Patients now have a way to reach a physician that no phone call ever offered: directly, in writing, at any hour, with no one screening it first. That was the entire point. It was also, it turns out, the entire problem.
A patient types “not feeling great today” into the portal. It could mean a cold. It could mean chest tightness they’re downplaying. The message looks the same either way, sitting in a queue next to a refill request and a question about billing, waiting for whoever opens it to figure out which one it actually was.
What the portal actually replaced
The honest answer is nothing. A national study published in JAMA this year, using Epic health record data from more than 140 million patients across over 2,000 hospitals and 47,000 clinics, found that portal messages rose 153 percent between 2020 and 2025, from about 1 message per patient per year to 2.5. Office visits didn’t fall over that same period. They rose 17 percent. Telephone calls dropped only slightly, about 6 percent.
If the portal were replacing calls and visits, those numbers would have moved the other way. Instead, patients gained an entirely new channel and kept the two they already had. The portal wasn’t a substitute for anything. It landed on top of a schedule that hadn’t gotten any lighter.
An MGMA Stat poll from early last year found seven in ten medical groups reporting an increase in portal message volume over the previous year alone, with almost none reporting a decrease. This isn’t a one-time adjustment settling into a new normal. It’s still climbing, and there’s no sign in the data of it leveling off.
Why this channel is different from every other one
A phone call goes through a receptionist, who decides whether it needs to interrupt a room or can wait for a callback. A fax lands in a stack, and someone works through that stack before anything reaches a desk. Even a patient standing at the front window has to check in before the physician knows they’re there.
The portal skips all of that. A message sent at midnight lands in the same inbox, formatted the same way, whether it says the refill ran out early or describes a symptom that needs attention tonight. Whatever sorting happens, happens after the physician has already opened it, because nothing upstream did the sorting first.
This isn’t specific to any one practice or any one physician’s habits. Research from the AMA, conducted with Epic, found that physicians in the U.S. receive roughly three times as many inbox messages as physicians in other countries, a gap tied partly to how early American patients gained portal access, and partly to the absence of any triage layer built into that access before it went live.
None of this argues the portal should go away. A patient who’d otherwise wait three weeks for an appointment just to ask a simple question is genuinely better served by a message than by a wasted visit. The channel has real value. What’s missing is anything standing between that channel and the person whose time is hardest to replace in the building.
What this actually costs, beyond the hours
The deeper cost isn’t the volume alone, though the volume is real and rising. It’s what happens when a system delivers something routine and something urgent at the exact same visual weight, with no signal to tell them apart before they’re read.
ADVERTISEMENT
Faced with that, a physician ends up doing one of two things. Reading every message as though it might be serious, which means the inbox never actually closes. Or learning, over months, which kinds of messages probably don’t need a close look, and sorting by instinct instead of certainty, under time pressure, with no second set of eyes on the call.
The second pattern is where the real risk sits, and it isn’t a character flaw. Nobody handed physicians a system built to help with this sorting, so they built their own, on top of a day that already had no slack in it.
On the screen, the message that genuinely needed a same-day response looks identical to the one that could have waited until Monday. The interface can’t tell the difference. Only a read can, and by the time that read happens, the sorting has already occurred in the physician’s head, not before it.
Why a smarter inbox doesn’t fix it
The obvious next move is to make the inbox itself smarter. Flag urgent keywords. Sort by subject line. Let an algorithm push certain messages to the top. Health systems have tried versions of this for years, and the honest result is that keyword filters catch what they’re told to look for and miss everything they weren’t.
A patient describing chest pain rarely writes the word “urgent.” They write about being tired, or a strange ache, or a question that sounds routine until someone with training reads the whole sentence in context.
That’s the part a filter can’t do and a person can. Sorting a portal message isn’t just pattern-matching for scary words. It’s judgment, the same kind a receptionist already exercises on the phone every day when they decide whether to interrupt a room or take a message.
Nobody would trust a keyword filter to answer the phones. There’s no clear reason the portal should be held to a lower standard, and every reason to think it’s been held to one simply because nobody built the alternative before the tool went live.
The missing piece was never the technology
Every other channel into a practice was built with a person in front of it before it ever reached the physician. The phone got a receptionist. The fax got someone who worked through the stack. The portal got nothing, because it was rolled out as a self-service tool for patients, and nobody thought to staff the other side of it the same way.
The practices closing this gap aren’t asking physicians to manage their inbox differently, and they aren’t waiting on a smarter algorithm either. They’ve put a trained person in front of the portal, the same role a receptionist already plays for the phone: someone who reads every message first, resolves what’s routine on the spot, and sends forward only what actually needs a physician’s judgment, with enough context attached that nothing has to be explained twice.
That person doesn’t need to be clinical staff. They need to be trained on the practice’s patients, its common questions, and where the line sits between routine and not, the same training any front-desk hire already receives for the phone.
Pull up your own inbox for a single day. Count how many messages needed you specifically, and how many just needed someone to read them and respond with the right answer or the right next step. If the second number is larger, the portal isn’t the problem. The missing front door is.
Karan Kanwar is the CEO of Central AI and Wing Assistant. Central AI provides an AI operating system for appointment-driven medical practices, handling calls, texts, web chat, and messages around the clock, qualifying patients and booking appointments directly into a practice’s existing systems. It is HIPAA compliant and designed to complement the tools practices already use.
Wing Assistant pairs that technology with skilled virtual assistants who take on the administrative work that pulls teams away from patient care. Kanwar writes on where automation and human support meet in health care operations, and how practice leaders can adopt both effectively. He shares updates on LinkedIn.
