In short
- HIPAA-secure patient messaging keeps every message behind authenticated sign-in inside your product, because each message is electronic protected health information.
- The Security Rule's technical safeguards map directly to messaging features: unique user IDs, emergency access, automatic logoff, encryption, integrity checks, authentication and transmission security.
- Route messages to named owners, keep message content out of push and text notifications, honor patients' requests for other channels, and track time on patient-initiated threads if you bill e-visits.
HIPAA-secure patient messaging means patients and their care team exchange messages inside your app or portal, behind authenticated sign-in, and the product treats every message as electronic protected health information (ePHI). The HIPAA Security Rule's technical safeguards at 45 CFR 164.312 tell you what the feature must do: identify each user uniquely, allow emergency access, log users off after inactivity, protect messages in storage and in transit, detect tampering and verify who is signing in. The rest is product design: who owns each message, what a notification may say, how messages reach the record, and how long a patient waits for an answer.
Why build messaging into the product instead of using text or email?
Text and email are where patients already are, but you do not control either channel. Messages land on shared devices and lock screens, sit on third-party servers and cannot be routed, assigned or attached to a chart. Messaging inside the product gives you identity, routing, history and the ability to write a conversation into the record.
Patient preference still counts. Under 45 CFR 164.522(b), a covered health care provider must accommodate reasonable requests from patients to receive communications by alternative means or at alternative locations, and may not require the patient to explain why. Plan for that in the design: in-product messaging as the default, text and email as notification channels the patient chooses, and a recorded preference when a patient asks for something else.
What does the Security Rule require of a messaging feature?
Section 164.312 lists five technical safeguard standards. Some implementation specifications are required and some are addressable, which means you implement them where reasonable and appropriate and document the decision if you do not. For messaging, each one becomes a feature:
- Unique user identification (required): every patient, proxy and staff member has their own account, and staff never share a clinic login.
- Emergency access procedure (required): a documented way for authorized staff to reach messages when the usual owner is unavailable, for example by reassigning a pool's inbox.
- Automatic logoff (addressable): sessions end after a set period of inactivity, on the web and in the app.
- Encryption and decryption (addressable): messages and attachments are encrypted at rest.
- Audit controls: the rule asks for mechanisms that record and examine activity in systems that contain ePHI. For messaging, that means recording who read, sent, forwarded or reassigned each message.
- Integrity: a mechanism to corroborate that messages have not been altered or destroyed in an unauthorized way. Messages should be append-only, with edits stored as new versions.
- Person or entity authentication: procedures to verify that whoever signs in is who they claim to be.
- Transmission security: integrity controls and encryption for messages sent over networks.
How strong should sign-in be?
NIST SP 800-63B-4 gives the federal benchmark. At authentication assurance level 2, it requires two distinct authentication factors and says verifiers shall offer at least one phishing-resistant option. It recommends an inactivity timeout of no more than 1 hour and an overall reauthentication timeout of no more than 24 hours at that level. It restricts one-time codes sent over the phone network and does not allow email as an out-of-band authenticator. A messaging feature is a good reason to bring your patient sign-in up to that level, because a compromised messaging account exposes the patient's whole conversation history.
How should messages be routed?
Unrouted messages are the main reason patient messaging fails. A patient writes to "the clinic" and nobody owns the reply. Build routing in from the first version.
- Patients choose a topic, and the topic decides who answers: a question for my doctor, a prescription, an appointment, billing, a technical problem. Each topic maps to a pool.
- Each pool has named staff, and each message has one owner at a time. Reassignment is an explicit, recorded action.
- Clinical messages go to the patient's own care team. In the systems we build, each clinician sees only their own patients, so the inbox reflects the panel.
- Billing and technical messages go to staff who never see clinical content.
- Out-of-hours replies tell the patient when to expect an answer and what to do in an emergency, before they finish typing.
- A worklist shows unanswered messages by age, so nothing waits silently over a weekend.
What may a notification say?
Push notifications, texts and emails appear on lock screens and shared devices. Keep the content of the message behind sign-in and send only the fact that something is waiting: "You have a new message from your care team." The same rule applies to staff notifications. If a patient has asked to be contacted only in a certain way, the notification service must check that preference before sending.
Mobile platforms add their own rules. Apple's App Store Review Guidelines say health apps may not store personal health information in iCloud, so check where message caches and attachment previews are stored on the device and what goes into device backups.
How should attachments be handled?
- Limit file types and sizes, and scan uploads before staff open them.
- Strip location metadata from photos before storing them.
- Store attachments in the same encrypted store as messages, served through short-lived links that require a session.
- Let staff file a useful attachment, such as a photo of a wound or a home blood pressure log, into the chart in one step.
How do messages reach the medical record?
A clinically relevant conversation belongs in the patient's record. Decide early whether every thread is copied to the EHR, or only threads a clinician marks as clinical. How the copy is made depends on the EHR: some vendors expose messaging or document APIs, others accept documents only. We work from the vendor's documentation and sandbox for each one. Keep the message store in your product as the system for the conversation itself, and send the EHR a readable, signed summary or transcript.
Caregivers need one more rule. When a proxy writes on a patient's behalf, the message must show it came from the proxy, both in the inbox and in the copy sent to the record.
Can patient messages be billed?
Medicare has paid for some patient-initiated digital communication. CMS's telemedicine fact sheet describes e-visits as non-face-to-face, patient-initiated communications with practitioners through an online patient portal. It lists CPT codes 99421 to 99423 for practitioners who can bill evaluation and management services independently, covering 5 to 10, 11 to 20, and 21 or more minutes of cumulative time over up to 7 days, and HCPCS codes G2061 to G2063 for clinicians who cannot. The patient must be established with the practice and must start the exchange. Check the current code descriptors and your Medicare Administrative Contractor before building billing on them, because code sets change.
If you plan to bill, the messaging feature needs a few extra pieces: a flag for whether the patient started the thread, a 7-day window from that first message, a timer for clinician time on the thread, and a link to the billing practitioner. These are cheap to add at the start and hard to reconstruct later.
Should you build messaging infrastructure or buy it?
Few teams should build the transport layer from scratch. Real-time delivery, email and text notifications are available as services. Any vendor whose service stores or transmits PHI on your behalf is a business associate and needs a BAA, so check that before you pick one. One safe pattern is to keep message content in your own encrypted store and use outside services only for notifications that contain no PHI.
For the white-label home-based care platform we built for a digital health company in Australia, we built chat with attachments, a calendar with task management and the integrations that make care coordination work in one place, using Twilio, SendGrid and Socket.io.
What should you test before launch?
- A patient and a proxy writing about the same patient, and how each appears to staff.
- Reassignment, emergency access and what happens when a staff member leaves.
- Notification text on locked phones, on the web and in email, for every message type.
- Session timeouts on all channels, and sign-out after a password change.
- Attachment upload limits, scanning and filing into the chart.
- An attempt to read another patient's thread by changing an ID in the request.
That last test belongs in the penetration test we run before launch. We also sign a BAA before anyone on the team sees PHI.
Adding messaging to your product?
Talk to an engineer at Onix Health about your routing, your EHR and your security model. Our free five-day technical audit ends with a written report on your messaging design and what to change before patients rely on it.
Sources
- 45 CFR 164.312: HIPAA Security Rule technical safeguards (Cornell LII)
- 45 CFR 164.522: rights to request privacy protection, including confidential communications (Cornell LII)
- 45 CFR 164.502: uses and disclosures of PHI, including disclosures to business associates (Cornell LII)
- CMS fact sheet: Medicare telemedicine health care provider fact sheet (virtual check-ins and e-visits)
- NIST SP 800-63B-4: Digital Identity Guidelines, Authentication and Authenticator Management
- Apple: App Store Review Guidelines