Large GLP‑1 review finds no bone or joint harm, and a lean mass drop the authors call low certainty

The World Health Organization asked a research center in Liege, Belgium to answer a question that comes up constantly for people taking GLP-1 medicines: what do these drugs do to bone, muscle and joints? The answer, published September 12 in the journal Drugs, is that the bone and joint worries mostly did not show up, and the muscle question is still open.

The researchers pooled 60 studies covering 1,250,717 people. Forty-six were randomized controlled trials, the design that assigns people to a drug or a comparison by chance and gives the most reliable answer. Thirteen were real-world studies that follow people already taking the medicine, and one was a database of side effect reports. The drugs included semaglutide, liraglutide, exenatide, dulaglutide and tirzepatide.

Bone density and fracture rates looked no different

Bone mineral density is a scan measurement of how much mineral is packed into bone. Lower density generally means higher fracture risk. Across five trials measuring the lower spine, five measuring the hip bone near the joint, four measuring the whole hip and five measuring the whole body, GLP-1 medicines made essentially no difference. The pooled changes were 0.01, 0.03, 0.04 and 0.00 grams per square centimeter, all small enough to be indistinguishable from zero.

Fractures told the same story. Across ten studies and 972,713 people, the odds of breaking any bone were 0.93, with a range running from 0.71 to 1.22. Because that range crosses 1.00, the value meaning no difference, the result does not show either an increase or a decrease.

One finding moved and then moved back. Spinal fractures looked less common on GLP-1 medicines in the raw numbers, at odds of 0.66. When the authors used each study's most carefully adjusted figures, which account for differences between the people being compared, the effect disappeared.

Lean body mass fell, on low certainty evidence

Twenty-eight studies covering 1,765 people measured lean body mass, which is roughly everything in the body that is not fat. Pooled together, people on GLP-1 medicines lost more of it than comparison groups did. The standardized difference was 0.52 below the comparison groups, with a range from 0.80 below to 0.23 below, all of it on the side of less lean mass.

The authors rated the certainty of this particular finding as low. The studies behind it were small, between 21 and 195 people each, and ran from 19 days to 72 weeks.

They also put the number in context. Lean mass made up roughly 15 to 25 percent of the total weight lost, which they describe as similar to what happens with calorie restriction and with bariatric surgery, and much smaller than the fat lost alongside it.

A more specific measurement weakened the signal

When the authors narrowed the analysis to the four studies that reported fat-free mass specifically, the difference shrank to 0.08 and was no longer statistically meaningful. They note that many of these measurements come from bioelectrical impedance, a method that passes a small current through the body and is sensitive to how much water a person is carrying. Early weight loss shifts body water, so some of the apparent lean mass drop may be fluid rather than muscle tissue.

Knee arthritis scores did not improve

Three studies used the WOMAC questionnaire, a standard scale for knee and hip arthritis that scores pain, stiffness and physical function. None of the three measures improved to a statistically meaningful degree. The three studies also disagreed sharply with each other, which the authors flag directly rather than smoothing over.

Muscle strength was barely measured at all

The gap the authors keep returning to is that almost no study measured whether people could actually do more or less: grip strength, walking speed, rising from a chair. Losing lean mass on a scan and losing physical function are not the same thing, and this review cannot say whether one led to the other. They call for trials that measure strength and performance directly.