It’s Not About Video. It’s About Always-on Triage.

Vineeta Agarwala

It’s Time to Heal is a special package about engineering the future of bio and healthcare. See more at:

The video visit has been one of the gifts of the otherwise merciless pandemic. As a physician, I’ve loved seeing patients access care so conveniently: Elderly patients who don’t have to waste two hours commuting and finding parking; brand new parents asking questions of their pediatricians from their living rooms; patients with mobility issues receiving pain medication without having to leave their homes. 

Ultimately, though, the focus on the video visit might be something of a red herring. We know that telemedicine will not replace most healthcare services. We’ve already seen a swing back towards in-person ambulatory care in many specialties—one gynecologist told me her mix went from 0% telemedicine visits pre-COVID, to >80% video in April, now back down to <10%, used only in a minority of cases where patients prefer video. Yes, providers will continue to offer both options: A psychiatrist told me his department is creating separate ‘video only’ clinic blocks and ‘in-person only’ clinic blocks for each provider; similar patterns are emerging from primary care groups. 

This mix begins to finally hint at what the real gift of digital medicine is. It’s not the format in which we deliver healthcare to patients, it’s much bigger: a massive shift in how we triage our patients to the right kinds of care. Continuous, always-on triage is coming to all parts of our health system.

It’s not the format in which we deliver healthcare to patients, it’s a massive shift in how we triage our patients to the right kinds of care.

What do we mean by triage? The word has its origins in the French verb trier (to sort), and is clinically defined as the “prioritization of patient care based on illness, severity, prognosis, and resource availability”[1]. When we think of clinical triage, images of a busy emergency department (ED) or a tent in a disaster zone come to mind. But most disease is chronic rather than acute, and most patients actually need to be triaged all the time and everywhere, not just during the short episode when they are in the hospital, or at a clinic visit (where they spend only a tiny minority of their time). Triage is essential to making sure that the right resources—always constrained—are used at the right time, in the right way, to the patient’s maximum benefit and to the healthcare system’s highest efficiency.

The COVID pandemic has forced clinics to contemplate, for the first time, the simple question of whether a patient needs an in-person vs video visit. Asking this question is itself a new layer of “always-on” digital clinical triage—in some ways even more valuable than the medium of the video visit itself. Triaging between a video vs in-person visit considers a lot of information about a patient’s condition: “He got new hearing aids, so a video visit checking they’re working would be great.” “Her son moved away recently, and she’s alone at home. I’d like to have this difficult conversation in-person if possible.” “His lesion is in a hard-to-see location, so let’s have him come in.” “She requested a video visit because travel to our clinic is really expensive for her.” It is these conversations—not the video visits themselves—that are bringing us a little bit closer to the ultimate goal of personalized triage that is always on, covering 360 degrees of the patient, 365 days of the year. These conversations are factoring in not only clinical acuity, but also what I call the ‘3Cs’: patient choice, context, and community. They are enabling us to finally see patients as whole people, with whole lives that are changing constantly in ways that should inform how we provide care.

We need a truly intelligent, “always-on” healthcare system which anticipates when patients will deteriorate, compares them to other patients who look like them, and monitors their health on a continuous basis.

A lot of digital health innovation over the past year has been framed around the narrow concept that reimbursed telemedicine episodes and asynchronous communication create new business models—but in fact these are just table stakes. We need a truly intelligent, “always-on” healthcare system which anticipates when patients will deteriorate, compares them to other patients who look like them, and monitors their health on a continuous basis. This is the promise of “always-on triage”—a much more ambitious, and more fundamentally transformative lens. 

Because those table stakes are finally in place, always-on triage is finally being built. These are some of the specific ways in which always-on triage can, and will, be materialized:

  • Proactive versus reactive triage. The vast majority of healthcare triage today (outside of primary care screenings) is reactive to patient complaints. While this makes sense for many acute care scenarios, for most patients with common conditions or care plans, we already know a lot from the body of reactive data that we have been generating historically. Technology is now being used to learn continuously from data, and push proactive care messages to patients. Think, for example, of patients discharged after a knee replacement; a woman sent home after giving birth; a patient started on a new medication with well-known side effects. In these scenarios, clinicians should be the ones reaching out, providing “push-based”, continuous care management to check for common complications, rather than the traditional pull-based, patient-initiated care (once something’s gone wrong, if they can even appropriately identify that).
  • Integrating remote data collection into triage. Let’s go back to that brand new mom. In some situations, we know that it would be great to monitor that mom’s vitals for the first few days after childbirth. We have no shortage of ways to do this with a proliferation of companies developing novel, next-generation remote patient monitoring (RPM) hardware—the problem is that seamless integration with existing clinical workflows for patient triage has lagged. More continuous streams of vital sign data are likely to enable more timely interventions, but only if we have the ability to analyze these data and alert providers in a sustainable, scalable way. You can’t just introduce a new dashboard for every tool. Instead, we need smart rules and systems that help providers seamlessly know how to review additional data with the limited time they have, which thresholds are actionable, and what they can do for patients in need. And interoperability initiatives are making it easier to import streams of data into the EHR, paving the way for more innovation in this area.
  • Triage between sites of care. For many types of care, we’re missing a key layer of infrastructure to triage between inpatient vs ambulatory care center vs home-based care. I’ve personally had to keep patients hospitalized, for example, because there was no other way to get them an IV infusion. Technology can help identify patients who are candidates for alternative sites of care, and also connect them with the support services they might need to receive that care: home-based physical therapy, or an oncology infusion at home, or a COVID antibody infusion at an urgent care clinic—antibodies that are just sitting on shelves right now, partially because there’s no tech infrastructure like this. The CMS Hospital At Home reimbursement framework is likely to further incentivize a lot of innovation in this area. 
  • Always-on access to specialist triage. A cornerstone of clinical triage is the ability to steer patients towards (or away from) relevant specialists. Today, this steerage happens via a “consult” in the inpatient setting, or a “referral” in the outpatient setting. When you’re hospitalized, inpatient consults happen fast, and in close collaboration with your primary team (like the cardiologist stopping by your hospital bed to review an unusual rhythm strip). But the downside is you get a consult from whoever is there in that clinical setting, at that time, and you’re unlikely to see that specialist again for future care and follow-up. In the outpatient setting, on the other hand, referrals from your PCP can take months to actually materialize, and the degree of connectivity with the primary care physician can vary widely. It is an artifact of how our payment systems were designed that these two modes of inpatient vs outpatient specialist triage are so distinct and separate from one another. Technology can support the development of a dynamic, reimbursed marketplace in which hyper-specialists anywhere (or specialists locally, who can also provide follow-up) could weigh in for both inpatient and outpatient needs.
  • Intelligent triage beyond medical care. To understand 360 degrees of the patient, we need to know who they live with, how they get around, what they eat, and how to best support all of that (e.g., their social determinants of health, or SDoH). I had a patient who missed multiple telemedicine appointments, until she was able to finally log on from the parking lot of a local elementary school with free WiFi. SDoH can be real barriers to healthcare, and can also result in futile healthcare spend. Value-based reimbursement regimes (such as Medicare Advantage plans) are already figuring out how to leverage tech to access more patient support outside clinic walls—from rideshare services to government food stamps to patient assistance programs that help patients pay for drugs. Technology can help identify the right patients for these services; create SLAs between these fragmented programs and the healthcare system; and share back info about adherence and outcomes—all of which will increase the return on medical expenditures and keep patients healthier.
  • Triage across multiple providers. It’s an unfortunate artifact of the complexity of the healthcare system that any one physician typically only sees their own practice’s history of that patient, or maybe one or two other practices, at best. And yet patients see dozens of doctors in all kinds of locations. Your PCP, for example, won’t get notified if you get prescribed a new medication from your psychiatrist—obviously crucial information. The new CMS interoperability rule mandates that providers share data with one another, with patients, and with health plans. This has already provided a tailwind for provider adoption of ADT (admit/discharge/transfer) event notifications, but this is only the beginning. The next transformative shift will come when the ecosystem of EHRs and data sharing APIs enable real-time, workflow-integrated data sharing as part of routine care.
  • Reimbursement regimes that reward smart triage. All of the above technologies today typically create extra burden (and often are new budget line items) for providers and plans. This isn’t sustainable; it can’t be the provider that pays for all of this. We need reimbursement regimes that allow providers and tech vendors to participate in shared cost savings. Bundled, risk-adjusted annual reimbursement for the care of a diabetic patient may be the future, but it can only be real if you get credit for doing smart triage all year long. Tech startups are going to lead the way in designing and executing on novel, competitive contracts with payors (including CMS).

These innovations in healthcare are all coming. While they will undoubtedly be buoyed by the adoption of telemedicine, focusing on the video visit would be missing the forest for the trees. It’s not about the medium of video — it’s about building muscle across our entire healthcare system to provide always-on triage.


  1. Sharon E. Mace MD, Thom A. Mayer MD. Pediatric Emergency Medicine. 2008.

Want more a16z Bio + Health?

Sign up for our bio + health newsletter to get the latest take from us on the future of biology, technology, and care delivery.

Thanks for signing up for the a16z Bio + Health newsletter.

Check your inbox for a welcome note.

MANAGE MY SUBSCRIPTIONS By clicking the Subscribe button, you agree to the Privacy Policy.