News · Heart & Metabolic
Semaglutide and tirzepatide tracked half the femoral fractures of another diabetes drug
The worry about weight-loss drugs is that fast loss costs muscle and bone, which should show up as broken hips. Comparing 27,896 matched over-65s against a rival diabetes medicine, fractures and falls both ran lower over a year.
- Older adults on semaglutide or tirzepatide broke a femur about half as often over a year.
- Falls were around a third lower too, measured against another diabetes drug.
- That runs against the worry that fast weight loss costs bone and muscle.
- In absolute terms the fracture gap is about 2 people in every 1,000 over one year.
- One year is short for bone, and doctors chose who got which drug.
The objection to giving weight-loss drugs to people in their seventies has always been physical rather than metabolic. Losing weight quickly costs lean tissue as well as fat, bone thins as the load on it drops, and an older person with less muscle and lighter bones is a person who falls and then breaks something.
A study in Osteoporosis International looked for that harm in the records of older adults with type 2 diabetes and did not find it. Femoral fractures were significantly lower with semaglutide than with a comparison diabetes drug, at roughly half the rate. Falls were lower too.
Why falls are the outcome that matters
For older adults, falls can be especially serious. They are at higher risk of falling, and more likely to fracture a bone when they fall, particularly where bone is already thin.
The consequence is not the break itself. A broken bone, especially when it is in a hip, may even lead to disability and a loss of independence for older adults, and a substantial share of people never return to living as they did before.
That is why a drug that might weaken an eighty-year-old’s legs is treated more seriously than the same question in a forty-year-old, and why this comparison was worth running.
What was compared with what
Everyone in the analysis was 65 or over, had type 2 diabetes, and carried extra weight. The comparison group is the important design choice: users of semaglutide or tirzepatide were matched against users of dipeptidyl peptidase-4 inhibitors, another class of diabetes tablet.
Comparing against an active treatment rather than against untreated people removes a large chunk of confounding at a stroke, because both groups have diabetes serious enough to be medicated and both are attending appointments to get it.
After matching, 27,896 patients were included in the semaglutide comparison and 12,808 in the tirzepatide comparison, followed for a year.
The numbers
Femoral fractures ran at 0.3% on semaglutide against 0.5% on the comparator, and 0.2% against 0.4% for tirzepatide. In people, that is roughly two fewer broken thigh bones per thousand patients per year.
Falls ran at 3.6% against 5.4% for semaglutide, and 3.6% against 5.7% for tirzepatide. Around eighteen fewer falls per thousand people per year, which is a larger absolute effect than the fracture figure and a smaller relative one.
Subgroup analyses showed consistent fall reduction across categories, while fracture reduction was more pronounced in patients with BMI ≥ 30, meaning the fracture benefit concentrated among the heaviest patients.
The internal mismatch worth noticing
Read those two results together and something does not line up.
If these drugs protect bones by preventing falls, the fracture reduction should be about the size of the fall reduction. Falls dropped by roughly a third. Fractures dropped by roughly half.
Fractures fell further than the falls that cause them. Either something else is going on, such as a change in bone quality or in how people land, or the fracture estimate is being pushed around by how few events there were: 0.3% of 13,948 people is a small number of broken bones to build a comparison on.
The falls finding rests on far more events and is correspondingly more trustworthy. It is also the one with a mundane explanation available, since carrying less weight makes a person steadier and easier to catch.
What a year cannot show
Bone does not respond quickly. Density lost during a year of weight loss shows up as fractures across the following decade, not within twelve months, so a one-year window is arguably too short to detect the harm the study set out to look for.
That cuts both ways. It means this result does not clear these drugs of causing fractures later, and it also means the observed benefit is unlikely to be a bone-density effect, because there has not been time for one.
Who gets which drug
The deeper problem is that nobody was randomly assigned.
A clinician deciding between a weekly injection and a daily tablet weighs cost, dexterity, cognition, frailty, whether someone lives alone, and how much they are likely to tolerate. The frailer, older, more complicated patient often ends up on the tablet, and frailty is among the strongest predictors of falling there is.
Propensity matching balances what appears in the record. It does not balance the impression a doctor formed in the room, and that impression is exactly what decides these prescriptions.
What it is reasonable to take from this
Not that these drugs strengthen bones. The evidence here does not support that and the timeframe could not show it.
What it does support is narrower and still useful: across almost 41,000 older adults over a year, the feared musculoskeletal harm did not appear. For someone weighing whether an older relative should be on one of these medicines, the absence of a signal in that direction is worth knowing, alongside the fact that a year is not long and the comparison was not randomized.
The trial that would settle it would need to follow people for five years with bone density measured directly. Until someone runs it, this is the shape of the evidence: reassuring, short, and about people who were never assigned anything.
People also ask
What did the study find?
Femoral fractures were significantly lower with semaglutide (0.3% vs 0.5%; HR 0.488; 95% CI 0.367-0.649) and tirzepatide (0.2% vs 0.4%; HR 0.452; 95% CI 0.280-0.729) compared with DPP-4 inhibitors. Fall risk was also reduced for both semaglutide (3.6% vs 5.4%; HR 0.663) and tirzepatide (3.6% vs 5.7%; HR 0.664).
Why would anyone expect the opposite?
Because rapid weight loss takes lean tissue as well as fat, and bone density falls as body weight falls. In an older person, less muscle means less balance and less protection when they land, which is the standard reason for concern about these drugs in the over-65s.
What were the drugs compared against?
DPP-4 inhibitors, another class of oral diabetes medicine. Comparing against an active treatment rather than against nothing removes some of the difference between people who are being treated and people who are not.
How big is the difference in real terms?
Femoral fractures affected 3 in every 1,000 on semaglutide against 5 in every 1,000 on the comparator, over one year. Falls are more common, at roughly 36 per 1,000 against 54.
Is one year long enough to judge bone?
Not really. Bone loss shows up as fractures over years, not months, and a one-year window may be too short to capture harm that accumulates. The fall finding is better suited to this timeframe than the fracture finding.
Could the two groups simply be different?
That is the central weakness. Doctors decide who gets an injectable weight-loss drug and who gets an oral tablet, and the frailer, older, more complicated patient often gets the tablet. Frailty predicts falls directly, and propensity matching cannot fully correct for a judgement that was never written down.
Should anyone start or stop a drug because of this?
No. These medicines are prescribed for blood sugar and weight, and the decision belongs with the clinician managing them. This is general information rather than medical advice.