s needed and what’s possible.

Clinical experience gap is widening

Market trends are changing the healthcare workforce composition. Increasing numbers of seasoned physicians are leaving the profession while less experienced clinicians (e.g. APPs) backfill positions in demand. As a result, we have a growing clinical experience gap. On-the-job training is the current solution to make up for deficiencies in experience. For a deeper dive on these trends, read about the evolving primary care workforce and Bloomberg reporting on clinical training.

Chronic disease is ubiquitous yet rarely seen

Residency earned its name from an era when doctors lived in hospitals, learning medicine at their patients' bedsides. These patient cases were their textbooks, teaching them how diseases unfold from admission to discharge. This model excels for acute conditions like heart attacks and strokes, where immediate intervention is key. But modern medicine faces a different challenge: chronic diseases that develop over months and years, not days. The feedback loops in chronic care are painfully long—a medication change today might not show its full impact for months, and a missed intervention might not reveal its consequences for years. Even programs emphasizing continuity of care struggle to show trainees the full progression of chronic diseases. This training gap is particularly concerning given that chronic and mental health conditions consume 90% of US healthcare spend. Witnessing enough variations of chronic disease to build true expertise is nearly impossible through traditional training alone—it requires years of practice, patient by patient. And as our population ages, these conditions rarely come alone—clinicians must learn to recognize and manage complex patterns of interacting illnesses, often through trial and error with real patients (Figure 1).

Unlock EMRs as textbooks, not just QuickBooks

The Gutenberg moment for clinical training has been in front of us all along — the EMR (Figure 2). Yes, EMRs were designed for billing and are far from perfect. But beneath their administrative facade lies an untapped educational goldmine. Using medical records as a medium for curriculum delivery is not a novel idea yet nobody has seized this opportunity at scale. Creating and operating high-quality synthetic data to simulate patient care will be at the crux of the next revolution in clinical education. It’s only a matter of time before EMRs become adopted as the superior learning management system (LMS) in medical education. Data on trainee clinical excellence (e.g. prescribing behaviors, referral patterns) and operational excellence (e.g. productivity, inbox management) can guide training interventions. Insights about trainee readiness for the real world practice can inform hiring and placement decisions. Being able to benchmark and predict performance of their workforce will offer employers an advantage. More importantly, educational institutions may begin to differentiate by bearing risk on the quality of their graduates.

The market is beginning to notice this opportunity. Last year, Epic launched Lyceum to accelerate medical student adoption of their systems, while startups like EHR Go and Scope have created proprietary teaching EMRs to operationalize case-based learning. VCs are funding companies like Soma Lab (YC S24) to reimagine medical training through virtual clinical experiences. However, success hinges on seamless integration into real clinical workflows. Until (if ever) we free medicine from EMRs, that's where training needs to happen. In 2025, I anticipate a surge of capital deployment to tech-enable clinical training, driven by an urgent need: preparing the least experienced (and least expensive) clinicians who increasingly care for the most complex patients.

Breaking the specialist ceiling

Specialists master organs, but patients have illnesses that cross these boundaries. Today's training spreads generalists too thin across multiple domains, limiting their ability to develop deep expertise. EMR simulations could change this - enabling generalists to become experts in the chronic disease patterns their patients actually have. Just as some developing countries leapfrogged landlines straight to cell phones, EMR simulations can elevate generalists to hyperspecialists. By simulating years of disease progression in weeks, generalists can master the real chronic illness patterns in their communities (Figure 3). This precision training approach would advance health equity. Simulated care pathways can help drive targeted interventions for vulnerable patients (e.g. managing the interplay of obesity, schizophrenia, and COPD for a patient living in a food desert). For risk-bearing organizations, this means finally having a systematic way to train clinicians in the precise competencies that drive quality metrics (e.g. deprescribing anticholinergic medications in elderly patients to excel at Star measures).

More discovery ahead

There’s a strong signal for EMR-based simulations of chronic conditions to be the future of clinical training. This could be something that remediates low-performers already in practice, facilitates practice transformation (e.g. transitioning from FFS to VBC), differentiates candidates in the hiring process, platforms next-generation CME, or integrates with curricula of training institutions to measure trainee competencies.

P.S. If anyone wants to riff on additional use cases and GTM strategies, happy to share my findings from discovery work with ICPs as I build in this space. You can reach me at paulius@xprimarycare.com

Thank you to Alex Kazberouk and Emily Lu for reading drafts of this.

Care gaps represent the delta between what clinical guidelines recommend and what patients actually receive. They exist because patients miss preventive screenings (shoutout USPSTF), don’t fill prescriptions or struggle to manage chronic conditions. Most people know that taking care of our health early pays dividends later, but here are some facts to drive home the point: 

• Treatments for later stage cancers are about $100,000 more expensive per year for every additional stage (Figure 1).

• Patients who develop diabetes cost $13,000 more per year than patients with pre-diabetes.

Despite the clear benefits to patients and growth of value-based objectives for payers and providers, the uptake of preventive care is trending down. From 2015 to 2020, US adults who received all recommended preventive services dropped from 8.5% to 5.3%. While some specific measures have improved, historical data from NCQA shows most population-level quality metrics have changed little over the last decade.

Though individual patients face a huge diversity of obstacles to accessing care, there are also larger structural factors playing out nationally that make it challenging for healthcare organizations to effectively organize efforts to improve patient adherence to preventive care.

Besides patients, who else is reaping the benefits of closing these gaps?

For most of the last two decades, health plans were the primary beneficiaries of quality bonuses associated with increasing the proportion of patients who receive preventive care. Perhaps most prominently, Medicare Part C (aka Medicare Advantage) uses a 5-star system that directly measures the population-level access and adherence of preventive care services to grade health plans. A half-star change in a rating can mean $40M less in revenue for a health plan of typical size (~250,000 members). It’s evident that health plans are incentivized to care about care gaps with so much revenue at stake.

In the last few years, we have seen a rapid increase in the fraction of medical groups participating in contracts with health plans that share in quality incentives, reflecting the increasing popularity of value-based care (VBC). In 2023, $1 in $4 distributed by payers is associated with risk-sharing contracts, nearly doubling since 2018 (Figure 2).

In principle, primary care physicians are well positioned to engage patients and deliver preventive care, both because of their trusted relationship and knowledge of community resources. In practice, PCPs get caught in the tangled web of quality measures across the payers they contract with. For example, in a recent study of PCPs from one integrated health system, medical groups were keeping track of over 50 distinct quality measures across 10 different contracts with health plans. Even if there was more alignment across payers, primary care teams, and patients on which measures to prioritize, PCPs would require over 14 hours per day to provide recommended preventive care for all their patients. With an ever increasing number of screening guidelines, PCPs cannot reasonably tackle care gap closure alone. 

Goodhart’s law warns us that dysfunction can arise from incentivizing improvement of measures that are imperfect representations of some true underlying objective. “Managing the measure instead of the patient” can incentivize the unintended behaviors. 

For example, for quality measures intended to improve the control of blood pressure, only the last blood pressure reading of the year counts. This can incentivize organizations to prioritize blood pressure reduction campaigns at the end of the year. Similarly, medication adherence measures only consider whether a patient fills their prescription on time, not whether the medicine is actually being taken, leading to campaigns focused on at-home delivery of medication or encouraging extended day supplies.

Another common fear of care gap closure incentives is that contracts with different quality objectives will drive physicians to treat patients from one health plan differently from another. This concern is why it’s so important for medical groups to be thoughtful when designing incentive programs that encourage their physicians to close gaps. In fact, some practices have deliberately avoided care gap incentive programs for physicians for this reason.

The frantic “Q4 Push”

While clinicians close care gaps with their patients all year, organizations only become aware of emerging population-level care gaps near the end of their annual VBC contracts. Only care gap closures that occur before the end of the contract year count towards that year’s quality measures. In contrast, organizations have several months after the start of the next year to document care delivery of gap closure from the previous year. The result is an unintended whiplash of changing priorities: deliver as much care gap closure as possible before December 31st, and then focus on challenging documentation in the new year.

This short, cyclic demand window for care gap closure undermines the intent of population-level quality measures and strains the limited resources of healthcare organizations, especially in the context of other year-end efforts (like HCC gap closure, CAHPS surveys, or re-enrollment). For example, staff tasked with calling patients and making appointments in Q4 get reassigned to tracking down and documenting proof of care delivery from the previous year.

As a result, an entire cottage industry has emerged of vendors who help healthcare organizations surface, close and document care gaps. In particular, the emphasis on reporting requirements from different health plans has led to the adoption of NLP and other advanced technologies to facilitate the Q4 push. Similarly, third party at-home visits, at-home tests, mammobuses and other services are in full swing in Q4 helping health plans rescue their at-risk care gaps.

Efforts to engage patients with important preventive care services during the Q4 push are often primitive and deeply frustrating for both the patient and the healthcare organizations. This is true for payers, clinicians and everyone in between.

Here are some horror stories from the trenches:

• One payer spent hundreds of thousands of dollars staffing a new call center with 20 full-time temp workers for the last 12 weeks of the year simply to call every single one of their members with a care gap. They also proactively sent every patient behind on their colorectal screening a FIT kit. 

• Providers often have 10+ VBC contracts with dozens of care gaps. Every health plan is faxing them “chase lists” of their members with outstanding gaps. In one case, a health plan sent a provider group a large stack of prepaid Walmart gift cards with vague instructions to use them to get patients to come in for their annual wellness visit. Unfortunately, most of the gift cards ended up last in desk drawers instead of in the hands of patients. 

• Patients describe this time of year as one that erodes trust with their healthcare teams. Some patients receive several FIT kits and encouragement from their health plan – whom they are reticent to trust. Other patients will try to get preventive care only to find that there are no appointments available for months due to increased demand.

The result is a fragmented mess of stakeholder organizations stepping on each other's toes trying to close gaps. The emergence of Accountable Care Organizations (ACOs) and Independent Physician Associations (IPAs) between payers and patients has further complicated care gap closure efforts.

With so many contracts and quality objectives, how do healthcare operators choose which gaps to go after? 

Because there are not enough resources nor enough time to achieve perfect scores for all quality objectives across all plans, VBC organizations will go after some care gaps and ignore others. Here’s a preview of some prioritizations:

• Health plan contracts will have obligatory quality minimums in addition to optional quality incentives. If a medical group is behind on contract obligations, especially for those representing a large portion of the patient population, then efforts to close those gaps will get prioritized.

• Among optional gaps, medical groups will focus on those associated with the largest incentive bonus; these often come from the VBC contract that represents most of their lives under care. 

• Once a contract is prioritized, operators focus on the most achievable measure objectives, based on the ease of improving them and how close they are to their goal. This can unintentionally result in deprioritization of the metrics that need the most attention. 

• Not all gaps are equally easy to close. Some gaps, like measuring blood pressure, can be closed by a medical assistant over the phone. Other gaps require an office visit with a physician. Priority gap closures also change as appointment availability shrinks and time available in Q4 runs out. 

While some large organizations, particularly academic medical centers, have developed effective care gap closure campaigns, the majority of small and medium-sized medical groups struggle with the infrastructure and resources needed to effectively pursue population health objectives. Several opportunities exist to quell the chaos of the Q4 push. For example, it may be necessary to revise the arbitrary reporting rules that govern care gap closures. Implementing a rolling measurement period for care gap closures could help smooth out the Q4 curve, allowing for more consistent, year-round efforts. Additionally, extending documentation periods or providing provisional credit for scheduled appointments could alleviate pressure on healthcare organizations, enabling them to distribute resources more efficiently throughout the year. 

The challenges underlying pursuit of population-level goals are not unique to the Q4 push to close care gaps. Just as value-based contracts are proliferating with primary care clinics, an increasing number of digital health companies specialized in specific clinical areas enter risk-based contracts with payers. As the network of organizations sharing risk and pursuing population-level objectives is growing, the patient relationship is splintering between more and more healthcare operators vying for their attention. Technology that connects care across organizations will be key to achieving population health goals.

Thank you to Ajay Haryani, Ian Perlman, Alex Kazberouk, Joel Gray, and Doug Murad for reading drafts of this.

“You’re too smart to go into primary care” has been the hidden curriculum across medical schools during my training. My medical school mentees confess not much has changed recently. Nonetheless, some (albeit fewer and fewer) primary care diehards persist against odds and end up in family medicine, pediatrics, or internal medicine. Anecdotally though, many graduates “back into” primary care because they were not competitive enough for other specialties. The 2024 residency match rate data tells the real story of where this trend is heading (Figure 1). Additionally, many physicians who transition into primary care from fields like hospital medicine or emergency medicine often do so not out of an original desire to practice primary care, but rather as a response to burnout or dissatisfaction in their previous career trajectory.

PCPs make significantly less compared to their colleagues (Figure 2). Medical schools have experimented with free tuition to potentially offset lower earning potential for those pursuing primary care specialties. Unfortunately, this has not proven to be true.

The Primary Care Collaborative illustrates the increasing outflow of PCPs in the last decade (Figure 3). Meanwhile, the AAMC predicts a shortage of up to 40 thousand primary care physicians over the next decade (Figure 4). The magnitude of this problem will be felt across half of the physician office visits in the US. Only a small and privileged sliver of the population can afford to ignore the problem of such epic proportions. 

What’s the big deal? Primary care matters. Plenty of evidence shows high ROI on primary care. For example, Oregon’s Patient Centered Primary Care Home (PCPCH) program found that $1 dollar investment in primary care returns $13 dollars in savings. Another study found that every 10 additional primary care physicians per 100, 000 population was associated with a nearly 2 months increase in life expectancy. 

How are the market forces correcting for the declining primary care physician supply? The rise of advanced practice providers (APPs i.e. NPs and PAs) are absorbing the demand and filling workforce gaps. It’s a practical solution, especially for those in the business of healthcare. APPs are usually half the cost of physicians and are much quicker to “produce” (Figure 5). 

To appreciate the context of the magnitude of the looming workforce challenges, it’s important to understand how vastly different physician and non-physician training actually is. Bringing up this distinction in training is not to label who’s “better” but to reinforce that clinical experience is at the heart of this discussion. I have learned a great deal of clinical pearls from those with more mileage than me, irrespective of their degree. I have also met physicians whom I would not recommend to my frenemies. Being good at medicine takes time. There are no shortcuts to high-quality care, even with AI copilots. 

Based on candid conversations with newly minted APPs, many share a common sentiment of feeling woefully unprepared to handle the breadth of undifferentiated patients and the complexities of chronic illness exacerbations across urgent care and primary care settings. Their frustrations include being expected to do the same work as physicians, i.e. see the same number of patients in a day, treat the same type of cases irrespective of complexity, and perform the same type procedures. Many have come forth with critique of their own training (Figure 7). A recent article has shed more light about “predatory institutions” that contribute to this trend.

The parallel conversations in the physician community have been disheartening to witness (Figure 8). If we perpetuate such silos in healthcare, we’ll only further entrench the gaping disconnect among those delivering patient care.

Let’s peel back the curtain on workforce composition at select large primary care and urgent care organizations based on billing NPIs (Figure 9). Not surprisingly, much of clinical care is delivered by non-physicians on aggregate. It’s worthwhile to note that ChenMed does not rely on NPs as much as others including its senior primary care counterpart Oak Street Health. Market trends indicate that urgent care centers must integrate primary care services to remain competitive. Patients now expect access to primary care at every healthcare “front door,” making it an essential service line rather than a unique selling point.

Note: Special shoutout to Michael Stratton and his team at Health Data Atlas for crunching the numbers and illustrating this data.

This provider distribution data reveals significant variability in workforce composition that is generally tackling similar “jobs to be done” across acute and chronic care settings. No single model has emerged as clearly superior. We're seeing a lot of experimentation. For instance, Firefly Health employs a virtual-first primary care model that relies on nurse practitioners as the clinical backbone of their interdisciplinary teams with physicians stepping in as partners on complex cases (Figure 10). This approach aims to match patients with the most appropriate provider for their specific health needs, potentially optimizing both care quality and resource utilization. However, we still don't know which approaches actually work better and what the ideal team composition should be.

Clay Christensen’s prediction has come to pass: “the host of technological enablers will fuel the disruption of specialists by primary care physicians in the future. In addition, these same technological advances will enable nurse practitioners and physician assistants to disrupt primary care physicians.” Unlocking effective collaboration between APPs and physicians is a key strategy for addressing the workforce challenges we are facing. This effort starts with reimagining medical education to make every clinician more effective and efficient. One approach is to integrate electronic medical records (EMRs) into the curriculum as a core educational tool, elevating clinical training to match the demands of the real world. XPC Clinic is leading by example. The future of primary care depends on our ability to innovate across education, technology, and care delivery models. By facilitating collaboration to enhance clinical training, we can build a primary care system that's not just functional, but truly thriving—one that we'd all feel comfortable aging into.

Thank you to Matthew Sakumoto and Michael Hobbs for reading drafts of this.

• Session 1: Opportunities in Primary Care

• Session 2: Pharmacist Education and Training

• Session 3: Optimizing Therapy

• Session 4: Clinical Superpowers

• Session 5: Pharmacist Care Operations

• Session 6: Metrics and Outcomes

• Session 7: Paying Pharmacists

• Session 8: Pharmacist Scope of Practice

• Session 9: Legislative Landscape for Pharmacists

• Session 10: Integrating Pharmacists

  ---

  ENROLL NOW

  ---

MEET THE INSTRUCTORS

SELECTED HIGHLIGHTS FROM THE COURSE

About half of all physician office visits happen in primary care. The concerning trend is the decreasing primary care workforce with a net loss of 10 clinicians per 100K people across the U.S. It’s troubling to see this because we have ample evidence that primary care delivers a high return on investment, with studies showing significant cost savings. For instance, a 2016 Oregon program documented that every $1 invested in primary care generated $13 in savings.

Between 1950 and the 1970s, the role of pharmacists expanded from dispensing medications to providing patient care services, reflecting a cultural shift within the pharmacy community. This led to the establishment of the PharmD programs, which particularly emphasized clinical education. The 1980s marked the beginning of the pharmaceutical care era, recognizing pharmacists as essential healthcare providers and leading to the universal adoption of the PharmD as the entry-level degree by 2005.

Becoming an ambulatory care pharmacist typically requires at least one year of postgraduate training, with additional certifications such as the Board of Pharmacy Specialties (BPS) or Board Certified Pharmacotherapy Specialist (BCPS) being helpful for job prospects. While it is possible to transition into this role without a residency, it usually requires more experience and connections.

The appointment-based model is a program designed to improve medication adherence and provide efficiencies for pharmacies. This model combines medication synchronization with an appointment with the pharmacist, ensuring that all chronic medications are refilled on the same day of the month. The key difference between this model and an autofill program is the preemptive outreach to the patient before the appointment date to update any changes to their medications. The impact of this model can be significant, with medication adherence increasing by 20% or more, making patients 3 to 6 times more likely to adhere to their medication regimens.

Pharmacists are also involved in point-of-care testing and the treatment of minor illnesses, which is particularly valuable for young adults and families who often seek urgent care for these conditions. For example, a program was set up in pharmacies where patients could receive a point-of-care test, have the results reviewed by a telehealth prescriber, and get a prescription while still in the pharmacy—all within an hour and at a lower cost than visiting an urgent care center. This model was found to save patients time and money, with the pharmacy visit costing about one-third of what an urgent care visit would cost.

Pharmacist salaries in non-retail settings, such as physician offices, are often higher than those in retail or hospital positions. For example, in 2023, pharmacists in physician offices earned an average salary of $142,000. In contrast, nurse practitioners and physician associates generally earn slightly lower salaries on average (e.g. approximately $130,000), with those in physician offices earning below the overall average for their roles. The demand for pharmacists in clinical positions, particularly in physician offices, has been increasing, driving up salaries in these areas.

Pharmacists can significantly impact healthcare quality measures, such as those defined by the Healthcare Effectiveness Data and Information Set (HEDIS) and Centers for Medicare & Medicaid Services (CMS). Ambulatory care pharmacists contribute to improving outcomes in areas like hospital readmissions, emergency department utilization, and potentially preventable complications. They also help reduce fragmentation within the healthcare system by serving as a bridge between different care silos.

In some cases, pharmacists can bill directly for their clinical services. For example, since March 1, 2024, Pennsylvania Medicaid recognizes pharmacists as mid-level providers, allowing them to bill for services within their scope of practice. The list of billable services is extensive, covering everything from Hemoglobin A1c testing to vaccinations and patient evaluation and management.

Some private payers also allow pharmacists to bill directly for clinical services. Pharmacists must be credentialed, enrolled, and contracted with insurers to qualify for billing. Credentialing verifies the pharmacist’s qualifications and ensures they meet the necessary standards to join a provider network, which is essential for receiving reimbursement from health plans.

As of now, 18 states have CPAs that allow pharmacists to prescribe medications based on rapid diagnostic tests. If you look at prescriptive authority for pharmacists with an advanced license in California, they are able to perform the bread and butter services of primary care: “initiate, adjust, and discontinue drug therapy, order and interpret tests, manage diseases in collaboration with healthcare providers, and refer patients to other providers.”

When evaluating favorable legislative environments for pharmacists, the key question is which states are willing to pay pharmacists for clinical services and recognize the full scope of their skills. Medicaid coverage for pharmacist services varies by state. Over 20 states have expanded Medicaid coverage to include pharmacist services, with 15 states granting full provider status, allowing pharmacists to bill for clinical services. Recently, Pennsylvania, New York, North Carolina, and Virginia have also offered pharmacist provider status.

Savvy primary care practice owners are beginning recognizing the immense potential of integrating pharmacists into their clinical teams. We predict that as practices scale, it will be common practice to bring on pharmacists as early hires over nurse practitioners or physician associates. The data and economics support this. Knowing how to operationalize this will be the differentiator for delivering high quality care. Reach out to us if you are interested to delve deeper into the topics we have covered.

P.S. XPC Clinic is open for collaborations 🚀

ENROLL NOW

XPC TALK THAT SPARKED THE MAKING OF THIS COURSE

Access Pre-Order Discount

XPC Retreats bring together thoughtful people to have candid conversations about hot topics in primary care. Healthcare conversations frequently face the Blind Men and an Elephant problem which we solve by making sure our retreat gatherings feature those from various healthcare silos (insurance, clinicians, brokers, health tech, private equity, etc.) who may not otherwise have the chance to speak freely with each other. The theme of our recent summer retreat was “Scaling Primary Care,” inspired by Dr. Eric Bricker of AHealthcareZ. He was kind enough to feature his takeaways in one of his famous white board videos. We share our own summary of the event with more detailed notes from the event for our paid subscribers, including thoughts from Gus Roman who helped us with organization and partnerships during our planning. 

First, a big thank you to Aasim Saeed, MD, MPA of Amenities Health, Kike Oduba, MD, MPH of WellnessWits and MassChallenge for supporting our event through sponsorship! As we look forward to future retreats, we are open to suggestions for future themes and welcome working with sponsors who like what we do and want to team up with us.

Paying for primary care

To stimulate and guide conversation throughout the retreat, we polled attendees on relevant issues. Scaling primary care requires obtaining patients and patients currently receive primary care from different payers. We first asked “Who should pay for primary care?”

Results reflected an overall belief that patients had the most to gain when taking charge, by way of payment, of their own primary care. Many of the current problems with the primary care system comes from misalignment of long term incentives when paid by third parties (a topic we cover in our Primary Care 101 course). Commercial insurance and employers only have incentives to keep patients healthy for a span of several years based on how frequently people change jobs and plans on average, but the government has an incentive to keep citizens healthy across their lifespan. One major difficulty brought up, among many, for the government to take over primary care payment was the fragmentation of the US government and uncertainty about which level (locality to federal) would take responsibility for payment and management of primary care. Those who picked employer and commercial insurance made a case about practicality based on how the health system exists today. Over half of Americans are covered under some form of employer sponsored or commercial insurance and this paradigm is unlikely to change any time soon. 

Patient acquisition markets

Many of our attendees were operators and clinicians. We wanted to see how they approached finding new patients for their services, so we asked  “What is the best market for patient acquisition?”

The responses suggest a bullish attitude towards the direct to consumer market, but conversation reflected otherwise. Different government models were discussed with some attendees having participated in CMMI models. Across the board, everyone appeared to agree that government programs have too many requirements and often fail to pay appropriately. Employers came next in discussion and, unlike the general distaste for government programs, just about everyone agreed on the potential for scale in the employer market. During the conversation about best markets, the impact of primary care came under scrutiny. Examples like employee wellness programs, which have had barely any impact on health or finances, were highlighted to question the validity of selling primary care. Ultimately, the discussion resolved with an agreement that different markets have different needs. 

The role of the PCP

On the other side of scaling is the supply of providers. We first wanted to see how people felt about the role of the primary care provider (PCP). “Is the role of the PCP primarily as a delegator/communicator or a ‘one stop shop’?”

Most of the immediate conversation after each poll question involved a bit of litigation about the wording and being forced into picking a single answer for complicated questions. This question raised more dissent than the others as most believed PCPs should participate equally in care navigation and medical care. But this line of dialogue was as fleeting as it was intense. The subsequent conversation tried to navigate the dichotomy and touched on a few interesting topics along the way.

PCP as Delegator (aks “referralist”)

In large cities and integrated payvider systems, referrals happen with ease and often with surprisingly helpful tech. References to the common analogy of PCP as quarterback came out during this section and the idea was raised of PCP as a “first stop shop” rather than a “one stop shop.” However, one attendee brought up the reality that most PCPs are referralists in hospital systems. Many of the clinicians brought up concerns around physicians having to see higher numbers of patients, forcing them to refer more than necessary. One participant mentioned people may not want to even attend medical school if they knew their future was simply to delegate care. 

PCP as “One Stop Shop”

The qualification “one stop shop for what you need most of the time” was more widely accepted than the moniker “one stop shop.” Different locations and stages of life require different expertise and local PCPs will always end up specializing to their population. Not to mention PCP as a one stop shop seems disingenuous when most PCP offices will carve out certain services such as mental health, vaccines, etc. Building trust lies at the core of the one stop shop conversation and a brief aside about tech came to an agreement that tech will always only augment the PCP because people will want a trusted person to confirm their chatbot/Google search/AI response. 

Primary Care and Medical Education

Certain tough points surfaced about medical education. Medical schools might be attracting the wrong people. People in the medical field propagate the myth of an obsolete physician archetype which prospective students seek to be when applying only to get disillusioned as their education progresses. Part of the issue comes from the knowledge base learned in school often not matching knowledge and skills needed in practice. PCPs start practicing in a largely administrative specialty without the proper administrative tools or training. Academic health centers came under fire as education in that default setting innately funnels students into hospital-based jobs, which often are the referral mills described earlier. 

Other Team Members

Nurse practitioners came up only briefly when discussing hospital systems beginning to replace physicians as PCPs. Pharmacists took center stage throughout the primary care team discussion. All the physicians had positive views of pharmacists they work with and have relied on their expertise in one way or another. Through bits and pieces of disparate stories about encounters with pharmacists, the group learned that pharmacists can do more than anyone individually thought, including patient education and, in some states, independently prescribing or refilling medications. Many of the operators and clinicians stated they would hire a pharmacist if their practices only had the budget to add one more person. 

Relationship with hospitals and health systems

Hospitals did not have friends in the room during this section. The 10.5% dissent reflected more of a difference in understanding the question. They read the question as “primary care should interact with hospitals” rather than “primary care should be owned and operated by hospitals.” Those who worked in hospital leadership offered up ways their hospital systems have implemented ACOs which supported primary care and ways systems have improved primary care quality and standardization. Everyone else countered with the misaligned incentives arising from hospitals relying on high reimbursement specialties for income while leaning on primary care as loss leaders incentivized to generate referrals into the system. 

Financial models of the future

Today’s primary care financing landscape allows three main models of sustainability to exist: direct to consumer (individual or employer DPC), value based contracting (in all its many forms), and loss leader to refer to lucrative procedures and specialties (current hospital model). Those who chose the downstream referral model were skeptical at how quickly the healthcare system would change based on how slow it has historically responded to changes. Several participants brought up stats on the growth of VBC and DPC which affirmed the rest of the group’s optimism in seeing either model dominate in the next 20 years. VBC contracting is largely driven by government payer mandates and features mostly hospital systems but has been trending towards including smaller groups and practices (e.g. ACO Primary Care Flex Model). The spread of DPC has been primarily driven by growth in the employer market.  

Having a PCP

Healthcare tends to be a “do as I say and not as I do” field. In a room full of healthcare leaders and innovators, only about half reported to have their own PCP. The results were met with some shock and some amusement, though in other events we’ve hosted, around 50% seems to be a standard percentage. Discussion after this question navigated around what having a PCP means. Some people treat companies as their PCP rather than individuals and others, namely those with chronic or rare disease, will use their specialist as a PCP. There was also a tangent about preventative medicine and what insurance will cover, but before the conversation devolved into clinicians arguing over population health and number needed to treat, we brought the focus back to care delivery. The individuals from the insurance companies talked about the difficulty of having their covered lives engage with PCPs as well as the messy process of PCP attribution and assignment. 

Our article continues for paid subscribers with highlights of additional discussion topics: the evolving role of urgent care centers and an investor’s perspective on scaling primary care plus bonus quotes from hallway conversations.

Read more

Primary care physicians (PCPs) have historically been described as quarterbacks or gatekeepers of care, but to our knowledge no one has conceptualized them as universal application programming interfaces (APIs) of healthcare. In this article, we illustrate

• why interoperability is their superpower that no other specialty can match 

• how solving primary care challenges impacts the entire healthcare ecosystem

• what types of collaborations are needed to build solutions that resonate with patients, clinicians, and payers alike

By bringing together clinical and technical domains, we aim to catalyze a new generation of primary care builders that traverse across healthcare silos. We demystify how clinical information flows in care delivery and identify missed opportunities for disruption. 

Let’s review the fundamentals. What exactly are APIs? 

An API is an interface between different software applications, allowing them to communicate with each other in an agreed upon format and exchange information. APIs come with codebooks (idealistically), maintained and updated to define the available calls and the parameters of the expected return. API calls are specific messages sent through this communicator. Common API calls include GET (gathering data), POST (adding new data), and PUT (updating existing data). 

What types of “API calls” do doctors make?

Physicians are the interface between patients and the healthcare ecosystem. On the surface, they GET(symptoms), POST(prescriptions), and PUT(treatment plans). But in practice, their role extends past simply translating from patient to health system and dives deep into the complex bureaucratic abyss that is navigating and operating health systems. Figure 1 illustrates the diverse array of API-like calls primary care doctors make, showcasing the multifaceted nature of their responsibilities in care delivery.

The superpower of interoperability across specialties

Most people do not understand what primary care is capable of. PCPs now manage a growing range of conditions that previously warranted specialist interventions. This is driven by a modern healthcare paradox: patients are much sicker yet live longer. To keep pace, PCPs leverage technological advancements in data consolidation, distribution, and search (e.g. UpToDate, DynaMed, AAFP). Such clinical decision support tools have effectively diluted information asymmetry in healthcare. What used to be in the hands of specialists is now at the fingertips of primary care doctors (and will eventually enter the knowledge domain of patients themselves allowing them to “practice at the top of their license”). Thus, symptoms or diagnoses that used to trigger immediate Referral APIs now route to the PCP Knowledge APIs. This increases the significance of PCP contributions in healthcare because primary care goes further into every field by managing increasing complex medical conditions across multiple organ systems. See Figure 2 to delve deeper into the Knowledge API calls to better understand the breadth of primary care.

Here are a few examples of formerly specialist care that is now commonly managed by PCPs:

• Cardiology: Cardiologists often defer the management of anticoagulation (e.g. warfarin therapy) or treating hypertension to PCPs unless the patient fails to respond to multiple treatments.

• Nephrology: Nephrologists prefer not to engage in managing acute kidney injuries (AKIs) that can be addressed in primary care or see patients with chronic kidney disease until it progresses to stage 4.

• Endocrinology: The management of conditions like hypothyroidism or type 2 diabetes is increasingly being left to PCPs unless the patient shows non-responsiveness to a range of treatments at the primary care level.

Several market and practice trends are also driving this change:

• Logistical challenges: Accessing most specialist care takes longer than getting an appointment with a PCP who tackle the burden of patient needs by extending their scope of practice. This phenomenon is particularly true in rural communities.

• Increased specialization: Specialists are increasingly focusing on highly complex cases and PCPs are managing the growing pool of moderate complexity conditions.

• Value-based care models: These models emphasize reducing unnecessary escalations in care to manage healthcare costs effectively. Consequently, the responsibility falls on PCPs to minimize referrals and manage more conditions in-house, thereby reducing the financial risk for patient populations.

The challenges of interoperability

As the scope of primary care has expanded to manage more complex patient needs and care offerings, so too has the influx of data into the practice. This entails handling incoming API feeds with lab and imaging results, reports from specialist referrals, request for DME orders and prescription refills, and also the inbound from patients and their families through phone calls, emails, and direct portal messages. The volume of these inbound signals can overwhelm PCPs, the equivalent of generating an HTTP error 429 due to “too many requests.” In medicine, we call that burnout.

Volume of API calls isn’t the only issue faced by PCPs. Many of the healthcare API calls are “sloppy,” with no maintained codebooks. PCPs are essentially functioning like analog switchboard operators trying to appropriately navigate this disorganized system. The reality is that AI cannot effectively replace this role if all entry and exit points within the healthcare network remain chaotic, ill-defined, and often not even digital. The connection points within this system are haphazard and heavily reliant on local or insider knowledge. For example, there's no universal codebook that guides interactions with specialists or diagnostic services across networks. When primary care does decide to refer to a specialist, the PCP is responsible to plug into the specialist system through the Care Coordination and Referral API calls, capturing both patients’ entry and exit points across disparate institutions. Much of this interoperability is achieved through labor-intensive manual processes (e.g. making phone calls, sifting through faxes, etc.) - a system with significant latency. This lack of standardized and accessible information creates a significant barrier to delivering high-quality care.

The problem of ill-defined API calls

Physicians often have a bad reputation for being too focused on the EMR when interacting with patients. This EMR-centric focus stems from the volume of EMR-based API calls they are expected to handle during an office visit (e.g. documentation and billing requirements, ordering of labs and tests, etc.). These API calls in the EMR are often the better defined ones. As a result, AI solutions are targeting the “easier” or less messy API operations, such as refilling meds (e.g. Prescryptive), interpreting labs (e.g. Elaborate), automating billing (e.g. Diagnoss), or synthesizing medical knowledge from journals (e.g. Open Evidence). These solutions tackle the issue of volume overload, but the opportunities to make a real difference lie elsewhere. 

API calls outside of EMR are less clearly defined. Admin API calls are worth a special mention because of the range of skills needed to handle this category of tasks within local contexts (e.g. paperwork for FMLA, DME, disability placards, or workers compensation). These still exist primarily as printed paper forms which are not standardized, and often are full of bureaucracy at best, and intentional obfuscation at worst. The completion of these admin tasks is invisible to medicine when completed as expected but becomes painfully prominent when the tasks (or “packets”) get dropped. For example, incomplete FMLA forms can mean no household income for a patient during a medical leave. Incomplete records retrieval for pre-op evaluations can derail elective surgeries and cause scheduling nightmares, create frustration for patients, and lead to avoidable lost revenue for the surgeons. It is no surprise that entire startups exist to fill the gap of guiding patients through the healthcare labyrinth for pre-ops (e.g. Orchestra Health) or during health scares (e.g. Carte Clinics). But these problems and approaches tend to rely on localized or regional system knowledge, are human driven, and are time and labor intensive. These tasks are not celebrated in a system that rewards interventions over incremental care.

We also continue to be painfully far away from mapping out API endpoints for interventions addressing social determinants of health, such as resources for food insecurity, domestic violence assistance, housing or transportation needs, and counseling services. While it’s not exactly a physician’s job to deal with the root causes of these predicaments, overlooking them undeniably contributes to worse health outcomes (e.g. we can’t prescribe insulin to a patient who doesn’t have access to refrigeration for storing the medicine). Additionally, outcomes like managed diabetes are often the very metrics that dictate physician pay (e.g. HEDIS measures or ACO performance targets). Thus, the primary care office becomes the de facto home for handling the social needs of patients, even if it is not empowered or resourced to do so. It’s hard to do this at scale because many of the solutions are local with the answers stored in physical 3-ring binders and pamphlets at social service departments rather than in digitally accessible formats like JSON databases.

It is no wonder these are not the target of technology or AI innovation as these problems are not sexy, don’t pay well, are poorly defined, and are hyper-localized and difficult to scale. Only visionary builders motivated by true impact will take on such challenges. Solving for primary care pain points will impact the entire healthcare ecosystem because PCPs are part of the majority of workflows, both upstream or downstream of specialty care. We believe that the most meaningful impact will come from operationalizing and scaling interoperability for the ill-defined API calls.

The path forward 

In light of these challenges, there are two pathways forward. Either we take significant strides to clean up and standardize these connection points, making them more like the meticulously documented APIs of tech companies, or we reinvent how information flows and break out of the legacy system. The first approach requires a concerted effort from all stakeholders – from frontline clinicians, payers, patients and local communities. This transformation demands a fundamental shift in how we view the multitude of touchpoints in healthcare. It calls for a clinical workforce that is much more tech savvy and leadership that is more tech literate. Aligning incentives at scale is going to be difficult. 

The second approach entails a more radical overhaul of healthcare interactions. This requires at least two changes: 1) stronger coupling of PCPs to patients and 2) placing patients, rather than physicians shackled to EMRs, at the center of API calls. Patients are poor navigators of healthcare on their own, but when paired with a trusted expert, chances of finding timely high-quality and low-cost care increase dramatically. Direct primary care is the poster child for this phenomenon. Further synergies will come from decentralizing healthcare API calls to yield leverage to patients similar to how Photon Health has reimagined medication prescribing around the patient experience. As a result, we imagine, just as hospital metrics birthed the field of health system informatics, a new crop of informaticist-PCPs will emerge to help broker API calls for patients – not dissimilar to how doctors help patients interpret lab results today. These roles will serve as the linchpin to managing human and digital care navigation resources, data streams, and systems of record. If digital therapeutics (DTx) make a comeback, informaticist-PCPs are poised to specialize as the platform operators.

This PCP-patient dyad model to co-manage an API hub essentially inverts the current structure where patients have limited visibility or control in healthcare. New companies can tap into these flows to provide supporting services while incumbent players are pressured to reorient business models around consumer preferences. The key is architecting an infrastructure where patients sit at the heart of operations rather than the periphery with proper admin support from their PCP offices. Favorable regulatory changes are equally necessary and maybe then personal health records (PHRs) will make a comeback with the right use case (i.e. rooting for Mere Medical). 

We reimagine primary care as the universal API because PCPs are the connectors of healthcare in its current form. It is exciting to see the progress of the quickly evolving AI technologies automating primary care tasks inside the EMR, but the most impactful work that’s yet to be done lies in the space of ill-defined API calls. We believe that the future of primary care innovation will be shaped by the technologies that extend beyond the clinic walls and strengthen the collaboration between patients and their PCPs. Solving for this will be the next frontier of innovation in healthcare.

Thank you for reading our musings on opportunities to improve primary care. We are excited to support builders in this space. Don’t hesitate to get in touch!

Health tech has reached maturity. It’s not totally clear when this happened but at some point mid-pandemic everyone on LinkedIn started sharing logo maps attempting to organize our suddenly in vogue field - by services offered, by number of patients served, by total funding raised. Some might argue that it was also the height of the unbundling pandemic. No longer just an interesting “sell the dream” space, health tech now represents the innovation engine of healthcare broadly. Along with this maturity new organizational structures and roles have developed within health tech organizations, including the emergence of the Clinical Innovation team. 

In this article, we sketch out the “Clinical Innovation Mindset Stack” and further empower those driving clinical innovation from within health tech organizations. Our perspective here is that of practitioners doing the work, trying out new approaches, and seeing what works. As we iterate on what’s working right now, we invite feedback on what works for others. We’ll describe the typical work of the clinical innovation team, the benefits of retaining a beginner’s mind with this work, “the stack” we’re using on a daily basis, and ways you can get started.

Let’s start with a quick working definition of clinical innovation: innovation and improvement that originates from within the clinical organization and advances at least one component of the Quintuple Aim of healthcare without compromising others. We are particularly interested in exploring clinical innovation from the perspective of healthcare delivery, where patients are the ultimate customer, though all of this advice is broadly applicable. 

A couple of disclaimers: 

• Clinical innovation can also refer to a bunch of things that we’re not trying to write about here, including entrepreneurialism in health tech, technology or research advances in healthcare, and the function of acquiring new clinical and operational capabilities for your organization through vendor arrangements, JVs, or M&A. All of that stuff is awesome, but it’s not our focus here.

• Innovation is of course happening everywhere throughout the organization. Innovation is a function of culture, not one team or role.

The People

The clinical innovation team may live in several different places within the organization. Here are some common arrangements: 

• Clinical Operations, reporting up through the Chief Medical Officer. This is probably the most common arrangement.

• Operations, reporting up through the Chief Operations Officer. In this arrangement, the team may drop the “clinical” descriptor in exchange for something more general, like Performance Innovation or Care Innovation. 

• Senior Leadership Team, reporting to the CEO directly. This is particularly an option if the CEO is an MD or other type of clinician. 

Clinical innovation team members could be a mix of clinicians (practicing in the organization or not), operations leaders with strong analytic and communication skills, and healthcare consultant-types who have found their forever home of a client. It is worth noting that the clinical innovation “team” may start off as one person, or even a fraction of one person’s role, and then develop as the organization grows. 

The Work

So what’s the actual work of the clinical innovation team? Generally, clinical innovation is about problem-solving. Our advice to anyone just getting started in this space is to actively embrace problems. Problems are the main input of the innovation process. The clinical innovation team’s core work starts with deeply understanding the problems of your patients, the care team, and your business. Some questions to get you started: 

• What are the top 3 patient complaints? 

• Where are there bottlenecks and friction in a core workflow? 

• What workflows have a high rate of failure? 

• What does the care team describe as their biggest challenges? 

• What new services could we add that would drive health outcomes for our patients?

Being “solution-oriented,” as one finds in many job descriptions, is only half of the equation. First, we recommend being “problem-obsessed” so that improvement efforts are informed by a deep knowledge about the problem or challenge. Our advice: Go see the problem. Or even better, experience the problem. Clinicians who still practice are particularly well positioned for problem (read: product and customer) discovery and can leverage their role to surface insights uniquely available to them. The added benefit of clinicians feeling the “pain” firsthand is that they are motivated to develop and implement improvements that will make their own lives easier. If you can’t experience the problem yourself, then the second best approach is speaking with the patients and the care team members experiencing the problem or service gap. Get forensic and do some chart dives. Document what you learn in a way that tells the story of the problem. This approach is not just about identifying issues but also understanding the context in which they exist. The space between the patient and their healer is steeped in tradition and ritual. These elements, part of the 'way we've always done it,' significantly influence the nature of the problems you observe and the resistance to change. Recognize that you're navigating a landscape shaped by centuries of practice, where historical inertia plays a crucial role in how problems are perceived and addressed.

Teams that embrace many types of problems develop an appetite like that of an omnivore - they are comfortable eating whatever lands on their plate. A diverse problem diet is an asset as it will deepen the team’s understanding of the overall health of the organization and where its problem areas are. In any given quarter, the team will likely support a mix of operations-focused efforts, like launching a patient scheduling process or redesigning the onboarding flow, as well as  clinically-focused efforts like launching new services and developing new protocols. The content of the challenge will change, but it’s all about solving problems. Developing a versatile, flexible problem-solving framework will help the team support innovation and improvement across the organization, no matter what’s on the menu. And context matters. Working for large corporations is like showing up at all-you-can-eat buffets while startups are more like potlucks where you BYOP (bring your own problems).

The Stakeholders 

Healthcare organizations often refer to themselves as “patient-centered.” To really understand an organization’s commitment to this ethos, observe how empowered the frontline care team is to drive improvement and innovation on behalf of the patient. One of our most strongly held beliefs is that the best improvement and innovation work comes from the care team because they are closest to the patient. One of the essential functions of the clinical innovation team is to facilitate problem-solving in partnership with care team members by bringing structure to problem-solving conversations. Some examples: facilitating ideation sessions, helping the team design an experiment, making a workflow more visual by sketching it in FigJam, and setting up prioritization exercises, like dot voting or an effort-impact matrix. 

Creating space for patients to add value as co-innovators is essential. The best innovation workshops often include all stakeholders, which can be hard to achieve in practice. There are a lot of ways to bring the patient’s voice into the room, like pre-recorded interviews or narratives from care team members about their patients. However, inviting an actual patient into the room/Zoom is often the right move and can be done with the right framing and support from the legal team. The value of patient representation is recognized by federally-funded programs. For example, Federally Qualified Health Centers (FQHCs) are funded by the Health Resources & Services Administration (HRSA) and require at least 51% of the board members to be patients who are served by the community center. Researchers are evaluating the impact of patient advisory contributions on healthcare outcomes. Such findings could further drive patient inclusion and participation in care delivery innovation.

Beyond one-off workshops, continuous feedback from patients throughout their journey can support the organization’s improvement culture. This approach goes beyond outcome metrics like NPS or other satisfaction measures and invites feedback on any part of the patient experience. Checking for satisfaction at the end of a patient’s journey risks recency bias and likely misses the opportunity for micro-improvements in upstream processes, like intake/onboarding. Cultivating feedback across internal and external stakeholders is a necessity for survival and adaptability. Making it easy for everyone to provide feedback means simplifying the process, being receptive to feedback at multiple touchpoints, and ensuring that their input leads to tangible changes. Simple changes such as embedding feedback buttons in EMRs and patient portals can be a game-changer. Ultimately, when clinical teams feel a sense of agency to affect their work and patients take on a proactive role in their care experience, care delivery organizations become more responsive, resilient, and attuned to the needs of those they exist to serve.

The Mindset

The Zen Buddhist concept of Shoshin refers to having the mindset of a beginner, no matter how expert or advanced you may be. Beginners are also learners. Learning requires openness, curiosity, and humility. This is the essential mindset of those driving innovation. 

In our experience, a key signal that you have a healthy innovation process is that there are moments of shared epiphany. It’s those moments where the room/Zoom volume gets loud with the team’s excitement about a breakthrough or a new idea that they want to try and everyone is smiling. This experience is the “runner’s high” of an innovative culture. These epiphanies don’t happen when everyone brings their “expert” mindsets to a challenge - or if they do, it’s way less fun to be a part of.

One way to practice this mindset is to ask questions that challenge your assumptions and invite exploration and curiosity rather than asking questions with the expectation that the answer will further prove your point or reinforce your current thinking. This may seem risky at first, exposing you as uninformed or naive. To start, practice it in low-risk settings and do it with purpose (e.g. what are you trying to understand more deeply?). Demonstrating your beginner’s mind also helps build a permission structure for others to do the same. 

Sample questions to have at your fingertips:

• How can I view this situation from another perspective?

• What is the core issue I’m trying to understand here?

• What am I not seeing that could be relevant?

• Could you help me understand why you built it that way?

• What’s the purpose of this step in the workflow or patient journey?

• What are you hoping the patient does next?

• What aspects of this workflow might be overlooked by those not engaged in it on a daily basis?

The Competencies

Establishing the right mindset for driving innovation is a prerequisite, in part because almost no one starts their career in clinical innovation with the entire stack under their belt. Much of this will be learned along the way. This stack is assembled from preexisting parts. The methods and frameworks we’re describing already exist and have proven effective in other settings. What we hope is unique and helpful is their synthesis and reframing in the health tech context. With that, let’s look under the hood of clinical innovation.

1. Lean Methodology: 

“Without standards, there can be no improvements” - Taiichi Ono, originator of the Toyota Production System

Lean is a methodology for delivering greater value to the customer through continuous improvement and the removal of waste in core workflows. Many organizations practice lean in some form, but its application is highly variable from the casual fan to the orthodox practitioner. If you dip your toe in, you’ll quickly learn that lean is a deep pool of highly developed approaches, rival schools of thought, and strong opinions about “the right way” to do lean. Our advice: put on your innovator’s floaties, jump in, and figure out what works for your organization.

Lean has a long history of driving improvement in healthcare, though its origins are in car manufacturing. In his seminal book, The Toyota Way, Jeffery Liker describes 14 principles for developing a culture of improvement and building an organization of empowerment where teams ingest problems and metabolize them into innovation. Start with The Toyota Way - read it as a field manual, ideally with a colleague or two. You can also cruise through the resources on the Institute for Healthcare Improvement website.

How is it relevant to health tech?

In healthcare, we have standards of care. Lean applies that same rigor of promoting the best known way to all workflows, whether clinical or operational. Standards are the baseline for improvement and form a foundation that we can build on. Without them it’s difficult to gauge how new changes stack up against the status quo. Standards also create stability for the team and the patient, and scalability for the organization. Working in health tech can feel like you’re drowning in pilots, MVPs, and proofs of concept. This culture of innovation is what makes the field exciting, but it can feel disorienting and unstable if workflows remain duct-taped and pilot-y even as the business scales. 

Lean’s focus on standards is a helpful counterweight to the dread teams may feel when the next “We’re launching another pilot! 🚀” email hits their inbox. The rapid iteration cycles common in startups are a feature and typically a signal of a healthy culture of innovation. Introducing standardization to that innovation cycle will support deployment and scaling of innovation work. An interesting parallel exists in how the DevOps methodology accelerated the speed and agility of software deployment in large organizations (“10+ Deploys Per Day” kicked off this movement and The DevOps Handbook is a modern guide for achieving this). 

How do I get started?

Our advice is to start with a core workflow that is performed with variation and develop Version 1 of the standard workflow. Variation implies that there is no standard for doing the workflow. One of the fastest ways to discover variation is to have team members watch one another work and discuss what they learned afterward (institutionalized as “facilitated practice” at ChenMed) - an experience that often disappears for physicians after residency. The variation you observe can be viewed as a short-term asset – it represents a series of unintentional experiments conducted by the team to determine the most effective workflow. Version 1 of the standard workflow is likely hiding in the team’s variation. The first step is to align on what the workflow needs to do for the patient or the care team. In other words, what value does this workflow need to deliver? Once you can answer that question, you’re looking for the workflow that delivers that value in the simplest, fastest, safest, and most repeatable way. The documented variation now becomes Version 1 of the workflow, which needs to be disseminated to build consensus for adoption.

2. Design Thinking:

“But people aren’t cars!” This knee-jerk critique of lean and its origins in car manufacturing comes up sometimes. It’s likely a reflection of someone’s experience with a poorly implemented lean program or just a lack of familiarity with lean. Nevertheless, design thinking pairs really well with lean, particularly in the context of human-centered sectors like healthcare where lean can seem like an incomplete approach. Lean and design thinking share many of the same principles even if the specific language is novel to each. A short list of examples: 

• The belief that those experiencing a problem hold the insight to unlocking progress.

• The importance of understanding the end-to-end workflow from the perspective of those doing the workflow and experiencing pain in the current state.

• The belief that progress happens through cycles of learning achieved typically through iterative experimentation and a mindset of curiosity. 

Design thinking translates the “patient-centered care” ethos into action by providing a framework for putting the patient at the center of the design process. Design thinking anchors the innovation process in active empathy for the patient by immersing innovators in real observations of the patient experience. The patient voice is not relegated to an end-of-funnel feedback survey, but rather an upstream, pre-design stakeholder voice. The design thinking method is fairly well established and IDEO’s method is considered the method.

Begin by framing a question to focus your team on understanding the real needs of your customers, and then seek inspiration through observation and discovery. Use this inspiration to generate innovative ideas, build and test prototypes, and upon finding the right solution, share the story with your stakeholders.

Like any method, there’s the version described in clear, discrete steps and then there’s the reality of its application. The design thinking method is not a set of instructions for “executing innovation” but rather a guide for orienting teams as they move through the phases of innovation. It’s helpful to know which step you are on and what is likely the right next step, but you may choose to move backwards as you learn more or you may (gasp!) choose to skip a step when the pressure of delivering in a startup environment requires it. Though, word to the wise, don’t skip the “Test to Learn” phase. You can shorten it, but be cautious in skipping it. 

How is it relevant to health tech?

Design thinking scales well. Regardless of how many patients an organization serves and how much revenue it generates, the design thinking method draws your attention to subjective patient experiences and observations about real patients and care team members. Growth and success can sometimes trick us into thinking that we already know what the patient needs and what they value in our product. Practicing design thinking and its active empathy will keep us close to the patient experience and help us avoid slipping into a closed, expert mindset. 

Design thinking also brings a fresh perspective to legacy problems. Is there a more legacy problem in healthcare than wait times? Waiting rooms are as synonymous with healthcare as white coats. Children playing “doctor” typically start the narrative in the waiting room. To reduce wait times, many organizations reach for the lean playbook, which helps reveal root cause drivers like schedule design, appointment start times, and patient wayfinding. 

Here’s where the design thinking + lean combo gets really cool. If we get closer to the patient in that waiting room, we might learn a little more about their experience. No one wants to wait, so there’s no reason to overthink the value of reducing wait times. Allowing ourselves to be curious about the needs of the patient in the waiting room could lead us to deeper learning about the patient’s experience. Perhaps we even figure out how to get rid of the waiting room?

Here are some questions a design thinking mindset may lead us to asking: 

• What is the patient worried about as they wait? 

• Is this their first time seeing their care team?

• Do they have concerns about insurance or cost? 

• Do they have someone with them in the waiting room? If so, who are they and why are they there?

• Did they just come from a blood draw or imaging and are carrying with them information for the care team?

Design thinking helps us explore problems more deeply, particularly when elements of those problems are difficult to measure. Wait times are a great problem for lean as they are highly measurable. Design thinking gives us permission (requires us even) to explore questions that are less measurable - feelings, perceptions, the why behind human behavior. 

How can I get started? 

To better understand the method, we recommend starting with IDEO’s generous list of resources for facilitating activities and explaining concepts. You can also download their field kit for free (how is this stuff free?!). We also love Oscar’s approach to collaborative storyboarding described in this Medium post. This is a great example of a team going through the design thinking process and expressing what they learned through a journey map. It’s also a relatable story from a well-known OG health tech organization. The Oscar team does a great job of describing their method. Also, here’s the IDEO guide for creating a journey map - with a healthcare example to boot.

The other really thoughtful thing that the Oscar team did is that they developed a fictional patient, Gina, and built the workflows around Gina. You don’t necessarily need to hire an illustrator like the Oscar team did, but sketching out visual representations of your patients is a good way to force yourself to really think about your patient and represent them visually. Empathy mapping is a great way to bring your patients into the room. Here’s IDEO’s empathy map guide.

Design thinking has become a social technology that helps us not only ask more impactful questions but also transforms us as innovators in the process of deploying it. By learning how to challenge our own assumptions, we guard against biases that constrain our perspectives.

3. Clinical Experience: 

Health tech innovations are changing the way healthcare is delivered. A hybrid experience is emerging where some things happen in person and others happen virtually. Even with these changes, most care is still delivered by a person in an exam room, surgical suite, or pharmacy anteroom. Every clinician working in health tech today was trained in person in a hospital setting. Plato’s Allegory of the Cave is relevant here. If we do not understand how traditional clinical medicine works, we’re inferring what it might be from the shadows on the wall. We need to get out of that digital cave and into the clinic. There is a lot of wisdom accumulated in the clinic setting: anticipating unintended consequences, downstream effects, and insights about patient behaviors and expectations. For example, primary care physicians have a distinct advantage in grasping the subtleties of care transitions due to their training in various environments. Their experience spans initiating referrals from outpatient clinics and urgent care centers to managing patients referred to them in hospital and emergency room settings. Such experiences give insight into when it may be premature and unnecessarily costly to send a patient to the ER while recognizing when it’s prudent to not delay escalation of care. Clinical decisions are often intertwined with decisions of capital and resource allocation in the real world where medicine is not immune to the laws of business. Having access to clinicians with expertise in end-to-end continuum of care can be a differentiating factor for startups building next generation tech-enabled healthcare services.

How is it relevant to health tech?

Health tech, at its best, is about disrupting thoughtfully evolving healthcare. If you are coming to health tech from another space, then it is particularly important to commit yourself to understanding the practices and patterns of the clinic. This will build trust and accelerate your work. It will also help you identify some of the friction points we’ve imported from the old days that still need creative problem-solving. Patient scheduling and capacity management are still necessary even if exam rooms are sometimes Zoom rooms. Patients who join appointments virtually may have the same expectations for care as those who arrive in person even if their needs are different. We still care how much patients wait before an appointment, even if there is no waiting room. And as much as health tech wants to eliminate the waiting room, it’s important to understand what it’s replacing. For example, in rural communities, waiting rooms are often the locus of socializing. It’s the rare opportunity for neighbors to catch up with each other or bring home-grown tomatoes to their doctor as a sign of gratitude for their trusted relationship. 

Non-clinical founders are strongly encouraged to team up with their clinical counterparts. Having the right industry insight from the beginning can prevent many early pivots and avoid unnecessary distractions (Primary VC hosted a fascinating discussion from seasoned founders on this very topic). Onboarding clinical talent brings not only their subject matter expertise, but also their clinical networks from years of training, practicing medicine, and experience in endless pain points in care delivery. Valuable lessons are learned by practicing medicine that cannot be gleaned from simply passing multiple choice exams that grant the “MD” title. Access to guidance from clinicians in the trenches is a competitive advantage to health tech builders. Some investors even argue that more physicians should be health tech CEOs.

How can I get started? 

Ideally shadow clinicians in-person to soak it all in. Talk to them about their training and identify gaps in resources and explore the principal-agent problem that is pervasive in healthcare. If medical shows are going to be your curriculum, Scrubs is the closest to real life. Lean heavily on the beginner’s mindset. Ask about anything you don’t understand: exam room layout, the sphygmomanometer ritual, the flow of money, documentation requirements, or any strange acronyms (e.g. what does SOAP or MEAT mean?).

Consider exploring low-tech and no-tech innovations that can be equally impactful. For example, the Patient Priorities Care (PPC) clinical framework optimizes patient care by hyperfocusing on patient-defined goals (known as “The One Thing” at Devoted Medical) to guide all healthcare decisions. This model does not require transformative technology and relies on millennia old innovation: communication. Standardizing a workflow like this across most EMRs or even paper charts is a piece of cake compared to navigating interoperability at Epic. Successful execution involves task management across the care team in supporting the priorities of patients. Improvements in objective metrics for diabetes, hypertension, and congestive heart failure management back up the innovation rhetoric for PPC.

4. Product Partnership

Understanding the role of the product team is crucial and may not be intuitive for those coming from traditional healthcare settings. It would be a mistake to think of your product team as equivalent to the IT Team or an internal EMR specialist team. The product team decides what does, and more importantly, does not get built in the first place. They are one of your key stakeholders and partners in driving innovation in your organization. 

Similar to the way product managers exert influence over internal stakeholders without having direct authority, clinical innovation teams leverage their unique insights and position to ship “innovation.” Both positions conduct user and stakeholder interviews. They also interface with software engineers and designers. This is critical because clinical and operational know-how guides what code is written and informs the ultimate design experience for patients or clinicians. You can’t automate a highly variable workflow. The clinical innovation team can help drive standardization, which then enables the product team to support automating parts of the workflow and return time to the care team.

How is it relevant to health tech?

Depending on what your organization is up to, the product team may be developing internal products (an internal EMR, revenue cycle widget or a communications platform) and externally facing products (e.g. an app for patients or caregivers). Consider the administrative burden in healthcare that plagues patients and clinical teams alike. Primary care presents some of the most interesting opportunities for care transformation for product-minded innovators. The entire specialty has become a catch-all for all patient related admin tasks. Most specialists turf responsibility for patient needs outside of their specialty’s organ system to “follow up with PCP.”  This heavily administrative “privilege” is not matched with access to project management tools for PCPs. Traditionally, the tools physicians use (e.g. EMR) have been designed to service billing requirements. Virtual primary care trailblazer, Dr. Jay Parkinson, has illustrated that many of the basic things we do in medicine can be thought of as projects. For example, UTI requires a few days to manage until resolution, pneumonia - a couple of weeks, and breast cancer - many months. Pairing insights to administrative needs with product expertise can accelerate the solutions that clinical innovators roll out. Having a strong product partnership ensures that such iteration cycles align to stakeholder value.

How can I get started? 

Meet people in product: designers, product managers, and chief product officers. Ask to shadow them to understand how they frame problems and spot opportunities. You’ll notice that product leaders anchor the innovation process by continuously reconnecting the day-to-day building work back to the core problem they are trying to solve for patients or care teams. Take Product Management courses, such as Product Management Basics from Pendo and Mind the Gap, to learn core concepts. Read industry classics such as Escaping the Build Trap and Inspired to appreciate the product development philosophy. Join industry communities like PMs in Healthcare to connect with fellow builders and innovators. Immerse yourself in the world of product, not necessarily to become a product manager, but to gain appreciation for the user-centered approach product teams bring. The goal is to learn how product pros think so you can better partner with them. Their skills will strengthen your innovation ecosystem.

5. Systems Thinking

“Today’s problems come from yesterday’s solutions.” - Peter Senge

It’s inevitable that some of the innovations we ship will cause unintended consequences. The only way to decrease the odds of unwanted future friction points is to apply a systems thinking approach to the totality of our interventions. This way we can guard against siloed thinking and short-term horizons. The key insight here is to consider second and third order effects of initial actions. We do so by examining interconnectedness of all components and meticulously thinking through how they might influence each other. In this process, we consider all stakeholders and the impact of our solutions on their experiences. Part of the strategy is proactive reflection that helps anticipate which groups of stakeholders might be adversely affected and which processes will require course correction. 

How is it relevant to health tech?

Systems thinking in health tech is about understanding and optimizing the complex web of interactions in healthcare delivery. Distribution and integration are the two major bottlenecks for many healthcare innovations and products. Identifying what forces (e.g. regulatory, technical, or financial) influence their adoption is part of the puzzle when operating in an opaque system. It’s rare for incentives in healthcare to be aligned, which causes frustration among innovators when their rational solutions don’t gain traction. Identifying entrenched behaviors through a systems lens is one way forward. Considering the ripple effects of each novel intervention among a multitude of stakeholders is what makes clinical innovation both bewildering and exciting. For those building point solutions, consider how each patient flows into and out of your ecosystem.

For example, value-based systems take on risk and are incentivized to reduce costs for avoidable expenses, such as inappropriate ER utilization. Innovations with a systems perspective would consider alignment across many domains impacting a patient’s decision to go to the ER, such as: 

• primary care access (PCP not available)

• care navigation needs (not knowing where else to go)

• medication costs (seeking rescue medications in ER due to inability to afford maintenance treatments)

• health literacy (not understanding what ailments can be managed outside of ER)

• financial literacy (not understanding cost implications of frequent ER visits)

• transportation limitations (ambulances take patients only to ER)

• food security and housing stability (ER offers a snack and temporary shelter)

How can I get started? 

Start by mapping the system you’re innovating within. While there are many frameworks for this, the Systems Thinking for Health Actions (STHA) is fairly comprehensive and approachable to beginners. The process begins by identifying system components and tracing feedback loops. This will provide an overview of the variables that your work will influence. 

Learn how to recognize and stratify leverage points. Donella Meadows’ framework illustrates how the deepest roots of a system shape its surface-level structures. Identifying higher leverage points such as value systems is integral to transformation work, but is much harder to affect compared to influencing lower leverage points such as information flows.

Finally, learn how to predict system changes over time through experimentation. If you get this far, you’re bound to be tempted to build simulation models to test your theories of change.

Parting thoughts

The Clinical Innovation Mindset Stack offers a framework for thinking differently about healthcare transformation. Here are the key takeaways to remember:

• Stay grounded in the lived experiences of patients and care teams. Continuously gather observations and insights from the frontlines.

• Take an iterative approach. Design thinking and lean methodologies promote cycles of rapid experimentation and learning.

• Foster strong partnerships with product teams. Align innovation efforts to core stakeholder needs.

• Immerse yourself in clinical environments. Draw inspiration from traditional care delivery, but challenge the status quo.

• Adopt systems thinking. Consider downstream effects and unintended consequences.

• While frameworks provide structure, innovation remains an art. It ultimately requires creativity, empathy and a willingness to rethink assumptions. 

Practicing the Stack in the madness of a startup

Even for folks working in clinical innovation roles, not every moment of work life is filled with the thrill of experiments, epiphanies, workshopping, and ideation. It’s almost always mixed in with the typical planning meetings, huddles, and operational projects. And for those in clinical roles, actually caring for patients. Speed and time (i.e. funding) is often highly valued in any startup, health tech included. You may not have the leadership support or the time to do things the ideal way. Drawing on your wisdom from using the Stack to pull the right tool, concept, or exercise is very much part of clinical innovation in a fast-paced startup. 

Beware that an alternative innovation pre-requisite may exist for mature organizations without a previous innovation role. While an innovation lead can be brought in to “innovate,” they may really be needed to fix all the duct tape that’s barely holding things together. Even the best innovation work won’t sustain unless there is a solid foundation to build on. Patience is key in this scenario.

Evaluating the Stack

As our field matures, leaders of clinical innovation teams will require tools to evaluate their colleagues across competencies of the Stack. While criteria to measure progress within each competency is yet to be determined, we propose scorecards to track individual and group performance. These insights can be used to refine curricular developments and identify emerging rockstars as well as those that will require extra coaching.

Teaching the Stack

At some point, an organization’s need to problem solve and innovate, will scale beyond the capacity of the current clinical innovation team. The organization can either continue to grow the team to keep pace, or create a learning and development strategy for empowering everyone in the organization to problem solve and innovate. This pivot may come later in the organization’s development (think Series D and beyond). Building agency and problem-solving skills across the organization is key to sustaining innovation.

Go forth and innovate! And let us know how it goes.

Paulius Mui, MD is the Director of Clinical Innovation at Next Level Medical. 

Ben Dolson is the Director of Clinical Performance and Innovation at Twin Health.

Including our bonus recap video of the XPC hackathon and the Primary Care 101 course:

Part 1: XPC Hackathon

In June, XPC entered a new frontier of engaging our primary care innovation community by trailblazing a couple of new initiatives. From hanging out in most healthtech Slacks and attending major healthcare conferences over the last year, it has become evident that the appetite for building with peers and learning along the way is unmatched by existing opportunities in primary care. It was frustrating seeing conferences as celebrations of work yet to be done with so much untapped and hungry talent in attendance. Naturally, the idea of a 48-hour XPC Hackathon was born and we set to manifest “collective intelligence” and capture value from open innovation.

The goal of our hackathon was to bring a group of individuals passionate about primary care who normally do not get to interact in the same space. They would build for one of the three tracks: Care Insights, Patient Empowerment, and Care Delivery. Plus $5,000 in prize money was a cherry 🍒 on the top. 

For some participants, our event was the catalyst to transition to healthcare from another industry (eg. fintech to healthtech). For others, it was the place to meet co-founders and start new ventures. We had a mix of seasoned founders as well budding entrepreneurs for whom this was their first ever hackathon. They were not alone.

We reached for the impossible and formed the most stellar advisory board that became fast friends. Our planning team represented the composition of hackathon dream teams consisting of engineering, clinical, product and strategy talent. 

We recruited a rockstar line up of mentors across many of the silos we set out to break down and match the ambition of hundreds of participants across more than half of the world’s time zones. We learned afterwards that a number of mentors became formal advisors to some of the newly formed teams because they were there from the beginning. 

We had a spectacular assembly of over 20 sponsors, big and small, including venture capital firms, open-source health tech startups, as well as EMR, data and insurance companies. They not only committed financial support and served as competition’s judges but also gave backstage access to their products and helped with mentoring visionary hackers and builders. 

The event culminated with the award ceremony on the 51st hour of the hackathon after a two-hour sprint of judging all submissions and doing live pitches for Spark of Potential: Most Promising Startup Award.   

Special mention of award winners:

• Rogue - Team 58 (AI Frontier: Best AI Application Award) - helping doctors evaluate performance of AI models and understanding their biases

• SimpleDoc - Team 33 (Patient Empowerment Award) - improving health literacy and reducing healthcare costs for patients

• Recco - Team 26 (Care Delivery Award) - facilitating medication reconciliation using natural language via chatbot

• NEXTpill - Team 54 (Care Insights Award) - synthesizing patient and payer data to improve clinical workflows

• Triage.ai - Team 28 (Spark of Potential: Most Promising Startup Award) - standardizing clinical triage

Even though the event was virtual, the energy throughout the weekend was palpable from all stakeholders: participants, mentors, sponsors, and hosts. Shortly after the event, we began receiving messages and started seeing LinkedIn posts going up about everyone’s experiences. 

Here are some of our favorite highlights:

• Turquoise Health wins best hackathon journalism award (their article is a must-read recap): “Regardless of the scope of the solution (as broad as decreased costs for patients, employers, and providers and as granular as discharge disposition planning), all the groups honed in on the fact that in order for healthcare to improve, we’re going to need to be smart about pulling data from a number of different resources.”

• Healthie, one of the most innovative EMR startups, celebrated with their community the sponsorship of their (and our) first Hackathon.

• Metriport interviewed several teams that incorporated their open-source APIs and data streams into product demos. 

• Morgan Mendis and his team showcased their mini-design sprint from the hackathon. Really cool to see under the “hood” of innovation.

• We loved Raji Reddy’s sincere reflection: “I’m famished by lack of food and sleep in last 48 hours, stress to perform was through the roof and now that our idea is submitted.. I’m trying to calm down, but it’s hard 😂 If you ask me if I will do this again.. oh yeah!!!”

• Being a first-time hacker did not stop Neha Katyal from going all in: “Although we were all first-time hackers, we experienced the full product lifecycle in one go - understanding pain points, ideating opportunities, recruiting talent, building and iterating an MVP, UX/UI, testing on real datasets, creating a market strategy, and pitching to investors and end users.”

• James Leonard experienced what we hoped to curate: “One of the coolest parts of the process was the opportunity to interact and pick the brains of all of the mentors.”

One of the unexpected and more fascinating outcomes of the hackathon was inspiring others to replicate this magical experience themselves. Several of our sponsors are launching their respective first ever virtual hackathons in the coming months (eg Health Universe Hackathon, AvoMD Hackathon). Stay tuned for another round of collective intelligence from the healthtech community.

Part 2: XPC Academy

Our second initiative in June came from our love for sharing what we know about primary care and our frustration of how much misunderstanding there exists about our specialty. Plus inbound of the same set of inquiries about pain points in care coordination of primary care patients or questions about opportunities in Direct Primary Care (see our landscape article) nudged us to package our knowledge into something that scales. 

We launched XPC Academy with our Primary Care 101 course over four 2-hour interactive sessions. Similar to the crowd drawn by the hackathon, our learners came from various backgrounds: policy, tech, insurance, sales, and clinical. The course covered a 10,000 foot view of primary care across fifteen sessions, spanning frameworks, payment structures, office operations, and workflows. 

In our true spirit of being tech-forward physicians, we employed the latest technology available, such as video editing with Descript and generating visuals with Midjourney.

A portion of the course was pre-recorded and played while learners interacted with us and each other on Slack in real time. At times, we could barely keep up with the pace of simultaneous conversations. Live discussions and polls took place during breaks in between recorded segments. The diversity of backgrounds in the audience generated many stimulating discussions.

This course served as the starting point for future 201 level courses. Each subsection could've been its own course. We are looking to work with partners to develop future courses. If you or your company has special knowledge on a certain topic and would like to do a deep dive with us in a 201 course, feel free to reach out. 

Hot off the press: XPC Fellowship

Created to empower founders reinventing primary care, XPC Fellowship launches today. The fellowship is a six month immersive virtual program through structured mentorship and self-directed learning. There is no financial or equity cost to the Fellows. We are different from an incubator or accelerator because we pour energy and resources into people, not ventures. We invest time, not money. We do this through curated connections to break down silos across engineering, product, design, and clinical domains. If you think you’re at the frontier of primary care, we would love to hear from you. Fellowship applications are on a rolling basis, and you can apply starting today.

Our first cohort of Fellows include Michael O’Brien, MD and Zac Schulman. Michael is a med-peds physician executive who is building a healthtech startup to improve care navigation for young people with complex chronic conditions, starting with sickle cell patients. Zac is a medical student at Case Western and winner of the Patient Empowerment Award at our hackathon. He is working on improving health literacy for patients by taking SimpleDoc to the next level. 

We are excited to collaborate with Sarah Gebauer, MD as our inaugural faculty and co-developer of the fellowship and its curriculum. Our inaugural partners include Metriport and Light-it who will serve as advisors to Fellows throughout the program.

XPC Team

To make this all possible, our team is growing with David Nichols, MD joining our core team as Partner and Celia Steigerwald as our Chief of Staff. 

Dave is also a family physician whom we met over a year ago at DPC Summit. He’s a systems thinker who ran a successful cardiac rehab clinic throughout medical school and residency. Celia has been an admin extraordinaire for a decade and has the chops to keep us all in line. 

XPC Virtual 48 hour Hackathon on AI + Healthcare kicks off in 2 weeks on June 9. We’ve brought together an incredible group of mentors, judges, and sponsors. $5K in prizes. Deadline to apply is June 1.

XPC Course: Many of the principles discussed in this post will be covered in our course launching at the end of June. Our target audience is non-clinicians and companies seeking to improve primary care knowledge for their workforce. 

Primary Care for America came together in 2021 as a collaboration among 11 of the top primary care organizations in the country. Last week, they held their third gathering of primary care leaders at Primary Care 23 in Washington DC to discuss all aspects of primary care. Thanks to Shawn Martin, EVP & CEO of the American Academy of Family Physicians, I had the opportunity to attend the event in person.  

While covering a large landscape, several themes stood out across all the talks:

• Fee-for-service (FFS) care does not make sense for primary care and all primary care should shift to a value-based model (VBC).

• Lots of friction exists in converting current practices from FFS to VBC.

• Primary care is central to a functional healthcare system. 

• Team-based care will supplant the single PCP to create a more comprehensive, holistic experience for the patient.

• Developing strong relationships with patients is core to primary care.

• Primary care continues to be underfunded despite numerous studies showing benefits as well as comparative studies with other countries.

• Degradation of primary care in payment and prestige has created a huge challenge in maintaining and growing the primary care workforce.

• Out of the 4 C’s of primary care, continuity and comprehensiveness received most of the emphasis and attention. 

• Telehealth, spurred by COVID, has been a mixed blessing and how to use the technology properly to the full extent of its potential has yet to be determined.

• Functional primary care improves health equity and reduction of primary care in a community increases health access discrepancy.

As a joint venture of organizations with a similar goal of improving and promoting primary care, hearing the recurrent themes across all the speakers came as no surprise. But even though the “foundational why” was consistent, the “how” varies widely. Individual talks were mostly panels organized by the different sectors and aspects of primary care.

Medicare: CMS innovation Center continues to produce and study various models to drive accountable care and shift resources to incentivize development in underserved areas. Some challenges they face are optimizing predictability with accuracy as well as data sharing between CMS and providers. While most CMS efforts have focused on larger systems, they recognize more collaboration needs to happen with community partners. New models take 18-36 months to develop and CMS plans to roll out 3-4 new payment models later this year. 

Operators who work heavily or primarily with Medicare continue to push for predictable and increased payments to primary care while understanding the budget neutrality constraint on CMS. Current outcome measures remain burdensome, while key benefits of primary care such as trust and access do not have metrics and are harder to measure. Working with specialists also is a bit of a friction point since they do well from fee for service. Developing a model where everyone wins requires a global aligning of incentives which has yet to happen. 

Medicaid: While Medicare is sometimes subject to the whims of political activity, Medicaid feels those effects at a 50x multiple due to political factors at each state. Overall, Medicaid patients continue to be the highest need and require the most care, but Medicaid continues to pay the least at a fraction of Medicare (which already is a fraction of commercial payers).

Employers: Employers exist on a spectrum of understanding health benefits. The ones engaging in value based arrangements are generally progressive employers with a decent understanding of healthcare benefits and the importance of a strong primary care foundation. Trusted PCPs can help drive employees towards certain medications or specialists better than any plan design can. Onsite and near site clinics improve engagement and positively affect outcomes. 

Primary Care Across the Lifespan: Environmental and social factors have disproportionate impact at the ends of lifespan (childhood and geriatric). Primary care has an increased role in screening and addressing social determinants of health in these populations. Even within a single age group, one type of care model may not solve problems for the entire age group. Building relationships helps understand patient needs and provide individual care. 

Primary Care in Different Communities: Working in specific settings requires understanding the culture of the community as well as the unique needs of the population. Using big data to map communities is easy, but more important is finding anchor communities to build partnerships and deepen local relationships. The type of community can determine strategy in building new clinics vs. supporting the existing primary care infrastructure. 

Workforce: Supply of primary doctors continues to decrease as more residents choose to subspecialize. Many primary care trained physicians also choose to go to other settings like urgent care or telehealth or move to non-clinical roles like admin or research. The demand for traditional PCPs keeps increasing with a rapidly aging population and more chronic conditions overall. Money is necessary but not sufficient to solve the supply issue. Current messaging about the woes of primary care is not helpful, the narrative needs to shift more optimistically to entice more students to choose primary care. 

Community Health Centers (CHC): Governance by community boards where at least 51% of members are patients make CHCs very responsive to community needs. Federal funding has been a major source to help these centers grow. Expansion has economic benefits as these centers hire many in the underserved community, decreasing poverty, which is the leading social driver of poor health outcomes. Some of these centers have also started residency programs for nurse practitioners and social workers and find a high retention rate of these providers after graduation. 

Health Tech: Most tech up to now has been built for episodic, fee for service medicine and focused on optimized billing rather than medical care. Data analytics and administrative task simplification are two high yield areas where tech can make a sizable difference in primary care. Uptake of new tech is often low because startups make widgets in a vacuum and then realize clinicians don’t want or need it when going to market. Primary care stakeholders need to create more awareness of their needs in the health tech community and health tech people should make more of an effort to understand day to day primary care challenges.

XPC Commentary 

Most VBC models require risk sharing by the PCPs, which in a sense simply moves the metrics from numbers to outcomes. For example, in a diabetic, instead of controlling A1c, the primary care team now focuses on decreasing hospitalizations and exacerbations of chronic disease caused by diabetes, both of which generally involve lowering A1c. PCPs regularly complain about metrics like the A1c because, even with the best medical management and counseling, they can’t control what patients do outside the office. This is where the team based discussion begins. 

The current primary care team (PCT) involves providers, clinical staff, coding staff, and billing staff. While vital in FFS, the last two don’t contribute to improved patient outcomes. Capitation enables a shift of resources from billing and coding personnel towards the formation of innovative teams that can enhance patient outcomes, including staff who engage the patient outside of the office visit. Many of the panelists showcased the flexibility of capitation-based team building by describing the unique PCTs they built which suited their community. But one ongoing challenge for capitated payments stems from its flexibility. Since value-based payments are prepaying for services, payors need to decide how much to pay out and how much risk sharing is involved. Exactly how much payment is required to develop an outcome improving PCT without creating redundant roles or excess profit will remain top of mind for all payers. 

Direct primary care (DPC) arrangements stand out as the only VBC model which doesn’t directly involve risk sharing. Most DPC employer arrangements involve a per-member-per month (PMPM) with the promise of savings for the employer without partaking in the savings. Retail DPC goes further and cares more about customer service than risk sharing since the individual patient pays for the service and market forces determine the capitated payment. 

On the topic of consumers and markets, absent from the wide-ranging discussion on primary care were companies from the growing direct to consumer (DTC) market. Comprehensive care and continuity took center stage, but growth of the DTC and urgent care markets suggest point of first contact needs more attention from establishment and larger primary care groups. Many people now consider urgent care centers as their primary care, and as DTC solutions grow, more people will have their first or only interaction with healthcare through a DTC company, which then has tremendous opportunity to develop more primary care services or even guide referral and health system navigation. Most users of DTC services and urgent care facilities tend to skew younger which stands in stark contrast to many discussions which focused on the Medicare population. 

VBC arrangements make sense for Medicare aged patients because their risk is more predictable (no unexpected new chronic diseases and less risky behavior) and a beneficiary will more likely stay with their Medicare product for the rest of their life compared to younger populations with employer-sponsored or commercial insurance. In the game of risk sharing, time horizons matter. The average person only stays with a commercial insurance plan for 2 years on average and 4 years with an employer. These short time horizons make payments for preventing chronic disease, which may not manifest for decades, hard to justify. Furthermore, chronic diseases and risk adjustment plays a big role in determining the amount of capitated payment. Finding the right capitation amount for younger, healthier populations, may prove a harder challenge and perhaps may not even be the best financial model of care for them. 

As organizations make the push towards VBC, they hope to bolster the dwindling trickle coming from the primary care pipeline. Awareness and payment are the top two challenges to improving it. Most students will rotate in FFS primary care offices or with hospital-based PCPs who don’t even know if they are in a VBC arrangement. Students also generally wouldn’t even know what VBC is if they saw it. Even if they saw a functional primary care practice, the underfunding and belittling of primary care still makes primary care specialties amongst the least competitive. VBC arrangements, if they hold up to their promise and potential, may help turn the tide as PCPs take more control over global patient care and receive commensurate compensation. Clinics which have their own residencies solve the awareness problem and could help develop a new generation of clinicians trained specifically in value-based care.

Today’s students are also computer-native and are training in a world of AI and rapidly advancing technology. The healthcare system is notoriously slow at adopting modern tools (as it continues to support the pager and fax industry). Part of the lack of tech adoption, or at least the appropriate adoption, was because of small practice consolidation into hospital systems where those who decided what to adopt were not the end users. While most “tech-enabled” practices are DTC ones, advancement of certain platforms has allowed smaller practices to start and grow by simplifying and decreasing the cost of practice management. Smaller practices can adopt new innovative solutions faster and fuel development of tools which actually improve patient care rather than just optimize billing. A feedforward process can begin as tech spurs growth of smaller practices and smaller practices fuel the need for more and better solutions. 

Primary care is hurtling down an unsustainable path of majority FFS where it delivers poor service and outcomes and students have low and waning interest in the field. Those gathered at PC23 all share the goal of trying to carve a new path and bring primary care to greener pastures. Primary care continues to be a topic of intense interest as everyone tries their own solutions and groups gather to push their philosophies, sometimes even trying to redefine primary care itself. In a sense everyone is solving for X, where X = Primary Care. 

Recordings from Primary Care 22 and the inaugural sessions are available, which means the recordings from this year’s sessions should be posted soon as well. 

The event kicked off with a wide-ranging breakfast panel which was not recorded or live streamed. Takeaways from the breakfast are available for paid subscribers. 

Read more

XPC Hackathon begins in less than a month! We’ve gathered mentors and judges from our favorite health tech startups and venture capital firms. Spots are limited so we encourage you to apply today! 

Primary Care 101 Course launching in late June. Sign up for our alpha cohort by May 31.

This past weekend we assembled a fantastic group of thinkers together in the mountains of Virginia for our Spring Retreat. Special thanks to Milliman Inc. for sponsoring the event and helping us select our retreat theme of Defining Value in Primary Care.

Dustin and Fritz from Milliman Inc. led us in a workshop where they taught us how to think like actuaries. We explored how to evaluate employer expenses, which levers direct primary care arrangements could pull on to reduce overall costs to employers, and how to value the per member per month fee. Phil Eskew, DO, JD, MBA joined us virtually and chimed in with his experience.

Jay Parkinson, MD, MPH joined us virtually from California. He shared his background and journey building multiple startups. We had a robust discussion over his views of primary care and the problems he’s currently tackling.

The legendary Eric Bricker, MD came straight from working the field on his farm in Texas to jump right into a deep dive on employer sponsored health care. We explored trends, predictions, motivations and got a masterclass in how employer plans work (Here's a 101-level overview).

Jonathan Bush joined us from his own retreat with his family. His VP of Business Operations & Finance Jason Bent also tuned in. We discussed Zus Health and the potential of the product for primary care. There was an interesting line of conversation about EMR data vs. claims data.

Paulius and I enjoy meeting interesting people and having thought provoking conversations with them. Our retreats allow us to bring together some of our favorite people to one place for deeper conversations and to build connections across our network. We plan to continue hosting invite-only retreats and are looking at late September for our next one. Feel free to reach out if you are interested in sponsoring. Sponsors help us take our retreats to the next level and we curate attendees and content based on sponsor interests.

We’ve compiled high yield retreat insights for our paid subscribers below.

RETREAT INSIGHTS

Primary Care

• Primary care is responsible for tens of millions of dollars in referrals. Affecting just a small percent could yield a couple million in savings. Proving these savings is incredibly difficult.

• Primary care currently has value until the point of referral. Improving care coordination where the primary care team can follow the patient to the point of referral and back to primary care is an important part of a reinvented primary care model.

Read more

FYI: June is going to be hot at XPC. Here’s why 1) Primary Care 101 course 2) XPC Hackathon

Our clinical perspective

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 often a large disconnect between clinicians and builders. Here are some of the companies whose work we’re following in this space.

The Arena of AI-Enabled Clinical Workflow

by Ben Lee from Quite a Few Claims

Quite A Few Claims

Some bold(?) claims and thoughts into white space opportunities in healthcare.

By Ben Lee

Yeah, it’s another AI post. First of all, a self-deprecative disclaimer:

After every claim or hypothesis made in this post, please assume that this image immediately follows it.

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

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 render eventually 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 AI/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 ~200 deployments on prem 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. Rohan D’Souza (former CPO at Olive) opines: “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?

My internal AI, working efficiently(?) for this post

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 :)

Diving into Ambient Documentation Financials 

by Blake Madden from Workweek

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.

Onboarding a new patient in a primary care office is a crucial but painful process. Currently it involves a combination of patient self-reported conditions on a clipboard and record retrieval by the primary care physician (PCP) office from wherever the patient was previously seen, assuming they can remember all the locations. The clipboard process is repetitive and often redundant, and the records retrieval process can be arduous.

Currently, the records process goes through these steps:

1. Patient fills out “authorization to release records” form or forms.

2. PCP office faxes the forms to the respective specialist or previous PCP offices.

3. PCP offices tracks whether or not the outside offices have responded to the request and follows up on those who don’t send records.

4. When records do come, they arrive as a packet of mostly useless and repetitive information. Most will arrive electronically by e-fax, but some come by physical paper fax or even regular post office mail.

5. Received records are scanned into patient chart.

6. PCP skims through pages of useless noise to try and find meaningful information about their new patient.

Fortunately, the world of interoperability has advanced significantly and has the potential to simplify this process.

Most patient data currently sits in three large Health Information Networks - Carequality (Epic focused), CommonWell Health Alliance, and eHealth Exchange (mostly government affiliated data, e.g., the VA). While some EMRs, such as Epic have access to these networks, allowing users to see data from outside their office immediately, most EMRs do not. Connecting to the Networks requires technical skill, understanding of regulation, and time, which most clinics, startup companies, and smaller EMR companies do not have to undergo the endeavor themselves. Catalyzed by the 21st Century Cures Act, companies have emerged to simplify the process by building connections to the HINs and then linking those via API to end users who want access to the networks.

Zus Health has focused on EMR integration, partnering with Healthie, Canvas, and Elation. Particle Health focuses more on data analytics for end user companies such as Oak Street and One Medical (initially as part of Iora, and now part of Amazon). Health Gorilla has completed the regulatory process to become one of the first Qualified Health Information Networks (QHIN). For a great intro to interoperability, check out Particle Health’s white paper.

Making care more interoperable is a game changer for PCPs. Better information at point of care allows for delivery of higher quality of care. Moving patient data across a multitude of healthcare silos is a huge challenge, and we’re rooting for everyone tackling it. 

Paulius and I attended ViVE this week in Nashville. We sat down at the Music City Center to review our thoughts and takeaways before heading home. As luck would have it, our friend Blake Madden from Hospitalogy was passing by and decided to join us for our conference review. He does great work for anyone not subscribed to his posts. Check out his full review of ViVE 2023.

Of course, one major perk of attending ViVE was networking and relationship building. Conferences bring awesome people together and provide a great excuse to meet up. We enjoyed seeing old friends, friends made at HLTH, and meeting new people who we are excited to continue seeing and sharing ideas with.

BIG ANNOUNCEMENT: XPC ACADEMY

We are launching XPC Academy by opening up an alpha version of our primary care 101 course launching June 2023. Made by doctors. Packed with insider knowledge. We designed this course for anyone who wants to have a foundational understanding of primary care. Perfect for sales teams, engineers, analysts, founders, investors, or other healthcare nerds. Sign up for the alpha cohort ends May 31. Questions? Hit us up on LinkedIn.

• Paulius Mui, MD

• Kenneth Qiu, MD

P.S. Scholarships available if your enthusiasm for primary care exceeds your funds.

When people imagine today’s doctor, most still picture a person in a white coat or scrubs with a stethoscope around their neck, writing prescriptions and encouraging better life choices. Healthcare has a way of staying decades behind in technology, but health tech has grown and continues to grow rapidly, pushing the whole industry forward. Doctors may soon replace their 1860’s era listening tube with a pocket-sized ultrasound. Instead of prescribing only pills, doctors could prescribe digital therapeutics. Technology even allows the patient to bypass the traditional doctor’s office all together in favor of direct-to-consumer telehealth platforms. All the developments in the health tech space can become overwhelming, and even with all the advancements, few clinicians know about most of them. That’s where Digital.Health comes in.

Paulius and I have been helping Dr. Daniel Kraft with his digital.health platform since we met about a year ago. The site collects all of health tech in one place, allowing for searching, comparison, and evaluation of thousands of products and services. Primary care has the widest scope of practice as the point of first contact and therefore can leverage more technology than most other specialties, which allows us to think globally about all the products and services listed on the platform. The platform has grown and changed a good amount in the last year. Recently, Dr. Kraft named Parisa Vatanka, PharmD as CEO of the project. We are excited to see how it continues to develop into a tool where clinicians can discover new diagnostic tools and perhaps even start prescribing tech therapies for their patients.

In addition to Digital.Health, Dr. Kraft has plenty of other projects going on at any given moment. He has an upcoming NextMed Health conference in San Diego from March 13-16 where he will be bringing innovators from across the industry to help catalyze a human-centric, tech-enabled health age. Join the action at $750 off the registration fee.

Understanding Direct Primary Care (DPC) begins with having consensus on what it means, not just for scholarly or marketing purposes, but also for legal implications as outlined by Dr. Phil Eskew: “1) charge a periodic fee. 2) not bill any third parties on a fee for service basis, and. 3) any per visit charge must be less than the monthly equivalent of the period fee.”

More terms to be aware of that may be misunderstood:

• Pure DPC = membership fees only, no insurance.

• Fee-for-service (FFS) = no membership fees, only bill insurance. If a patient doesn’t have insurance, then they have to pay out-of-pocket for medical services each time.

• Hybrid = one portion of patients pays only monthly membership fees while a different portion of patients are billed only through insurance (e.g. Rockwell Physicians). This can also be a transient state when practices are transitioning from FFS to pure DPC.

• Concierge = double-dipping in membership fees and insurance billing (e.g. MDVIP). Some DPC practices market themselves as concierge practices despite not taking any insurance to associate with high-touch experience.

DPC features

Applies to traditional brick-and-mortar practices.

• Smaller panel of patients, as low as 200 to 1200 per clinician (traditional panel size is about 3000 patients)

• Longer visit times 30-60 minutes or more

• Routinely use telehealth and text-based care

• Health system advocacy and navigation via discounted lab and imaging prices. Helping find cash pay specialists or using in house e-consult services.

Mappers

The DPC Frontier mapper has largely been considered the go to mapper. However, the mapper requires self reporting and exists more as a platform to display practices rather than an actively managed product, meaning there are no accuracy checks or quality control. The Alliance mapper only includes members of the DPC Alliance. Since health shares partner with DPC practices, the health shares have developed their own mappers for their members (e.g. Sedera, Zion HealthShare).

DPC origin story

Dr. Garrison Bliss is largely credited with starting the first DPC practice in 1997 in

Seattle, Washington. During the first few years, Dr. Bliss fought with the then Insurance Commissioner of Washington state over whether or not his new model of practice constituted an insurance product. He eventually won that battle and went on to found Qliance, a multi-clinic DPC practice which famously piloted seeing Washington Medicaid patients on a per member per month basis. Jeff Bezos and Michael Dell were notable investors. Qliance ended in fiery failure for a number of complicated reasons. Dr. Bliss also founded the DPC Coalition, a lobbying organization, to help advance DPC legislation at the state and federal level and coined the term “direct primary care” for policy discussions. Wins for the organization include DPC language in the Affordable Care Act as well as favorable DPC legislation in a majority of states.  

Other physicians independently created insurance-free practices in the early 2000s, and there was some convergent evolution with many physicians making their own DPC practices, like Dr. Bliss did, without previously hearing about the model. These physicians met at concierge conferences back in the day and eventually became some of the original members of the DPC Coalition. In 2012, FMEC spearheaded the first DPC summit and in 2018, the DPC Alliance formed. 

Since independent physicians created the DPC model, most physicians feel the independent, small practice embodies the true spirit of DPC. As the DPC model grows and changes, these physicians believe new entrants, especially corporate and Virtual Primary Care, do not truly reflect the core philosophy of DPC and prefer to refer to them as DINOs (DPC In Name Only). Interestingly enough though, many of the corporate entities who do subscription model primary care do not generally call themselves DPC.

Subscribe now

Who is DPC?

We categorize the DPC landscape based on customer bases. In the retail model (B2C), memberships are sold directly to patients and can be led by individuals or corporations. In the employer model (B2B), memberships are sold to employers who offer DPC as a benefit to their employees. Both of these models can be led either by individual-owned practice networks or large corporations. DPC practices range in size from tiny to massive.

• Micro-practices (e.g. Defiant Direct Primary Care, EuDoc)

• Multi-clinician groups (e.g. Antioch Med, Direct Access)

• DPC physician networks (e.g. Nextera Healthcare, Primary Health Partners)

• Corporate (e.g. Everside Health, Crossover Health).

Individual-led retail practices

Practices founded by independent physicians, usually family medicine trained, range from micro-practices to multi-clinician practices which generally remain in their geographic location. A few of these are virtual with house calls, but the vast majority are traditional brick-and-mortar practices. There is a saying “if you’ve seen one DPC, you’ve seen one DPC.” Every clinician brings a unique take on their practice. We illustrate a variety of niches within DPC practices.

• Lifestyle and wellness - Dr. Lifestyle Clinic

• Bilingual - Dandelion Health

• LGBTQIA+ - Seven Hills Family Medicine

• Pediatrics and Lactation - Bloom Pediatrics & Lactation

• Full Spectrum with Obstetrics - Antioch Med

• Emergency Medicine Led - Thrive Direct Care

• Hospitalist Led - Fischer Clinic

• High-End - Priority Physicians

• House Calls - My Happy Doctor

Corporation-led retail practices

Tech-forward approach with broader geographic reach than individual-led retail practices. Most founders are not physicians. Generally limited in scope and primarily virtual (e.g. not all offer physical exams or office-based procedures that are included in traditional brick-and-mortar DPC memberships). For example, K Health and Curai are purely text-based care. Some are also working with health plans (e.g. Galileo and United Healthcare). From a patient’s point of view, this arrangement shifts the perceived identity of primary care physician from an individual to a corporation “your PCP with the United Healthcare Virtual First plan will be Galileo Health, rather than an individual provider.” Needs of traditionally underserved communities have attracted new entrants like Zócalo Health, which is focusing on the Latino community. Lemonaid Health expanded on-demand single-problem offerings to include virtual primary care membership through a partnership with SteadyMD. Forward stands out from the batch because it has physical locations for in-person care in addition to virtual and text-based care on top of perks such as in-home phlebotomy and prescription delivery. Crossover Health does have a consumer-facing DPC option, but we’ll group them with employer practices which aligns with their dominant business model.

Individual-led employer practices

Some independent DPC practices have found success working with businesses. They generally start as single or multiple physician practices which then grow to larger networks as they obtain larger contracts. While many of these practices have multiple locations and a handful of physicians, their networks involve contracts with other independent DPC practices to handle higher numbers of employees across a larger geographic location. Self-funded and level funded employers have benefitted from DPC arrangements. These practices work with benefit advisors and/or third-party administrators (TPAs) to help land more business accounts. 

Some examples include:

• Nextera Health

• Primary Health Partners

• Persona Healthcare Direct

• Frontier Direct Care

Corporate-led employer practices

On-site/near site companies have existed for a while and never historically were considered DPC. Several newer companies made specifically for employers use a per member/employee per month model which technically makes them fit the legal definition of DPC.

• The growth of Crossover Health (XO) was catalyzed by Apple and Facebook onsite clinic contracts. Acquiring Sherpaa in 2019 further expanded their virtual care reach. The interdisciplinary approach to care gives XO patients access to latest tech-enabled primary care coupled with behavioral health, physical therapy, health coaching, care navigation and more. Aetna Advanced Primary Health, a zero-dollar employee benefit, has emerged from a recent partnership with Aetna.

• Everside Health has over 300 clinics in more than 30 states with focus on serving employers and unions. The company started out as Paladina Health, which was part of DaVita, and grew by subsequent acquisitions of Activate Healthcare, Healthstat and R-Health.

• CirrusMD serves “deskless workers” in customer service and blue-collar industries through text-first virtual primary care with access to specialists.

• Eden Health has a presence in-person and virtually with experts in insurance benefit navigation alongside their multidisciplinary clinical teams. 

Hint Health recently launched Hint Connect which seeks to obtain larger employers by contracting across individual-led DPC practices to create a vast network under one umbrella. Their model makes them a bit of an anomaly which doesn’t fit neatly into any of the categories. 

Special mentions

• At the academic center level, Hopkins is trialing a DPC model for their employees at their primary care clinics. University of Houston announced a DPC clinic a year ago which has closed due to unclear and unpublicized reasons. 

• One Medical's core business model by volume is concierge. They charge a very low membership fee and bill insurance with every visit. The DPC parts of this company’s portfolio include Iora, which started as DPC before pivoting to Medicare Advantage (MA), and several employer based DPC practices, such as the individual employer DPC group PeakMed. Interestingly, One Medical makes 17x more revenue per member with the MA business line.

• A polarizing topic for physicians that cannot be ignored is the increase of nurse practitioners (e.g. Nice Healthcare) and physician assistants (e.g. Smartypants Medicine) in independent primary care practice.  

How does DPC interact with health insurance?

DPC physicians believe primary care, as a routine, inexpensive service, does not need and should not need insurance coverage. Most will advocate for a DPC membership to be combined with a high deductible plan or an alternative product like a health share (e.g. Zion HealthShare) or CrowdHealth. Some companies, however, feel DPC can live in harmony with insurance, and can look like simple capitation plans which historically have had different incentives. 

• Taro Health is a startup insurance company on the ACA marketplace which has DPC as its foundation. They pay negotiated membership fees for in-network PCPs which are all DPC practices. They also strive to significantly reduce additional administrative burden compared to the incumbent insurance companies. 

• Flume Health has partnered with Frontier Direct Care, allowing Frontier to build out their own personalized health plan.

• Decent operates as both a professional employer organization (PEO) and a health insurance company, providing a full suite of services for employers. They also have DPC as a foundation for their primary care, and recently added One Medical locations to their offering.

• North Carolina-based Healthgram pairs in-house expertise as a TPA with their DPC physicians to cut costs for employers. 

Major hurdles

• Health Savings Account (HSA) - Despite multiple state laws and federal letters of clarification, the IRS still believes DPC practices are health plans. The HSA issue remains muddy with one faction claiming HSAs are allowable and encouraging their use with and for DPC payments, while others, fearing audit, caution against pairing the two. Ultimately, this is an extremely esoteric discussion at a high level and left mostly for lawyers and accountants to decide at the individual level. 

• Medicare - Medicare does not allow opted-in physicians to charge cash prices to beneficiaries. Theoretically, none of the retail corporate DPC companies can engage the Medicare population because of this rule. Most mature DPC practices have physicians who have opted out of Medicare. A much smaller number of practices are hybrid with only billing Medicare alongside their DPC practice. Since Medicare recipients are often the best candidates for DPC, removing the requirement to bill Medicare would help younger DPC practices start seeing Medicare patients sooner. Some companies like Pearl Health are trying to make Medicare arrangements more DPC-like through direct contracting. In the employer space, there is a gray zone argument where since the employer pays the membership, Medicare beneficiaries can see non-opted out physicians as a benefit rather than a traditional doctor visit, bypassing the billing requirement. 

DPC tech stack favorites

• EMR - AtlasMD, Elation, Cerbo, AeroDPC (recently acquired by Hint Health)

• Practice management - Hint Health (billing), Spruce (patient communication)

• eConsults - Thea Health, RubiconMD

• Clinical tools - ButterflyiQ (handheld ultrasound), SkinIO (AI skin screening), Smartheart (EKG)

… many more out there.

DPC communities

Physicians in the DPC community come to these virtual watering holes (Facebook groups) for support and guidance. These are intentionally very hard to find and join. Almost all are invite-only.

• DPC Docs - Core DPC group.

• Pediatricians who do DPC

• DPC Women

• Direct Primary Care Alliance Group

• Regional DPC groups

• Direct Primary Care - Public group. Questionable content.

Organizations

• DPC Alliance

• DPC Coalition

• Docs 4 Patient Care

Notable reports

• Direct Primary Care: Evaluating a New Model of Delivery and Financing by Society of Actuaries (2020)

• Hint Health Trends Report (2022)

Media

• Blogs -  DPC News, DPC Frontier

• Podcasts - My DPC Story (interview with Dr. Qiu), plenty of single episodes about DPC on various podcasts

Popular Books by and for physicians

• The Official Guide to Starting Your Own Direct Primary Care Practice by Doug Farrago, MD

• Slowing the Churn in Direct Primary Care by Doug Farrago, MD

• Magic, Pixie Dust, and Miracles by Shane Purcell, MD

• Medical Answers Now!: How Direct Primary Care Guarantees Fast Access to Your Doctor by Troy Burns, MD

• Startup DPC: How To Start And Grow Your Direct Primary Care Practice by Paul Thomas, MD

• Sparks Start Fires: A Guide for Dreamers Who Are Also Doctors by Julie Gunther, MD

Tools for starting DPC practices

• The DPC University by DPC Alliance

• The AAFP DPC Toolkit

Conferences

• DPC Summit - main gathering for physicians interested in pursuing DPC or are currently practicing or growing DPC

• Hint Summit - health tech with a DPC focus

• Nuts and Bolts - smaller alternative to DPC Summit for clinicians

• You Powered Benefits Symposium and Health Rosetta Summit - benefit advisor conferences with strong DPC presence

• Free Market Medical Association Annual Conference - cash pay (i.e., non-insurance) based medical services.

Further Recommended Reading

• Best Practices from Virginia Quarterly Review by Emily Silverman, MD

• DPC Generations from DPC News by Kenneth Qiu, MD

• Side Effects May Include... from DPC News by Kenneth Qiu, MD

Special thanks to Celia for helping put together the visuals.

Disclaimer: Dr. Kenneth Qiu founded and runs EuDoc Direct Primary Care; serves on DPCA Advisory Board; advises Taro Health; and writes for DPC News.

Very few people understand our healthcare system as well as Eric Bricker, MD. Fewer still can teach and explain the various moving parts as well as he can. Dr. Bricker trained as an internal medicine physician, attending medical school at University of Illinois College of Medicine, followed by residency at Johns Hopkins. In addition to his top-rate medical training, he also has a fascinating and unusual professional background which he narrates in his Healthcare Money Campfire Stories.

Dr. Bricker currently works as Medical Director of SimplePay Health and continues to publish daily videos through his popular blog and YouTube channel, AHealthCareZ.

Below are some of his videos about primary care:

• Direct Primary Care Overview

• Primary Care Innovation at Scale

• Virtual Primary Care: It’s Time

• Medicare Value Based Payments Explained

• Primary Care: US vs Other Countries

• Value Based Care: Business Case Against It

• Patient Must Come First in Value Based Care

• Value Based Care Happened 40 Years Ago

  Subscribe now

XPC Interview Topics with Dr. Bricker:

• Value Based Care and Claims-Free Care [01:30-06:54]

• Insurance Based VBC vs Direct Pay [06:55-09:20]

• Why VBC has not worked previously. [09:21-11:30]

• Virtual First Primary Care [11:32-16:03]

• Virtual Only Primary Care [16:04-19:57]

• Advice to Employers [19:58-24:48]

• Role of Primary Care Physicians in Employer Benefits [24:49-29:28]

• Changes to Medical Education [29:29-31:41]

• Thoughts on the Future of Primary Care [31:42-35:11]

• How Can Innovators Create New Opportunities [35:12-37:10]

Subscribe now

XPC Virtual Rounds

A gathering of primary care minds at XPC Virtual Rounds sparked some fascinating conversation about many topics. Here are some highlights:

What does primary care mean in the changing health care landscape?

• Relationships at the core

What can we unlearn about primary care?

• The function of referrals

• Undoing fragmentation of care

• Incentivizing preventive behaviors differently

What would you do with $1m to improve primary care?

• Educate patients to be better consumers of health

• Bundling DPC physicians with insurance

• Enabling patients and care givers to practice at the “top of their license”

Which policy changes have a meaningful impact on primary care?

• Health savings accounts in DPC

• Health data and privacy

• Health system ownership of PCPs

XPC Scholarship

Innovation requires thinking about the future. For us, the future includes those in the medical education pipeline. We announced our scholarship at retreat and presented it to its first recipient, Zac Schulman, a first-year medical student at Case Western Reserve University School of Medicine. Our goal with the scholarship is to empower medical students and residents who have a desire to learn about the changing world of primary care and challenge the limitations of our current system.

XPC Virtual Social Hours

We also had a few virtual sessions which weren’t recorded to allow for a more relaxed conversation.

Phil Eskew, DO, JD, MBA, one of the smartest people we know and all-around awesome person, flexed all his degrees by guiding us through stimulating discussion on medical, legal, and business topics.

The notorious Nikhil Krishnan dropped by with some memes, wisdom, and wisdom through memes. He’s been doing incredible work over at Out of Pocket and both Paulius and I are big fans of his content.

Looking Forward

As our network has grown, we’ve found more and more people are looking for a space to network and discuss all aspects of primary care. Therefore, we plan to open the door even wider for Spring Retreat by having more in person attendees as well as virtual discussion. Tentative plan: May 5-7 in Virginia. Feel free to reach out if you have any interest in sponsoring or joining. Stay tuned for more info!