Continuous glucose monitors for people without diabetes: one trial with an outcome

A continuous glucose monitor is a small sensor in the arm reporting interstitial glucose every few minutes. In type 1 diabetes it is transformative and the evidence is not in question. Sold to a metabolically healthy adult as a window into their metabolism, it becomes a different proposition, and the literature supporting that proposition is smaller than the marketing implies.

A systematic review searched January 2020 to August 2025 for studies of CGM in non-diabetic individuals for cardiovascular prevention. It found seven. Six carried a low risk of bias, which is respectable. Only one dealt with an actual outcome measure.1

Seven studies and one outcome is the whole evidential foundation of a product category currently sold by subscription.

What the good result actually shows

The clearest positive finding is worth taking seriously precisely because it is so specific.

CGM was used to personalise the timing of exercise, starting a walk before the individual’s own postprandial glucose peak rather than at a fixed interval after eating. This significantly reduced postprandial glucose, insulin and C-peptide.1 That is a real effect from a real mechanism, and it is the sort of thing only continuous measurement can enable, because the peak arrives at a different time in different people.

CGM also functioned as a motivational device, increasing readiness for physical activity.1 Behavioural effects are easy to dismiss and probably should not be: a measurement that gets somebody walking after dinner has done more than most biomarkers.

Figure 1

The missing number

Here is the question a consumer CGM cannot answer, and it is the one the whole product depends on.

What is a normal glucose excursion after a meal in a metabolically healthy adult, and above what level does it matter?

There is no clinically validated threshold. Consumer apps display target ranges and flag spikes, but those ranges derive largely from expert opinion or from diabetic care targets, not from outcome data in people with normal HbA1c and fasting glucose. A rise after eating is normal physiology. Without a validated cut-off, an app telling somebody their oatmeal caused a spike is reporting a number and inventing the interpretation.

This is why the surrogate evidence, though genuinely interesting, does not close the gap. Observational work links higher glycaemic variability to surrogate markers of cardiovascular risk such as blood pressure variability, and detects subclinical dysregulation in menopause and in obstructive sleep apnoea.1 Those associations may well be pointing at something real. They do not establish that watching your own variability and eating differently in response improves anything.

A 2026 systematic review of CGM in non-diabetic populations put it plainly: effectiveness for glycaemic parameters, body weight and behavioural outcomes remains unclear.2

Why grade C rather than D

Grade C means mechanistically plausible and widely used, with no trial evidence of net benefit in the population being screened. Grade D would mean the measurement is unreliable or the inference unsupported.

CGM does not belong in D, and the distinction is worth defending. The sensor measures what it claims to measure, the physiology is not in doubt, and there is at least one demonstrated mechanism by which the data changes a decision usefully. That is a better position than several more expensive things sold alongside it.

What keeps it out of B is the absence of an outcome. One study with an outcome measure, in a literature of seven, is not a foundation for a permanent subscription.

How to buy it, if you are going to

Two framings, and only one of them survives the evidence.

As a time-limited experiment, the case is reasonable. Two to four weeks, with a specific question: which of my regular meals produces the largest excursion, and does walking twenty minutes after that meal flatten it? That is close to the intervention the positive study tested, it produces an answer you can act on permanently, and then the sensor comes off.

As a continuous subscription for a metabolically healthy adult, there is nothing in the literature to support it. You will accumulate a great deal of data with no validated threshold to interpret it against, and the most likely outcome is either that you ignore it or that you narrow your diet on the basis of numbers nobody can tell you are abnormal.

What would change the grade

A randomised trial in metabolically healthy adults, comparing CGM-guided lifestyle change against the same advice without a sensor, with HbA1c or incident dysglycaemia at twelve months. That trial is cheap by the standards of this field, the sensors are already manufactured at scale, and it would settle the question in a year.

Separately, and more urgently: somebody should establish reference ranges for postprandial excursion in people without diabetes. Until that exists, every consumer app is flagging deviations from a range that was never validated for the person reading it.