Continuous glucose monitors have moved well beyond specialist clinics. They now appear in television ads, pharmacy aisles, social feeds, and conversations about everything from insulin dosing to “metabolic optimization.” That wider visibility creates a problem: a tool that can be genuinely useful for diabetes management is easily presented as if more data must always produce better health.
The better question is not whether continuous glucose monitoring is good or bad. It is whether the information will help a particular person make safer, clearer, clinically meaningful decisions.
A large study published October 7, 2026, adds useful long-term evidence for adults with type 2 diabetes at high risk because of poor glucose control or serious hypoglycemia. It strengthens the case for sustained real-time monitoring in that population. It does not prove that everyone with prediabetes, insulin resistance, or simple curiosity needs a sensor.
In this article
- What a CGM measures
- What the new long-term study found
- Why the findings need restraint
- Who is most likely to benefit
- How to use the data without chasing every spike
- Practical limitations and safety issues
- A better decision question
What a continuous glucose monitor measures
A continuous glucose monitor, or CGM, uses a small sensor placed under the skin to estimate glucose in interstitial fluid every few minutes. Depending on the system, readings appear automatically or after scanning. The display can show a current estimate, direction of change, alerts, and patterns across meals, activity, sleep, medication, and illness.
That is different from a laboratory glucose test or a traditional finger-stick reading. CGM readings lag behind blood glucose, especially when glucose is changing quickly. Sensors can also be inaccurate, compressed during sleep, affected by placement, or temporarily unreliable. Symptoms and unexpected readings sometimes need confirmation with a blood glucose meter, particularly when treatment decisions could cause hypoglycemia.
The real value is not the sheer volume of numbers. It is the ability to see patterns that are difficult to capture with occasional testing: overnight lows, repeated post-meal rises, time spent in or out of a target range, and whether a treatment change is helping across days rather than at a single moment.
What the new long-term study found
The new JAMA Network Open study examined 25,648 adults with type 2 diabetes in Kaiser Permanente Northern California. Participants qualified because their A1C was above 8% or they had needed emergency or inpatient care for hypoglycemia. Researchers compared 2,771 people who started and sustained real-time CGM use with 22,877 controls who had not started CGM.
This was not a randomized trial. The researchers used a target-trial emulation, a sophisticated observational design intended to make real-world data resemble the comparison a trial would have made. After statistical adjustment, sustained CGM use was associated with an A1C about 0.8 percentage points lower than the control group at three years. The CGM group also had a later reduction in urinary albumin-to-creatinine ratio, a marker of kidney damage. Estimated kidney filtration, however, did not meaningfully improve.
Those distinctions matter. The study provides reasonably strong real-world evidence that people with high-risk type 2 diabetes who continue using real-time CGM may achieve better long-term glucose control. The kidney finding is more tentative. A change in albumin in the urine is encouraging, but it is not the same as demonstrating preserved kidney function, fewer cases of kidney failure, or fewer cardiovascular events.
Why the findings need restraint
Observational studies can adjust for many measured differences, but they cannot eliminate every important difference between people who choose and sustain a technology and people who do not. Motivation, clinical support, insurance coverage, comfort with devices, and the ability to respond to data may all influence outcomes.
Attrition was substantial. About 44% of the CGM group was censored after discontinuing use, while nearly 19% of controls were censored after starting CGM. Everyone received care in one integrated health system, which may limit how well the results transfer to settings with less education, follow-up, or coverage. The study also focused on real-time CGM; it cannot establish that every device or monitoring approach produces the same result.
Funding deserves notice as well. Dexcom supported the research through an independent investigator award, alongside Kaiser Permanente and federal grants. The authors reported that funders had no role in the design, analysis, or publication decision. That does not invalidate the work, but it belongs in the evidence assessment—especially when a device manufacturer benefits from favorable findings.
The result is evidence-sharpening, not a universal mandate. It fits with randomized trials showing that CGM can improve glucose control in selected adults with type 2 diabetes, while extending the observation window and adding kidney markers. Long-term randomized trials are still needed before concluding that CGM itself prevents kidney disease or other complications.
Who is most likely to benefit
The American Diabetes Association’s 2026 technology standards recommend CGM for people using insulin, for some people taking noninsulin drugs that can cause hypoglycemia, and more broadly when CGM data can help management. The wording is important: the device should support a management need.
- People using insulin. Trend information and alerts can help detect impending highs and lows and support safer dose decisions when used with appropriate education.
- People with recurrent or hard-to-detect hypoglycemia. Overnight patterns and alarms may reveal risks that intermittent checks miss.
- People whose A1C remains above target despite treatment. CGM may show whether the problem is fasting glucose, meals, medication timing, missed doses, illness, or another pattern requiring attention.
- People making a defined treatment or lifestyle change. A clinician-guided short trial can sometimes answer a specific question without committing someone to permanent monitoring.
- People able and willing to act on the information. Data are useful when they lead to a sensible decision. More readings without a plan can become noise.
By contrast, evidence does not support using CGM to diagnose diabetes or prediabetes. It also remains unclear whether routine monitoring improves meaningful health outcomes in metabolically healthy people who are not taking glucose-lowering medication. A sensor can make normal biological variation look like a problem in need of correction.
How to use the data without chasing every spike
CGM creates an unusually persuasive stream of numbers. The line rises after a meal and falls after a walk; it feels as if the device has delivered a verdict. In reality, a single rise rarely means much by itself. The food, portion, time of day, preceding activity, sleep, stress, medication, sensor lag, and the person’s starting glucose all matter.
Useful review usually looks for repeated patterns rather than isolated peaks. Clinicians may consider time in range, time below range, average glucose, variability, and whether patterns are consistent enough to justify a change. A1C, symptoms, medications, kidney function, cardiovascular risk, and personal priorities still belong in the same conversation.
This is also where ordinary health headlines can mislead. A study may report a large relative improvement while the practical change is modest—or the reverse. RDJ’s guide to reading relative and absolute risk offers a useful framework for judging claims built around devices, medications, and screening.
Food responses should be interpreted with the same care. The goal is not to flatten every glucose curve by eliminating entire food groups. Diet quality, nutritional adequacy, sustainability, medications, and the person’s overall clinical picture matter. The discussion in Low-Carb Diets: Why Food Quality Matters Beyond the Carb Count applies here: one metric cannot carry the whole meaning of a diet.
Practical limitations and safety issues
Cost and coverage remain major barriers. So do skin irritation, alarm fatigue, information overload, device compatibility, and unequal access to professional support. People who are less comfortable with technology or who have limited English proficiency may need better training and follow-up, not merely a prescription.
CGM readings should not be treated as infallible. The National Institute of Diabetes and Digestive and Kidney Diseases advises users to understand their device and work with their healthcare team on targets and interpretation. The FDA has also warned that phone settings, operating-system changes, connected accessories, or notification permissions can interfere with diabetes-device alerts.
For people taking insulin or drugs that can cause hypoglycemia, acting on a false reading can be dangerous. Confirm a reading when symptoms do not match the sensor, when the device instructs you to check, or when a treatment decision carries meaningful risk.
A better decision question
“Should I get a CGM?” is too broad. A more useful question is: What decision would this information help me make, and what will I do if the data show a problem?
For someone using insulin, missing nighttime lows, or struggling with persistently high A1C, the answer may be clear. For someone without diabetes who is mostly curious about food responses, the benefits are less established and the risk of unnecessary restriction or anxiety is more real.
If you are considering CGM, bring a defined goal to your healthcare professional: detecting lows, evaluating medication timing, understanding fasting glucose, testing a practical change, or reviewing why A1C remains high. Agree on how long to monitor, which patterns matter, when to confirm with a meter, and what would count as a useful result.
Technology earns its place when it clarifies a decision. It does not become valuable merely because it can measure something.
Related reading
- GLP-1 Drugs in 2026: Every Major Medication, What the Research Shows, and What Comes Next
- Intermittent Fasting: When It Helps, When It Hurts, and When It’s Dangerous
- Risk Cut in Half? How to Read the Numbers Behind a Health Headline
Sources and further reading
- Karter AJ, et al. Continuous Glucose Monitoring and Long-Term Glycemic and Kidney Outcomes. JAMA Network Open. Published October 7, 2026.
- American Diabetes Association. Diabetes Technology: Standards of Care in Diabetes—2026. Diabetes Care. 2026.
- National Institute of Diabetes and Digestive and Kidney Diseases. Continuous Glucose Monitoring.
- U.S. Food and Drug Administration. Check Diabetes-Related Smartphone Device Alert Settings.
Disclaimer
Content on Real David Julian is provided for educational and informational purposes and may include research-based discussion, personal experience, and opinion. It is not medical advice and is not a substitute for individualized care from a qualified healthcare professional. Always discuss your specific health needs, diagnoses, medications, supplements, treatments, or major health decisions with an appropriate healthcare professional. See the full Disclaimer for additional information.