How to Run RPM in a Primary Care Practice: A Step-by-Step Workflow

Remote patient monitoring is easy to describe and harder to operationalize. This is a practical walkthrough of what a primary care team actually does, week to week, once a patient is enrolled — from candidate selection through closing the loop on an escalation.

Most of the friction in RPM isn't clinical, it's operational. A practice can have a clear rationale for monitoring a patient and still stall out on who calls whom, when, and how any of it gets documented for billing. Below is one reasonable way to structure the workflow. Your practice's exact process should be built with your billing and compliance team, since coding and time requirements are set by payers and can change.

The workflow, step by step

1. 1. Identify candidate patients

   Start with patients who have a chronic condition that benefits from data between visits — diabetes, hypertension, and other conditions where day-to-day readings inform medication or lifestyle adjustments are common starting points. Many practices build a simple registry query (diagnosis code, recent A1C or blood pressure trend, upcoming visit) rather than relying on individual clinicians to remember to ask.

2. 2. Have the enrollment conversation

   This is a consent conversation, not just a device handoff. The patient needs to understand what's being monitored, who looks at the data, how often, what happens if a reading is concerning, and any cost-sharing that may apply. Written consent is typically required and should be documented in the chart. Practices that treat this as a rushed afterthought tend to see higher device abandonment later.

3. 3. Set up the device and educate the patient

   Device setup usually happens same-day or at a dedicated onboarding visit: pairing the device, confirming data is flowing to the practice's platform, and walking the patient through the basics — how to wear or use it, what a normal reading looks like for them, and who to call with technical problems versus clinical concerns. A short teach-back ("show me how you'd check your reading") catches setup errors before they become weeks of bad data.

4. 4. Establish the monitoring cadence

   Define, in writing, who reviews incoming data and how often. A common structure is a designated nurse, medical assistant, or care coordinator doing a daily or every-other-day pass over flagged or out-of-range readings, with the supervising clinician reviewing a summary weekly or at the next scheduled visit. The clinician sets the thresholds and escalation rules up front so staff aren't making clinical judgment calls outside their scope.

5. 5. Document interactive time as it happens

   Time-based RPM billing codes generally require a documented, live interaction with the patient about their data during the month — not just passive data review. Best practice is to log each contact in real time: date, duration, who made the contact, what was discussed, and any resulting action. Reconstructing this at month-end from memory is a common source of denied or downcoded claims.

6. 6. Close the loop when a patient needs escalation

   Every RPM program needs a defined escalation path: what triggers an outreach call, what triggers a same-day appointment, and what triggers a message to the clinician versus a call to emergency services. The person reviewing data day to day should never be guessing where the line is — that's a decision the clinician makes in advance, in a written protocol, and revisits as needed.

The administrative burden is real

Even a small RPM panel generates a continuous stream of data that someone has to look at, triage, and act on between visits — on top of the enrollment paperwork, device logistics, and time-tracking described above. For a panel of even a few dozen patients, manually scanning every reading for every patient every day is not a sustainable use of clinical staff time.

This is the specific problem automated triage and decision-support software is built to address. Rather than a staff member scrolling through raw glucose or blood pressure traces for each patient, software can apply clinician-defined rules to surface only the readings and trends that actually need a human look — flagging a pattern instead of a single noisy data point, and leaving the clinical judgment and any resulting action to the care team. Endobits works this way for CGM data: it's decision-support that helps a practice manage the volume, not a system that reviews or manages patients on its own.

Related: Remote patient monitoring: the complete guide · RPM patient consent & enrollment · All resources