Is a continuous glucose monitor without diabetes worth it?

A continuous glucose monitor without diabetes - a sensor worn on the skin that estimates blood sugar every few minutes - did not improve blood sugar in people whose readings were already normal. That is the finding of a 2026 systematic review pooling 23 studies, 7 of them randomised trials, covering 1,074 people without diabetes in 11 countries [3].

Averaged across everyone in those studies, the sensor group did end up with modestly lower average glucose than the comparison groups. The review's own breakdown puts that gain in one place: people with prediabetes improved, and people with normal glucose showed no appreciable benefit [3]. Prediabetes means blood sugar sitting above the normal range but below the line for a diabetes diagnosis [7].

That group has the most room to move. Each year, 5 to 10% of people with prediabetes go on to develop diabetes and a similar share drift back into the normal range, and changing diet and activity cuts the risk of progression by 40 to 70% [7].

Weight did not shift. Across the same pooled studies, people wearing a monitor finished no lighter than those who were not, and the reviewers concluded the device is unlikely to produce weight loss on its own without a structured programme around it [3].

The readings are estimates, not measurements of blood. In 9 healthy adults measured against finger-prick blood samples, a sensor on the upper arm sat 12.4% away from the blood value on average, one on the thigh 18.3%, and a thigh sensor with the skin deliberately warmed 23.5% [6]. A hospital study of 40 critically ill patients found a similar average gap of about 12% against blood drawn from an artery, but individual readings ran from roughly 27 below to 48 above the blood value in milligrams per decilitre, the unit used for the amount of sugar in blood [5].

The one use the research supports is about timing. A review of 7 studies in people without diabetes reported its clearest result when participants used their own readings to start walking before their glucose peaked after a meal, which lowered the glucose and insulin rise that followed [4]. The same review found monitors picking up early glucose irregularities in groups not usually screened for them, including women going through menopause and people with obstructive sleep apnoea, a condition where breathing repeatedly stops during sleep [4]. Direct evidence that any of this prevents heart attacks or strokes does not yet exist [4].

There was no large reference set for what a normal sensor trace even looks like until CGMap collected readings from more than 7,000 people without diabetes aged 40 to 70 [1].

The sensor never touches blood. It reads the interstitial fluid, the watery layer between cells just under the skin, and glucose has to move out of the capillaries into that fluid before the sensor registers anything [6].

That handoff runs on blood flow, which is why the same person wearing two sensors gets two different numbers. In the exercise study, warming the skin over the thigh site raised the skin temperature there by 3.1 degrees Celsius against 1.1 at the arm, and the thigh readings dropped and drifted further from the blood value [6]. During cycling the arm-versus-thigh gap closed, because working muscle pulls more blood through the leg [6].

The behavioural half is simpler. A monitor turns an invisible number into a graph you check, and that is enough to change what people eat over the short term and to keep them engaged with a programme they are already following [3].

What it does not do is change how the pancreas releases insulin or how muscle takes up glucose. When that machinery is working normally there is nothing for the feedback to correct, which fits the split the review found: improvement in prediabetes, nothing in people whose glucose was already normal [3].

The evidence base is small. All 23 pooled studies together held 1,074 people, only 7 of them randomised trials, and the behavioural changes were measured over short periods rather than years [3]. No trial has followed people without diabetes long enough to see whether wearing a sensor changes anything that matters, and the cardiovascular review says so directly: the effects on heart attacks and strokes are unmeasured [4].

The accuracy numbers come from populations that look nothing like a person wearing a sensor at a desk - 9 healthy adults on an exercise bike [6] and 40 patients in intensive care [5].

Who this evidence does not speak for:

• Adults under 40 and over 70. The reference set for normal readings covers ages 40 to 70 only, so there is no comparison group outside that band [1].
• Anyone with diagnosed type 1 or type 2 diabetes. That is a separate question with separate evidence, and in type 1 diabetes monitoring is already part of treatment rather than an optional extra [15].
• Athletes reading their trace during hard exercise. Where the sensor sits changes the number by several percentage points, and the arm and thigh disagree least when the leg is working [6].
• Anyone buying a sensor to lose weight, since the pooled studies found no weight difference [3].

Glucose readings are also not a diagnosis. Prediabetes and diabetes are defined by blood tests ordered by a doctor, not by what a skin sensor draws on a phone screen [7].

Longevity clinics list continuous glucose monitoring as a diagnostic service, and a first visit means having a sensor applied - conventionally on the back of the upper arm, the placement the accuracy research uses [6] - then wearing it over a stretch of days while eating and moving normally, with the trace reviewed afterwards. The research supports one narrow use of what comes back. In the review of 7 studies, the clearest result came from people who read their own trace to find when glucose peaked after a particular meal, then took their walk before that point instead of after [4]. The lowering came from the walk, not from wearing the sensor.