Screening, graded
A glucose monitor without diabetes: real data, no outcome evidence
Evidence grade C-. Arm-worn interstitial glucose sensors (Dexcom Stelo, Abbott Lingo, and app-layer programs) sold to adults without diabetes as metabolic insight.
- Outcome evidence: none. No randomized trial shows benefit in people without diabetes, and the 2026 JAMA Internal Medicine review found no scientific evidence of health improvement or disease prevention in this population.
- Accuracy: error of about 8 to 10% against reference, worse early in wear, with no validated interpretive thresholds for people without diabetes. Buyers pay about $89 a month to react partly to sensor error.
- Cost: About $89/month subscription (Stelo or Lingo, vendor figures), about $1,070/year continuous; app-layer programs add fees on top.
- Repeat cadence: the defensible pattern is a time-limited 2 to 4 week experiment, then removal; continuous wear is the vendor's pattern, and the record labels the basis that way.
Already have a result that worries you?
That includes a glucose trace that has you worried about diabetes. A sensor trace cannot diagnose it; the diagnosis is made on laboratory blood tests. Your next step is the clinician who ordered the test, or your own doctor, before anything else on this page. Bring the result itself, not a grade from a website.
The grade, axis by axis
Every screening grade on this site is built from the same five axes. The headline grade is outcome-change evidence discounted by harms: for this test, Zero outcome-change evidence caps at C+. Discounts below the cap: systematic over-reading in the direction that creates false alarm, no validated thresholds to act on, and a default cadence that converts one-off curiosity into about $1,070/year. Out of the D range because physical harm is minimal and a short, question-driven experiment is a legitimate self-measurement use.
1. Does early detection change outcomes?
NONE. No randomized trial shows CGM use by people without diabetes improves any health outcome or prevents diabetes or cardiovascular disease; a 2026 JAMA Internal Medicine evidence review concluded there is no scientific evidence of benefit in this population. The ADA Standards of Care make no recommendation for people without diabetes. FDA clearance (Stelo, March 5, 2024; Lingo, 2024) permits marketing and is not an outcome finding; Lingo is expressly not intended to diagnose diabetes.
2. How well does it detect, in people without symptoms?
MARD around 8 to 10% mid-range in filing data, worse at the extremes and in early wear (Lingo: 11.5% days 1 to 3, 8.5% by days 5 to 7), with interstitial lag up to about 15 minutes. In healthy users the physiology sits near the noise floor: healthy adults spend about 96% of time between 70 and 140 mg/dL (Shah et al., 2019), normal post-meal excursions overlap ranges labeled prediabetic (Hall et al., 2018), and a 2025 University of Bath randomized comparison found the FreeStyle Libre 2 systematically overestimated glucose in healthy adults, enough to make users believe they were prediabetic. There are NO validated interpretive thresholds for people without diabetes.
3. What does the harm ledger look like?
| Harm | Rate or scale | Source |
|---|---|---|
| Behavioral: over-restriction of nutritious foods, food anxiety, and orthorexia-pattern behavior in response to normal excursions | Review-level evidence (flagged at review level, 2024 reviews), not trial-quantified | Longevity Lab Report screening evidence review, 2026-10-18 |
| False alarm from systematic over-reading, and false reassurance from a flat trace (a trace cannot rule out insulin resistance and does not replace HbA1c) | Structural | University of Bath randomized comparison, Am J Clin Nutr, 2025 |
| Skin and adhesive reactions at the insertion site | The main documented device harm; uncommon | Longevity Lab Report screening evidence review, 2026-10-18 |
4. Does a positive result connect to an action that helps?
LOW. Nothing a healthy trace shows maps to an intervention with proven outcome benefit beyond generic advice (walk after meals, fiber first, sleep). Zeevi et al. (Cell, 2015) supports curiosity and short, question-driven self-experiments; it does not establish that continuous monitoring changes outcomes. Any diabetes-suggestive pattern still requires laboratory confirmation.
5. What does it cost, including the follow-through?
Per screen: About $89/month subscription (Stelo or Lingo, vendor figures), about $1,070/year continuous; app-layer programs add fees on top
Downstream: Mostly the user's own spending: foods and supplements bought or abandoned in response to the trace.
Repeat cadence: Continuous indefinite wear as sold; the defensible pattern is a time-limited 2 to 4 week experiment answering a specific question, then removal (basis: vendor-claim for continuous wear (subscription model); expert practice consensus, labeled as such, for the time-limited pattern)
On the same scale
This is the second measurement sold to healthy people on a subscription cadence. The first, the DEXA body composition scan, grades B as a measurement tool because its error is small, characterized, and stable on one scanner. The glucose monitor's error is larger, its targets for this population do not exist, and no outcome evidence stands behind it; that gap is the distance between B and C-. Both sit in the screening comparison table with the imaging tests.
Grades are research summaries, not medical advice. Decisions about your own stack belong with your clinician.
These pages grade a test in people without symptoms, as a population question. They never grade a reader, and they cannot read your result.
Longevity Lab Report
Screening news, graded when it happens.
The grades on this page stay free. The email is the week's longevity and biotech news in plain English: biotech news, expert moves, what is working, and new supplements and peptides, each story linked to the grades on this site.
Questions, answered plainly
Does a glucose monitor help if you do not have diabetes?
Graded C-. No randomized trial shows a continuous glucose monitor improves any health outcome in people without diabetes, and a 2026 evidence review in JAMA Internal Medicine concluded there is no scientific evidence that CGM improves health or prevents disease in this population. The American Diabetes Association makes no recommendation for it, and standard obesity guidance recommends against routine use outside research.
How accurate is the sensor?
Mean absolute relative difference runs about 8 to 10% against reference measurement, worse in the first day or two of wear and at high and low extremes, with interstitial lag of up to about 15 minutes behind blood glucose. In one trial in healthy adults, the main US system over-read glucose by about 56% at normal levels. There are no validated glucose targets for people without diabetes, so part of what buyers react to is sensor error, not physiology.
How long should you wear one?
The defensible pattern is a time-limited experiment: wear one for 2 to 4 weeks with a question written down in advance, then take it off. Continuous wear, the pattern the subscriptions sell, costs about $89 a month, roughly $1,070 a year, and no evidence supports monitoring past the point where your personal patterns are mapped.
Can a CGM diagnose diabetes or prediabetes?
No. Regulatory clearance for these devices is permission to sell them, not an outcome finding and not a diagnosis. Diabetes is diagnosed on laboratory blood tests. A trace that worries you is a reason to get the lab test with your clinician, not a result in itself.
Who is CGM actually proven for?
People with diabetes, insulin users especially, where trials show better control and fewer dangerous lows. That evidence is the reason the devices exist, and it does not transfer to people with normal glucose regulation, whose post-meal rises are ordinary physiology.
Sources
- Evidence review of CGM in people without diabetes. JAMA Internal Medicine, 2026 (University of Gothenburg group). (2026)
- University of Bath randomized comparison of FreeStyle Libre 2 in healthy adults. American Journal of Clinical Nutrition, 2025. (2025)
- Shah VN et al. Continuous glucose monitoring in healthy adults. 2019. (2019)
- Hall H et al. Glucotypes. PLoS Biology, 2018. (2018)
Last reviewed October 18, 2026
- 2026-10-18|Initial grading from the 2026-10-18 screening evidence review.