Small CGM trials show promise; glucose-guided meal scores showed no clear advantage#
Can a continuous glucose monitor help you lose weight if you don't have diabetes? The evidence to date is thin and split. Two small randomized trials published in 2026 found that people given more monitor use lost more weight than comparison groups getting the same counseling. A larger 2022 trial found that a diet built to flatten each person's glucose spikes showed no clear weight-loss advantage over a plain low-fat diet. And separate lab work suggests that a single recorded meal response may not reliably predict your response to that food next time.
Those results are less contradictory than they look, because they are testing different things. The positive trials tested wearing a monitor as feedback. The negative one tested eating to minimize glucose spikes. The difference matters. Reduced intake is one plausible explanation for the positive results: in the Glasgow trial, monitor users' energy intake fell by about 235 calories a day from baseline. For the wider question of how any tracking number compares with what you actually ate, start at how accurate calorie counting is.
Where the idea came from: the same meal, very different curves#
The case for glucose-guided eating rests on a real finding. When an Israeli group monitored 800 people for a week and logged their responses to 46,898 meals, they found high variability in how different people responded to identical meals, and an algorithm using blood tests, habits, body measures, activity and gut microbes predicted each person's response well enough to lower post-meal glucose in a blinded dietary trial1. That study, and what it does and doesn't mean for the glycemic index, is covered in the glycemic index explained.
The UK PREDICT 1 study scaled up the observation. In 1,002 healthy adults, including twins, the post-meal glucose response to identical meals varied with a population coefficient of variation of 68 percent. Meal macronutrients explained 15.4 percent of the variance in glucose responses, more than person-specific factors such as the gut microbiome at 6.0 percent, and a machine-learning model predicted glycemic responses with a correlation of 0.772. Many of its authors were employees of or consultants to a company that sells a personalized-nutrition program, which the paper discloses.
Notice what neither study measured: weight loss. PREDICT 1 is a mapping study, not a diet trial, and Zeevi's intervention outcome was post-meal glucose over a week. Both established that glucose responses differ between people. Neither showed that acting on those differences makes anyone thinner.
The trial that tested glucose-guided eating for weight#
That test came later. The Personal Diet Study randomized 204 adults with overweight or obesity and prediabetes or mild type 2 diabetes to one of two six-month programs with identical counseling — 14 sessions — and dietary self-monitoring in both arms. One group followed a standard low-fat diet. The other got color-coded meal scores from an app, generated by an algorithm that predicted each person's glucose response to specific foods, so they could choose meals that kept their personal spikes low3.
At six months the low-fat group had lost 4.31 percent of body weight and the personalized group 3.26 percent. The difference, 1.05 percentage points in favor of low-fat, was not statistically significant (95% CI −0.40 to 2.50; P = .16). Body composition and adaptive thermogenesis didn't differ either. The authors' conclusion is plain: a personalized diet targeting lower post-meal glucose "did not result in greater weight loss compared with a low-fat diet."
Two limits keep this from being the last word. The population had abnormal glucose metabolism, so it is not a trial in people with normal blood sugar. And it tested advice derived from glucose predictions, not the experience of wearing a monitor and watching your own curve. It answers "does eating for a flatter curve beat a standard diet?" — with no clear advantage found — rather than "does a monitor on your arm help?"
The trials that tested wearing one#
Two 2026 trials did test the device as feedback, and both came out positive.
| Trial | Who | Design | Result |
|---|---|---|---|
| Popp et al., 2022 | 204 adults with overweight or obesity and prediabetes or mild type 2 diabetes | Glucose-predicted meal scores vs low-fat diet, 6 months | −3.26% vs −4.31% body weight, no significant difference |
| Hatta et al., 2026 | 49 adults who had just lost >5 kg | Dietitian-led maintenance with or without a CGM, 24 weeks | −2.5 vs +0.4 kg (intention to treat) |
| Gradiser et al., 2026 | 35 women with overweight or obesity | Same exercise and diet counseling; CGM at start, midpoint and end vs start and end only, 12 weeks | −5.5 vs −0.2 kg |
In the Glasgow maintenance trial, people who had recently lost weight were randomized to a dietitian program with or without a monitor worn for 24 weeks; they wore it on 92.6 percent of days. The monitor group lost a further 2.5 kg while the comparison group regained 0.4 kg in the intention-to-treat analysis. The telling number is energy intake: it fell by 235 calories a day in the monitor group and rose by 23 in the comparison group4. The authors call it a feasibility trial and ask for a fully powered one. One author reports consulting fees from companies that include a glucose-monitor manufacturer.
The Croatian trial is smaller still — 35 women over 12 weeks — and both groups wore a monitor at the start and end, the intervention group also at the midpoint, and the control group lost almost nothing despite supervised exercise three times a week, which makes the size of the gap hard to generalize5.
If a glucose monitor helps with weight, eating less is a more likely route than finding a magic food.
Put the Glasgow numbers next to the Personal Diet Study and a consistent picture forms. Choosing food by its predicted glucose curve didn't beat a standard diet; wearing a monitor that makes eating visible went along with eating less in the one trial that measured intake. That is the same mechanism researchers credit for food logging, and it doesn't require the glucose reading to be telling you anything special about fat storage.
Why a single spike is a weak signal in people without diabetes#
If the device's main value is feedback, the quality of that feedback matters, and here the lab evidence is sobering.
At the US National Institutes of Health, researchers combined monitor data from two inpatient feeding studies involving 30 adults without diabetes, who were served the same meals twice about a week apart; one study used two device types and the other used one. Across 1,189 duplicate-meal responses, the correlation between a person's response to a meal and their response to the identical meal a week later was only about 0.45, and within-person reliability was low (intraclass correlation 0.28 and 0.17 for the two devices). A person's variability between identical meals was about as large as their variability between different meals. The authors concluded that individual post-meal responses "were highly variable in adults without diabetes" and that advice based on them "requires more reliable methods involving aggregated repeated measurements"6.
A separate crossover study at the University of Bath compared a monitor with fingerstick capillary samples in 15 healthy adults. The sensor read fasting and post-meal glucose about 0.9 mmol/L higher than the capillary reference, overestimated time spent above 7.8 mmol/L roughly fourfold, and the size of the bias differed from person to person and food to food7. One of its authors sits on the scientific advisory board of a personalized-nutrition company.
Neither study says the devices are useless. Both say that "this food spiked me, so this food is bad for me" is a conclusion one reading can't support — especially in people whose glucose stays in the normal range, where the spikes being chased are modest to begin with.
A glucose curve isn't a calorie count#
The deeper problem is that a flatter curve and fewer calories are different targets that sometimes point in opposite directions. Cooking oil barely moves blood glucose, yet it is among the most calorie-dense things in a kitchen; fruit moves glucose more and carries far fewer calories per bite. Choosing foods only for a flatter curve can favor calorie-dense options such as fat, the most energy-dense macronutrient.
A controlled inpatient trial showed how far those two targets can part. Twenty adults ate freely from a plant-based, low-fat, high-glycemic-load diet and an animal-based, ketogenic, very-low-glycemic-load diet for two weeks each. On the low-fat diet — the one with the higher glucose and insulin responses — they ate 689 fewer calories a day8. What that says about insulin and appetite is taken up in insulin and hunger; for this question, the lesson is narrower: participants ate more on the diet with lower glucose responses, though the diets differed in several ways.
A monitor is also a different kind of instrument from the ones that count the other side of the ledger. Like a wrist tracker's calorie estimate, it measures a signal and invites you to infer something it doesn't directly measure — see how accurate fitness-tracker calories are. If you wear one to lose weight, the evidence is most consistent with using it as a nudge that makes snacks and portions visible — while keeping the actual scoreboard where it has always been: intake, and a weight trend over weeks.
FAQ#
Do glucose spikes cause weight gain in people without diabetes?#
The trials here don't show that. A diet designed to minimize each person's predicted spikes produced 3.26 percent weight loss versus 4.31 percent on a standard low-fat diet, a non-significant difference, and in an inpatient trial the diet with the larger glucose responses led people to eat 689 fewer calories a day.
How accurate are continuous glucose monitors in people without diabetes?#
In one lab comparison with fingerstick samples, a sensor read about 0.9 mmol/L high and overstated time above 7.8 mmol/L roughly fourfold. Separately, responses to the same meal served twice a week apart correlated at only about 0.45 within a person, so single-meal readings are a shaky basis for food rules.
Is a CGM worth trying for weight loss?#
The best case for it is as feedback: in a 49-person maintenance trial, monitor users lost 2.5 kg while controls gained 0.4 kg, a 2.9 kg difference, and their intake fell about 235 calories a day. That trial and a 35-person trial are small and short, so treat the benefit as plausible but unconfirmed.
Sources#
- Zeevi D, Korem T, Zmora N, et al. Personalized nutrition by prediction of glycemic responses. Cell. 2015;163(5):1079-1094.
- Berry SE, Valdes AM, Drew DA, et al. Human postprandial responses to food and potential for precision nutrition. Nat Med. 2020;26(6):964-973.
- Popp CJ, Hu L, Kharmats AY, et al. Effect of a personalized diet to reduce postprandial glycemic response vs a low-fat diet on weight loss in adults with abnormal glucose metabolism and obesity: a randomized clinical trial. JAMA Netw Open. 2022;5(9):e2233760.
- Hatta NZM, Gray CM, Harvie M, et al. Optimizing weight loss maintenance: can continuous glucose monitoring play a role (OWL-CGM)? A randomized controlled feasibility trial. Clin Nutr. 2026;64:106738.
- Gradiser M, Bilic-Curcic I, Okun I, et al. Intermittently scanned continuous glucose monitoring enhances weight loss and adherence in women with obesity: a randomized controlled trial. Diabetes Technol Ther. 2026;28(9):907-915.
- Hengist A, Ong JA, McNeel K, Guo J, Hall KD. Imprecision nutrition? Intraindividual variability of glucose responses to duplicate presented meals in adults without diabetes. Am J Clin Nutr. 2025;121(1):74-82.
- Hutchins KM, Betts JA, Thompson D, Hengist A, Gonzalez JT. Continuous glucose monitor overestimates glycemia, with the magnitude of bias varying by postprandial test and individual — a randomized crossover trial. Am J Clin Nutr. 2025;121(5):1025-1034.
- Hall KD, Guo J, Courville AB, et al. Effect of a plant-based, low-fat diet versus an animal-based, ketogenic diet on ad libitum energy intake. Nat Med. 2021;27(2):344-353.
Source: BurnWeek — "Do continuous glucose monitors help with weight loss?", https://burnweek.fit/blog/continuous-glucose-monitors-for-weight-loss/. Licensed CC BY 4.0: free to quote or reuse with a link to this page.
