Continuous glucose monitors were built for people with diabetes, where they prevent genuine emergencies. They are now sold to people without it, on the promise that watching your blood sugar will reveal something important about your metabolism.
Here is the finding that should frame everything you see on that app: in a study of 153 healthy, non-diabetic people wearing blinded sensors, the median time spent above 140 mg/dL (7.8 mmol/L) was 2.1% of the day — around 30 minutes. Not zero. Healthy metabolisms go above that line every day.
So the question is not whether you spike. You do, everyone does, and it is what a working pancreas looks like. The useful question is what, if anything, the pattern tells you.
This article is general information, not medical advice. It is not about managing diagnosed diabetes, where CGMs are a well-established clinical tool. If you have symptoms of high or low blood sugar, a diabetes diagnosis, or concerns about your metabolic health, talk to a qualified healthcare professional.
What a CGM Actually Measures
The first thing worth knowing is that a CGM does not measure your blood.
The sensor filament sits in the interstitial fluid — the fluid between cells just under the skin — not in a blood vessel. Glucose reaches that fluid from the bloodstream by diffusion, which takes time. When your blood glucose moves quickly, the interstitial reading trails behind it.
That chain matters when you are interpreting a single reading. The number is a processed estimate of a delayed proxy — reliable enough to show patterns over hours, much weaker as a precise value at a single moment.
What "Normal" Actually Looks Like
This is the part most CGM marketing leaves out, and it is the most useful thing in this article.
A multicentre study published in The Journal of Clinical Endocrinology & Metabolism put blinded Dexcom G6 sensors on 153 healthy, non-diabetic participants aged 7 to 80 across twelve centres, for up to ten days. Blinded matters: the participants could not see their readings, so the data shows normal physiology rather than behaviour changed by watching.
What a Healthy Person's Glucose Actually Does
Reference values from 153 blinded, non-diabetic participants aged 7–80 (Shah et al., JCEM, 2019).
| Measure | What they found | What it means for you |
|---|---|---|
| MeasureMean glucose | What they found98–99 mg/dL (5.4–5.5 mmol/L) | What it means for you104 mg/dL in those over 60 |
| MeasureTime in 70–140 mg/dL | What they found96% of the day (median) | What it means for youNot 100% — gaps are expected |
| MeasureTime above 140 mg/dL | What they found2.1% — about 30 min a day | What it means for youGoing above 140 is normal |
| MeasureVariability (CV) | What they found17% ± 3% | What it means for youGlucose is supposed to move |
A healthy person spends roughly half an hour a day above 140
If your app flags every excursion above that line as an event, it is flagging normal physiology.
Read the third row again. The median healthy participant spent about half an hour a day above the threshold that consumer apps often treat as a warning.
And the fourth row is just as important. A coefficient of variation of 17% means healthy glucose moves around — it is not meant to be a flat line. An app that rewards flatness is rewarding something physiology does not do.
So Is That Spike a Problem?
Probably not, and there is research pointing the same way.
A 2018 study in PLOS Biology from a Stanford group fitted CGMs to people spanning normal glucose regulation through to diabetes. Their finding about the healthy group was striking: "even individuals considered normoglycemic by standard measures exhibit high glucose variability using CGM, with glucose levels reaching prediabetic and diabetic ranges 15% and 2% of the time, respectively."
People with entirely normal blood tests spent a meaningful share of their day in ranges that, taken as isolated readings, look alarming.
One honesty note on that paper. Its headline contribution was a system for sorting people into "glucotypes", and that classification drew a formal published critique, to which the authors responded and stood by their results. The variability observation is less contested than the clustering method, but it is fair to know the paper was argued over.
How Accurate Is the Number?
Manufacturers publish accuracy as MARD — mean absolute relative difference — and lower looks better. It is less comparable than it appears.
A review in the Journal of Diabetes Science and Technology examined exactly this, concluding that although comparing MARD figures between systems "appears to be straightforward on the first glance", a number of complex factors make the manufacturer-published numbers difficult to interpret. Their recommendation is direct: "The MARD should not be used as the sole parameter to characterize CGM systems."
The practical consequence for a non-diabetic user is simple. Your reading of 142 may correspond to a blood value meaningfully above or below it. Treating 139 and 145 as different events is reading noise as signal.
What a CGM Can Genuinely Tell You
None of this makes the device useless. It makes it useful for different things than the marketing suggests.
Your own responses, compared with each other. The same meal produces different responses in different people, and a CGM can show you your pattern. Noticing that white rice reliably does more to you than oats is a real, personal observation that no population average provides.
The effect of movement after eating. This one is consistently visible and consistently actionable — a walk after a meal usually blunts the curve, and seeing it happen is more persuasive than reading it. Our guide to desk-job fitness covers the trial evidence behind that effect.
Sleep and stress showing up in the data. Poor sleep and acute stress raise glucose through hormonal routes that have nothing to do with food, which is genuinely illuminating if you assumed diet was the only input. See how much sleep you actually need for what the evidence supports there.
Patterns, not points. Everything above requires looking at days and weeks. Nothing in that list requires reacting to a single reading.
Where It Can Mislead
Who Gets Something Out of This
Based on what the data can and cannot support, rather than on what is marketed.
Likely to be useful
- You want to compare your own meals
- You will look at weeks, not moments
- You are curious and not anxious
- You accept spikes are normal
- You have a specific question to answer
A time-limited experiment
Likely to mislead
- You expect a flat line
- You will cut foods on one reading
- You already worry about eating
- You are chasing an app score
- You expect it to detect disease
Cost without information
The common failure is narrowing the diet for no reason
Fruit, oats and whole grains all produce visible rises. Removing foods with good evidence behind them because a sensor drew a hill is a worse outcome than not measuring at all.
Two risks deserve naming plainly.
Food anxiety. A device that scores your meals, in a culture already loaded with dietary guilt, can push people toward restriction that no evidence supports. If watching the number makes eating stressful, that is a cost, and it is not a small one.
False reassurance, in both directions. A CGM is not a diabetes test. Normal-looking readings do not rule out a problem, and alarming-looking ones do not establish one. Diagnosis uses fasting glucose, HbA1c or a tolerance test, interpreted by a clinician — which is also what should follow if you genuinely suspect something is wrong.
If Not Spikes, Then What?
It is fair to ask what you should look at, if a wobbling line is not it.
The answer is that the established tests exist precisely because they avoid the problem a CGM has. Each one removes the variability rather than displaying it.
HbA1c measures the proportion of your haemoglobin that has glucose attached. Because red blood cells live around three months, the result reflects your average exposure over roughly that period. One restless night, one large dinner, one stressful morning — none of them move it. That is the entire point: it is deliberately immune to the noise a CGM shows you in real time.
Fasting plasma glucose takes a single reading under one standardised condition, which makes it comparable between people and across years in a way that "my glucose hit 160 after lunch" simply is not.
An oral glucose tolerance test gives everyone an identical challenge and measures the response, so differences reflect physiology rather than what happened to be on the plate.
All three have agreed interpretations set by national clinical guidelines, and all three are validated against actual long-term outcomes. A CGM trace in a healthy person has neither property — there is no established threshold for "too many spikes" because no one has shown that spikes within a normal range predict anything in people without diabetes.
There are useful non-blood signals too. Body composition is one, and waist-to-height ratio is a better everyday proxy than BMI. Sleep, activity and alcohol all affect glucose regulation, and all are easier to change than a reading.
If you want to know whether your metabolic health is fine, those tests answer that question. A sensor answers a different, narrower one.
So Should You Wear One?
If you are curious and the cost is comfortable, a time-limited trial is a reasonable thing to do. Two to four weeks, with a question in mind — how do I respond to my usual breakfast?, does a walk after dinner change anything? — gives you most of the available insight.
What is far harder to justify is wearing one indefinitely, hoping that continuous surveillance of a normal physiological variable will eventually reveal something. For a healthy person, the curve is mostly reporting that the system works.
If the question you actually want answered is "is my metabolic health okay?", the honest route is a blood test and a conversation, not a sensor and an app. For related wearable evidence, see do sleep trackers actually work and can a smartwatch detect heart problems; more in our health tech hub.
The Bottom Line
A continuous glucose monitor measures interstitial fluid, not blood, with a lag and a margin of error, and presents the result as a precise-looking number.
In people without diabetes, that number does what healthy physiology does: it rises after meals, spends about 30 minutes a day above 140 mg/dL, and varies by roughly 17% around its own mean. Research has repeatedly found that people with entirely normal blood work spend part of every day in ranges that look worrying out of context.
Used as a short, curious experiment to learn your own patterns, a CGM can teach you something real. Used as a continuous verdict on each meal, it mostly manufactures anxiety about a system that is working exactly as designed.
Frequently Asked Questions
Is it normal for blood sugar to spike after eating if I don't have diabetes?
Yes. A rise after eating is what a healthy metabolism does. In a study of 153 blinded, non-diabetic participants, the median time spent above 140 mg/dL (7.8 mmol/L) was 2.1% of the day — roughly 30 minutes. Healthy glucose also varies by about 17% around its own average. A flat line is not the goal and is not what healthy people produce.
Does a continuous glucose monitor measure my blood sugar?
Not directly. The sensor sits in interstitial fluid under the skin, not in a blood vessel, and glucose reaches that fluid from the blood by diffusion. That creates a lag, so rapid changes show up late. The device also measures an electrical signal that an algorithm converts into a glucose figure, so the number on screen is a processed estimate rather than a direct measurement.
Can a CGM tell me if I have diabetes or prediabetes?
No. Diagnosis uses fasting glucose, HbA1c or an oral glucose tolerance test, interpreted by a clinician. CGM readings in people without diabetes routinely enter ranges that look abnormal in isolation — one study found normoglycemic individuals reaching prediabetic ranges around 15% of the time — so the data cannot be read as a diagnosis in either direction.
How accurate are consumer glucose monitors?
Accuracy is usually published as MARD, but a review in the Journal of Diabetes Science and Technology concluded that manufacturer MARD figures are harder to compare than they appear and that MARD "should not be used as the sole parameter to characterize CGM systems." Practically, treat small differences between readings as noise rather than as distinct events.
Should I cut out foods that spike my glucose?
Be careful here. Fruit, oats, legumes and whole grains all produce visible rises while having good evidence behind them. Narrowing your diet because a sensor drew a hill is a common and avoidable mistake. If you want to use the data, compare your own options against each other over weeks rather than eliminating foods after a single reading.
Is it worth wearing one if I'm healthy?
Possibly, as a short experiment with a specific question — how you respond to your usual breakfast, or whether walking after dinner changes your curve. Two to four weeks will show you most of what there is to see. Wearing one indefinitely in the hope of detecting something is harder to justify, and carries a real risk of food anxiety for some people.
Sources
- Shah et al. — Continuous Glucose Monitoring Profiles in Healthy Nondiabetic Participants: A Multicenter Prospective Study, The Journal of Clinical Endocrinology & Metabolism (2019)
- Hall et al. — Glucotypes reveal new patterns of glucose dysregulation, PLOS Biology (2018)
- Breschi, Perelman & Snyder — Response to Hulman and colleagues regarding "Glucotypes reveal new patterns of glucose dysregulation", PLOS Biology (2021)
- Heinemann et al. — Benefits and Limitations of MARD as a Performance Parameter for Continuous Glucose Monitoring in the Interstitial Space, Journal of Diabetes Science and Technology (2020)
