No method has been tested to prevent it, because the one measured predictor is how much weight comes off. In 20 people scanned before and after starting a GLP-1, midface volume fell by a median 9.0% [1]. That worked out to about 7% of midface volume for every 10 kg lost. No trial has tested slower loss, protein or exercise against the face.
That makes Ozempic face mostly a question about weight loss rather than about one drug. The same pattern appears with the hair shedding that follows rapid loss, where the subtype that rises points at the weight lost rather than the molecule.
What was actually measured
A facial plastic surgery group at one academic center searched its records for GLP-1 prescriptions [1]. It kept patients with head and neck CT or MR imaging both before and after the prescription. Twenty qualified, with a median age of 54.
They had been on the drug for an average of 321 days and lost an average of 11.0 kg. Total midface volume fell by a median 9.0%, with an interquartile range of 3% to 14%. Superficial fat fell 11.0%. Deep fat fell 7.0%, but its range ran from minus 20% to 15%.
Weight lost correlated with superficial volume loss, at rho 0.590. It did not correlate with deep volume loss. The authors describe it as one of the first quantitative assessments of the phenomenon, which is accurate, and it is also most of what exists.
Why age changes the trade
Facial fat is not only a cost of excess weight. In 186 pairs of identical twins, a panel rated age from photographs [2]. A 4-point higher BMI made a twin look older before age 40 and younger after it.
An 8-point higher BMI made a twin look older before 55 and younger after. Smoking and sun exposure made twins look older at every age. The study measured appearance at one moment, not change during weight loss, so it explains the direction rather than the size.
The practical reading is narrow. The same weight loss is likely to show more in the face of a 60-year-old than a 30-year-old. That is a reason to expect it, not a reason to avoid treatment.
The levers, and what each rests on
| Lever | What is measured | Measured on the face? |
|---|---|---|
| Losing less weight | About 7% midface volume per 10 kg lost, in 20 people | Yes, once |
| Losing it more slowly | More fat-free mass lost during fast loss, mostly gone after 2 to 4 weeks of stability | Not measured |
| Protein and resistance training | Lean mass held in trials of the body | Not measured |
| Hydration and nutrition | Listed as risk factors in a review of 40 studies, without effect sizes | Not measured |
Losing it more slowly
Slowing loss is the most repeated advice, and its evidence comes from the body. A review gathered 13 studies that produced the same percentage loss at slow or fast rates [3]. Rates ranged from 0.2 to 3.2 kg a week.
Faster loss may take more fat-free mass during the losing phase. Those differences shrank after 2 to 4 weeks of holding the new weight. The review judged the remaining differences unlikely to matter clinically.
One randomized trial in that literature shows the size. In 57 adults, a 500 kcal diet for 5 weeks and a 1,250 kcal diet for 12 weeks produced similar losses, of 9.0 kg and 8.2 kg [4]. Fat-free mass loss was 8.8% on the fast diet against 1.3% on the slow one.
None of these studies used a GLP-1, and none imaged a face. On a GLP-1 the rate is set largely by the dose schedule. That is covered in the four-week titration steps, and the trade between more loss and more symptoms is in whether a higher dose is worth it.
Protein and resistance training
These protect lean mass, and the case for them on a GLP-1 is set out in how much muscle goes during treatment. The single-trial detail is in what happened to muscle mass and grip.
The difficulty is anatomical. The measured facial loss sat mainly in superficial fat, and resistance training builds muscle rather than facial fat. Nobody has shown that holding lean mass protects the cheeks. It is still worth doing for the body, and the face is not the reason.
Nutrition and hydration
A 2026 review of 40 studies on GLP-1 drugs and skin quality listed risk factors for more severe changes [5]. They were advanced age, a long history of obesity, rapid weight loss, poor hydration and insufficient protein intake.
The included studies mixed clinical trials, case reports and observational work. The abstract gives no effect size for any factor. A list like this is where to look, not a measured risk. What a reduced intake does to micronutrients is covered in whether you need vitamins on a GLP-1.
Prevention plans are still hypotheses
A 2026 paper proposes a risk-stratification model and a four-phase prevention algorithm keyed to the pace of weight loss [6]. Its own methods call it hypothesis-generating. It rates itself Level V evidence and says future studies are needed to validate it.
Most of the remaining literature concerns injectable correction after the fact, much of it written by practitioners who provide it. That is treatment, not avoidance, and this page does not assess it. The mouth has a parallel gap, described in what has been measured about Ozempic teeth.
What nobody has measured
No randomized trial has imaged the face while testing any intervention. No study compares facial change on a GLP-1 with the same loss by diet. Nobody has measured how much volume returns after stopping, or at what rate.
So the answer a reader can act on is limited. The amount of weight lost drives most of the measured change. Age makes it more visible. Slower loss and lean-mass protection are sensible for the body, and untested for the face.