Continuous glucose monitoring for cholesterol
A 14-day CGM is the most direct feedback available on what a meal does to your glucose — and no trial has yet shown that acting on it changes any outcome in people without diabetes.
Why glucose monitoring helps with cholesterol
Every glucose spike triggers an insulin surge. Insulin activates hepatic de novo lipogenesis — your liver literally converts excess glucose into fatty acids, packages them into VLDL particles, and exports them into your bloodstream. This is the direct biochemical bridge between your diet and your triglyceride number. By identifying which foods spike your specific physiology, you can cut the supply line to elevated triglycerides at the source.
The conventional 2-hour OGTT cutoff for impaired glucose tolerance is 7.8 mmol/L — and keeping postprandial peaks below this is associated with reduced cardiovascular risk in non-diabetic populations.
🎓 A CGM is a learning tool, not an alarm system
The single most important thing to understand before you wear a CGM: it shows you a continuous picture, but a single spike does not damage your arteries. The clinical signal that actually predicts cardiovascular harm is your HbA1c — a 3-month average of how much glucose has been bound to your red blood cells. That is the cumulative exposure number doctors use to assess risk; a CGM is the behavioural-learning tool that tells you which habits drive that average up or down.
Think of it like a fitness tracker for your blood sugar. The point isn't to obsess over every notification — it's to see your patterns clearly enough that you change them, then take the tracker off having permanently upgraded how you eat. Most people only need to wear one for 14-28 days once in their life to recalibrate their understanding of what their body actually does with the foods they eat every day. The insights are sticky; the device is temporary.
The "amazing to see" moments that change behaviour for years afterward: watching a 10-minute walk after dinner cut a glucose spike in half in real time; discovering that the "healthy" smoothie spikes harder than a chocolate bar; realising that switching the eating order of your meal (vegetables → protein → carbs) flattens the curve by 30-70%; seeing that quinoa and lentils barely move the line while a baked potato sends it through the roof. These observations don't come from reading about glucose — they come from seeing your own data, and once seen they can't be unseen.
What a CGM is also genuinely good at: identifying undiagnosed metabolic issues early. If your fasting numbers run above 7.0 mmol/L, or your post-meal peaks routinely hit 10+ mmol/L and stay elevated for hours, that's a medical conversation worth having — well before HbA1c crosses any diagnostic threshold. The CGM gives you the early signal; the HbA1c and your GP confirm what to do about it.
| Metric | Target |
|---|---|
| Peak postprandial glucose | < 7.8 mmol/L (ideally < 7.0) |
| Spike magnitude (rise from baseline) | < 1.7 mmol/L |
| Time to return to baseline | < 2 hours |
| Glucose variability (SD over 24h) | < 1.0 mmol/L |
| Time in optimal range (4.0-7.0) | > 90% of the day |
| Average glucose over 14 days | 5.0-5.8 mmol/L |
Choose your CGM device
Three UK-available CGM options, compared side-by-side below. Our overall pick for self-knowledge use is the Dexcom ONE+ (marked ★ in the table) — the reasoning sits under the table. The "Buy direct" row links straight to the manufacturer's UK page.
Side-by-side comparison
All three sensors share the same core idea — a small filament under the skin reading interstitial glucose — but the practical differences matter. Lower MARD = closer to actual blood glucose.
| Spec | Dexcom ONE+Dexcom | Libre 2 PlusAbbott | LingoAbbott (wellness) |
|---|---|---|---|
| Age range | 2+ years | 2+ years | 18+ only · no insulin |
| Accuracy (MARD) | 9.8% adult / 7.7% paediatric | 8.2% — best in this comparison | ~9.3% — least accurate |
| Sensor life | 10 days + 12 h grace | 15 days — longest | 14 days |
| Reading frequency | Every 5 min | Every 1 min | Every 1 min |
| Sensor placement | Arm, abdomen + buttocks (ages 2–6) | Upper arm only | Upper arm only |
| Sensor size | 60% smaller than prior Dexcom (smallest Dexcom yet) | ~21 × 5 mm disc | ~21 × 5 mm (Libre platform) |
| Glucose reading range | 2.2–22.2 mmol/L | 2.2–27.8 mmol/L — widest | Wellness range only (~2.8–16.7 mmol/L) |
| Water resistance | Waterproof to 2.4 m for 24 h | Water-resistant (showering OK) | Water-resistant (~1 m / 30 min) |
| Alarms | High/low + predictive low + Delay 1st | Optional high/low | None — wellness only |
| API / data sharing | Share, Glooko, retroactive event API | LibreView, LibreLinkUp | Apple Health / Health Connect only |
| UK price (approx) | From £35 ex VAT per 10-day sensor | ~£52 per 15-day sensor | ~£89 per 4-week subscription |
| Best for | Kids and adults, anyone who can't easily wear arm-only sensors, swimmers, widest age band | Fewest sensor changes per month, anyone already in the Abbott Libre ecosystem | Healthy adults wanting food-impact learning — not for insulin users |
| Buy direct | 🛒 Dexcom UK | 🛒 Abbott UK | 🛒 Amazon UK |
Our pick — Dexcom ONE+, on price and availability rather than on accuracy. Checked September 2026; sensors, prices and UK availability change without the packaging changing, and the criteria above outlast the pick. For a 14–28 day self-knowledge trial of the kind described in this section, the Dexcom ONE+ wins on the dimensions that matter most for non-diabetic learning use: it's the only CGM that lets you wear it somewhere other than the back of your arm (useful if you sleep on one side, lift weights, swim regularly, or simply find arm sensors uncomfortable), it's the only fully waterproof option (no taking it off to swim), it works from age 2 upward, the alarms are genuinely useful for catching reactive-hypoglycaemia patterns, and per-day it's cheaper than Lingo with the same low-investment commitment. The 5-minute reading interval is also plenty for learning purposes — minute-by-minute readings are only meaningfully different if you're managing insulin doses. If you specifically need 15-day sensor wear (fewer changes per month) or are already in the Abbott ecosystem, Libre 2 Plus is the close runner-up.
How to actually use a CGM (14-day protocol)
- Day 1-3 (baseline): Eat normally. Don't change anything. Just observe what your typical glucose curves look like — you'll be surprised which foods cause the biggest spikes.
- Day 4-10 (isolation testing): Test foods individually. Eat a single food with nothing else, and watch the curve. Wait 2-3 hours between tests to return to baseline. Test each food at least twice, on separate mornings, and take the average. A single reading is not a result: in the trial described below, the same person eating the same standardised meal on two different days produced correlations between replicates ranging from 0.26 to 0.73 depending on the food. The researchers averaged repeats precisely because one test is too noisy to act on, and repeatability was worst for the foods that barely moved the line.
- Day 11-14 (modification testing): Re-test the foods that spike — but modified. Add fat (olive oil, nuts), fibre (chia, psyllium), protein, or vinegar. See how the curve flattens.
- What commonly causes unexpected spikes: standard GF bread (rice flour), supermarket GF cereal, white rice, smoothies with fruit juice base, "healthy" granola, dates, dried fruit, ripe bananas, sushi rice, even some "diet" snacks.
- What commonly stays flat: meat and fish without sauce, eggs, leafy vegetables, full-fat Greek yoghurt with berries, nuts, avocado, hummus, lentils, chickpeas, quinoa (often surprisingly mild).
- Pair foods strategically: food order, food pairing, post-meal walking, and meal timing all dramatically flatten the curve — see the four evidence-based techniques below for the specific studies and effect sizes.
- Track sleep and stress: poor sleep raises baseline glucose; chronic stress shows up as a flatter but elevated profile. Both contribute independently to cardiovascular risk.
📖 How to interpret what you see — context matters more than peak numbers
Once your sensor is on and the data starts flowing, you'll see numbers that may worry you at first glance. Most of them are fine. Here's how to read your data the way a clinician would — focusing on shape and duration, not single readings.
📈 What "above 7.8 mmol/L" actually means
First, what a sensor can and cannot tell you. A CGM reads interstitial glucose, not blood, so it lags a rising or falling blood level by several minutes, and typical mean error against a laboratory reference sits around 8–9%. At a reading of 7.8 that error alone spans roughly 7.1 to 8.5. So a single value on either side of any line below is not a result — the same rule as the test-each-food-twice instruction above, applied to thresholds instead of foods. Patterns across repeated days are the signal; individual crossings are not.
7.8 mmol/L is the official threshold separating normal post-meal glucose from impaired tolerance — but only at the 2-hour mark. The same number 1 hour after eating means something very different.
| When it happens | Reading | What it usually means |
|---|---|---|
| 1 hour after eating | 7.9 – 9.5 mmol/L | Normal — temporary spike. Healthy adults regularly peak to 8-9 mmol/L 45-60 min after carb-containing meals. The key is what happens next. |
| 2 hours after eating | Below 7.8 mmol/L | Normal — your insulin response cleared the glucose efficiently. |
| 2 hours after eating | 7.9 – 11.0 mmol/L | Impaired Glucose Tolerance (pre-diabetes) — sustained elevation suggests insulin resistance. Worth a GP conversation, especially if seen repeatedly. |
| 2 hours after eating | 11.1 mmol/L or above | Diabetic range if confirmed on separate occasions. See your GP. |
| Fasting / between meals | Above 7.0 mmol/L | Insulin resistance signal — the liver is releasing glucose your body can't clear. This is the most concerning pattern; not a spike, a baseline. |
The shape of the curve matters more than the peak. Brief spike to 9.0 that crashes back to 5.5 within 90 minutes = your body works. The same peak that plateaus at 8-9 mmol/L for hours = the worrying pattern. Look for the return, not just the rise.
📉 When the line dips below 4.0 mmol/L
Seeing readings in the high 3s is far more common than people realise — especially overnight — and usually means nothing. Context tells you whether to worry.
Between meals (daytime): 30-90 minutes coasting in the high 3s while you feel completely fine is just your body idling at baseline. It should plateau, not keep drifting downward.
Overnight: less often than this page used to say. The reference data in healthy adults (Shah et al., JCEM 2019) puts 96% of the day between 3.9 and 7.8 mmol/L, with a median of about 15 minutes a day below 3.9. Some people do log an hour or two, mostly overnight — but a large share of those runs are compression artefact rather than real glucose. And note this, because it decides what to do next: symptoms caused by a genuine glucose in the high 3s are unusual in people without diabetes, which is one more reason a laboratory value is needed before anyone concludes anything.
Reactive dip after a high-carb meal (2-4 hours after eating): brief, 15-45 minutes, then the body restores balance. Common after sugary or refined-carb meals as insulin overshoots slightly.
What's actually worth a GP call: a downward drift (numbers continuously falling rather than flattening), prolonged daytime lows while you feel symptomatic (fatigue, brain fog, irritability), or any low accompanied by genuine symptoms. Compression artefact is worth knowing about too: if you see a sudden sharp overnight drop that lasts an hour then snaps back, you probably slept on the sensor — physical pressure pushes fluid away and causes a falsely low reading.
The reactive hypoglycaemia pattern — a pattern the standard tests don't measure
There is one CGM pattern worth understanding, because the standard tests do not measure it: HbA1c is an average over months and fasting bloods are taken before you have eaten, so neither of them sees what happens after a meal. That is a gap in what the tests cover, not a failing of the person ordering them. This is not a “low reading is bad” alarm — single dips are common and benign, as covered above. What matters is the specific shape and timing of a recurring pattern.
- Lows when fasting or overnight that are not compression artefact — genuine low glucose away from meals is a different problem from a post-meal dip, and a more serious one.
- Confusion, seizure, or collapse at any point. This is not something to investigate with a sensor.
- Previous stomach or bariatric surgery. Post-surgical hypoglycaemia is a recognised and specific condition, and it is managed differently.
- Any glucose-lowering medicine, or other medicines that can lower glucose. Everything on this page assumes you are not taking one.
The signature pattern — recognise this on your CGM
- A high carbohydrate meal (refined carbs, sugary drinks, white bread, pastries, large portions of rice/pasta) — often with relatively little fat or protein to slow absorption.
- A sharp spike 30–60 minutes after eating, often reaching 8.5–11.0 mmol/L.
- A steep crash 2–5 hours later — glucose drops below your starting baseline, often into the high 3s or low 3s mmol/L.
- Symptoms during the crash: shakiness, sweating, light-headedness, palpitations, sudden intense hunger, irritability, brain fog, anxiety, or in some cases near-syncope. Eating sugar resolves them within minutes.
- It happens repeatedly after similar meals — not a one-off.
This pattern has a clinical name: reactive (postprandial) hypoglycaemia. The usual explanation offered is that the pancreas, in someone with early insulin resistance, secretes a delayed and excessive insulin response to the carbohydrate load — too much insulin arrives, too late, and overshoots. Glucose plummets, the body releases adrenaline and cortisol to push it back up, and the adrenergic symptoms above are largely from those counter-regulatory hormones.
How to tell them apart. A compression low is usually flat-bottomed, arrives without symptoms, occurs while you are asleep and on the side the sensor is on, and snaps back to the previous trend within minutes. A real reactive low follows a meal by two to five hours, comes with the symptoms listed above, and recovers gradually. If you did not feel it, it did not wake you, and it happened at 4am, treat it as an artefact until it repeats while you are awake. A sensor reading alone is not a diagnosis of anything, which is what the next box is about.
Two corrections to what this page used to say. First, the threshold. In people without diabetes the Endocrine Society's criterion is a plasma glucose below 3.0 mmol/L (55 mg/dL). The 3.9 figure this page previously quoted is the diabetes threshold and does not apply here — using it would label a great many normal people as hypoglycaemic.
Second, and more importantly: a CGM cannot document the triad at all. It reads interstitial fluid rather than blood, it lags, and it over-reads lows — the three things that matter most at exactly this end of the range. The sensor can raise the question. Only a laboratory glucose drawn while you are symptomatic can answer it.
And if it is investigated, the right test is not the obvious one. For post-meal symptoms the appropriate investigation is a mixed-meal test in secondary care — not an oral glucose tolerance test, which produces false positives in people who turn out to be perfectly well. Worth knowing before you ask for one by name.
Why this connects directly to your lipid profile
This is the key insight that the test case you may have heard about exposed — and that conventional screening systematically fails to catch. The same hyperinsulinaemia that produces the post-meal crash is also driving high triglycerides and the atherogenic lipid pattern. Here's the mechanism:
1. Excessive postprandial insulin stimulates the liver to overproduce VLDL (the triglyceride-rich lipoprotein), because insulin is fundamentally an anabolic "store this energy" signal — and in insulin-resistant livers this VLDL-secretion brake fails. Hepatic de novo lipogenesis ramps up.
2. Elevated triglycerides appear in your fasting lipid panel — often as the first abnormality, sometimes the only one.
3. Cholesteryl ester transfer protein (CETP) then shuffles triglycerides from VLDL into your HDL and LDL particles in exchange for cholesterol — triglyceride-enriched HDL is rapidly cleared by the kidneys (HDL drops), and triglyceride-enriched LDL becomes small and dense (the dangerous, atherogenic kind).
4. The result: the "lipid triad" of atherogenic dyslipidaemia — high TG, low HDL, small-dense LDL — appears years before HbA1c crosses any diagnostic threshold.
This is exactly why a person can have a "normal" HbA1c, a passing lipid panel from the GP, and yet be actively developing atherosclerotic damage. The dyslipidaemia is real; the postprandial hyperinsulinaemia is driving it; the CGM crash pattern is the visible fingerprint. None of the conventional screening tools will see it because they measure fasting state and three-month averages — they aren't watching the meal-by-meal hormonal chaos.
If you see this pattern on your CGM — what to actually do
- Document it. Note the timing, the meal, the glucose values (peak and nadir), and the symptoms. A few screenshots from your CGM app showing the spike-then-crash shape is far more persuasive to a GP than describing it.
- Ask your GP for fasting insulin (or a HOMA-IR calculation). HbA1c can be normal in early hyperinsulinaemia; fasting insulin almost never is. This is the test that catches what HbA1c misses. UK NHS availability varies — private labs (Medichecks, Thriva, Randox) offer it for ~£30 if your GP declines.
- Check your full lipid panel with particular attention to the TG/HDL ratio — this is exactly the marker your Integrated Risk card in Section 3 highlights. A TG/HDL above 0.87 mmol/L in someone with reactive hypoglycaemia is the atherogenic-dyslipidaemia signature.
- Consider hs-CRP and apoB. The combination of postprandial hyperinsulinaemia + elevated TG/HDL + elevated hs-CRP + high apoB is the high-risk picture that the conventional panel never assembles.
- Lifestyle response works fast. Lowering refined carbohydrate intake, eating protein/fat/fibre before carbs at meals, post-meal walking, and resistance training to expand skeletal-muscle glucose disposal all directly reduce postprandial insulin demand — typically within 2–4 weeks the CGM crash pattern softens or disappears.
- If symptoms are severe or include neuroglycopenic features (confusion, loss of consciousness, seizures), or if low readings persist despite dietary change, this needs proper specialist workup — including ruling out rare causes such as insulinoma. Don't self-diagnose if anything feels neurologically wrong.
The takeaway: a CGM doesn't just teach you which foods spike your glucose. It can reveal a pattern of metabolic dysfunction that no standard blood test will detect, and that pattern is mechanistically the same thing driving any elevated triglycerides or unfavourable lipid pattern you may have. If your lipids are odd and conventional explanations don't fit, this is worth investigating.
⏳ Why a single spike doesn't damage your arteries — but a pattern does
Artery disease (atherosclerosis) is not triggered by single high readings. It is a slow, cumulative process that typically requires 5 to 10 years of sustained or frequent elevation to manifest as detectable structural disease. Think of it like water damage to wood — a splash that you wipe up does nothing; standing water for months rots the floor.
The damage mechanism stacks three processes that all depend on prolonged exposure, not peaks:
1. Glycation — excess glucose binds to proteins in the bloodstream, forming Advanced Glycation End-products (AGEs) that stiffen the endothelial lining of arteries.
2. Oxidative stress & inflammation — sustained high glucose drives free-radical release, making vessel walls "sticky" and inflamed.
3. Plaque formation — once the lining is inflamed, circulating LDL slips beneath it, oxidises, and accumulates as fatty plaque that calcifies over years.
This is why so many people are diagnosed with type 2 diabetes and coronary artery disease at the same time — asymptomatic mild hyperglycaemia had quietly been damaging vessels for a decade before anyone tested. The clinical metric for this cumulative exposure is HbA1c, not your CGM. An HbA1c persistently at 42 mmol/mol or higher is the threshold where cumulative risk becomes meaningful; a CGM peak that touches 8.5 and returns to 5.5 is not.
Glucose damage also doesn't act alone — it accelerates exponentially when combined with elevated LDL/apoB, high blood pressure, or smoking. Lowering any one of those reduces the cumulative artery hit even if the others stay the same.
What people are surprised by — the moments that change behaviour
- A 10-15 minute walk after a meal can cut the spike in half in real time. Watching it happen on your phone screen is the moment most people decide post-meal walks are non-negotiable.
- Eating order matters — though how much depends on who you are. Vegetables → protein → carbs at the same meal can drop the spike substantially versus eating the carbs first. Same food, same calories, different curve. But the size of that effect is not the same for everyone, and the people who benefit least appear to be the ones with insulin resistance. See the section on individual variation below before treating a percentage as your percentage.
- "Healthy" foods often spike harder than treats. Fresh-pressed orange juice, smoothie bowls, granola, dried fruit, dates, sushi rice, and ripe bananas frequently outrank a small chocolate bar. The fibre and protein in real food blunt sugar release; juicing and refining strip both away.
- Quinoa, lentils, and chickpeas barely move the line. The protein + fibre + slow-digest starch profile is genuinely different from rice or potato, and you'll see it on the graph.
- Poor sleep raises tomorrow's baseline. A bad night will show up as a higher fasting reading and a flatter, elevated curve all day. The link to cardiovascular risk is real, not a wellness platitude.
- Stress alone can spike glucose. No food required — a stressful work call or argument can produce a 1-2 mmol/L rise from cortisol and adrenaline. Once seen, it's hard to dismiss "stress affects your health" as vague advice.
🧬 Why your curve is yours: what a 2025 trial actually showed
The case for wearing a CGM rests on a claim that is easy to state and was, until recently, thinly evidenced: the same food does not do the same thing to everybody. A 2025 Nature Medicine trial from Stanford tested it properly, and the result is more interesting — and more qualified — than the headline.
What they did. Fifty-five adults with no diagnosis of diabetes, about half of them with prediabetes, ate seven different carbohydrate meals, each containing exactly 50 g of carbohydrate: jasmine rice, white bread, shredded potato, macaroni, black beans, mixed berries and grapes. Each meal was eaten at least twice, on separate mornings, after an overnight fast, with nothing added and no exercise for three hours. Every participant also had gold-standard laboratory testing for insulin resistance and for how well their pancreas was working.
Rice raised glucose most on average — but the average described almost nobody. Sorted by whose glucose rose highest on which meal, 35% were rice-spikers, 24% were bread-spikers and 22% were grape-spikers, with the rest peaking on potato or pasta. Your worst food is not reliably the population's worst food.
And here is the part that matters most, which the coverage tends to skip. Not one of the 55 participants had their highest spike on beans or on mixed berries. Not one.
So the individual variation is real, but it sits inside the group of foods that were already the problem. The variation is in which starchy or sugary food is worst for you — not in whether beans and berries beat rice and bread. General advice was not overturned by this study. It was confirmed, and then refined. That is an argument for wearing a CGM to find your own worst offenders, not an argument that the basics do not apply to you.
The pattern tracks your physiology, not your preferences. People whose glucose rose most on potato were more insulin resistant and had poorer pancreatic beta cell function; people who peaked on grapes were more insulin sensitive. The ratio between someone's potato response and their grape response separated the insulin-resistant from the insulin-sensitive. This is the real finding: a curve on a phone screen is picking up something true about your metabolism, not just about the food.
The negative result, which is the most useful thing here. The same trial tested whether a preload — 10 g of fibre, 10 g of protein, or 15 g of fat, eaten 10 minutes before rice — flattened the spike.
Averaged across everyone, all three worked, and all three worked only slightly. The effect sizes were small by any standard. About two thirds of participants saw a reduction; some saw no change, and a few spiked higher with the preload than without it.
And the benefit was concentrated in the people who needed it least. Preloads reduced the spike in participants who were insulin sensitive or had normal beta cell function. In those who were insulin resistant, they did very little. The authors' proposed explanation is that a preload works through two mechanisms at once — slowing absorption, and prompting an insulin response — and if the second mechanism is impaired, slowing absorption alone is not enough.
The hypothesis that follows — and it is a hypothesis, from six or seven people per group. If you are the person most likely to be told “just eat your protein first”, you may be the person it helps least. That is not established, and a larger trial could overturn it. It is a reason to check on your own trace rather than assume, and a reason not to rely on food ordering as a substitute for reducing the carbohydrate load itself.
One finding this page previously left out, and should not have. The paper's own abstract reports that rice-spikers were more likely to be Asian participants. This site initially omitted it because the association did not survive correction for multiple testing (pFDR 0.21) — but readers will meet the claim elsewhere, usually without that caveat, so stating it with the caveat is more useful than staying silent. It is consistent with older glycaemic-index work reporting higher rice responses in Chinese than European participants, and the authors note it could reflect genetics, habitual rice intake, or both. Treat it as a hint about which food to test first, not as a prediction about you. A separate association between bread-spikers and higher blood pressure is labelled exploratory by the authors themselves and is still not stated here.
Four limits worth holding on to before you over-read any of this.
- The subgroup numbers are very small. The preload comparison rested on roughly six or seven people per group, and fewer still in the beta cell analysis. This is a direction of travel, not a settled quantity.
- Repeatability was moderate at best. Correlation between a person's two attempts at the same meal ranged from 0.26 to 0.73, which is why the researchers averaged repeats. It is also why the protocol above now tells you to test twice.
- These were single foods eaten alone on an empty stomach. The authors state plainly that the preload results cannot be extrapolated to a normal mixed meal. This is a laboratory-grade answer about rice, not a rule about dinner.
- Participants could not see their own readings — glucose was blinded until the study ended — and no clinical outcomes were measured at all. The trial shows that curves differ and that the differences mean something physiologically. It does not show that watching your own curve changes your health. Nothing yet does. That gap is the honest reason this chapter frames a CGM as a tool for self-knowledge rather than as treatment.
Where this leaves the chapter. Individual variation is real, it is large, and it reflects measurable physiology — which strengthens the case for testing yourself rather than working from a glycaemic index table. But it strengthens it within the existing advice, not against it. Beans and berries were mild for everyone. The starchy and sugary foods were the problem for everyone; only their ranking changed. Wu et al., Nature Medicine, June 2025 — 55 participants. Position as of September 2026.
🛠️ Four evidence-based techniques to flatten your glucose curves
You've seen what your curves look like. These four interventions have specific peer-reviewed studies behind them and produce the biggest visible effects on a CGM trace. Try them, watch the curve change in real time, and you'll have learned more about your own metabolism in two weeks than years of fasting blood tests can teach you.
Change your food order
Eat protein and vegetables before carbohydrates at every mixed meal. Same food, same calories, same total carbs — different sequence.
Shukla et al., Diabetes Care (Weill Cornell, 2015): in adults with type 2 diabetes given identical meals in different sequences, glucose was 29% lower at 30 min, 37% lower at 60 min, and 17% lower at 120 min when protein and vegetables were eaten before carbs vs. the reverse order. Insulin response was also significantly reduced. The mechanism is mechanical and hormonal: fibre and protein slow gastric emptying, blunting the rate at which glucose enters circulation. Read the population carefully: that trial was in people with established type 2 diabetes, and the effect in people without diabetes has looked considerably smaller — see individual variation below.
How to apply it: at lunch and dinner, eat the salad/vegetables first, then the meat/fish/tofu, then the rice/pasta/bread last. The order matters more than the proportions.
Never eat sugar alone
Always pair refined carbs with protein, fat, or fibre. A cookie eaten alone produces a much sharper spike than the same cookie eaten at the end of a meal.
Multiple food-sequencing and meal-composition studies show that combining carbohydrates with protein and fat blunts the postprandial glucose response, with figures around 30–50% often quoted, compared with the same carbohydrate eaten in isolation. Those are the higher end of a wide range, and a 2025 trial that tested preloads directly found much smaller average effects — see below. The mechanism overlaps with technique 1 but applies even to snacks and isolated sweet foods: protein and fat delay gastric emptying and stimulate GLP-1, which moderates the glucose surge.
How to apply it: if you're going to eat dessert, eat it after a real meal — not on an empty stomach. If you crave something sweet between meals, pair it with a handful of nuts, a piece of cheese, or yoghurt. Same dose of sugar, far smaller spike.
Walk after eating
A 10-minute walk immediately after a meal cuts the glucose peak by about 10%. Even 2–5 minutes of light walking helps.
Hashimoto et al., Scientific Reports (July 2025): in a randomised crossover trial of 12 healthy young adults given a 75 g glucose load, a 10-minute walk immediately afterwards reduced the peak glucose from 181.9 mg/dL → 164.3 mg/dL (10% reduction). A delayed 30-minute walk reduced overall glucose exposure but did not significantly lower the peak — meaning the timing matters more than the duration. Working muscles pull glucose directly out of circulation via GLUT4 transporters, an insulin-independent mechanism. (Caveat: small study in young healthy adults using a glucose drink rather than a mixed meal — the principle is well-established in larger literature but exact percentages vary by context.)
How to apply it: after dinner especially, take a 10-minute walk within 15 minutes of finishing. Pacing around the kitchen while washing up counts. The walk doesn't need to be brisk — a comfortable pace works.
Avoid late-night carbs
The same meal eaten at 8 PM spikes glucose more than at 8 AM — your body becomes physiologically more insulin-resistant in the evening, irrespective of how long since you last ate.
Morris et al., PNAS (2015): under controlled circadian-misalignment conditions with identical test meals at 8 AM vs. 8 PM, the postprandial glucose AUC was up to 17% higher in the evening — a circadian-driven effect independent of behavioural factors. Pancreatic β-cell function is well-documented as approximately 20% lower in the evening, and early-phase insulin secretion (the critical first-30-minutes response) is blunted. Prediabetic populations show even larger evening vs. morning differences. Mechanistically: CLOCK and BMAL1 genes regulate β-cell insulin biosynthesis on a daily rhythm, peaking in the morning and declining through the evening.
How to apply it: front-load carbohydrates earlier in the day. A rice-and-curry lunch will produce a smaller spike than the same dish at dinner. If you have late evening events with carbs (restaurants, social meals), apply techniques 1–3 with extra discipline — you have less metabolic margin for error.
Stacking these techniques compounds the effect. Eating vegetables and protein first and walking 10 minutes after and doing it at lunch rather than dinner doesn't just add the percentages — it can convert a meal that would have spiked you to 9 mmol/L into one that barely crosses 7 mmol/L. The same food, the same calories, dramatically different metabolic impact. This is exactly the kind of insight a CGM exists to reveal.
The bottom line. If your post-meal peaks come back down within 2 hours and your fasting numbers sit in the 4.0-5.9 range, your arteries are not actively being damaged — your metabolism is working. If your peaks plateau, your fasting numbers run high, or your overall average drifts up, that's the signal worth acting on — and HbA1c will confirm it. The CGM is the magnifying glass; HbA1c is the verdict.