A practice that moves its diabetic panel onto continuous monitoring does not get a little more data. It gets roughly two thousand times more. Here is what that does to charting, review ownership, coding and specialty configuration, and how to set it up before the volume arrives.
Take a primary care or endocrinology practice with 400 patients on continuous glucose monitors. Under quarterly A1c testing, that panel generated about 1,600 lab values a year, and every one of them arrived through a lab interface, landed in the chart, and got reviewed at a visit. The workflow for that has been solved for twenty years.
A sensor reading every five minutes produces 288 readings a day, roughly 8,600 a month per patient. Across 400 patients that is on the order of 3.4 million data points a month. Nobody reviews that. Nobody should try.
This is the operational problem underneath a clinical shift that has already happened, and most practices meet the clinical shift first and the operational one about six months later, usually when a monitoring program has quietly stopped being reviewed.
Why the Standard Moved, Briefly
A1c is an average across roughly three months. A steady 154 mg/dL works out to about 7.0%. So does a day that spends the morning at 58 and the afternoon at 250. The arithmetic averages out and the patient’s day does not.
In 2019 an international working group published consensus targets in Diabetes Care for time in range, the share of the day glucose sits between 70 and 180 mg/dL. For most adults the goal is above 70% of the day in range with under 4% below 70 mg/dL. The ADA Standards of Care carries those targets, with a gentler set for older adults and anyone at high risk from hypoglycemia: above 50% in range and under 1% below 70 mg/dL.
The practical consequence for a chart is this. Two patients with an identical A1c can look completely different once you measure days instead of averages. The example below is constructed rather than patient data, but the shape of the gap is well documented.
| Measure | Patient A | Patient B |
| A1c | 7.0% | 7.0% |
| Time in range, 70 to 180 mg/dL | 78% of the day | 51% of the day |
| Time below 70 mg/dL | 1% | 9% |
| Highest reading that week | 212 mg/dL | 341 mg/dL |
Patient B is dropping into hypoglycemia most days. Under quarterly A1c review, that never surfaces, because the one value in the chart is the one value that cannot see it. Continuous data also separates problems that get treated differently from each other: overnight lows, a dawn rise between 4 and 8 a.m. before anything has been eaten, a post meal spike that peaks at 240 and resolves by hour two, and glucose that is simply elevated all the time. Four different interventions, and a fingerstick schedule distinguishes none of them.
That is settled. What follows is the part that determines whether your practice can act on it.
The Real Question Is Triage, Not Interpretation
At panel scale the clinical question is no longer what does this patient’s glucose look like. It is which eleven of my 400 patients had something change this week.
Practices where continuous monitoring succeeds are not reading more graphs than practices where it stalls. They have built four things: data that lands in the chart, a named owner for between visit review, thresholds that surface the patients whose pattern shifted, and documentation that is a byproduct of the workflow rather than a separate task. Get those four right and the program runs. Get them wrong and you have 3.4 million readings a month that nobody opens.
1. The data has to land in the record
Sensor summaries that live in a device manufacturer’s cloud, behind a separate login, get reviewed enthusiastically for about three weeks. Then they do not. The three values that need to reach the chart alongside the labs and the medication list are time in range, time below 70 mg/dL, and the glucose management indicator. Once those are discrete fields in the record rather than a PDF attachment, they are trendable, they are reportable for quality programs, and they can drive a task or a flag.
That is ordinary electronic health record integration work, and it is the difference between a program that runs and a program that lapses. Ask specifically whether the values arrive as structured data or as a document, because the answer determines everything downstream.
2. Somebody has to own the between visit review
This is where chronic care management stops being a billing category and becomes the thing that catches Patient B. It needs a named owner, a defined monthly review window, documented outreach, and a care plan that actually gets revised when the data says to revise it.
Both chronic care management and remote patient monitoring have established CPT families built for exactly this work, and they are the reason the review time is fundable rather than absorbed. In broad terms you are documenting against the CCM management codes, the RPM setup, device supply and treatment management codes, and the CGM placement and interpretation codes. All of them turn on two things: documented time and a documented plan. Confirm current descriptors, time thresholds and rates against the applicable fee schedule and your payer mix before you build a program around them, because the details move.
The operational point is that if capturing that documentation requires your staff to open a second system and retype what they just reviewed, it will be captured for a month and then it will not, and the program will look unprofitable when what actually failed was the data entry path.
3. The workflow has to match the specialty
An endocrinology panel where most patients wear a sensor needs a different default review cadence, different threshold defaults and a different note template than a primary care panel where a dozen patients do. Configuring that is what specialty EHR setup is for. For practices running the diabetes side of this at volume, an endocrinology EHR built around insulin titration and sensor review will beat a general template bent into shape, mostly because the titration history and the sensor trend need to sit on the same screen.
4. The patient has to be in the loop
A patient who can see their own time in range, message about a pattern they noticed, and get a response without waiting for the next available slot is a patient whose data is worth collecting. Patient engagement tools are what close that loop. Without it, remote monitoring becomes surveillance that generates documentation and changes nothing, which is the version that gets cancelled at the next budget review.
And something has to do the surfacing
The triage itself is the piece practices most often try to do manually and most often abandon. Threshold based review at this volume is what an AI-powered remote care platform is for: it flags the patients whose pattern moved and leaves the steady ones alone until their next visit. Whatever you use, the requirement is the same. The output has to be a short worklist in the same place your staff already works, not a dashboard somebody has to remember to visit.
What to Settle Before You Enroll the First Patient
- Which sensor brands you will support, and whether each one delivers structured values or documents into your record.
- Who reviews, and in what window. Name the role, not the person, and put the window in the schedule.
- What triggers an outreach. Time below 70 crossing a threshold is the usual first rule, because it is the finding A1c hides and the one with acute risk.
- Which target applies to which patient. The general adult target and the older adult target are different numbers, and reviewing everyone against the tighter one causes harm in the older cohort.
- How review time gets captured. If it is not a byproduct of the review itself, assume it will not be captured.
- What the escalation path is when the reviewer is not the prescriber.
- What you will report, and to whom, for quality programs and for your own read on whether the program is working.
None of this replaces clinical judgment. Whether a 51% time in range means a basal adjustment, a meal timing conversation or a sensor that is not being worn correctly is the clinician’s call and always will be. The job of the software and the workflow around it is narrower: make sure the right eleven charts are on the screen when that judgment gets applied, and make sure the work of applying it is documented and funded.
The Short Version
A1c tells you where a patient has been on average. Time in range tells you what their days actually were, and it is the only one of the two that catches the lows. Clinically that argument is over.
Operationally, the practices that benefit are the ones that treated it as a workflow project rather than a device rollout. Structured values in the chart, a named review owner with protected time, thresholds doing the triage, documentation captured as a byproduct, and configuration that matches the specialty. Settle those seven questions above before the first enrollment and continuous monitoring changes outcomes. Skip them and it produces 3.4 million readings a month, a stalled program, and a reasonable looking conclusion that remote monitoring does not pay.
