Place · Level 3 · Metabolic health
CGM for the Metabolically Healthy
非糖尿病血糖生理 · 餐后波动是正常的 · Shah 2019 健康人 96% 时间 70-140 · 拉平曲线 营销拆解 · 何时真有用
Story path
- 1CGM & the normal curveCGM & the normal curve
- 2Brake & gas · glucose controlBrake & gas · glucose control
- 3What variability predictsWhat variability predicts
- 4Flatten the curve · what's realFlatten the curve · what's real
- 5When CGM truly helpsWhen CGM truly helps
- 6The cost of over-monitoringThe cost of over-monitoring
Chapter 1
CGM & the normal curve
CGM & the normal curve
You stick a small disc on your upper arm and a glucose curve pops up on your phone — eat a bowl of noodles and it climbs, then a while later it settles back down. Many people see that rising line and panic: is my blood sugar too high? Am I quietly becoming diabetic? Put the conclusion up front: for a metabolically healthy person, glucose going up after a meal is what a normally working body looks like, not a malfunction.
That little disc is a continuous glucose monitor (CGM). It doesn't prick your finger for blood; it leaves an ultra-thin probe under the skin and measures glucose in the interstitial fluid, giving a reading every 1-15 minutes that joins into a curve. Because it reads under the skin rather than in a vessel, there is a lag of several to fifteen-odd minutes versus true fingertip blood, and the number is not exactly your blood glucose — a point we'll return to, and an important one.
So what does a normal curve actually look like? A multicenter study of 153 healthy non-diabetic people wearing CGM (Shah 2019) gives a clear baseline: they spent about 96% of the time with glucose between 70-140 mg/dL (3.9-7.8 mmol/L); mean glucose was about 98-99 mg/dL; time above 140 across the whole day was only about 2%, roughly 30 minutes; and post-meal peaks rarely exceeded 140 and almost never 160. In other words: a healthy person's glucose rising to 120-140 after a meal and coming back down is something this system does every day, and does steadily.
So this island isn't about don't let your glucose move — it's about how much movement is normal, how the body reins it in, which flatten-the-curve claims have real substance, which are marketing overreach, and whether a healthy person should wear one at all.
That little disc is a continuous glucose monitor (CGM). It doesn't prick your finger for blood; it leaves an ultra-thin probe under the skin and measures glucose in the interstitial fluid, giving a reading every 1-15 minutes that joins into a curve. Because it reads under the skin rather than in a vessel, there is a lag of several to fifteen-odd minutes versus true fingertip blood, and the number is not exactly your blood glucose — a point we'll return to, and an important one.
So what does a normal curve actually look like? A multicenter study of 153 healthy non-diabetic people wearing CGM (Shah 2019) gives a clear baseline: they spent about 96% of the time with glucose between 70-140 mg/dL (3.9-7.8 mmol/L); mean glucose was about 98-99 mg/dL; time above 140 across the whole day was only about 2%, roughly 30 minutes; and post-meal peaks rarely exceeded 140 and almost never 160. In other words: a healthy person's glucose rising to 120-140 after a meal and coming back down is something this system does every day, and does steadily.
So this island isn't about don't let your glucose move — it's about how much movement is normal, how the body reins it in, which flatten-the-curve claims have real substance, which are marketing overreach, and whether a healthy person should wear one at all.
Chapter 2
Brake & gas · glucose control
Brake & gas · glucose control
If a healthy person's glucose goes up and comes back down after a meal, then who pulls it back? It's a two-way hormonal brake-and-gas system.
Insulin is the brake: after a meal, glucose enters the blood, pancreatic beta cells release insulin, and muscle, liver, and fat move glucose into cells while the liver stops releasing glucose and stores the surplus as glycogen. Glucose then falls.
Glucagon is the gas: when glucose runs low (fasting, exercise), pancreatic alpha cells release glucagon, and the liver breaks down stored glycogen to put glucose back into the blood, holding it up so it doesn't drop too far.
A set of incretins (such as glucagon-like peptide-1: A gut hormone released after eating that makes you feel full and helps lower blood sugar.) is also released from the gut when you eat, prompting insulin, suppressing glucagon, and slowing gastric emptying — making the post-meal wave smoother (Dimitriadis 2021).
The default result of this system: a healthy person's post-meal glucose peaks about 1 hour after the first bite and returns to baseline within 2-3 hours. That rise-and-fall hillock you see on a CGM is evidence of the brake and gas working in real time — not a signal that they're broken.
Here's the key distinction: diabetes is this system failing — not enough insulin (type 1) or cells not listening to insulin (type 2) — so glucose climbs and can't come back, sitting high for the long term. A healthy person's curve rises but returns cleanly. Same rising line, but one is a broken system and one is a system doing its job — you can't judge from the climb alone.
Insulin is the brake: after a meal, glucose enters the blood, pancreatic beta cells release insulin, and muscle, liver, and fat move glucose into cells while the liver stops releasing glucose and stores the surplus as glycogen. Glucose then falls.
Glucagon is the gas: when glucose runs low (fasting, exercise), pancreatic alpha cells release glucagon, and the liver breaks down stored glycogen to put glucose back into the blood, holding it up so it doesn't drop too far.
A set of incretins (such as glucagon-like peptide-1: A gut hormone released after eating that makes you feel full and helps lower blood sugar.) is also released from the gut when you eat, prompting insulin, suppressing glucagon, and slowing gastric emptying — making the post-meal wave smoother (Dimitriadis 2021).
The default result of this system: a healthy person's post-meal glucose peaks about 1 hour after the first bite and returns to baseline within 2-3 hours. That rise-and-fall hillock you see on a CGM is evidence of the brake and gas working in real time — not a signal that they're broken.
Here's the key distinction: diabetes is this system failing — not enough insulin (type 1) or cells not listening to insulin (type 2) — so glucose climbs and can't come back, sitting high for the long term. A healthy person's curve rises but returns cleanly. Same rising line, but one is a broken system and one is a system doing its job — you can't judge from the climb alone.
Chapter 3
What variability predicts
What variability predicts
Marketing's favorite line: even if your labs are normal, CGM can reveal you're quietly losing control. It has a real origin, but it's been inflated.
The real origin: Stanford's Hall 2018 study (57 people) proposed the concept of a glucotype — some people who are normal by standard tests still, on CGM, spike into the prediabetic (~15% of time) or even diabetic (~2%) range. This got used to argue that everyone is secretly dysglycemic.
Where it's inflated, point by point:
The classification itself drew serious peer critique: a 2021 formal comment in PLoS Biology (Hulman 2021) argued that glucotypes' reproducibility, interpretability, and relationship to traditional CGM metrics are all shaky — it reads more like an exploratory finding than a mature tool for labeling healthy people.Consumer CGMs systematically read high: a 2025 randomized crossover trial (Hutchins 2025, University of Bath) compared head-to-head — CGM-estimated fasting and post-meal glucose were about 0.9 mmol/L higher than fingertip blood; it overestimated time above 7.8 mmol/L (140 mg/dL) by roughly 2-4 fold; and for the same smoothie, CGM reported a glycemic index of 69 (medium) versus 53 (low) by the standard method. In other words, a good share of the scary spikes a healthy person sees are the device reading high, not the body actually swinging that much.
So what does glycemic variability predict in a healthy person? The most honest answer today: the evidence is weak and mixed. A 2024 systematic review and meta-analysis (Hjort 2024) pooled non-diabetic data and found that variability was not clearly associated with insulin sensitivity, fatty liver, adiposity, blood lipids, blood pressure, or oxidative stress; only some coronary-atherosclerosis markers showed a hint of signal, and variability was higher in people whose glucose regulation was already impaired. More to the point: there is no evidence that a healthy person deliberately flattening their variability buys fewer diseases or a longer life. Variability itself is not a disease.
In one line: seeing variability is not the same as finding disease. A healthy curve naturally undulates, the device tends to read high on top of that, and treating every little peak as an alarm is being framed by marketing.
The real origin: Stanford's Hall 2018 study (57 people) proposed the concept of a glucotype — some people who are normal by standard tests still, on CGM, spike into the prediabetic (~15% of time) or even diabetic (~2%) range. This got used to argue that everyone is secretly dysglycemic.
Where it's inflated, point by point:
The classification itself drew serious peer critique: a 2021 formal comment in PLoS Biology (Hulman 2021) argued that glucotypes' reproducibility, interpretability, and relationship to traditional CGM metrics are all shaky — it reads more like an exploratory finding than a mature tool for labeling healthy people.Consumer CGMs systematically read high: a 2025 randomized crossover trial (Hutchins 2025, University of Bath) compared head-to-head — CGM-estimated fasting and post-meal glucose were about 0.9 mmol/L higher than fingertip blood; it overestimated time above 7.8 mmol/L (140 mg/dL) by roughly 2-4 fold; and for the same smoothie, CGM reported a glycemic index of 69 (medium) versus 53 (low) by the standard method. In other words, a good share of the scary spikes a healthy person sees are the device reading high, not the body actually swinging that much.
So what does glycemic variability predict in a healthy person? The most honest answer today: the evidence is weak and mixed. A 2024 systematic review and meta-analysis (Hjort 2024) pooled non-diabetic data and found that variability was not clearly associated with insulin sensitivity, fatty liver, adiposity, blood lipids, blood pressure, or oxidative stress; only some coronary-atherosclerosis markers showed a hint of signal, and variability was higher in people whose glucose regulation was already impaired. More to the point: there is no evidence that a healthy person deliberately flattening their variability buys fewer diseases or a longer life. Variability itself is not a disease.
In one line: seeing variability is not the same as finding disease. A healthy curve naturally undulates, the device tends to read high on top of that, and treating every little peak as an alarm is being framed by marketing.
Myth · 'glucose spikes cause all disease'
Glucose spikes are the root of all disease — inflaming you, aging you, storing fat, causing diabetes is the core line of this business. It transplants something true in diabetes onto healthy people, swapping the premise along the way.In diabetes, chronic, repeated, can't-come-back high glucose genuinely has a causal link to vascular, nerve, kidney, and eye complications — that evidence is solid and not in dispute. But a healthy person's post-meal peak of 120-140 that falls back within two hours is a completely different thing: brief, controlled, happening every day, the system working normally.
The three swaps:
Swapping chronic, uncontrolled hyperglycemia is harmful for any post-meal rise is harmful;Swapping variability correlates with complications in diabetics for variability is quietly causing disease in healthy people too — when Hjort 2024 shows precisely that this link is weak in healthy people;Calling correlation causation, then reversing it to sell you a promise that flattening it prevents disease — a promise with no outcome evidence in healthy people.
The real, usable stance: if you already have prediabetes or diabetes, controlling post-meal glucose is a meaningful clinical goal (next scene). But if you're metabolically healthy, your glucose hitting 140 after a mango isn't you getting sick — it's the mango having sugar and your body handling it normally. What actually shapes long-term metabolic health is weight, activity, fiber, sleep, not smoking — the big levers — not the height of any single meal's curve.
Chapter 4
Flatten the curve · what's real
Flatten the curve · what's real
Glucose marketing does contain real things — and precisely because there's real signal, it's easy to inflate its weight. Take the most common flatten-the-curve hacks and honestly mark which part is real, and how real.
1. Food order (vegetables and protein first, carbs last) — this is a real effect. Shukla 2015 (Diabetes Care) had people with type 2 diabetes eat the same meal, only putting vegetables and protein before the carbs, and the area under the 120-minute post-meal glucose curve was about 73% lower; it's been reproduced in prediabetes and healthy young adults. The mechanism is protein and fiber slowing gastric emptying and pulling on incretins. But: this is a number over a few post-meal hours, the sample was tiny (Shukla was 11 people), and there's no evidence a healthy person buys fewer diseases by reordering their food. It's a good habit, not a medicine.
2. Walk after meals — also a real effect. Buffey 2022 (Sports Medicine) meta-analysis showed that interrupting sitting with standing, and especially light walking, clearly lowers post-meal glucose and insulin; even 2-5 minutes of walking after a meal does a little. And walking's benefits go far beyond glucose — this advice holds even if you never look at a CGM.
3. Different people respond differently to the same meal — this observation is true. The PREDICT study (Berry 2020, Nature Medicine, 1,002 people) found that on identical meals, people's post-meal glucose varied by about 68%, driven by meal composition, the individual, and partly the gut microbiome. This is the scientific bedrock of the personalized nutrition business.
But sort out what's built on that bedrock: individual variability is real, but that doesn't mean every healthy person needs a CGM to eat well. The one place personalization has been built into a full intervention with harder outcomes is ZOE's METHOD randomized controlled trial (Bermingham 2024, Nature Medicine): a personalized dietary program improved triglycerides and self-rated wellbeing versus general dietary advice — note, that's the effect of a whole lifestyle program (not a CGM alone), the comparator was generic advice, and the outcomes were intermediate markers, not disease rates. It shows personalization has some signal, but it's nowhere near proving a healthy person must wear a CGM.
To close: what these hacks share is a real but brief, small effect on the post-meal curve; and the part of them with genuine long-term value (eat more vegetables and protein, walk more) is already good advice that doesn't need a CGM to approve it. At most, a CGM lets you see these effects — but seeing them isn't the same as needing to.
1. Food order (vegetables and protein first, carbs last) — this is a real effect. Shukla 2015 (Diabetes Care) had people with type 2 diabetes eat the same meal, only putting vegetables and protein before the carbs, and the area under the 120-minute post-meal glucose curve was about 73% lower; it's been reproduced in prediabetes and healthy young adults. The mechanism is protein and fiber slowing gastric emptying and pulling on incretins. But: this is a number over a few post-meal hours, the sample was tiny (Shukla was 11 people), and there's no evidence a healthy person buys fewer diseases by reordering their food. It's a good habit, not a medicine.
2. Walk after meals — also a real effect. Buffey 2022 (Sports Medicine) meta-analysis showed that interrupting sitting with standing, and especially light walking, clearly lowers post-meal glucose and insulin; even 2-5 minutes of walking after a meal does a little. And walking's benefits go far beyond glucose — this advice holds even if you never look at a CGM.
3. Different people respond differently to the same meal — this observation is true. The PREDICT study (Berry 2020, Nature Medicine, 1,002 people) found that on identical meals, people's post-meal glucose varied by about 68%, driven by meal composition, the individual, and partly the gut microbiome. This is the scientific bedrock of the personalized nutrition business.
But sort out what's built on that bedrock: individual variability is real, but that doesn't mean every healthy person needs a CGM to eat well. The one place personalization has been built into a full intervention with harder outcomes is ZOE's METHOD randomized controlled trial (Bermingham 2024, Nature Medicine): a personalized dietary program improved triglycerides and self-rated wellbeing versus general dietary advice — note, that's the effect of a whole lifestyle program (not a CGM alone), the comparator was generic advice, and the outcomes were intermediate markers, not disease rates. It shows personalization has some signal, but it's nowhere near proving a healthy person must wear a CGM.
To close: what these hacks share is a real but brief, small effect on the post-meal curve; and the part of them with genuine long-term value (eat more vegetables and protein, walk more) is already good advice that doesn't need a CGM to approve it. At most, a CGM lets you see these effects — but seeing them isn't the same as needing to.
Chapter 5
When CGM truly helps
When CGM truly helps
Debunking CGM into a healthy person doesn't need to chase the curve does not mean it's useless — on the contrary, for the people who should use it, it's one of the most substantial tools in diabetes care of recent years. The point is using it on people whose system has already failed.
Per the ADA 2025 Standards of Care (Diabetes Technology chapter), CGM is explicitly recommended for:
Type 1 diabetes — even from the moment of diagnosis. Their bodies make almost no insulin and must use injected insulin to make the body's decisions for it; CGM's real-time numbers and high/low alarms bear directly on safety.Anyone with diabetes on insulin (type 1 or 2) — CGM helps titrate doses, catch nighttime lows, and see which meals spike hardest.Type 2 diabetes on non-insulin glucose-lowering drugs — from the 2025 edition, this is also included as consider, to reach individualized glucose targets.
Other genuinely valuable settings:
Prediabetes — using it to see which foods and activities best pull glucose back into the normal range is clinically meaningful feedback.Research — CGM is a fine instrument for studying post-meal metabolism, individual variation, and drug effects (many of the studies cited here rely on it).Diagnostic uncertainty — clinicians sometimes use it to catch high/low patterns labs miss.
See the distinction? What these people share is a glucose system that is already broken or being treated, so the information CGM gives actually changes a decision (how much insulin, whether to switch drugs, whether to seek care). For a metabolically healthy person, however pretty the curve, it rarely changes anything they should do — eat more vegetables anyway, walk more anyway. Information is only worth collecting when it can change an action.
Important reminder: if you already have diagnosed diabetes or prediabetes, this island is about healthy people and is not aimed at you; follow your clinician and these guidelines, and do not stop prescribed monitoring or medication just because you read a popular-science piece.
Per the ADA 2025 Standards of Care (Diabetes Technology chapter), CGM is explicitly recommended for:
Type 1 diabetes — even from the moment of diagnosis. Their bodies make almost no insulin and must use injected insulin to make the body's decisions for it; CGM's real-time numbers and high/low alarms bear directly on safety.Anyone with diabetes on insulin (type 1 or 2) — CGM helps titrate doses, catch nighttime lows, and see which meals spike hardest.Type 2 diabetes on non-insulin glucose-lowering drugs — from the 2025 edition, this is also included as consider, to reach individualized glucose targets.
Other genuinely valuable settings:
Prediabetes — using it to see which foods and activities best pull glucose back into the normal range is clinically meaningful feedback.Research — CGM is a fine instrument for studying post-meal metabolism, individual variation, and drug effects (many of the studies cited here rely on it).Diagnostic uncertainty — clinicians sometimes use it to catch high/low patterns labs miss.
See the distinction? What these people share is a glucose system that is already broken or being treated, so the information CGM gives actually changes a decision (how much insulin, whether to switch drugs, whether to seek care). For a metabolically healthy person, however pretty the curve, it rarely changes anything they should do — eat more vegetables anyway, walk more anyway. Information is only worth collecting when it can change an action.
Important reminder: if you already have diagnosed diabetes or prediabetes, this island is about healthy people and is not aimed at you; follow your clinician and these guidelines, and do not stop prescribed monitoring or medication just because you read a popular-science piece.
Chapter 6
The cost of over-monitoring
The cost of over-monitoring
Finally, one thing the marketing page never writes: for a healthy person, wearing a CGM has its own downsides, and they're not trivial.
A 2024 narrative review of CGM use in people not living with diabetes (Oganesova 2024, Diabetic Medicine) puts it bluntly: there is almost no evidence that CGM is accurate in healthy people or delivers any health benefit; instead it can bring a set of real harms:
Misreading normal fluctuation as a defect: many people believe you shouldn't rise at all after eating, so a normal post-meal peak triggers anxiety — when we already know that rising and falling back is exactly what healthy looks like.Starting to fear food: fearing spikes, some people avoid fruit, oats, carrots — healthy foods — and eat an ever-narrower diet. That's backwards.Number addiction and anxiety: some get captured by the curve, check it dozens of times a day, and hang their self-worth on a reading.Fueling disordered eating / orthorexia: in those already inclined, a CGM can push the pursuit of pure eating into a pathological obsession. The review therefore calls for tighter regulation of these devices sold to healthy people.
So what should a healthy, curious person do? A few practical rules:
Default stance: you don't need a CGM to prove you're healthy. Rising to 120-140 after a meal and falling back is the body doing its job — no need to flatten it.If you really want to wear one, treat it as a short experiment, not a verdict: two weeks to satisfy curiosity and confirm that walking and vegetables really do flatten the curve is plenty; don't wear it long-term, don't turn every meal into an exam.Recognize the warning signs: if it makes you fear food, check numbers compulsively, or eat ever more narrowly, take it off immediately — by then the harm outweighs the little information.Spend your effort on the big levers: what actually decides long-term metabolic health is weight, activity, fiber and overall diet quality, sleep, not smoking — all of which you can do without a CGM, and which pay off far more.
Disclaimer: this is popular science, not medical advice, and no substitute for a clinician. It discusses metabolically healthy ordinary people; if you have, or suspect you have, diabetes, prediabetes, or another metabolic condition, follow your doctor's guidance, use monitoring tools per guidelines such as the ADA's, and do not change a plan your doctor gave you because of one popular-science article.
A 2024 narrative review of CGM use in people not living with diabetes (Oganesova 2024, Diabetic Medicine) puts it bluntly: there is almost no evidence that CGM is accurate in healthy people or delivers any health benefit; instead it can bring a set of real harms:
Misreading normal fluctuation as a defect: many people believe you shouldn't rise at all after eating, so a normal post-meal peak triggers anxiety — when we already know that rising and falling back is exactly what healthy looks like.Starting to fear food: fearing spikes, some people avoid fruit, oats, carrots — healthy foods — and eat an ever-narrower diet. That's backwards.Number addiction and anxiety: some get captured by the curve, check it dozens of times a day, and hang their self-worth on a reading.Fueling disordered eating / orthorexia: in those already inclined, a CGM can push the pursuit of pure eating into a pathological obsession. The review therefore calls for tighter regulation of these devices sold to healthy people.
So what should a healthy, curious person do? A few practical rules:
Default stance: you don't need a CGM to prove you're healthy. Rising to 120-140 after a meal and falling back is the body doing its job — no need to flatten it.If you really want to wear one, treat it as a short experiment, not a verdict: two weeks to satisfy curiosity and confirm that walking and vegetables really do flatten the curve is plenty; don't wear it long-term, don't turn every meal into an exam.Recognize the warning signs: if it makes you fear food, check numbers compulsively, or eat ever more narrowly, take it off immediately — by then the harm outweighs the little information.Spend your effort on the big levers: what actually decides long-term metabolic health is weight, activity, fiber and overall diet quality, sleep, not smoking — all of which you can do without a CGM, and which pay off far more.
Disclaimer: this is popular science, not medical advice, and no substitute for a clinician. It discusses metabolically healthy ordinary people; if you have, or suspect you have, diabetes, prediabetes, or another metabolic condition, follow your doctor's guidance, use monitoring tools per guidelines such as the ADA's, and do not change a plan your doctor gave you because of one popular-science article.
Practical · three people, three answers
Compress should you wear one into three lines:Diagnosed diabetes / prediabetes / on insulin: yes, wear it, per your doctor and the ADA guidelines. This island's debunk is not aimed at you.Metabolically healthy, purely curious: optional. If you want to, treat it as a two-week experiment — confirm that walking and eating vegetables first really do flatten the curve, satisfy the curiosity, and stop; don't wear it long-term, don't make it the judge of every meal.Metabolically healthy but anxiety-prone / a history of disordered eating: better not to wear one. For you, the curve is more likely to become a new source of anxiety, while the information it gives can barely change anything you should already be doing.
One overarching test: information is only worth collecting when it can change an action and doesn't harm you. For diabetics, a CGM usually meets both; for healthy people, it usually meets neither.