CGM vs HbA1c: What a Glucose Monitor Can and Cannot Tell You

1. What does a CGM actually measure?

Not your blood. That surprises most people who wear one.

A continuous glucose monitor sits a few millimetres under the skin and samples interstitial fluid. That is the fluid sitting between your cells. Glucose reaches it after it reaches your bloodstream, so the reading trails your actual blood glucose by roughly five to fifteen minutes.

Most of the time that lag does not matter much. It matters when levels are moving fast, which is exactly when people screenshot the graph. A sharp curve after lunch is partly the food and partly the sensor catching up.

Sensors also drift. Each one is calibrated by the manufacturer, and two worn at the same time on the same person can disagree. That is tolerable when you are tracking a trend. It is a problem when you are treating a single number as a verdict on a slice of toast.

2. Do CGM numbers match an HbA1c test?

In people with diabetes, closely. In everyone else, barely at all.

That is the finding of a Mass General Brigham study of 972 adults aged 40 and over, published in Diabetes Technology and Therapeutics in October 2025.

The researchers compared sensor averages against the standard lab test. In people with type 2 diabetes, the two tracked each other well. In people with pre-diabetes the link weakened. In people with normal blood sugar it essentially vanished.

So the sensor is not a cheaper version of the blood test. If your glucose is in the normal range, the two measurements are telling you genuinely different things.

3. What does HbA1c tell you that a sensor cannot?

The long view, and a number your GP will recognise.

Glucose in your blood sticks to haemoglobin, the protein that carries oxygen inside red blood cells. Those cells live for about three months. HbA1c measures how much sugar has attached, so it reflects glucose levels over the preceding two to four months. No sensor, no wearing period, no diary.

It also comes with agreed thresholds. UK laboratories report HbA1c in mmol/mol. Below 42 is the low-risk band. Between 42 and 47 is the range clinicians treat as pre-diabetes. A result of 48 or above is the level used to diagnose diabetes, and that diagnosis is made by a clinician, not by an app.

A CGM has no equivalent. There is no agreed spike height that means anything about your future health. That is the gap the graphs quietly paper over.

4. Is there evidence CGMs help people without diabetes?

Not yet. The researchers who reviewed it were blunt about that.

A 2024 narrative review in Diabetic Medicine, led by researchers at UCL and Birmingham Children’s Hospital, looked at sensor use in people not living with diabetes.

They found accuracy in this group is under-researched. They found little evidence that the data changes health outcomes. They raised the risk of obsessive tracking and disordered eating, and called for commercial claims about the benefit to non-diabetic users to be treated as misleading.

One point is easy to miss. Exercise can push interstitial glucose readings around in ways that do not reflect a problem. A hard session can produce an alarming-looking trace and a completely normal person.

This is also why the NHS does not fund sensors for people without diabetes. Diabetes UK sets out who qualifies, and it is built around type 1 and insulin-treated type 2 diabetes.

5. Should you wear one anyway?

It depends what you want from it. Be honest about which.

As a feedback tool, a CGM can be genuinely useful. Watching a breakfast of white bread behave differently from the same breakfast with eggs teaches you something no lab report will. Rodriguez made exactly that point: in people without diabetes, sensors may work as behavioural biofeedback, but they do not directly reflect longer-term blood sugar control.

As a health assessment, it is the wrong instrument. It cannot tell you where you sit against a clinical threshold, and it cannot see anything except glucose.

If two weeks of graphs would make you anxious about food, that is a reason to skip it. The review above flagged that outcome specifically.

6. Which markers give you the real picture?

Start with the three-month average, then read it next to the markers that share the same root cause.

HbA1c is measured in our Advanced Heart Health panel (£159), alongside the full lipid panel, ApoB, Lp(a) and hs-CRP. That pairing is deliberate. Blood sugar and cardiovascular risk tend to drift together, and reading one without the other leaves half the story out.

Two honest limits. Fasting insulin and HOMA-IR, the markers that can flag insulin resistance early, need a venous sample rather than a finger-prick, so they are not part of our home kits. And a blood panel reads far more than sugar: our guide to the full blood count covers the markers a glucose sensor will never see.

A test measures. It does not diagnose. Any result that lands outside a reference range is a conversation to have with a qualified clinician, and our reports say so.

Frequently asked questions