GLP-1s, exercise and bone density: two trials explained

Start with the outcome the study actually measured

Bone density, bone-turnover markers and fractures answer different questions. A study can detect a change in a scan measurement without showing how often people fracture a bone. Before interpreting a claim about GLP-1 treatment, look for the outcome in the methods rather than relying on a reassuring or alarming headline.

This evidence explainer revisits two human trials published in 2024. It is not a report of a new discovery. One studied liraglutide with or without an exercise programme after initial diet-induced weight loss; the other studied semaglutide in adults selected for increased fracture risk. Their different designs matter. [1,2]

For a practical reading habit, complete this sentence: the researchers measured ____ in ____ people, compared ____ with ____, over ____. If the headline mentions an outcome that does not fit the first blank, slow down. That missing link is often where a plausible interpretation becomes an unsupported promise.

The liraglutide trial: an exercise programme and a comparison

The June 2024 JAMA Network Open paper analysed bone outcomes in 195 adults aged 18–65 with obesity and without diabetes. After an eight-week diet, participants entered 52 weeks of exercise, liraglutide, both or placebo. Density changes included the diet period. [1]

Liraglutide alone reduced hip and spine density more than exercise alone. The combined group did not differ significantly from placebo at those sites despite greater weight loss. Exercise included cycling and circuit activity, not just lifting. [1]

Table 2 shows that hip density fell within the combined group. Its nonsignificant comparison with placebo does not mean zero loss or prove equivalence. [1]

Translate a headline such as exercise protects bones into a question: protects compared with what, at which site, and on which measure? The answer determines whether you are reading about a relative difference, stable measurements or fewer clinical events. Those possibilities should not be exchanged for one another.

The semaglutide trial: a different starting population

The May 2024 eClinicalMedicine phase 2 trial randomized 64 adults with increased fracture risk to semaglutide or placebo for 52 weeks. Most were postmenopausal women; mean age was 63. Eligibility involved low bone density and/or a recent low-energy fracture. This was a different population from the liraglutide exercise trial. [2]

The primary bone-formation marker outcome did not differ significantly between groups. Secondary findings included higher bone resorption and lower hip and lumbar-spine density with semaglutide than placebo. The authors suggested weight loss could help explain the resorption finding; that possible mechanism was not proved. [2]

The trial does not establish that semaglutide causes fractures. It also does not tell you what would happen if those participants followed the exercise programme from the other trial. Combining two papers in a paragraph cannot create a treatment comparison that neither paper performed.

When someone says one trial contradicts another, first compare eligibility and interventions. A result in older adults selected for skeletal risk is not automatically the same question as a result in younger adults with obesity after a supervised diet. Both can inform a discussion while leaving important uncertainty.

Separate useful guidance from proof of an outcome

The joint society advisory on GLP-1 nutrition, available in a corrected May 2026 version, discusses appropriate diet and resistance training as part of muscle and bone support. It is expert guidance informed by the literature, not another trial demonstrating fracture prevention. [3]

That distinction still leaves a practical next step: discuss an appropriate, sustainable activity and nutrition plan with your existing care team. The evidence does not support copying a study's entire exercise programme without considering your own function, experience and medical circumstances.

Keep expectations specific when seeking support. A trainer can help explain an exercise plan and its adaptations. Your clinician can interpret your medical history and decide whether a bone-health assessment is appropriate. Neither role should turn a small trial into a universal prediction about an individual scan or fracture.

Bring a decision-focused note to your clinician

You do not need to arrive with a conclusion about your medication. Bring the question that affects your next decision, plus the relevant information you already have. The following prompts organise the conversation; they are not screening criteria or instructions to request a scan.

• What does my existing bone-health history mean for this discussion? Bring any prior scan report or fracture history rather than trying to reinterpret it yourself.
• If we monitor something, what question would that measurement answer, and what would change in my care as a result?
• Which exercise options fit my current function and experience, and who can help me adapt them?
• If appetite or symptoms make eating difficult, who can review my intake and existing nutrition advice?
• Which parts of these studies resemble my situation, and which important parts do not?

Read the next headline with three boundaries

First, keep the measured outcome in the conclusion. A scan finding should remain a scan finding unless clinical-event evidence supports a stronger statement. Second, retain the comparator: exercise alone, placebo and combination treatment are different reference points. Third, retain the population and time window.

The liraglutide analysis was secondary, the sample size was based on weight change, and analyses were not adjusted for multiple comparisons. Both trials covered about a year of treatment. These limits discourage a universal rule for GLP-1 users. [1,2]

Use this article alongside the explanation of lean mass versus muscle mass if a scan report includes several body-composition measures. Bone density, lean mass and muscle strength should each be described with their own measurement. A single reassuring label such as preserved body composition can conceal which outcome actually changed.

Do these results prove that GLP-1 medicines cause fractures?

No. Changes in bone density and turnover markers do not establish a change in fracture incidence. These trials cannot supply a long-term fracture-risk prediction for an individual.

Does the combination group prove exercise prevented all bone loss?

No. Hip bone density decreased within that group. Its comparison with placebo was not statistically significant; that does not establish zero loss or equivalent outcomes.

Should everyone on GLP-1 treatment request a bone scan?

These studies do not establish universal screening criteria. Discuss your history and the decision a measurement would inform with your existing clinician.