Glucose Variability: Why Two People With the Same A1C Live Very Different Lives

Two people both average 154 mg/dL over a fortnight. Both get an A1C around 7.0%. One spends the whole time drifting between 120 and 190. The other swings from 45 to 300 and back, several times a day, treating lows at work and correcting highs at midnight. The average cannot tell them apart. Standard deviation can.

What an Average Hides

An average is a single number standing in for hundreds of readings, and it is deliberately indifferent to how those readings were arranged. A 40 and a 268 average out to 154, which looks identical on paper to a steady 154 and is nothing like it in real life.

This is why A1C alone is an incomplete picture. It is built from average glucose, so it inherits the same blind spot. Two people with matching A1Cs can have completely different day-to-day experiences, different hypoglycemia risk, and different amounts of the exhausting mental work that swinging blood sugar demands.

Standard Deviation, Without the Statistics Lecture

Standard deviation (SD) measures spread. Loosely: how far a typical reading sits from your average. A low SD means your readings cluster; a high SD means they scatter.

In mg/dL, the rough interpretation looks like this:

• Under 30 — tight and stable. Readings mostly sit close to your average.
• 30 to 60 — typical range for many people managing diabetes with reasonable consistency.
• Over 60 — meaningful swing. Highs and lows in the same day are likely, and so is hypoglycemia.

Those bands are the ones SweetLife uses when it flags a period as stable or variable in your Insights, and they are a reasonable starting frame — but SD on its own has a flaw worth understanding.

Why Coefficient of Variation Is the Better Number

Standard deviation does not account for where your average sits, and that turns out to matter a great deal.

Consider an SD of 50 mg/dL. If your average glucose is 110, an SD of 50 means readings routinely reach into the 60s — you are living close to hypoglycemia. If your average is 220, that same SD of 50 keeps you between roughly 170 and 270. Same spread, entirely different risk, because the distance to a dangerous low is what actually threatens you.

Coefficient of variation (CV) fixes this by expressing the spread relative to the average:

CV = (standard deviation ÷ average glucose) × 100

The two examples above give a CV of 45% and 23% respectively — which correctly identifies the first as the unstable one, even though SD called them identical.

The 36% Threshold

International consensus guidance settled on CV of 36% as the dividing line between stable and unstable glucose. Below 36% is considered stable. Above it, hypoglycemia risk rises appreciably.

It is a genuinely useful target because it is achievable and because it points at something you can act on. Unlike A1C, which takes months to move and averages away the detail, CV responds to specific fixable causes within weeks.

It is also worth knowing that CV and Time in Range measure related but distinct things. You can have a decent Time in Range and a poor CV — ricocheting through the target zone on your way between extremes, technically in range a lot of the time without ever being stable. That combination is common and worth catching.

What Actually Drives Variability

High variability usually traces back to a handful of causes, and the useful move is finding which of them is yours.

Correction stacking. Treating a high before the previous dose finished working, going low, treating the low with fast carbs, and rebounding. Each correction is reasonable; the sequence is not. Checking insulin on board before correcting breaks the loop.

Meal bolus timing. Insulin given at the first bite arrives after the carbohydrate does, producing a spike followed by a fall as the insulin catches up too late. The spike and the subsequent low are one event.

Over-treating lows. A genuine low feels urgent, and urgency invites eating far more than the 15 grams needed. The rebound high then earns a correction, and the cycle continues.

Inconsistent timing. Meals, insulin and activity at wildly different times each day give your body a different problem to solve daily. Variability in the inputs shows up as variability in the output.

Basal that does not fit. If your background insulin is wrong for part of the day, everything layered on top has to compensate. This is worth investigating with your care team rather than adjusting alone.

Finding Yours

SweetLife computes standard deviation and coefficient of variation over whatever period you select in Insights, alongside Time in Range and average glucose. Comparing a 14-day window against the 14 before it shows the direction of travel, which is more useful than any single reading of the number.

Insights also flags periods as stable or variable automatically and surfaces the day-of-week and time-of-day patterns underneath. Discovering that your variability is concentrated in weekday evenings is far more actionable than knowing your CV is 41%, because the first one names a cause and the second only names a symptom.

The Short Version

Average glucose and A1C describe where you sit. Standard deviation and coefficient of variation describe how much you move, and moving a lot is both harder to live with and riskier than the average suggests.

Aim for a CV under 36%. Then look for the specific cause of your swings rather than trying to be generally more careful — the causes are usually few, specific, and fixable, and your logged data will point straight at them.