News · Sleep
Melatonin use in children tracked with less REM sleep
A JAMA Network Open study matched 342 pairs of children at a sleep clinic. Those prescribed melatonin had a lower share of REM sleep, and no other measure differed.
Based on a cross-sectional, propensity score-matched analysis of overnight sleep-lab recordings in 684 matched children
- Melatonin is the most widely used sleep aid in children, and this is the first large sleep-lab test.
- Children on melatonin had a median 16.7% REM sleep against 19.0% in matched children.
- None of the other 14 sleep measures differed, including total sleep time and breathing.
- The design is cross-sectional, so it cannot say which came first.
- The children all attended a sleep clinic, so this is not the general child population.
Start with what this study cannot do, because it changes how much weight the finding carries.
It looked at children at a single point in time, comparing those on melatonin with those not on it. It did not give anyone melatonin, did not follow anyone forward, and cannot tell you whether the drug changed the sleep or the sleep is why the drug was prescribed.
With that established, the finding is still worth knowing, and it is the first of its kind.
Why nobody had measured melatonin in a sleep lab
Melatonin is the most widely used sleep aid in children, and the researchers behind this study, published in the journal JAMA Network Open, point out that no large study has examined its association with objective sleep architecture measured by polysomnography.
Sleep architecture means the shape of a night: how much time is spent in each stage, and in what order. Polysomnography, or PSG, is the overnight lab recording that measures it directly, with electrodes tracking brain waves alongside breathing, oxygen and movement.
So a drug given to very large numbers of children had never been checked against the one measurement that shows what their sleep actually looks like.
What the melatonin comparison showed
The researchers used a database of sleep studies from a single academic pediatric sleep laboratory, covering children seen between 2017 and 2019. Of 3392 children with usable recordings, 346 were on melatonin.
Comparing those 346 with the rest directly would be close to meaningless, because children prescribed a sleep aid differ from children who are not. So each melatonin user was paired with a similar non-user on age, sex, body mass index percentile, comorbidity burden, obstructive sleep apnea, and epilepsy, producing 342 matched pairs.
One measure differed. Melatonin users had a lower median percentage of REM sleep than their matched counterparts, 16.7% against 19.0%. REM is the stage most associated with dreaming, named for the rapid eye movements that accompany it.
Nothing else did. Total sleep time, sleep efficiency, non-REM sleep stages, respiratory indices, arousal index, periodic limb movements, and oxygen desaturation index did not differ across the fourteen other measures examined.
How solid the REM sleep finding is
Better than a single comparison usually is, for two reasons the study built in deliberately.
The first is that testing fifteen things and reporting the one that came out different is a well-known way to find nothing. The researchers corrected for that, and the REM difference survived it.
The second is that the matching was re-run in a different form. The REM association was robust to 1:k matching, meaning each melatonin user was compared against two or three similar children instead of one, and it held when repeated recordings from the same child were averaged.
The size is modest. In the standard measure of effect the study reports, the gap sits in the small-to-moderate band: real, and not dramatic.
What the melatonin sleep study cannot separate
The core limitation is not a detail. It is the shape of the study.
Everything was measured in one snapshot, with no before and after. A child prescribed melatonin has a sleep problem, and that problem might itself alter REM sleep. The finding is equally consistent with melatonin lowering REM and with lower REM being a feature of the children who end up on melatonin.
The authors made an effort here, adjusting for several psychiatric diagnoses, including attention-deficit/hyperactivity disorder, that commonly accompany both. That narrows the gap. It does not close it.
There is also the population. These were children referred to a sleep laboratory, which is a group with more sleep pathology than children in general. And the study is paywalled, so this account rests on its published summary rather than the authors’ full discussion of what limited them.
What parents can take from the melatonin finding
Not a warning, and not reassurance either.
The honest reading is that a medicine given to a great many children has now been measured properly for the first time, one difference showed up, and its cause and consequence are both unknown. A difference in REM percentage has no established meaning for how a child feels or functions.
What the study does is remove an assumption. Melatonin has been treated as inert enough to buy in a supermarket, and the first careful look at what it coincides with in a sleep lab found something rather than nothing. That is a reason to treat it as a medicine and involve a doctor, which is where a child’s sleep problem belonged anyway.
People also ask
What did the study actually measure?
Polysomnography, the overnight sleep-lab recording that tracks brain waves, breathing, oxygen and movement. The researchers used the Nationwide Children's Hospital Sleep DataBank, covering children evaluated at a single academic pediatric sleep laboratory between 2017 and 2019. The prespecified primary outcome was the percentage of rapid eye movement (REM) sleep; 14 further measures were exploratory.
How big was the REM difference?
Melatonin users had a median 16.7% REM sleep (IQR, 11.6%-22.0%) against 19.0% (IQR, 14.6%-23.2%) in matched nonusers, a standardized effect of Hedges g = -0.22 (95% CI, -0.32 to -0.11), FDR-corrected P = .003. That is a small-to-modest effect by convention, and it survived correction for testing 15 outcomes at once.
How were the two groups made comparable?
By propensity score matching: 1:1 nearest-neighbor on age, sex, body mass index percentile, comorbidity burden, obstructive sleep apnea and epilepsy, giving 342 matched pairs. After matching, all 6 covariates achieved standardized mean differences below 0.10, which is the conventional threshold for calling groups balanced. The finding also held when each melatonin user was matched to two or three nonusers instead of one.
Does this mean melatonin suppresses REM sleep in children?
It cannot show that. Everything was measured at a single point in time, so the design cannot separate an effect of the drug from the reason the drug was prescribed. Children given melatonin differ from children who are not, and the underlying sleep problem could itself alter REM. The authors adjusted for several psychiatric diagnoses, which helps and does not settle it.
Why does REM sleep matter?
REM is the stage most associated with dreaming, and in children it is thought to play a part in memory consolidation and emotional processing. What is not established is whether a difference of this size, in this population, has any consequence a parent would notice. Nobody has tested that.
Should parents change anything?
This is general information rather than medical advice, and melatonin for a child is a conversation with a pediatrician regardless of this study. What the finding adds is a reason to have that conversation rather than treat melatonin as a harmless default, particularly given how widely it is now used.