A physician finishes a full clinic day and opens the EHR inbox to find dozens of patient portal messages: a refill request, a question about a lab result, a billing complaint, a photo of a rash. Most practices never designed a workflow for this work. It simply arrived, one message at a time, and landed on the provider.
The short answer: patient portal message management works best when the practice treats messages as a defined workload, not an add-on to visits. That means sorting messages by type before anyone reads them, routing each type to the right role, setting clear escalation rules, and protecting time for the clinical messages that remain. Technology can help sort and route, but a named person or team still has to own the queue.
How Fast Is Portal Message Volume Growing?
Faster than office visits. A JAMA study published June 22, 2026, analyzed Epic Cosmos data from 2020 to 2025 and found patient-authored portal messages rose 153%, from 0.99 to 2.50 per active patient per year, while office visits rose 17% and telephone encounters fell 6%, as reported by AJMC. Messages per sender rose from 2.2 to 5.4 a year.
The same study, covered by Healio on July 29, 2026, concluded that messaging adds to visits rather than replacing them. In other words, the inbox is a second channel of care, and it has grown without a matching change in schedules or staffing.
Volume is also uneven. MGMA's October 28, 2026 MGMA Stat article cites a Health Affairs analysis in which the top 5% of patients at one academic health system generated more than half of patient-initiated advice messages. A few high-need patients can define a provider's inbox.
Who Handles Portal Messages in Most Practices Today?
Usually medical assistants and nurses, often without dedicated time. In MGMA's October 27, 2026 poll (238 responses), 71% of medical groups said someone is designated to handle portal messages for other providers. Among those who named a role, nearly two-thirds cited medical assistants, nurses or both.
The same MGMA article reports two warning signs:
- About 71% of "yes" respondents who addressed time or pay said there is no separate schedule time or added pay for the work.
- In groups with no designee, about six in ten said the treating provider handles their own messages.
The work gets done, but it is often invisible in schedules and pay, which makes it hard to manage.
How Should a Practice Redesign Inbox Work?
Start by measuring the queue, then design roles around it. The six steps below suit most independent and multi-specialty groups.
1. Audit Several Weeks of Inbox Data
Pull EHR inbox logs by message type, sender, handling role, time to first action and resolution. Many practices may find that a sizable share of messages are administrative: scheduling, forms, billing and refills.
2. Sort Messages Into Pools Before Anyone Reads Them
Create shared pools for administrative, refill and clinical messages instead of routing everything to the provider's personal inbox. Texas Medicine reported on August 4, 2026 that a Fort Worth physician group uses a shared messaging pool so several team members can work the same queue without filling physicians' private inboxes.
3. Match Each Pool to a Role
Front-desk staff handle scheduling and forms. Medical assistants handle refill preparation under protocol. Nurses handle clinical triage within their scope. Providers handle messages that need medical decision-making.
4. Write Escalation Rules
Define what staff can resolve, what needs licensed review, what should become an appointment and what is urgent. Publish response-time expectations in the portal itself. An MGMA policy expert quoted by the AMA suggests disclaimers that state reply times and ask patients to call for urgent questions.
5. Protect Time for Clinical Messages
A JAMA Network Open study covered by Medical Economics on December 10, 2025 gave internal medicine physicians one reserved slot per half-day for asynchronous EHR work. After-hours EHR use fell by nearly 25 minutes a week, with an early productivity dip that did not worsen.
6. Review Monthly and Adjust Pay or Schedules
Track volume per provider, time to first response and after-hours EHR time. Where pay is production-based, consider administrative time, fewer scheduled visits or a stipend for the people carrying the load, as MGMA suggests.
Illustrative scenario: a five-physician family medicine practice audits four weeks of messages and finds most are refill or scheduling requests. It creates three pools, assigns a medical assistant to refills under a written protocol, and gives each physician one protected slot per half-day. The physicians now see a smaller, clinical-only queue. This is a hypothetical example, not a client result.
Can Practices Bill for Patient Portal Messages?
Sometimes. Online digital evaluation and management services (CPT 99421–99423) can apply when a patient-initiated message from an established patient leads to a response involving medical decision-making, with consent and documentation. MGMA notes these codes require at least five minutes of physician or qualified professional time over seven days, and medical assistant or nursing time does not count.
Texas Medicine lists further conditions, including that billing is not allowed if the patient had a visit for the same condition in the previous seven days, or if the message leads to an appointment for that condition within the next seven days. Payer rules vary, so confirm policies with your coding and compliance team before billing.
Billing is not the main lever. Much inbox work, such as scheduling, forms and refills, is not billable as an e-visit. The bigger gain comes from routing work correctly.
Where Does AI Fit in Inbox Triage?
AI is most useful for sorting and flagging, with people deciding. The Texas Medicine article describes an EHR-integrated tool that flags messages by complexity and identifies those medical assistants can handle, with a required physician sign-off before clinical replies reach patients. MGMA cites a Southern California Permanente Medical Group study in which a message classification tool cut median time to first read a high-acuity message from about 22 hours to 5.
Inbox task | Automation fit | Human role |
|---|---|---|
Classify message type and urgency | Strong, with monitoring | Review misroutes weekly |
Route to the right pool | Strong | Own each pool's queue |
Draft replies to routine admin requests | Moderate | Staff review before sending |
Draft clinical replies | Limited | Licensed clinician reviews and signs |
Decide if a visit is needed | Not suitable alone | Clinician decides |
Flag high-acuity language | Useful as a safety net | Clinician acts on every flag |
Two cautions apply. First, avoid pasting protected health information into general-purpose AI tools; the physician quoted by Texas Medicine prefers EHR-integrated tools for this reason. Second, smaller practices may not have access to built-in AI, so the protocol work above matters more than any tool. If your practice has not yet set rules for AI use, our article on whether your practice has an AI policy is a practical starting point.
What Are the Risks and Implementation Notes?
- Bottlenecks: a single designated person can become the new choke point. Cross-train at least two people per pool.
- Duplicate review: if physicians re-read every message staff have handled, the redesign saves nothing. Define what physicians do not need to see.
- Scope of practice: escalation rules should match state scope-of-practice rules and your clinical protocols.
- Data quality: inbox reports are only as good as message categories. Standardize categories in the EHR before measuring, or health data analysis will reflect labeling noise rather than real workload.
- Patient expectations: state response times clearly and give an alternative for urgent needs.
How Softnotions Can Help
Inbox redesign is mostly operational, but it often runs into technical limits: messages that cannot be categorized cleanly, reports that require manual exports, and AI features that do not connect to the practice's actual queues.
Softnotions can support practices and digital health teams with engineering work such as:
- Building message classification and routing workflows using AI agents designed for healthcare operations, with human review built into each step.
- Connecting portal, EHR and reporting data so leaders can track volume, response time and after-hours work without spreadsheets.
- Placing engineers inside the practice's real environment through Forward Deployed Engineering, so workflows are shaped around how staff actually work the inbox.
Every engagement is scoped to your systems, your protocols and your compliance requirements.
Portal messages are now a permanent channel of care, and they are growing faster than visits. Practices that measure the work, sort it into pools, assign clear roles and protect clinical time can reduce after-hours burden without cutting patients off. Tools help, but ownership comes first.
If you want a second opinion on your inbox workflow or the data behind it, book a free 30-minute Expert Connect and we will walk through where routing or reporting could change first.
