A year-long study of more than 2,300 U.S. Veterans using continuous glucose monitors has quietly redrawn the boundary between compliance and care. Researchers found that how steadily someone wears a device matters far more than how often — a distinction that standard adherence metrics have long obscured. In the broader story of medicine's relationship with technology, this work asks a deeper question: not whether a patient follows instructions, but whether the rhythm of their engagement actually reaches them.
Study identifies three CGM usage patterns, revealing inconsistent users miss glycemic benefits
Consistent use was especially beneficial for patients with more complex diabetes management needs.
Mark
So the study tracked over 2,300 Veterans for a year. What made the researchers think that looking at day-to-day patterns would matter more than just counting total wear time?
Mimi
They noticed that real-world use doesn't follow a simple on-or-off pattern. People interrupt their use, restart, interrupt again. A patient might wear the device 78 percent of the time but in a very choppy way—on for a few days, off for a few days, on again. Another patient might wear it 78 percent of the time but in a much more stable pattern. The question was: does that difference in stability affect whether the device actually helps control blood sugar?
Luke
And they found it does. But I want to be careful here. They're showing association, not causation. It's possible that people whose diabetes is harder to manage are the ones who struggle with consistent use, and their worse glucose outcomes are because of the harder-to-manage diabetes, not because of the inconsistent use itself.
Mimi
That's fair. They tried to adjust for that using a statistical method called double machine learning, which accounts for a lot of baseline differences—prior glucose control, medications, healthcare utilization, and so on. But you're right that you can't fully rule out unmeasured confounding in an observational study.
Mark
The fluctuating group is interesting because 73 percent of them still met the standard 80 percent adherence threshold. So they looked compliant by the usual metric.
Mimi
Exactly. That's the key finding. The standard metric—"Did you wear it at least 80 percent of days?"—doesn't capture whether that wear was stable or fragmented. The fluctuating users had moderate overall coverage but high day-to-day variability. And their glucose outcomes were significantly worse than the consistent users, even though both groups had similar or overlapping overall wear percentages.
Luke
But the confidence intervals on some of these estimates are pretty wide, especially for the low-engagement group. They had to exclude a lot of people from the outcome analysis because they didn't have enough CGM readings at the end of the year. That makes the estimates less precise.
Mimi
True. They used inverse-probability weighting to try to account for that missing data, but they acknowledge it's a limitation. If someone stopped using the device because they were hospitalized or had other health problems, that's not random missingness, and the statistical adjustment might not fully correct for it.
Mark
What about the subgroup findings? They found that insulin pump users benefited more from consistent use.
Mimi
Yes. Pump users are managing their insulin more intensively, so they have more to gain from real-time glucose data. When they used their CGM consistently, they saw actual improvements in glucose control. Fluctuating users in that same group didn't see those improvements. It suggests that for patients with more complex management needs, the stability of CGM use becomes even more critical.
Luke
But again, the sample sizes in some of those subgroups are smaller, and the confidence intervals are wider. The pattern is suggestive, but I'd want to see it validated in another cohort before making strong clinical recommendations based on it.
Mark
The study also looked at what predicts who will be a consistent user versus a fluctuating user. What did they find?
Mimi
People who started with worse glucose control were less likely to maintain consistent use. People with more prior healthcare visits and hospitalizations were also less likely. And missed clinic appointments predicted lower consistent use. It paints a picture of patients who are already struggling—either with their disease or with engagement in their own care—and who face additional barriers to sustaining device use.
Luke
That's important context, but it also highlights a limitation of the study. It's mostly older male Veterans. The patterns might be different in younger populations, women, or people in commercial insurance settings. The researchers acknowledge this, but it's worth keeping in mind when thinking about how broadly these findings apply.
Le Pouls
- Nearly three-quarters of so-called 'adherent' CGM users were quietly falling behind — their glucose control worsening even as they technically met the 80% wear-time threshold.
- Advanced trajectory analysis exposed three distinct user profiles — consistent, fluctuating, and low-engagement — revealing that erratic use carries real physiological costs invisible to conventional metrics.
- The stakes are highest for the most complex patients: those on insulin pumps or glucagon therapy gained the most from consistent use, yet were also among those most likely to struggle with sustained engagement.
- Patients already burdened by poor glucose control, frequent hospitalizations, or missed appointments were disproportionately likely to fall into the fluctuating or low-engagement groups, compounding existing disadvantage.
- The study calls on clinicians, device makers, and health systems to shift focus from threshold-checking to stability-building — designing support that keeps people engaged continuously, not just cumulatively.
A year-long study of more than 2,300 U.S. Veterans using continuous glucose monitors has quietly redrawn the boundary between compliance and care. Researchers found that how steadily someone wears a device matters far more than how often — a distinction that standard adherence metrics have long obscured. In the broader story of medicine's relationship with technology, this work asks a deeper question: not whether a patient follows instructions, but whether the rhythm of their engagement actually reaches them.
When researchers tracked more than 2,300 U.S. Veterans through their first year of continuous glucose monitor use, they weren't just measuring how often people wore the devices — they were mapping the shape of that use over time. Published in PLOS Digital Health, the study challenges a foundational assumption: that an 80% wear-time threshold is a meaningful proxy for whether a patient is truly engaged.
Using trajectory clustering rather than simple averages, the team identified three distinct usage patterns. Consistent users wore their monitors nearly all the time, averaging 93.5% coverage with few interruptions. Fluctuating users achieved 78.2% overall coverage, but their engagement was choppy — cycling on and off in ways that aggregate numbers couldn't capture. Low-engagement users barely used their devices at all, averaging just 33.5% coverage.
The health consequences followed the same gradient. Consistent users saw almost no change in average glucose over the year. Fluctuating users experienced a 7.1 mg/dL rise in average glucose and a meaningful drop in time spent in the healthy range. Low-engagement users fared worst. The most pointed finding: 73% of Fluctuating users would have been deemed 'adherent' by standard criteria — compliant on paper, but measurably worse off in practice.
The benefits of consistency were sharpest for patients managing the most complex cases. Those using insulin pumps or prescribed glucagon — markers of advanced diabetes management — showed actual improvements in glucose control when they used their monitors steadily. Their fluctuating counterparts, even with comparable overall wear time, did not.
The study also traced who was most at risk of inconsistent engagement: patients with worse baseline glucose control, more hospitalizations, and a history of missed appointments. These are people already navigating difficulty, for whom sustained device use may require more than a prescription.
The researchers acknowledge the cohort's limits — older, predominantly male Veterans may not reflect all populations — and note that observational data cannot prove causation. But the signal is clear: stability of use predicts outcomes in ways that raw percentages cannot. For clinicians, the implication is that monitoring adherence thresholds is not enough. For the broader field of wearable health technology, it is a reminder that how people engage with tools — not merely whether they do — is where the real story lives.
More than 2,300 U.S. Veterans who started using continuous glucose monitors were tracked for a year, and what researchers found challenges a basic assumption about how we measure whether people actually use medical devices. The study, published in PLOS Digital Health, reveals that simply counting the percentage of days someone wears a CGM misses something crucial: whether that use is stable or erratic.
Continuous glucose monitors are small sensors that measure blood sugar every five minutes, offering real-time feedback that can help people with diabetes avoid dangerous swings in glucose levels. Clinical trials have shown they work—they reduce complications, lower hospitalizations, and improve overall control. But those benefits depend on consistent, near-daily use. In practice, people's engagement with these devices varies wildly, and it often declines over time. Previous research suggested that only about half of new users maintain what's considered "adherent" wear—more than 80 percent of days—over their first year.
The researchers, analyzing data from the Department of Veterans Affairs, took a different approach. Instead of reducing each person's year of use to a single percentage, they mapped out the actual day-to-day pattern of how people wore their devices. They used an advanced clustering method to identify distinct usage trajectories, and three groups emerged. The first group, labeled Consistent users, wore their monitors nearly all the time—an average of 93.5 percent coverage—with few interruptions. The second group, Fluctuating users, showed moderate overall coverage of 78.2 percent, but their use was choppy and irregular, marked by frequent on-and-off cycles. The third group, Low-engagement users, barely used their devices at all, averaging just 33.5 percent coverage.
Here's where the findings get interesting. When the researchers looked at what happened to people's blood sugar control over the year, they found a clear gradient. Consistent users saw their average glucose rise by only 2.5 milligrams per deciliter, while their time in the healthy range stayed essentially flat. Fluctuating users, by contrast, saw their average glucose jump by 7.1 milligrams per deciliter and their time in range drop by 3.56 percentage points. Low-engagement users fared worst, with glucose rising 14 milligrams per deciliter. The gap between Consistent and Fluctuating users was especially striking because 73 percent of the Fluctuating group would have been classified as "adherent" by the standard 80 percent threshold. They looked compliant on paper, but their actual glucose control suffered.
The study also found that the benefits of consistent use were not uniform across all patients. For people using insulin pumps—a sign of more complex diabetes management—consistent CGM use was associated with actual improvements in glucose control. The same held for people prescribed glucagon, a medication for severe low blood sugar. These are the patients with the most to gain from stable monitoring, and they were the ones who benefited most when they used their devices consistently. Fluctuating users in these high-need groups did not see comparable benefits, even when their overall wear time was high.
The researchers also looked at which patients were more likely to fall into the Fluctuating or Low-engagement groups. People who started with worse glucose control were less likely to maintain consistent use. Those with more healthcare visits and hospitalizations in the prior year were also less likely to stick with it. Missed clinic appointments predicted lower consistent use as well. These patterns suggest that patients already struggling with their diabetes or dealing with complex medical situations face additional barriers to sustaining device engagement.
The work introduces a framework for analyzing wearable device use that goes beyond simple adherence percentages. Rather than asking "Did the patient wear the device 80 percent of the time?" it asks "How stable was that use, and did it actually translate to better health outcomes?" The researchers acknowledge limitations: the cohort was predominantly older, male Veterans, so the patterns might look different in younger populations or commercial insurance settings. Some patients in the low-engagement group had incomplete follow-up data, which widened the uncertainty around those estimates. And like all observational studies, it cannot definitively prove that fluctuating use causes worse outcomes, only that the two are strongly associated.
But the core message is clear: consistency matters more than the raw percentage. A patient who wears a CGM steadily, even with occasional gaps, gets more benefit than one whose use is fragmented, even if both hit the same overall coverage number. For clinicians, this suggests that simply checking whether a patient meets an 80 percent threshold is not enough. For patients and device makers, it points to the value of support systems that help people maintain stable, uninterrupted engagement rather than cycling on and off.
Citations marquantes
Many people in the fluctuating group would still be considered adherent by standard definitions, while they experienced worse glucose outcomes than consistent users, highlighting the limitations of common adherence thresholds.— Study authors, PLOS Digital Health
Envie de l'histoire complète ?
Lire l'original sur PLOS (Public Library of Science) ↗
Health & Medicine