Name: CGMs for Non-Diabetics: A Biofeedback Toy, Not a Diagnostic
Creator: Realised
Published: 2026-06-01

Summary

For a healthy non-diabetic, a continuous glucose monitor is an n-of-1 biofeedback toy, not a diagnostic — it accurately shows that brief post-meal glucose rises are normal physiology, but there is no evidence that acting on those readings improves any hard health outcome in people with normal glucose, no validated interpretive thresholds inside the normal range, and a real risk of manufacturing spike-anxiety or orthorexia; useful as a time-limited awareness experiment (Tier 3), while the wellness claims that "spikes harm healthy people" or that "flattening your curve" makes you healthier are u

Why Emerging

Tier 3 (Emerging) for the load-bearing claim — a CGM is a useful behavioural-awareness tool for some healthy people — because the supporting evidence is a motivational/biofeedback signal and short-term self-experimentation, not outcome trials. The descriptive physiology (normal-range data) and the prediabetes benefit are stronger (Tier 1–2), but those aren't the claim this entry adjudicates for non-diabetics.

NOT Tier 2 because there is no hard-outcome or durable glycaemic benefit demonstrated in normoglycemic people; the value rests on awareness/motivation, which is genuinely emerging.
NOT Tier 4 because the descriptive accuracy (it does reveal real, normal physiology) and the prediabetes benefit are solid; this isn't speculative.

(The wellness claims this entry rejects — "spikes harm healthy people," "flattening your curve makes you healthier" — sit at Tier 4 / unsupported. The risk evidence, conversely, is more concrete than the benefit evidence: the orthorexia downside is documented while the optimisation upside is not.)

Practical takeaway

• If you're curious, run it as a one-off 2–4 week experiment, then stop. Use it to learn your own meal, exercise, and sleep responses — it is an awareness tool, not a vital sign to track forever. Permanent monitoring of a normal number is the failure mode the subscription is built on.
• Expect normal. Brief rises that settle within 2–3 hours are healthy. A reading touching 140–160 after a big carbohydrate meal is not a red flag in a non-diabetic — there are no validated "you must fix this" thresholds inside the normal range.
• Treat the number as approximate. Sensor error is ~10%+ and the reading lags real blood glucose, so don't agonise over small differences between meals — much of it is noise.
• Highest-value use: confirm a behaviour you already suspect helps, rather than hunting for foods to eliminate. Watching a 10–15 minute post-meal walk blunt your rise (see post_meal_walks_glucose) is a constructive, motivating use; building a list of "bad" foods to avoid is not.
• Watch the failure mode. If the device is making you anxious, avoid foods, or moralise about eating, it is harming more than helping — take it off. This is a real orthorexia risk, not a caution to recite and ignore.
• If you have genuine dysglycaemia risk (prediabetes, strong family history, PCOS), the evidence is better — but pair the CGM with a clinician or a structured program, which is the only context where the benefit actually showed up.

Evidence detail

Why This Entry Exists

A £100–200/month sensor that used to be a prescription device for diabetics is now sold over the counter to healthy people as a window into their "metabolic health." The pitch is seductive: see your glucose in real time, find the foods that "spike" you, flatten your curve, optimise. An entire subscription industry — Levels, Nutrisense, Lingo, Stelo, often bundled with an app and a dietitian upsell — has been built on reframing normal glucose physiology as a problem that needs continuous monitoring.

The honest picture is less exciting. A CGM works as advertised in one narrow sense: it accurately reveals the shape of your glucose, and for a healthy person that shape is reassuringly normal. What it does not do is improve anything. There is no trial showing CGM use makes a normoglycemic person healthier, no validated way to interpret a reading inside the normal band, and a concrete downside that independent academics — not just skeptics — keep flagging: the device is "expensive, appealing and addictive," and it can push some users toward disordered eating. So this entry exists to separate the real, modest value (a short biofeedback experiment that can motivate a behaviour you already suspect helps) from the marketing (continuous monitoring as a vital sign for the well).

What bad advice this protects against, in both directions:
• Treating a CGM as a health-optimising vital sign to track forever → you pay a subscription to monitor a number that, in a healthy person, the device most reliably proves is fine.
• Dismissing it as useless gimmickry full stop → as a 2–4 week experiment it genuinely can teach you your own meal/walk/sleep responses, and in real dysglycaemia (prediabetes) the evidence is better.
• Reading any value above 140 mg/dL as a "spike to fix" → there are no validated non-diabetic thresholds; brief excursions are normal physiology, and chasing them is where the orthorexia risk lives.

It does not own the underlying metabolic biology (see insulin_resistance_and_metabolic_dysfunction), the lifestyle-first glycaemic levers (blood_sugar_regulation), or whether spikes themselves matter (postprandial_glucose_spikes). It owns the device decision for a healthy person: does a CGM help you, what is it actually good for, and where it does harm.

Evidence

1. The shape it reveals is reassuringly normal (Tier 1 descriptive). In 153 healthy non-diabetics wearing CGMs (Metabolism, 2023), mean glucose was 98–99 mg/dL, median time-in-range (70–140 mg/dL) was 96%, and time above 140 was only ~2.1% — roughly 30 minutes a day. Transient rises that peak around an hour after eating and settle within 2–3 hours are normal physiology, not a malfunction. The single most reliable thing a CGM tells a healthy person is that their glucose is fine.

2. No glycaemic benefit in normoglycemic people (Tier 1–2). A 2026 systematic review and meta-analysis (European Journal of Medical Research; 23 studies, 1,074 non-diabetic participants, 7 RCTs) found CGM improved mean glucose only in prediabetes (SMD −0.54). "No appreciable glycemic benefit was observed in healthy normoglycemic populations," and there was no BMI benefit (SMD −0.25, not significant). The benefit is concentrated in dysglycaemia, not in normal glucose.

3. Zero hard-outcome evidence (Tier 1, absence). A separate systematic review of CGM for cardiovascular prevention in non-diabetics (32 studies) found no study showing CGM reduces cardiovascular events, mortality, or disease — only an associational "exercise motivation" signal. The authors' own words: "scientific proof… is lacking." There is no outcome trial behind the optimisation pitch.

4. The regulatory clearance is for safety-to-sell, not benefit (Tier 1, governance). Johns Hopkins Bloomberg School of Public Health (2026) notes the FDA cleared over-the-counter CGMs as consumer devices without requiring proof they improve health outcomes in healthy people. "FDA-cleared" here means "safe to sell," not "shown to help" — a distinction the marketing elides.

5. The disordered-eating risk is flagged by independents, not just skeptics (Tier 2). Because there are no established interpretive values inside the normal 70–140 mg/dL band, non-diabetics "may misinterpret and over-emphasise" normal readings, leading to "disordered eating behaviours or the complete avoidance of some foods" (Johns Hopkins). The scoping review (PMC12569367) raises "glucose-centricity" and orthorexia explicitly and calls the device "expensive, appealing and addictive." This is a documented downside, not a hypothetical.

6. Real value as a time-limited awareness tool (Tier 3). The genuine n-of-1 uses are seeing your own response to a specific meal, the effect of a post-meal walk, or how sleep and stress move your glucose — feedback that can motivate behaviour change. The best-supported framing (2026 meta-analysis) is "a precision biofeedback tool integrated within structured lifestyle programs," not a standalone monitor worn indefinitely.

Mechanism

What the device measures, and the lag. A CGM samples glucose in the interstitial fluid, not blood, via a subcutaneous sensor. Interstitial glucose trails blood glucose by several minutes because the molecule has to diffuse out of the capillaries — so the curve you watch is a delayed, smoothed proxy, sharpest exactly when you care most (right after eating).

Why the normal range is hard on accuracy. Even validated sensors run a mean absolute relative difference (MARD) of roughly 9–14% against a lab reference (FreeStyle Libre 3 ~8.9% vs Dexcom G7 ~13.6%, measured in diabetics, where the glucose range is wide). A non-diabetic operates in a narrow ~70 mg/dL band. A 10%+ error plus the diffusion lag means a good fraction of the "spikes" a wellness user agonises over are partly sensor noise, and some studies show CGMs overestimate glucose, especially post-meal. The signal-to-noise ratio is worst precisely in the population being sold the device.

Why a transient rise is not damage. Glucose rising after a carbohydrate meal and returning to baseline within 2–3 hours is insulin doing its job — normal physiology. The harm associated with glucose is chronic, repeated hyperglycaemia (sustained elevation, as in diabetes), not isolated normal postprandial excursions. The scoping review found that the harm claims circulating in wellness/grey literature — fat storage, brain fog, energy crashes, skin ageing, anxiety, cancer — are largely absent from the medical literature, and that 63.6% of the supporting mechanistic studies were done in endothelial cells, not living humans.

Why "flattening the curve" lacks an outcome. There is no demonstrated causal chain from "a healthy person's post-meal rise is smaller" to "that person is healthier." That link is the core sales claim and it is exactly the link with no trial behind it. (Whether postprandial spikes matter at all is covered in postprandial_glucose_spikes.)

Risks And Contraindications

• Disordered eating / orthorexia — the headline risk. With no interpretive thresholds in the normal range, the device invites over-reaction to normal readings, food avoidance, and "glucose-centric" eating. People with a history of, or tendency toward, disordered eating should generally avoid wellness CGM use.
• False reassurance and false alarm both. Sensor error (~10%+ MARD) and interstitial lag mean a non-diabetic's narrow range is mostly noise; readings can over- or under-state true blood glucose, especially post-meal. Acting on small differences is acting on noise.
• Cost with no demonstrated payoff. A recurring ~£100–200 / US$129–199 monthly subscription buys monitoring with no proven health-outcome benefit in healthy people. The opportunity cost (money, attention) is real.
• Not a diagnostic. A wellness CGM cannot diagnose diabetes or prediabetes; concerning patterns warrant proper clinical testing (HbA1c, fasting glucose, OGTT), not self-interpretation of a consumer sensor.
• Over-medicalising normal physiology. Turning every meal into a monitored event can manufacture a health problem where none exists — the thing the device most reliably reveals is that you are fine.

Controversy

Nature: commercial / wellness-marketing, with overstatement to defend against at both poles.

Position A — "A CGM is essential for optimising your metabolic health; spikes are harmful and you should flatten your curve." The wellness-subscription pitch.
• Best evidence: a CGM does accurately show your own glucose responses, and that biofeedback can motivate genuinely good behaviours (a post-meal walk, less late-night sugar).
• Where it's wrong: there is no outcome trial showing benefit in healthy people; "spikes harm the well" is largely absent from the medical literature and leans on cell studies; and the orthorexia/anxiety downside is concrete. The "FDA-cleared" halo means safe-to-sell, not shown-to-help.

Position B — "It's a useless gimmick; ignore it entirely." The reflexive-skeptic take.
• Best evidence: in normoglycemic people there's no glycaemic or BMI benefit and no hard outcomes; the marketing is overblown.
• Where it's wrong: as a short n-of-1 experiment it can genuinely teach behaviour, and in prediabetes the RCT signal is real (SMD −0.54). Dismissing the device wholesale over-corrects past the evidence.

The funding/bias dimension — cui bono, both ways. OTC/wellness CGM is a recurring-revenue subscription business (Levels, Nutrisense, Lingo, Stelo), often bundled with an app and dietitian upsell; the incentive is to reframe normal glucose physiology as a problem requiring continuous monitoring, and the "FDA-cleared" framing borrows medical credibility the benefit data don't support. The counter-skeptic point: the harm is not the device — it's accurate enough and physically safe — it's the narrative sold around it. Independent academics (Harvard Health, Johns Hopkins) are skeptical precisely because the healthy-person benefit evidence is scant while the disordered-eating downside is concrete; none of them is selling a sensor or a fasting app.

Realised Position: For a healthy non-diabetic, a CGM is a biofeedback toy, not a diagnostic. Worth a one-off 2–4 week experiment if you're curious and not prone to food anxiety — to learn your responses and confirm a habit like a post-meal walk — then take it off. It will not make you healthier to wear forever, the "spikes harm you / flatten your curve" claims have no outcome evidence, and the orthorexia risk is real. The honest verdict is "an interesting short experiment for some, a permanent vital sign for no one" — not "essential," not "useless."

Cross-Pillar Connections

• Diet (blood_sugar_regulation): the lifestyle-first glycaemic levers that actually move glucose with no device or subscription — the foundation a CGM should serve, not replace.
• Diet (insulin_resistance_and_metabolic_dysfunction): the underlying metabolic biology; this entry owns the device decision, that entry owns the dysfunction a CGM is genuinely useful for monitoring.
• Diet (postprandial_glucose_spikes): whether the spikes a CGM displays actually matter — the question upstream of "should I buy the monitor."
• Diet (post_meal_walks_glucose): the highest-value real CGM use — confirming that a short post-meal walk blunts the rise — and a behaviour worth doing whether or not you own a sensor.
• Diet (biomarker_tracking): the general principle of which numbers are worth tracking and when measurement helps versus manufactures anxiety; CGM-for-the-well is a case study in over-measurement.

What would change our mind

Falsifiability: explicit upgrade/downgrade criteria from source

• We'd upgrade toward Tier 2 if RCTs in normoglycemic people showed CGM use produced durable improvements in a meaningful outcome (weight, cardiometabolic markers, established behaviour change), not just self-reported motivation. Current trials show benefit only in prediabetes.
• We'd take the harm claims seriously if human (not cell-line) studies demonstrated that isolated, normal-range postprandial rises cause measurable harm in healthy people. Current support is dominated by endothelial-cell work and absent from clinical outcomes.
• We'd soften the accuracy caution if non-diabetic-validated sensors reached a MARD low enough that within-normal-range differences exceeded sensor noise. Current MARD (~9–14%) is too high for the narrow band.
• What would NOT move us: the interstitial lag, the absence of validated normal-range thresholds, and the documented orthorexia risk — these are structural to the device-and-population fit, not artefacts of immature data.

Industry bias note

Structural incentives the evidence base may reflect

This is a topic with commercial pressure mostly at one end, which is why the independent academic voices are the anchor.
• The wellness-CGM end: the subscription model (Levels, Nutrisense, Lingo, Stelo) profits from reframing normal glucose as a problem needing continuous, recurring-fee monitoring. The "FDA-cleared" line is commercially convenient because it borrows clinical authority the benefit data don't earn — clearance was for safety-to-sell, not proof of help. The "find your spike foods, flatten your curve" loop is designed to keep you wearing (and re-buying) the sensor.
• The reflexive-skeptic end: a smaller pull toward "useless gimmick" that over-corrects past the real prediabetes signal and the genuine biofeedback value.
• The clean signal: independent academic-clinical commentary — Harvard Health ("Is blood sugar monitoring without diabetes worthwhile?") and Johns Hopkins (2026) — plus the independent meta-analyses and the scoping review. None sells a sensor or a fasting program; all land on "evidence for healthy-person benefit is scant, the orthorexia downside is concrete, save it for dysglycaemia or a short experiment." Realised weights those over both the subscription marketing and the wholesale dismissal.

CGMs for Non-Diabetics: A Biofeedback Toy, Not a Diagnostic