Lemborexant linked to fewer hip fractures than Z-drugs in older adults, though the estimate is uncertain

Sleeping pills and broken hips have been linked in study after study. A study from Japan has now asked whether a newer kind of sleeping pill carries less of that risk than the Z-drugs many older people are prescribed.

For one of the two newer drugs, lemborexant, the answer was a tentative yes. Older adults who started it were hospitalized for hip fracture less often over the next year than those who started a Z-drug. For the other, suvorexant, the difference was not clear. The study, published on October 8 in the Journal of the American Geriatrics Society, was not a randomized trial, and the range around its main estimate is wide.

Why sleeping pills and hip fracture are studied together

A broken hip in old age is a serious injury. One review of the subject cites reports that one third of older people die within the year following hip fracture. The same review gives a US estimate of 24% for deaths from any cause in that year. Most hip fractures follow a fall, and anything that makes an older person unsteady at night raises the stakes.

Sleeping pills are suspected of being one such thing. The older ones, benzodiazepines, were followed by a group known as Z-drugs. A 2017 review lists them as zolpidem, zopiclone, eszopiclone and zaleplon. They were widely thought to be gentler than benzodiazepines.

The pooled evidence did not bear that out. Another analysis the same year found that people taking Z-drugs had a 90% higher risk of hip fracture, and concluded that between benzodiazepines and Z-drugs there is little difference between their respective risks. The danger was greatest at the start. Patients newly prescribed these medicines are at the greatest risk of hip fracture, the authors wrote. The review that listed the Z-drugs reached a similar result for fractures of any kind, while cautioning that studies included were observational and susceptible to confounding. Confounding means that some other difference between the groups, such as poorer health in those given the pills, accounts for part of the result.

The proposed reason lies in how these drugs work. They act on receptors in the brain that calm it, and some of those receptors also relax muscles. A Japanese research group wrote that binding to these receptors is believed to cause lightheadedness and falls.

The newer pills work differently. Orexin is a brain chemical that keeps people awake. Lemborexant and suvorexant are widely used dual orexin receptor antagonists (DORAs), the new paper says. An antagonist is a drug that blocks a receptor, and “dual” refers to the two orexin receptors both drugs block. Because they block a wake signal, the same Japanese group noted, orexin receptor antagonists do not have muscle relaxant effects.

That makes them look safer on paper. The new paper’s authors grant that they are generally considered safer than the older drugs, but say that evidence from direct comparisons remains limited.

How the hip fracture study worked

The researchers, with Yuichiro Matsuo as first author and Hideo Yasunaga as last author, conducted a cohort study using databases from Tochigi Prefecture, Japan. A cohort study follows groups of people forward in time and compares what happens to them.

They designed it as a target trial emulation. A target trial emulation means that the researchers first set out the randomized trial they would ideally run, and then apply its rules as closely as they can to existing records. Here the imagined trial would take people aged 65 and over who needed a sleeping pill and assign them to one drug or another.

So the study included only people who newly initiated either lemborexant or Z-drugs, and separately, people newly starting either suvorexant or Z-drugs. Following new users from the day of their first prescription avoids a known trap: people who have taken a drug for years without trouble are survivors, and comparing them with new users flatters the old drug.

The primary outcome was hospitalization for hip fracture within 1 year of cohort entry. A primary outcome is the one result a study is designed to answer.

Two statistical adjustments did the work that random assignment would do in a real trial. The first dealt with the fact that doctors choose different drugs for different patients. The paper says such differences at the start were handled using inverse probability of treatment weighting based on propensity scores. A propensity score is a person’s estimated chance of being given one drug over the other, based on their recorded characteristics. Weighting by it makes the two groups resemble each other on those characteristics.

The second adjustment dealt with people who did not stay on their pill. The paper explains that per-protocol effects of continuous treatment with the assigned drugs were estimated. Per-protocol means the analysis asks what happens to people while they keep taking the drug they started. A person was dropped from the count on stopping or switching, and further weighting corrected for the possibility that those who stop differ from those who continue.

What the study found for lemborexant and suvorexant

The lemborexant analysis cohort included 22,655 individuals. Of them, 6,879 started lemborexant and 15,776 started a Z-drug.

After adjustment, the rate of hospitalization for hip fracture over the year was 69% lower in the lemborexant group. The range around that estimate ran from 41% lower to 84% lower. The whole range lies on the side of fewer fractures, which is why the result counts as clear in statistical terms. It is also a broad range, which usually means the number of fractures behind it was limited.

The suvorexant comparison covered 22,377 people, of whom 6,699 started suvorexant and 15,678 a Z-drug. Here the paper reports that hip fracture incidence was lower in the suvorexant group, by 43%. But the range ran from 77% lower to 44% higher. An estimate whose range includes both a large benefit and a sizable harm cannot be told apart from no difference.

The results are relative. They say how the rate in one group compared with the rate in the other, not how many people in each group broke a hip. Hip fracture in any single year affects a small minority of older people, so a large relative difference can correspond to a small difference in the number of fractures.

The authors’ conclusion is worded with care. The two drugs, particularly lemborexant, may be safer alternatives to Z-drugs with respect to hip fracture risk, they write.

What earlier research on orexin blockers and fractures showed

For suvorexant, earlier Japanese studies disagreed with each other.

One, from Shizuoka Prefecture in 2023, pointed the opposite way from the new study. The suvorexant group comprised 6860 patients, set against 50,203 people on benzodiazepines or Z-drugs. Before any adjustment, the suvorexant patients were older and had more other illnesses, so the researchers paired each with a similar patient from the comparison group. In the matched groups, the rate of hip fracture was 48% higher with suvorexant, with a range from 20% to 82% higher. The authors concluded that among people in the Japanese regional population who use sleep drugs, those on suvorexant could be at higher risk. They acknowledged that residual unmeasured confounders may remain despite efforts to align the groups.

A second study that year compared 389 older adults hospitalized with hip fracture against 1,509 without. The risk of hip fracture was not increased in patients treated with suvorexant, it found.

The designs differ in ways that could matter. The Shizuoka comparison group mixed benzodiazepines with Z-drugs. The new study compared against Z-drugs alone and counted people only while they stayed on their first drug. Which approach comes closer to the truth is not settled by any of the three.

For the class as a whole, a pooled analysis appeared in the same journal less than two weeks before the new study. It combined 19 studies, 10 of them randomized trials, with 183,595 participants, and compared people taking orexin blockers with people not taking them. No association was observed between DORAs and fractures. The authors’ summary was that DORAs do not appear to increase the risk of falls or fractures in adults. They added that limited data in older adults warrant cautious interpretation.

That fits the new result loosely. If Z-drugs raise fracture risk and orexin blockers do not, people on orexin blockers would be expected to have fewer fractures than people on Z-drugs.

Safety in trials has not been spotless. A 2026 analysis of trials in older patients found that lemborexant and suvorexant were associated with higher rates of treatment-emergent adverse events than placebo, without an increase in serious adverse events. Treatment-emergent adverse events are unwanted effects that appear after a drug is started. And a large comparison of insomnia drugs in The Lancet in 2022 judged that lemborexant worked over the long term but that safety data on lemborexant were inconclusive.

What the hip fracture study cannot show

Patients were not assigned at random. Doctors chose each person’s pill. Weighting can balance the two groups only on what the records hold. Frailty, unsteadiness, memory problems and a history of falls are often poorly recorded, and any of them could sway both the choice of pill and the chance of a fracture.

A wide range. The lemborexant estimate could mean a fracture rate anywhere from about a sixth to about three fifths of the rate with Z-drugs.

No absolute numbers in the headline result. A relative reduction does not say how many fewer fractures there were per thousand people treated.

Same direction, different certainty. Lemborexant and suvorexant act on the same receptors. Both estimates point toward fewer fractures, and the suvorexant range overlaps the lemborexant one. The study does not show that the two drugs differ from each other.

It conflicts with one earlier study. The Shizuoka analysis found more hip fractures with suvorexant, not fewer.

No untreated group. Everyone in the study started a sleeping pill. The results compare one pill with another and say nothing about how either compares with taking none.

One prefecture, one year. The data come from a single region of Japan, where prescribing habits and doses may differ from those elsewhere, and follow-up stopped at 12 months. Only hip fractures that led to a hospital stay were counted, not falls or other fractures.

People who stayed on their drug. The main analysis describes continuous treatment. Many people stop or switch sleeping pills, and the statistical correction for that depends on assumptions.

Do not stop or change a prescribed medicine without talking to your doctor.

Whether a particular sleeping pill suits a particular older person is assessed by their doctor.

The study offers a sign that lemborexant goes with fewer hip fractures than Z-drugs in older adults, and leaves the size of the difference to be confirmed.