The numbers behind the numbers
Opening a diabetes report can feel a little like being handed a school report written in another language. There are percentages, averages, coloured graphs and abbreviations everywhere. Even when the results look good, it is not always obvious what they actually say about everyday life with Type 1 diabetes.
My latest 90-day Glooko report showed:
-
GMI: 6.3% (44.9 mmol/mol)
-
Average glucose: 6.8 mmol/L
-
Median glucose: 6.5 mmol/L
-
Time in range: 92%
-
Time high: 6%
-
Time very high: 1%
-
Time low: 1%
-
Time very low: 0%
-
Standard deviation: 1.7 mmol/L
-
Coefficient of variation: 25%
They are strong results, but no single number tells the whole story. Here is what each one adds to the picture.
GMI is an estimate, not a laboratory HbA1c
GMI stands for Glucose Management Indicator. It uses average CGM glucose to estimate the HbA1c that those readings might correspond to.
My GMI was 6.3%, or 44.9 mmol/mol. That does not mean a laboratory HbA1c test is guaranteed to give exactly the same result. CGM data and a blood test measure different things, and individual biology can create a difference between them.
I find GMI useful as a trend. If it changes over time, it can show whether my overall glucose level is moving up or down. I do not treat it as a replacement for an HbA1c arranged by my healthcare team.
Average glucose can hide the highs and lows
My average was 6.8 mmol/L, which sounds reassuring. The problem with an average is that very different days can produce the same result.
A day spent steadily around 6.8 mmol/L and a day swinging repeatedly between 3 and 11 mmol/L could end with a similar average. The experience—and potentially the risk—would be quite different.
That is why I would never look at average glucose alone. It becomes much more meaningful alongside time in range and glucose variability.
Median shows the middle of my readings
My median was 6.5 mmol/L. The median is the middle value when all readings are placed in order. Half are above it and half are below it.
Because a few unusually high or low readings can pull an average around, the median gives another useful view of where glucose usually sits. My median being close to my average suggests the report was not being dramatically distorted by a small number of extreme results.
Time in range shows where I spent my day
My report used a target range of 3.9–9.5 mmol/L and showed 92% time in range. That works out at just over 22 hours of an average day inside that range.
Time in range is one of the most relatable measurements because it shows how much of life is being spent within the chosen glucose boundaries. But it still needs context. A high time in range achieved through frequent or serious hypos would not represent safe control.
That is why the 1% low and 0% very low matter so much to me. More time in range is valuable, but not if it is bought by pushing glucose dangerously low.
Standard deviation measures the spread
My standard deviation was 1.7 mmol/L. In simple terms, it describes how widely glucose readings tend to spread around the average.
A smaller standard deviation usually points towards a steadier glucose profile. A larger one suggests more movement and bigger swings. However, standard deviation also tends to rise when average glucose rises, so it is best considered alongside coefficient of variation.
CV tells me how variable my glucose is
CV stands for coefficient of variation. It compares standard deviation with average glucose and expresses the result as a percentage.
My CV was 25%. International consensus guidance generally treats a CV of 36% or below as a marker of more stable glucose, although personal targets should always be agreed with a diabetes team.
CV is valuable because it helps distinguish between a reasonable average created by steady readings and one created by repeated highs and lows cancelling each other out.
The report is information, not a grade
The biggest thing I try to remember is that a diabetes report is not a judgment of effort, character or worth. Glucose is affected by far more than food and insulin: hormones, stress, sleep, illness, exercise, weather, absorption and sheer unpredictability can all play a role.
Rather than asking, “Did I pass?”, I find it more useful to ask:
-
Are lows occurring at a particular time?
-
Are there repeated rises after certain meals?
-
Is glucose stable overnight?
-
Are the same patterns appearing on several days?
-
Is improving one number making another less safe?
The value of a 90-day report is not that it produces a score. It helps reveal patterns that are almost impossible to see while dealing with Type 1 one decision at a time.