A meta-analysis of 60 studies and more than 1.25 million people, published in Drugs on 12 September 2026, gives the fullest picture yet of what GLP-1 drugs do to muscle, bone and joints. The bone finding is reassuring: no measurable effect on bone mineral density, and none on fractures. The muscle finding is the one to sit with — lean body mass falls consistently, driven mainly by semaglutide and liraglutide. But the authors rate their own certainty as low, and nothing in the review measured whether any of it makes people weaker.
What the analysis pooled
The team, led by Charlotte Beaudart at the University of Namur and including researchers from the WHO Collaborating Centre for Epidemiology of Musculoskeletal Conditions and Ageing, searched MEDLINE, Cochrane CENTRAL and Embase from inception to March 2025. From 1,148 candidate references, 60 articles met the inclusion criteria: 46 randomised controlled trials, 13 real-world evidence studies and one pharmacovigilance study, together covering 1,250,717 individuals taking semaglutide, liraglutide, exenatide, dulaglutide, tirzepatide and others. The protocol was registered with PROSPERO in advance (CRD420251024082) and the evidence was graded using GRADE.
On bone, the pooled models found no effect on bone mineral density at any site and no effect on fractures at any site, once the most adjusted effect estimate from each study was used. On joints, there was no significant change in WOMAC pain, physical function or stiffness — the standard osteoarthritis index.
On muscle, 28 studies contributed to a pooled estimate showing a consistent fall in lean body mass or fat-free mass, a standardised mean difference of −0.52 (95% confidence interval −0.80 to −0.23). A confidence interval sitting entirely below zero means the direction of the effect is dependable; the width of that interval means its size is not. Heterogeneity was high — an I² of 88%, meaning the individual studies disagreed substantially with one another. The finding survived every sensitivity analysis the authors ran, including leaving studies out one at a time, and subgroup analysis traced it mainly to semaglutide and liraglutide compared against placebo.
This sharpens rather than overturns what we have reported before. A Barcelona cohort we covered in August measured body composition directly and found semaglutide patients lost about 1.1 kg of fat-free mass across a year while their handgrip strength held steady. The new meta-analysis says the lean-mass half of that result is real and general, not a quirk of one hospital’s patients.
What it cannot say is whether the strength half generalises too. Lean body mass is a measurement of tissue, not of what that tissue can do. Losing fat-free mass while losing a large share of body weight is partly expected physiology: a smaller body needs less muscle to move it, and some of the “lean” compartment on a body-composition scan is water and organ mass rather than contractile muscle. The authors say as much, concluding that the changes appeared largely related to the weight loss itself.
What it does not settle
The authors rate the certainty of the lean-mass evidence as low under GRADE. That is their own assessment of their own headline finding, not a critic’s.
Three limits matter for reading it. The literature search closed in March 2025, roughly eighteen months before publication, so more recent trials are not represented. Heterogeneity of 88% means the pooled figure averages across studies that disagreed with each other, so it describes a general direction better than it describes any one patient. And no outcome in the review measured muscle function, strength or physical performance — the authors explicitly call for studies that do, which is an unusual and creditable thing for a paper to conclude about its own result.
One disclosure is worth knowing. The senior author is president of ESCEO, a non-profit that receives unrestricted educational grants from corporate partners, and sits on the editorial board of Drugs, the journal that published the paper; the paper states he was not involved in selecting peer reviewers or in the subsequent editorial decisions. One other author reports speaker fees from a nutrition company. The remaining authors reported no conflicts relating to this work.
Sources: Beaudart C, Malréchauffé Y, van Heden S, et al. “GLP-1 Receptor Agonists and Musculoskeletal Outcomes: A Systematic Literature Review and Meta-Analysis.” Drugs, 12 September 2026, doi:10.1007/s40265-026-02365-3 · PubMed record, PMID 42730869. Review materials deposited at the Open Science Framework.
