If you've had a sleep study — a polysomnogram, or PSG — you've probably received a report with a breakdown of your sleep stages. There's N1 (light sleep), N2 (core sleep), REM, and then there's N3: slow-wave sleep, or what most people call deep sleep.
For patients with Narcolepsy or Idiopathic Hypersomnia (IH), that N3 percentage isn't just a number. Emerging research suggests it may be one of the most important biomarkers we have — not only for understanding what's going wrong in your brain during the night, but potentially for predicting how well you'll respond to some of the most powerful treatments we have, including the oxybate medications: Xyrem, Xywav, and Lumryz.
Here's the part that surprises most of my patients: the story isn't as simple as "more deep sleep is better."
What N3 Sleep Actually Does
Slow-wave sleep — named for the large, rolling delta brainwaves that dominate it — is the stage during which your brain does its most critical maintenance work: clearing metabolic waste, consolidating memory, releasing growth hormone, and restoring the neurochemical balance that determines how alert you'll feel the next day.
When N3 is fragmented, insufficient, or disorganized, you wake up feeling unrefreshed regardless of how many hours you spent in bed. This is a hallmark complaint from both narcolepsy and IH patients — and it's not in your head. It shows up on the PSG.
What the Research Shows About N3 in These Conditions
Narcolepsy: Not Enough Deep Sleep, Poorly Organized
A landmark 2021 meta-analysis by Zhang and colleagues (Sleep Medicine Reviews, PMID 33934047) analyzed over 100 polysomnography studies in narcolepsy patients. The findings were clear: people with narcolepsy — particularly Narcolepsy Type 1 — have significantly less N3 sleep than healthy individuals. Their sleep is characterized by more time in the lightest stage (N1), more awakenings, reduced sleep efficiency, and an unstable architecture that shifts between stages throughout the night.
The reason is biological. In NT1, the brain has lost the orexin/hypocretin neurons that act as a stabilizing force on the switch between sleep stages. Without that stabilizer, the brain drifts constantly — too light, too fragmented, never settling deeply.
Idiopathic Hypersomnia: A Completely Different Problem
Here's where it gets interesting — and where I think the field is due for a significant rethinking.
For years, we assumed IH patients simply had "too much" deep sleep. Some early studies suggested elevated N3, which seemed to fit the clinical picture of long, unrefreshing sleep. But a landmark 2026 study by Adam, Barateau, and Dauvilliers (Sleep, PMID 41761764) — using a 32-hour continuous PSG protocol in over 200 IH patients — revealed something far more nuanced.
It's not that IH patients have too much N3. It's that their N3 never finishes its job.
In healthy sleepers, slow-wave activity (SWA) — the delta brainwave pressure that accumulates during waking — dissipates over the course of the night. By morning, it's spent. In IH patients, mathematical modeling showed that sleep pressure failed to dissipate even after 9 hours of sleep. The brain keeps generating the drive to sleep, even while the person is sleeping. The SWA machine is stuck in the "on" position.
This is not a quantity problem. It's a timing and regulation problem — a failure of homeostatic control.
"In idiopathic hypersomnia, the brain's slow-wave machinery keeps running past its natural stopping point. The problem isn't too little deep sleep or too much — it's that the regulatory system that should turn it off appears fundamentally dysregulated." — Dr. Vishal Saini
Why This Matters for Oxybate Treatment
The oxybate medications — Xyrem (sodium oxybate), Xywav (low-sodium oxybate), and Lumryz (once-nightly sodium oxybate) — are among the most powerful sleep-architecture agents in clinical medicine. They work primarily as GABA-B receptor agonists in the thalamus, driving thalamocortical circuits to generate deep, consolidated slow-wave sleep.
The results are dramatic. In a 2014 PSG study by Plazzi and colleagues (Sleep Medicine, PMID 25087195), narcolepsy patients showed an immediate and profound increase in N3 on the very first night of sodium oxybate administration. The 2026 DUET Narcolepsy study (Schneider et al., Neurology Therapeutics, PMID 42060037) found N3 duration increased by an average of 45 minutes per night. That's not a small effect; it fundamentally restructures the night.
For IH patients, a Phase 3 randomized controlled trial published in 2025 by Dauvilliers and colleagues (Neurology, PMID 40359459) found sodium oxybate significantly reduced daytime sleepiness (Epworth Sleepiness Scale improvement of nearly 7 points). A 2026 DUET IH analysis (Plante et al., CNS Drugs, PMID 41831073) showed oxybate also reduced nocturnal awakenings by over 3 per night.
The Big Unanswered Question: Does Your Baseline N3% Predict How Well Oxybate Will Work?
Here is what I find most compelling — and most frustrating — about the current literature: no trial has yet stratified oxybate response by baseline N3 percentage.
We know oxybate increases N3. We know it works clinically. But we don't yet know whether a patient with very low baseline N3 responds differently than one with near-normal N3.
My hypothesis — supported by the mechanistic evidence, though not yet tested prospectively — is that the answer likely differs between NT1 and IH:
In NT1, where N3 fragmentation is caused by the loss of orexin-stabilizing neurons, oxybate essentially fills the stabilization gap. Patients with more severely disrupted N3 architecture may have the most to gain.
In IH, where the problem is dysregulated homeostatic control rather than simple fragmentation, oxybate may be working by normalizing the temporal organization of SWS rather than simply adding more of it. A patient's N3 percentage may be less predictive than their SWA dissipation rate.
This distinction has real clinical implications. It suggests that the right PSG biomarker for predicting oxybate response in IH may not be N3% at all — it may be a more sophisticated EEG metric measuring how SWA unfolds across the night.
What This Means for Your Sleep Study
If you have narcolepsy or IH and are considering oxybate therapy, here's what I'd encourage you to ask about your PSG results:
What is my N3 percentage, and how does it compare to normal?
Is my N3 fragmented — are there frequent transitions out of deep sleep?
Is there evidence of cyclic alternating pattern (CAP) abnormalities in my deep sleep stage?
Has my sleep architecture been assessed over a full night, or was the study truncated?
These aren't just academic questions. They're the start of a personalized biomarker conversation that the field is only beginning to have systematically. At MWCSD, we routinely perform full-night in-lab polysomnography with detailed sleep architecture analysis, and our team can walk you through exactly what your N3 data means in the context of your specific diagnosis and treatment history.
The Bottom Line
The deep sleep percentage on your sleep study isn't just a number on a report. In narcolepsy and idiopathic hypersomnia, it reflects something fundamentally different in each condition — and as our understanding of how oxybate works deepens, it may become one of the most actionable biomarkers in personalized sleep medicine.
We're not there yet. But the research trajectory is clear, and at MWCSD we're committed to staying at the frontier of it — both in clinical practice and through our active clinical trial program.
Dr. Vishal Saini, M.D., FAASM is the Research & Medical Director at Mid-West Center for Sleep Disorders, and Principal Investigator on multiple Phase II/III sleep medicine clinical trials. He specializes in narcolepsy, idiopathic hypersomnia, and rare sleep disorders.
Have questions about your sleep study results or whether an oxybate therapy might be right for you? Book a consultation at mwsleep.com/contact-us or call (517) 887-6733.
References: Zhang et al. Sleep Med Rev 2021 (PMID 33934047); Adam et al. Sleep 2026 (PMID 41761764); Plazzi et al. Sleep Med 2014 (PMID 25087195); Schneider et al. Neurol Ther 2026 (PMID 42060037); Dauvilliers et al. Neurology 2025 (PMID 40359459); Plante et al. CNS Drugs 2026 (PMID 41831073).