The Five-Minute Appointment: How Healthcare Turned Care Into Throughput

In 2006, Dr. Itzhak Brook — a pediatric infectious disease specialist at Georgetown University School of Medicine, a man who had spent his career diagnosing infections in other people’s children — went to a Maryland hospital with pain in his throat. He was seen by experienced head and neck surgeons. When he was examined, physicians told him it was acid reflux. They changed his medication and sent him home.

But this happened more than once, and it kept happening until a resident physician, using a basic endoscopic procedure the senior doctors had never bothered to try, found a peach pit-sized tumor that had been growing undetected for seven months. By then, the cancer had progressed far enough that Brook’s only remaining option was surgery to remove his voice box. He now speaks in a whisper and he considers himself lucky to be alive.[1]

Dr. Brook is a physician. He knew what questions to ask. He had institutional access most patients will never have. He had the vocabulary, the relationships, the credentials. And the system still failed him — not because the doctors were malicious or incompetent, but because the system they were operating inside was not designed to look closer. It was designed to move faster.

And that difference is what matters more than almost anything else in this conversation.

The System Is Not Overwhelmed. It’s Optimized.

There is a durable narrative in American healthcare that doctors are simply too busy. That the system is strained. That with more resources, more staff, more funding — things would improve. It’s a sympathetic framing. It’s also largely a deflection.

The five-minute appointment — or its slightly more generous cousin, the fifteen-minute appointment — is not a symptom of a system under duress. It is the intended output of a system designed around a specific set of incentives. When you understand what the system is actually optimizing for, the appointment length makes perfect sense. It is not a bug. It is a feature.

The central mechanism is the Relative Value Unit — RVU — a metric introduced by CMS in 1992 under the Resource-Based Relative Value Scale (RBRVS) as a standardized way to reimburse physician services under Medicare.[2] The concept was reasonable on its face: create a consistent method for comparing the work involved in different procedures and pay accordingly. What it became in practice is something else entirely.

Today, work RVUs (wRVUs) are the dominant productivity metric in American medicine. Most physicians employed by hospital systems or large group practices have compensation packages tied directly to wRVU generation.[3] The implication of that structure is not subtle: the more billable encounters you generate, the more you earn. The system has literally quantified the value of your time, and it has decided that the most productive thing you can do with it is to see the next patient.

This is by no means a critique of the physicians living inside that system. I suspect many of them are acutely aware of the trap. But it is a critique of its architecture specifically, of the payers, hospital systems, and CMS payment policy that built the elaborate trap and continue to maintain it.

The Appointment Is Shorter Than the Problem

Let’s be precise about what fifteen minutes actually contains. Research tracking primary care visits found a median of six topics addressed per appointment, with the primary topic receiving about five minutes of discussion and each remaining topic receiving roughly one minute.[5] One minute per topic. In a visit covering six competing concerns.

More than half of all clinicians in one study reported needing more time than allotted for follow-up appointments, and two-thirds felt that deficit for new patient visits, where the gap was closer to eleven minutes.[6] These are not physicians who don’t care. These are physicians operating inside a scheduling architecture that was not built around the complexity of the patients in front of them. The schedule was built around the billing model.

And the billing model was not built around what patients need. It was built around what payers are willing to reimburse — which is itself built around what CMS has decided a physician’s time is worth. A calculation that was never designed to account for a patient with four comorbidities, a layered symptom history, and a presentation that doesn’t map cleanly to any single ICD-10 code.

The Invisible Population: Complexity as a Design Flaw

If you have one acute problem — a sinus infection, a sprained ankle, a straightforward blood pressure check — the fifteen-minute model works reasonably well. The system was designed with you in mind. You fit.

But a substantial and growing portion of the patient population does not present that way. Patients with autoimmune disorders, chronic pain, multi-system conditions, neurological complexity, or co-occurring mental health diagnoses are not arriving with single-variable problems. They’re arriving with layered histories, contradictory symptoms, evolving presentations, and often — after years of being shuffled between providers — a guarded, depleted energy that makes it harder to communicate what’s actually happening in the few minutes they’ve been given.

The system does not classify these patients as complex. Rather it classifies them as time-consuming, and that reveals everything about what the system is actually measuring.

“The diagnostic journey is really not a single decision at one point of time. It’s an odyssey that unravels overnight, in some cases, days, weeks, months, even years. It cuts across multiple care settings and different types of doctors.” — Dr. Daniel Yang, Gordon and Betty Moore Foundation, quoted in CNN’s coverage of the 2023 Johns Hopkins diagnostic error study[7]

The fragmentation Dr. Yang describes is not accidental. It is the product of a system where providers don’t have a full picture of patient history, where records from different encounters are scattered across disconnected systems, and where the incentive at each individual visit is to address what’s in front of you — billably — and move on. No single provider is accountable for the pattern. And said pattern, for chronically ill patients, is often the entire diagnosis.

A 2023 study published in JAMA Health Forum found that shorter visit duration was directly associated with higher rates of potentially inappropriate prescribing — and that this association held even after controlling for visit complexity.[8] In plain terms: the less time a provider has with a patient, the more likely they are to make a prescribing decision they really shouldn’t. Unfortunately, the system creates the conditions for the error, and then has no mechanism to attribute the harm back to the conditions that caused it.

Accountability That Doesn’t Track Outcomes

Here is the accountability architecture of American healthcare, stripped to its functional reality: the system tracks what was documented, what was billed, and what category the visit falls into. It does not, however, track whether the patient improved, whether the diagnosis was correct, or whether the care provided actually addressed the problem that existed.

CPT codes, ICD-10 diagnosis codes, and RVUs form the accountability infrastructure of modern medicine.[2] These systems are precise instruments for measuring transactions, not for measuring outcomes. And because what gets measured gets managed, the system has — over decades — become extraordinarily good at generating clean documentation of billable encounters and genuinely mediocre at catching what everyone else missed.

Large hospital systems have compounded this by layering RVU-based productivity dashboards, EHR-driven workflows, and algorithmic scheduling on top of a payment model that already rewarded throughput.[3] The result is a physician workforce that is not merely time-pressured but instrumentally discouraged from doing the thing that diagnostic medicine actually requires: sitting with ambiguity long enough to recognize a pattern.

The National Academy of Medicine’s landmark 2015 report on diagnostic error estimated that most Americans will experience at least one meaningful diagnostic error in their lifetime.[9] A 2023 study from Johns Hopkins estimated that diagnostic errors contribute to approximately 795,000 deaths or permanent disabilities in the U.S. annually.[7] Nearly 40% of those severe outcomes are linked to errors in diagnosing just five conditions — stroke, sepsis, pneumonia, venous thromboembolism, and lung cancer — all conditions where time is the variable that determines whether the patient lives or loses function.

The system does not count these outcomes as failures of throughput optimization. It counts them as clinical judgment calls, unfortunate outcomes, and statistical inevitabilities. The accountability gap between those two characterizations is where people die.

The Spreadsheet Problem, Made Concrete

Every complex system eventually makes its underlying logic visible. In healthcare, that logic surfaces in the scheduling grid and the productivity dashboard. Patients become rows. Visits become time slots. Complexity becomes a scheduling note — “time-consuming patient,” “difficult case” — the kind of language that signals disruption to throughput rather than a signal about what the provider should actually attend to.

There are some things you cannot put in a spreadsheet, like a patient’s pain trajectory over three years. Diagnostic ambiguity that doesn’t resolve cleanly. The trust that makes a patient tell the truth about what’s actually happening. The pattern of deterioration that only becomes legible when you look across fifteen visits instead of one.

What you can put in a spreadsheet: visit length, billing level, RVU count, provider productivity score. So that is what the system tracks. And that is what it optimizes.

CMS pays for documented encounters. Payers reimburse for coded visits. Hospital systems measure physician performance in wRVUs. None of these levers are connected to diagnostic accuracy or longitudinal patient outcomes in any meaningful, enforceable way. Not because the people running these systems are indifferent to suffering — many of them are not — but because the payment infrastructure gives them no other legible signal to follow.

What Breaks First

When a system is optimized for the wrong variable, failure doesn’t announce itself. It distributes across the population of people the system was never designed to serve well in the first place.

These are the patients who don’t improve. Who cycle through providers without accumulating a coherent diagnostic picture. Who fall through the gaps between specialists, each of whom addressed their piece of the problem with no visibility into the whole. Who stop seeking care because the experience of seeking care has become its own kind of harm — the re-telling of the same history, the same dismissal, the same prescription for the same symptom with no investigation into what’s actually driving it.

From the system’s perspective, these are outliers. Edge cases. The statistical noise around an otherwise functional model. From any other perspective, they are the signal — the evidence that the model is not functional for a substantial portion of the people it purports to serve.

Dr. Brook, with every institutional advantage available to a senior physician at a major research university, still needed a resident who had the curiosity to try a basic procedure that his senior colleagues had skipped. The thing that saved him was not a sophisticated diagnostic technology. It was an extra few minutes and a willingness to look closer. The system does not budget for that.

The Design Question With Ethical Consequences

Strip everything else away, and this becomes a question about what a healthcare system is actually for. Because the answer to that question is not found in mission statements or hospital marketing. It’s found in the payment architecture — in what the system measures, what it compensates, and what it ignores.

Right now, the payment architecture compensates speed and documentation volume. It does not compensate diagnostic accuracy, longitudinal care quality, or the kind of listening that catches what a code can’t capture. And so the system produces what it is designed to produce: a high volume of documented transactions with no reliable mechanism for tracking whether any of them actually made the patient better.

This is not a critique of individual clinicians. Most of the physicians inside this system are aware of its limitations. Many of them entered medicine to practice differently than the system permits. The critique belongs at the level of CMS payment policy, hospital system administration, and the payers who have every financial incentive to keep visits short and every structural power to enforce it.

The National Academies have called for primary care reform that centers patient outcomes over transactional billing models.[11] Researchers at the Milbank Memorial Fund documented in 2024 that primary care in the United States is in systemic crisis — not for lack of clinician effort, but for lack of structural support for doing the work that patients with complex needs actually require.[12]

The five-minute appointment is a policy outcome. Treat it like one.

The Promise of Value-Based Care — And Where It Breaks Down

At a design level, value-based care (VBC) is trying to solve exactly what this analysis describes. Move away from fee-for-service. Stop rewarding volume and speed. Start rewarding outcomes and cost efficiency. In theory, that means longer visits when clinically necessary, better chronic disease management, fewer unnecessary procedures, and actual accountability for what happens to the patient after they leave the room. That directly addresses the core structural critique: the system optimizes for what it measures, and VBC is attempting to change what gets measured.

It is a real attempt at correction. It deserves credit for that. It also deserves honest scrutiny — because in several critical ways, it doesn’t escape the same logic it’s trying to replace.

Problem 1: It Still Relies on Proxies, Not Reality

Under fee-for-service, the abstraction is the CPT code. Under value-based care, the abstractions are quality metrics, risk scores, HEDIS measures, and cost benchmarks. These are more sophisticated proxies. They are still proxies. They still don’t measure diagnostic accuracy, patient trust, complexity over time, or whether someone finally got the right answer after years of cycling through the wrong ones. The system still optimizes for what’s measurable — it just has a more elaborate spreadsheet.

Problem 2: It Shifts the Pressure. It Doesn’t Remove It.

Under fee-for-service, the pressure is: see more patients faster. Under value-based care, the pressure is: hit metrics and reduce cost. That is a different incentive with a different failure mode. It can produce under-utilization instead of over-utilization, avoidance of high-risk patients whose outcomes are harder to control, algorithmic care pathways that satisfy the metric without addressing the person, and a new version of documentation-driven care where providers are checking quality boxes rather than billing codes. Same structural constraint. New label.

Problem 3: The Infrastructure Isn’t There Yet

Even where VBC is implemented with genuine intent, the data infrastructure required to make it work as designed largely doesn’t exist. EHRs are not built for longitudinal insight — they are built for encounter documentation. Attribution models that determine which provider is accountable for which outcome are methodologically contested and operationally messy. And outcomes, by definition, take time to measure in a system that still runs on quarterly reporting cycles and annual contract renewals. So systems default to what is available and immediate: short visits, standardized workflows, documentation-heavy care. The same behavior, under a new payment framework.