Key takeaways
- Glucose variability is the size and frequency of the swings, as distinct from the average HbA1c reports.
- A spike is food meeting whatever first-phase insulin response is available, and the crash afterwards is the informative half — the signature of a system compensating rather than regulating.
- Glucose variability is the amplitude and frequency of the swings, as distinct from where the line sits on average.
- Endothelial dysfunction is an early step in vascular disease, so the pathway to a hard outcome is at least plausible.
- CGM was built for insulin dosing in type 1 diabetes and has since been pointed at everyone else.
HbA1c is an average, and averages hide shape. Two people can share an identical A1c while one runs a flat line all day and the other swings from 60 to 170 and back several times. Those are not the same metabolic situation, and the person swinging usually knows it — crashes, cravings, post-lunch fog — while the bloodwork says everything is fine. Continuous monitoring made that visible for the first time, and also produced a great deal of overclaiming.
What variability actually is
Glucose variability is the amplitude and frequency of the swings, as distinct from where the line sits on average. A healthy pattern holds most of the day in a narrow band — roughly 70-120 mg/dL — with modest post-meal rises to around 130-140 that resolve within two hours. A less healthy pattern shows frequent excursions past 160, drops into the 60s accompanied by hunger and shakiness, and rapid rebounds afterwards.
A1c cannot see this for arithmetic reasons: it reflects average glycation of haemoglobin over the red cell's lifespan, so a steady 110 and a wildly oscillating average of 110 produce a similar figure. It also inherits any problem with red cell turnover — anaemia, recent blood loss, haemoglobin variants (reading HbA1c properly).
What the swings report on is capacity. A spike is not the food acting alone; it is the food meeting whatever first-phase insulin response is available. Early insulin resistance shows up as a blunted, delayed early response — glucose climbs higher and stays up longer, then the delayed insulin arrives at once and overshoots, producing the crash. The crash is the informative half: the signature of a system compensating rather than regulating.
Why swings might matter on their own
The mechanistic case is oxidative stress. Experimental work found that oscillating glucose produced more endothelial dysfunction and more oxidative stress than the same average delivered as sustained elevation — cells appear to handle a steady insult better than a repeatedly changing one (Monnier et al., JAMA 2006; Ceriello et al., Diabetes 2008). Endothelial dysfunction is an early step in vascular disease, so the pathway to a hard outcome is at least plausible.
The immediate effects are less contentious and more useful day to day. A large excursion followed by a rapid fall produces exactly what people describe as blood sugar problems: the crash ninety minutes after lunch, the craving for fast carbohydrate, the irritability, the need to sleep after a large meal. Those are real and individually verifiable.
Now the honest limits. Most of the outcome evidence linking variability to complications comes from people with diabetes, where the swings are larger and the follow-up is long. In people without diabetes, the evidence that variability predicts anything independently of average glucose, insulin resistance and body composition is thin. No target range for variability in healthy people has been validated against outcomes. Anyone quoting one is quoting a convention, not a finding.
What continuous monitoring revealed
CGM was built for insulin dosing in type 1 diabetes and has since been pointed at everyone else. Three findings have held up.
Non-diabetic adults spike more than expected. Detailed profiling found distinct patterns of dysregulation in people whose fasting glucose and A1c were entirely normal, some spending meaningful time in ranges that would be flagged as impaired on a formal test (Hall et al., PLoS Biol 2018).
Responses are individual. The same food produces reproducibly different responses in different people — microbiome, prior meal, time of day, sleep, activity — which is the strongest argument that generic glycaemic index tables are a poor guide for any one person (Zeevi et al., Cell 2015).
Non-food inputs move the line. Stress alone can produce a rise of 30-50 mg/dL without anything eaten. A poor night's sleep worsens the next day's responses to identical meals. Identical food eaten late in the evening produces a larger excursion than the same food eaten earlier, because insulin sensitivity follows a circadian pattern.
What CGM has not established is that wearing one improves health in people without diabetes. It also has real interpretive traps: readings lag blood by several minutes, accuracy is poorest exactly when glucose is changing fastest, compression during sleep produces false lows, and two sensors on the same person can disagree meaningfully. Treating one alarming reading as fact is the commonest error (fasting glucose versus CGM).
What reliably produces a spike
- Refined carbohydrate eaten alone, without protein, fat or fibre
- Liquid sugar — juice, soft drinks, sweetened coffee — which arrives faster than solid food
- Late-evening meals, against declining evening insulin sensitivity
- Poor sleep the night before, independent of what is eaten
- A sedentary day — muscle that has not contracted takes up less glucose
- Acute stress, mobilising glucose with no food involved
- Alcohol followed by a carbohydrate-heavy meal
What flattens the curve, in order of return
- Walk after meals. Ten to fifteen minutes is enough, and it is the single most effective non-pharmaceutical lever available — contracting muscle takes up glucose through a route that does not require insulin (Buffey et al., Sports Med 2022). Breaking up sitting across the day works through the same mechanism (frequency of interrupting sitting time, Diabetes Care 2021).
- Change the order of the meal, not the meal. Vegetables and protein before the starch measurably blunts the excursion from identical food (Shukla et al., Diabetes Obes Metab 2019). It costs nothing and requires no restriction, which is why it survives contact with real life.
- Build muscle and train it. Skeletal muscle is the largest glucose sink in the body — the slow lever that changes capacity rather than managing one meal.
- Sleep. One short night measurably worsens next-day glucose handling. It is upstream of appetite, food choice and the response itself.
- Move the largest meal earlier where the day allows it, since insulin sensitivity is higher earlier.
- Address insulin resistance directly if it is present — see the reversal protocol and the insulin sensitivity guide.
- Medication where clinically indicated, which smooths the curve by acting on gastric emptying and glucose-dependent insulin release. That is a physician's decision after evaluation, not a first step.
Vinegar before a carbohydrate-heavy meal has a modest, repeatedly demonstrated effect. Real, and small — it belongs at the bottom of the list rather than the top, where it usually gets placed.
What to measure, and what to do with it
- Fasting insulin — the earliest of these markers to move, because compensation precedes failure (why fasting insulin comes first)
- HOMA-IR, calculated from fasting glucose and insulin together
- HbA1c, with its limitations understood
- A two-hour glucose tolerance test, which remains the standard way to see a meal response with an insulin value attached
- CGM for a defined period — two weeks, with a question in mind
If you wear a sensor, use it as an experiment rather than a scoreboard. Run the same meal twice under different conditions — with and without a walk after, early and late in the day, after a good night and a bad one. Watching the line all day and reacting to every rise produces anxiety and, in some people, a disordered relationship with food. Two weeks with three questions beats six months of monitoring.
The clinical pearl: two people with an HbA1c of 5.5 can be in different metabolic positions, and variability is part of what separates them. But the honest use of that fact is diagnostic, not obsessive. Large repeated swings with a normal A1c are a prompt to check fasting insulin, not a condition in themselves — and the fixes are a post-meal walk, a change in meal order, more muscle and better sleep, all of which are worth doing whether or not you ever wear a sensor.
Bottom line
Average glucose and glucose shape are different measurements, and A1c only sees the first. The case that swings cause more oxidative and endothelial damage than steady elevation is mechanistically reasonable, and the immediate symptoms are real. What is not established is that variability independently predicts outcomes in people without diabetes, or that wearing a monitor improves anything by itself. Use one as a diagnostic instrument for a fortnight, check fasting insulin alongside it, and act on the four things that flatten the curve: walk after meals, eat protein and vegetables before starch, build muscle, sleep. The 60-second assessment is a starting point if the pattern suggests something worth investigating properly.
Educational content, not medical advice. Laboratory interpretation and any treatment decision are made by a licensed physician after individual evaluation. Individual results vary.
