Remote patient monitoring grew fast in Medicare because the rules made it easy to start. A practice signed with a vendor, the vendor shipped blood pressure cuffs and glucometers, the vendor's nurses made the monthly calls, and the practice billed 99453, 99454, 99457 and 99458 under the physician's NPI. Some of those programs improved care. Some of them were a device in a drawer and a monthly call nobody remembered. CMS has been signaling for two years that it intends to tighten this, and the CY 2027 physician fee schedule proposed rule, released July 14, 2026, does.
The comment deadline is September 14, 2026, three weeks from today. If your practice bills remote monitoring, this article is meant to help you decide what to write and what to change regardless of what CMS finalizes in November.
Key takeaways
- Two of the three proposals, an initiating visit and an established-patient requirement for RTM, are already good practice. Most well-run programs meet them today.
- The third, that clinical staff time must come from staff the practice employs, would end the vendor-staffed model as it is commonly structured.
- CMS also proposes to revalue the codes downward and asks whether to replace the 17-code family with four bundled monthly G-codes.
- Before September 14: count your program, read your vendor contract, audit ten patients, and write a comment with numbers.
The current code family, briefly
Remote physiologic monitoring: 99453 (setup and patient education), 99454 (device supply with daily recordings or alerts, 16 or more days in 30), 99457 (first 20 minutes of treatment management with an interactive communication in the month), 99458 (each additional 20 minutes) and 99091 (collection and interpretation, 30 minutes). For 2026 CPT added shorter-duration codes, 99445 for device supply of 2 to 15 days and 99470 for 10 to 19 minutes of management. Remote therapeutic monitoring has a parallel set: 98975 (setup), 98976 and 98977 (device supply for respiratory and musculoskeletal systems), 98980 and 98981 (treatment management), plus its own shorter-duration additions. Seventeen codes in all, by CMS's count.
What CMS proposes
| Proposal | Today | What it would mean |
|---|---|---|
| RTM limited to established patients | RPM already requires an established patient; RTM has not | RTM programs that enroll at first contact (common in physical therapy and behavioral health) need a prior relationship |
| Separately reportable initiating visit, face-to-face or telehealth, before RPM or RTM begins, covering the monitoring plan, the clinical determination that monitoring is warranted and documented consent | An initiating visit has been expected for new patients under RPM; not consistently for RTM | Monitoring cannot start from a marketing call or a portal sign-up; a documented visit comes first |
| Payment only when clinical staff are employed by the billing practice, not third-party contractors | Incident-to rules allow contracted clinical staff under supervision, which is how most turnkey vendors operate | The vendor-staffed model, as commonly structured, would not be billable |
| Revalue the codes downward on practice expense inputs, especially device supply | Values based on device cost assumptions CMS now considers too high | Lower payment per patient per month for 99454 and its RTM equivalents |
| Request for comment on replacing the family with four bundled HCPCS G-codes, each monthly code requiring device supply, at least two days of data, 20 minutes of management and one real-time interaction | Separate codes for setup, device supply and time-based management | Simpler billing, fewer line items, and almost certainly a different total payment per patient per month |
The employed-staff proposal is the one that matters
The first two proposals are, honestly, good practice already. A monitoring program for a patient the practice has never seen is hard to defend clinically, and a documented initiating visit is where consent, device education and the care plan belong. Most well-run programs already do both, and a program that does not should start regardless of the rule.
The third proposal is different. A large share of Medicare remote monitoring today runs on a model where the vendor supplies the devices, the platform and the nurses or medical assistants who review the data and make the monthly calls, under the general supervision of the practice's physician. The practice bills, pays the vendor a per-patient fee, and keeps the difference. CMS is proposing that the clinical staff time counted toward 99457, 99458 and the RTM equivalents be furnished by staff the billing practice employs. If that is finalized as written, the vendor-staffed model does not survive in its current form.
We think CMS is responding to real abuse and to a real question about whether monitoring performed by someone who has never met the patient or the physician is "incident to" anything. We also think the proposal will catch well-run programs in small practices that cannot afford a dedicated nurse for forty monitored patients. That tension is exactly what a comment letter should describe, with numbers: how many patients, how many staff hours a month, what the program costs to run in-house versus contracted, and what happens to the patients if it stops.
A worked example
A two-physician internal medicine practice monitors 70 patients with hypertension and heart failure through a vendor. In a typical month it bills 99454 for about 60 of them (the rest miss the 16-day threshold) and 99457 for about 55, with 99458 on perhaps 15. At approximate 2026 national rates that is somewhere around $9,000 a month in Medicare revenue before the vendor fee, which for this practice runs about 45 percent of collections. The vendor's two nurses spend roughly 60 hours a month on this panel.
Bringing the work in-house means about 0.4 of a full-time nurse or a well-trained medical assistant, plus a platform license, plus devices. For 70 patients the arithmetic is tight and depends on the final code values. At 150 patients it works comfortably. At 40 it does not. The practice's comment letter should say exactly that: our program meets the initiating-visit and established-patient proposals today, the employed-staff test would end it at our current size, and here is the size at which it would survive. That is the kind of comment CMS can do something with.
What to do before September 14
- Know your program. Count monitored patients by code, monthly billing, and who actually performs the management time: your employee, a vendor employee, or a contractor working under your supervision agreement.
- Read your vendor contract. Look for the staffing clause, the termination terms and anything that says the vendor's staff are "leased" or "dedicated" to your practice. Then ask the vendor, in writing, how they intend to comply if the proposal is finalized. Some will offer to place staff on your payroll; some will not have an answer.
- Audit ten patients. Confirm each has an initiating visit documented, 16 or more days of readings for 99454 months, a documented interactive communication for 99457, and evidence the data changed something in the care plan. If you cannot show that, the employed-staff question is not your biggest problem.
- Write the comment. Describe your program factually, state whether it meets the established-patient and initiating-visit proposals (most will), and quantify the employed-staff impact. Suggest an alternative if you have one, such as a requirement that the same named staff member be assigned to the practice's patients under a written supervision protocol, rather than an employment test.
- Model the bundled G-codes. CMS has not published the four codes or their values. Take your current monthly revenue per monitored patient and ask what a bundle would need to pay to keep the program viable. That number belongs in the comment too.
What we would do regardless of the final rule
Require an initiating visit for every enrollee. Stop enrolling patients the physician has not seen. Document the clinical action taken on the data every month, not just the minutes. Make sure the physician, not only the vendor, can produce the monitoring record on request, because audit letters for remote monitoring are already arriving at practices we know, and the practice, not the vendor, signs the claim. And review the 16-day threshold monthly: billing 99454 in a month with eleven readings is the most common remote monitoring error we see, and it is the easiest for an auditor to find.
Questions we hear
If we bring the staff in-house, does the program still make sense?
Run the numbers as in the example above. A full-time nurse or a trained medical assistant can manage a substantial panel of monitored patients, and the 99457 and 99458 time is real clinical work. For a practice with a few dozen monitored patients the economics are tight; for one with several hundred they usually work. The proposed values, when CMS finalizes them, will decide it.
Does this affect our commercial payers?
Not directly. Commercial coverage of remote monitoring varies by plan and contract. Many plans follow Medicare's policy over time, so plan for that, and read your commercial contracts for any clause that ties monitoring coverage to Medicare rules.
Can you review our program?
Our RCM audit includes a remote monitoring review when a practice bills these codes, and our billing team will be adjusting claim workflows for whatever CMS finalizes in November.
What to do this week
- Pull a report of monitored patients by code for the last three months and note who performed the management time.
- Send the vendor a written question about compliance with the employed-staff proposal and ask for an answer before September 10.
- Pull ten charts and check the initiating visit, the 16-day count and the documented clinical action.
- Draft the comment with your patient count, staffing hours and the size at which an in-house program would work.
- Calendar September 14 and decide who files the comment.
