Is a CGM Worth It If You Don't Have Diabetes?

You can buy one now without asking anyone’s permission. In March 2024 the US regulator cleared the first over-the-counter continuous glucose monitor, aimed at adults who don’t use insulin — including, in its own words, people without diabetes who want to understand how diet and exercise affect their blood sugar. The same announcement carried a line that almost never survives into the marketing: users should not make medical decisions based on the device’s output without talking to a health care provider.

That sentence is the whole problem in miniature. A device sold on the promise of self-knowledge, shipped with a warning that you can’t act on what it tells you alone.

So, should you wear one? If you don’t have diabetes and you have no symptoms bothering you, no. There is no good evidence that wearing a sensor improves the health of a person who doesn’t have diabetes, and there is no validated number for you to aim at. Every institutional health page currently ranking for this question says a version of that, and then stops.

Which is not much use to the reader who already bought one and is looking at a graph right now. So here is the rest of it.

The rise is the machinery working

Your blood glucose is meant to go up after you eat. You have a carbohydrate-containing or mixed meal, digestion breaks the carbohydrate down, glucose crosses into your blood, the number climbs. That is normal, expected, well-documented physiology across healthy people and across the diet patterns that have been studied. If the line stayed flat after lunch, that would be the thing to investigate.

What happens next is the part worth caring about. Your pancreas releases insulin — first a fast burst of what it already had stored, then a steadier second wave for as long as glucose stays up — and that insulin moves glucose out of your blood and into muscle, liver and fat. It is a feedback loop, not a battle. In people without diabetes the post-meal pattern has a recognisable shape. It climbs, peaks somewhere around the hour mark, and is generally back near where it started within two to three hours, the exact timing depending on what the meal was.

The word “spike” is doing enormous unearned work here. It sounds like a fever or a crash. It describes food being absorbed.

The numbers nobody hands you

Here is what actually gets measured when researchers put blinded sensors on people without diabetes and don’t try to sell them anything.

Study	Who	Time 70–140 mg/dL
Shah 2019, multicentre	153 healthy, ages 7–80	median 96%
Framingham 2025	560 normoglycaemic adults, mean age 59	87%

Shah’s group found participants spent a median of about 30 minutes a day above 140 mg/dL (7.8 mmol/L), and that half of them spent 2% or more of the day up there — while under 1% of them spent that long above 180 mg/dL (10.0 mmol/L). The Framingham community cohort, older on average, found normoglycaemic adults sitting above 140 for roughly three hours a day, and above 180 for just 1.3% of it.

Read those two rows again. Both are careful studies in people without diabetes. They differ by nine percentage points on the headline metric, because they measured different populations of different ages with different devices. That gap is the point. If two good research groups can’t converge on what a normal day looks like, an app cannot possibly tell you your Tuesday was a bad one.

And notice what the Framingham figure does to the standard wellness framing. Three hours a day above the line that makes your phone glow orange — in people whose glucose handling is, on formal testing, entirely normal.

”Time in range” was built for somebody else

The percentage your app shows you is a real clinical concept. It came out of diabetes care, where international consensus targets were set for how much of the day a person with diagnosed diabetes — usually insulin-treated — should spend inside a defined glucose band. Those targets were calibrated against that population’s complication risk. That is what makes them meaningful.

Lift the same phrase, apply it to someone without diabetes, and you have imported a benchmark that was never derived for them. There is no equivalent established target range for a person without diabetes to aim at. When an app tells you that you scored 71% today, the honest follow-up questions are: in range for whom, against what target, tracking which health outcome? Usually the answer is a number built for a different problem. Why it gets reused anyway, I can’t tell you with a citation. My own reading is that a daily score you can win or lose is a good way to keep someone opening an app — but that is me reading an incentive, not a finding, and you should hold it that loosely.

A colour-coded score is a design decision wearing a lab coat.

The thing the device cannot do

This is the limitation everything else hangs off, and no firmware update fixes it.

Two very different situations can draw a similar line on your screen. One is a healthy pancreas handling a large or fast-digesting meal — up sharply, back down promptly, exactly as built. The other is genuinely impaired glucose handling, where the fast first insulin wave is blunted or the tissues have stopped responding properly, so the peak runs higher and the return drags. Same-shaped squiggle. Completely different meaning. A consumer sensor has no way to tell them apart, because the difference lives in mechanism and clinical context, not in the shape of the curve.

That is also why the sensor is not a screening tool. Prediabetes and type 2 diabetes are diagnosed on fasting plasma glucose, HbA1c, or an oral glucose tolerance test — none of which is a wearable. A meaningful number of Australian cases are undiagnosed at any given moment, which is an argument for getting the actual blood test, not for buying a sensor to guess with. If something is genuinely worrying you — unexplained weight loss, constant thirst, waking repeatedly to urinate, fatigue that won’t shift, repeated shaking and sweating after meals — the move is a GP appointment, not a longer subscription.

The cost that isn’t the sensor price

Sensors are consumables. The one the FDA cleared runs up to 15 days, then you buy another. The money is a recurring line item, and the honest per-insight arithmetic is unflattering: a fortnight of data, no validated benchmark to compare it against, and no model linking today’s curve to any outcome years from now.

The other cost is less discussed and, I think, more serious. The 2025 scoping review that pulled this literature together noted that the gap between the popular material on glucose spikes and the medical evidence could confuse readers and possibly lead to disordered eating behaviours, and it picked up published concern about an obsessive culture forming around food. Repeated, obsessive glucose-checking converts food from a thing you eat into a thing you’re graded on. That is not a claim that monitors cause eating disorders. It is a claim that a device turning every bite into a score is not neutral, and it never appears on the box.

Standing on a scale doesn’t make you lighter. Watching a graph doesn’t make your pancreas work better. The graph is a mirror, not a lever.

Two things this article’s own fact-check changed

The first draft of the book behind this piece said blood glucose rises after carbohydrate, protein, or a mixed meal in every healthy person, in every culture, on every diet pattern ever studied. The check flagged it. The carbohydrate and mixed-meal part is uncontroversial physiology; the protein part and the sweeping “every, every, every” were reaching past what the sources support. The corrected version is the one above, and it is narrower. The same paragraph also had the post-meal elevation running one to two hours, which was short. Peak at about an hour, back to baseline inside two to three, is what the literature actually describes.

The same draft called the scoping review a 2026 paper and attached a quote to it that belonged to a different article. It was published in October 2025, and the quote came from elsewhere. Both were fixed before publication, and the sentence you read above is now built from the review itself rather than from anything borrowed.

I’m telling you this because a machine wrote these words and the check is the only reason you should give them any weight. The machine matched each claim against its primary source, the check caught the machine’s own overreach, and a practising registered Australian doctor reviewed the results and is accountable for what stayed. Not a doctor reading every reference — that would be a nicer sentence and a false one.

What is still open

Whether glucose variability itself — the size and frequency of the swings, in someone without diabetes — predicts anything about long-term health is unresolved. Some researchers think there is a signal worth chasing. The current evidence is too thin to say it forecasts future disease in a person whose glucose regulation is working.

The reviews that have looked hardest at this land somewhere modest: there may be a narrow behavioural benefit from wearing a sensor, mostly around paying attention to what you eat, but the evidence for the outcomes actually being marketed — weight loss, metabolic health, disease prevention — runs from limited to unclear. The 2025 scoping review came down on uncertain: it could not say that wearing a sensor benefits the health of someone without diabetes, and judged that whatever real signal exists is more likely to sit in prolonged, frequent elevation than in any single post-meal rise. Which is not what a daily score is measuring.

What isn’t open: post-meal movement, adequate sleep, pairing refined carbohydrate with fibre and protein, going easy on large amounts of free sugar and refined starch, and a healthy weight range all have real evidence behind them for general metabolic health. All free. None of them requires you to know your glucose number at any point in the day.