A patient sends a portal message at 9:17 p.m. on a Tuesday. “Hey, I’ve been having some chest tightness when I climb stairs. It comes and goes. Also can you refill my metformin and check if my last A1C was OK? I don’t want to come in if I don’t have to.” This message requires clinical triage. It requires chart review. It requires a decision about whether to initiate a workup, message back, or call the patient directly. It involves a refill that must be processed, a lab result that must be retrieved and interpreted in context, and a patient who needs to understand whether she should, in fact, come in. In most practices, none of it is compensated. And none of it is unusual. This is a Tuesday night for most primary care providers.

The portal was marketed as a convenience. It became a liability

When patient portals became standard in primary care, the pitch was elegant: Patients could access their records, request refills, and communicate with their care team without navigating a phone tree. Less friction. More access. Better outcomes. That framing was patient-centered. It was also incomplete. What the portal actually did was remove the gatekeeper. Before the portal, a patient with a question after hours had two options: Call the office and get routed to on-call coverage, or wait until the next business day. The structure created natural triage. Urgent things went through urgent channels. Everything else waited. The portal removed that structure. Now a patient with a low-grade concern and a high-grade anxiety sends a message at 9 p.m. on the same platform where she sends a message at 9 a.m. The format feels like texting. The expectation of response has migrated accordingly.

A 2025 study published in JAMA Internal Medicine, drawing on EHR data from more than 280,000 outpatient physicians, found that patient portal message volume increased substantially at the onset of the pandemic and has not returned to pre-pandemic levels. The growth has been greatest among primary care physicians. Among those seeing more than 40 patients per week, off-hours EHR time reached more than 15 hours per week. A separate study published in the Journal of the American Medical Informatics Association found that clinicians in the top quartile for message volume had more than six times greater odds of high exhaustion compared to those in the lowest quartile. The time required to handle those messages was not built into provider schedules. It was absorbed silently, off the clock, in the margins of everything else. Absorbed silently. That is the operating phrase.

Patient-centered care became a mechanism of extraction

Health care culture has long held that the measure of a good provider is her willingness to put the patient first. This is not wrong. Patient welfare is why most of us entered this work. But the phrase “patient-centered care” has been stretched past its clinical meaning. In practice, it functions as a behavioral directive that places the cost of access entirely on the provider. Being patient-centered, in the culture of primary care, has come to mean: Be available. Be responsive. Be thorough. Do not make the patient wait. Institutions benefit from this framing. A provider who internalizes the patient-centered mandate as a personal obligation will handle portal messages without flagging the labor as uncompensated. She will not ask for protected time to manage her inbox. She will not push back on a message volume that has grown substantially since the portal launched. She will log in after dinner because she cares about her patients, and because not logging in feels wrong, negligent, and selfish.

Picking up the slack for this uncompensated work is not a character flaw. It is a rational response to a cultural system that has been designed to praise self-sacrifice and stigmatize boundary-setting. The problem is that the system benefits from this design far more than the provider does. Every hour of portal work completed off the clock is an hour of labor the organization did not pay for. Every evening the provider handles what did not fit in the day is an extension of the workday that does not appear on a timesheet. The patient-centered mandate is real. The cost structure behind it is invisible. And the provider absorbs both.

The message that looks simple rarely is

Most portal messages arrive in a form that suggests a quick resolution. A refill request. A question about a lab result. A symptom that the patient describes as minor. But refill requests require chart review to confirm the medication is still appropriate, the dose is current, and no interactions have emerged since the last visit. Lab results require interpretation in the context of the patient’s full clinical picture, not just the flagged value. And a symptom the patient has minimized (chest tightness on exertion, in a patient with poorly controlled diabetes and a family history of coronary artery disease) requires the same clinical rigor as a scheduled visit. The cognitive load does not scale with the message length.

A two-sentence portal message can require 10 minutes of careful clinical work. Multiplied across 15-20 messages per day, that is 2-3 hours of unscheduled, uncompensated clinical labor. It does not feel like a full workday because it arrives in fragments. But the math is the same. That is what makes portal management so difficult to contain. It does not feel like the work is expanding. It feels like you are just answering messages. The structure obscures the volume.

Three parties sustain this system. None of them intend to

Patients are not behaving badly when they use the portal the way they use a text message. They were told the portal is their connection to their care team. They were given a platform with no visible friction and no posted hours. The design implies availability. They are responding to the design. Organizations are not acting maliciously when they expand portal access without expanding staffing. They are responding to patient satisfaction metrics, reimbursement pressures, and the competitive pressure to be the practice that is easiest to reach. The unpaid labor required to sustain that accessibility does not appear on the balance sheet. Providers are not failing themselves when they respond to messages after hours. They are practicing exactly as they were trained: patient first, completion before rest, thoroughness as the mark of competence. They were not taught to audit how much of their labor is off the clock. That analysis was never part of the curriculum. Three parties, none acting in bad faith, producing a labor arrangement that is unsustainable for one of them.

How to actually contain this, without abandoning your patients

The solutions that work here are structural, not motivational. They do not require providers to care less about their patients. They require providers to renegotiate how patient access is delivered, and to stop absorbing the cost of a system design that was never their responsibility to fund.

• Distinguish between access and availability: Access means patients can reach their care team. Availability means their message will receive a response within posted hours. These are not the same thing. Most practices conflate them. The portal can offer genuine access 24 hours a day, with an explicit expectation that responses come during a defined window. Patients who understand this do not experience it as abandonment. They experience it as a clear standard.
• Apply the same triage logic to portal messages that you apply to calls: Not every portal message warrants a provider response. Refill requests, appointment scheduling, insurance questions, and administrative tasks do not require clinical judgment. They require a trained team member. In practices where this delegation is explicit, provider portal time drops significantly. The messages that reach the provider are the ones that actually need the provider.
• Name what the portal message is actually asking for: Before responding to any portal message, identify whether the message is a clinical question, an administrative request, or a visit substitute. Clinical questions that require examination, workup, or differential diagnosis belong in a visit. Responding to them as portal messages is not more efficient than scheduling an appointment. It is less safe and less reimbursable.
• Close the portal during defined hours and communicate this clearly: This is the most contested recommendation and the most effective one. Providers who establish defined portal response windows and communicate them explicitly to patients report sustained reductions in after-hours message volume. Patients adjust. The adjustment requires a period of transition, and it requires organizational support. It is not a unilateral action. But the conversation with leadership starts with the provider naming the problem.
• Document the labor: If portal management routinely extends past the end of your scheduled hours, that is data. Time your portal work for two weeks. Calculate the average. This is not self-pity. It is the evidence base for a conversation about workload distribution, protected time, or staffing. Providers who have this data are substantially better positioned to advocate for structural change than providers who describe a feeling of overwhelm.

The compassion argument runs in both directions

There is a version of the patient-centered care argument that says: Providers who limit their availability are prioritizing themselves over their patients. This argument has a flaw. It treats the provider’s well-being as a zero-sum trade against the patient’s. It is not. A provider who is processing portal messages at 10 p.m. is not practicing at her best. She is tired, her clinical judgment is degraded by end-of-day fatigue, and she is reviewing complex questions in a context that was not designed for careful clinical work. The 9 p.m. message about chest tightness on exertion deserves better than that.

Sustainable practice is patient-centered practice. The provider who is rested, working within compensated hours, and managing a workload that was actually designed to fit inside a normal week will give better care to more patients over a longer career than the provider who is running at 120 percent until she cannot. Patient-centered care has to include the structural sustainability of the people delivering it. A primary care system that depends on providers absorbing unlimited off-the-clock labor is not a sustainable system. It is a system that extracts from providers until they leave, and then wonders why access collapsed.

Candice Elam is a nurse practitioner.