She Felt Better on CPAP. Then She Didn't. Here's Why That's Clinically Important.
By Dr. Vishal Saini, M.D., FAASM — Mid-West Center for Sleep Disorders
Picture this: a 39-year-old woman is diagnosed with obstructive sleep apnea. She starts CPAP. For the first few months, she feels genuinely better — more rested, sharper, less exhausted by noon. Her sleep study was abnormal, she treated it, and it worked. Case closed.
Except it isn't. A few months in, the fatigue starts creeping back. She's sleeping with the mask, the device data shows a controlled AHI, and yet the hypersomnia has returned. She wonders if she's doing something wrong.
This scenario is one of the most common things I encounter in a sleep medicine clinic — and one of the most underappreciated. The return of sleepiness after initial CPAP improvement is not a compliance failure. It's a diagnostic signal.
How Common Is This?
First, let's establish that residual excessive daytime sleepiness (EDS) after adequate CPAP therapy is not unusual. A 2024 Danish national registry study of over 1,100 CPAP patients (Jennum et al., Sleep Medicine, 2024, PMID 38657350) found that 15.6% of patients still had clinically significant EDS — defined as an Epworth Sleepiness Scale score above 10 — even after treatment with confirmed adequate adherence. A major European consensus review (Craig, Pépin et al., European Respiratory Review, 2022, PMID 35613742) puts the range at 5 to 50%, depending on the population and how you define residual sleepiness.
That is not a rounding error. That is a substantial proportion of patients who are told they have OSA, treat it, and still feel like they're dragging themselves through the day.
The question is: why?
The CPAP Honeymoon — And Why It Ends
Here is something the sleep medicine literature has documented carefully but patients almost never hear about: part of the initial improvement on CPAP is not from the CPAP itself.
A 2023 systematic review and meta-analysis by Labarca and colleagues (Sleep Medicine Reviews, 2023, PMID 36495752) analyzed sham CPAP controlled trials — studies where one group received real CPAP and the other received a fake device delivering no therapeutic pressure. The finding was striking: sham CPAP produced a statistically significant reduction in subjective sleepiness of approximately 2 Epworth Sleepiness Scale points, even though it did nothing physiologically.
Two points on the ESS may sound small, but it falls within the clinically meaningful range. What this tells us is that the early "honeymoon" on CPAP — the first weeks where everything feels better — contains a real component of non-specific expectancy effect. The novelty of treatment, the attention from a care team, the act of doing something about the problem: these things measurably improve how patients feel, independent of what the device is doing to their airway.
When the honeymoon wears off, some patients are left with their underlying, unaddressed sleep problem fully visible for the first time.
"The initial improvement on CPAP can be real and meaningful. But when it fades, that's not a sign that CPAP failed. That's a sign we need to ask a harder question." — Dr. Vishal Saini
The Harder Question: Was OSA the Whole Story?
In sleep medicine, we have a concept I think of as the gateway diagnosis problem. OSA is by far the most commonly diagnosed sleep disorder — and because it's common, it tends to get treated first and sometimes treated as the only explanation. But several other conditions can co-exist with OSA, and some are actually unmasked by successfully treating it.
The two most important are idiopathic hypersomnia (IH) and narcolepsy type 2.
A 2023 expert review by Arnulf, Dauvilliers, and colleagues (Sleep Medicine Reviews, 2023, PMID 36921459) — representing some of the world's foremost hypersomnia specialists — addresses this directly. When OSA is present and causing fragmented sleep, it suppresses and complicates the clinical presentation of an underlying intrinsic hypersomnia. Patients focus on the apnea; clinicians treat the apnea; the more fundamental disorder stays hidden. MSLT workup — the objective daytime sleep test that diagnoses narcolepsy and IH — requires stable, consolidated sleep to be interpretable, which OSA disrupts. So the formal hypersomnia workup gets deferred until after CPAP succeeds.
Then CPAP succeeds. Sleep consolidates. And the intrinsic wake-system deficit — which was always there — finally shows itself.
A 2024 case report by Martinet and Bruyneel (Sleep Science, 2024, PMID 38846592) documents exactly this sequence: a CPAP-treated patient with controlled AHI and increasing EDS underwent MSLT and was found to have a mean sleep latency of 4.2 minutes with two sleep-onset REM periods — meeting diagnostic criteria for narcolepsy type 2. The authors note that among CPAP-treated OSA patients referred for MSLT because of refractory EDS, narcolepsy or IH is identified in 20–35% of cases.
The temporal pattern matters too. A patient who never responds to CPAP is different from one who improves and then regresses. The "improvement then return" pattern may actually be more typical of unmasked IH or narcolepsy type 2 — CPAP partially helped by consolidating sleep, but couldn't fix an intrinsic deficit in the arousal system.
What Else Can Explain It?
Unmasked hypersomnia isn't the only possibility. The clinical workup for residual EDS after CPAP needs to be systematic, because the causes are genuinely heterogeneous.
Treatment-emergent central sleep apnea (TECSA) is one important and often missed cause. When CPAP eliminates obstructive apneas, some patients develop central apneas that weren't visible before — not because CPAP caused them, but because they were hidden beneath the obstructive events. TECSA prevalence ranges from 5–20% of CPAP initiators (Zhang et al., Chinese Medical Journal, 2020, PMID 33009018), confirmed in a 2024 prospective cohort finding 12.4% (Berik Safçi, Sleep & Breathing, 2024, PMID 38308750). The problem: most home CPAP devices report obstructive AHI reliably but may undercount or miss central events entirely. A patient with significant TECSA may look "controlled" on their device download while experiencing significant arousal-driven sleep fragmentation.
Other contributors include: residual REM-related obstructive events (women in particular tend to have REM-predominant OSA, which can be undertreated by fixed-pressure CPAP), increasing mask leak over time, pressure mismatch from weight changes, and co-morbid periodic limb movement disorder. Depression should also be specifically evaluated — it both causes and is worsened by sleepiness, and is independently prevalent in OSA populations.
A Note on Women Specifically
The 39-year-old in this scenario isn't incidental. The sex differences in CPAP response are clinically real and recently quantified.
A 2026 longitudinal sex-stratified cohort study (Chen et al., Respiration, 2026, PMID 41269910) found that women showed a slower and less complete trajectory of improvement in sleep quality, psychological outcomes, and daytime function over the first 12 months of CPAP compared to men — even with comparable adherence. The SHINE patient survey (Mallampalli et al., Sleep & Breathing, 2026, PMID 41902991) found women with OSA carried higher burdens of fatigue and cognitive impairment than men at the same AHI severity, and were more likely to attribute their symptoms to depression, menopause, or stress — delaying diagnosis by an average of two years.
At 39, perimenopause deserves consideration as an independent contributor. Hormonal shifts beginning in the mid-to-late 30s affect sleep architecture directly, independently of OSA. Treating the apnea doesn't address the hormonal component.
"When a woman's CPAP stops working, the temptation is to assume non-compliance or lifestyle factors. The data says we should be looking harder at the diagnosis, not the patient." — Dr. Vishal Saini
What the Evidence-Based Workup Looks Like
The most recent French national clinical guidelines on residual EDS after CPAP (Barateau, Arnulf et al., Respiratory Medicine and Research, 2024, PMID 38861872) — co-authored by narcolepsy expert Isabelle Arnulf and endorsed by the French Sleep Research Society — provide a clear, evidence-based framework:
First, confirm the CPAP is actually working. Review the detailed device download: not just summary AHI, but central event counts, pressure histogram, leak data, and hours of use by night.
Second, rule out medical confounders: thyroid function, blood count for anemia, depression screening, and a specific assessment for restless legs/PLMD (which is underdiagnosed and very common in this population).
Third — and this is the part most primary care practices skip — if the CPAP data looks adequate and no confounders emerge, the next step is an in-laboratory PSG followed by daytime MSLT. Not a home sleep test. Not just a repeat apnea screen. A full overnight polysomnogram to assess sleep architecture, followed the next morning by a standardized nap protocol to measure objective sleep latency and whether the patient enters REM sleep abnormally fast.
This test distinguishes idiopathic hypersomnia from narcolepsy from normal objective alertness — and the distinction matters enormously for treatment.
If MSLT is abnormal, the patient has a diagnosable central hypersomnia disorder that needs to be treated on its own terms, not with a CPAP adjustment. If MSLT is normal but clinical sleepiness persists, a 2025 network meta-analysis of 14 RCTs (Tanayapong et al., CNS Drugs, 2025, PMID 40208562) demonstrates that adjunct pharmacotherapy is well-supported: solriamfetol (Sunosi) showed the strongest efficacy across both subjective ESS and objective wakefulness testing, outperforming modafinil and armodafinil.
What This Means If You Recognize This Pattern
If you — or someone you know — had initial improvement on CPAP that has faded, the right response is not to adjust the mask and wait. The right response is a more complete evaluation.
That evaluation should include a detailed review of CPAP device data, a conversation about other symptoms (cataplexy-like episodes, sleep paralysis, hypnagogic hallucinations, sleep inertia, unrefreshing sleep), depression and mood screening, basic labs, and a clear plan for in-laboratory testing if the initial review doesn't explain the sleepiness.
The goal is to answer the question that should have been on the table from the beginning: is this all OSA, or is OSA part of a larger picture?
At MWCSD, we perform these evaluations routinely. We have specific expertise in idiopathic hypersomnia and narcolepsy — conditions that are frequently missed in OSA populations precisely because OSA is diagnosed first. If your CPAP data looks fine but you don't feel fine, that gap deserves investigation, not reassurance.
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 clinical trials in sleep medicine across Michigan.
Experiencing residual sleepiness despite CPAP therapy? We can help figure out why. Book a Consultation: https://www.mwsleep.com/contact-us | (517) 887-6733
References: Jennum et al. Sleep Medicine 2024 (PMID 38657350); Craig, Pépin et al. Eur Respir Rev 2022 (PMID 35613742); Labarca et al. Sleep Med Rev 2023 (PMID 36495752); Arnulf, Dauvilliers et al. Sleep Med Rev 2023 (PMID 36921459); Martinet & Bruyneel. Sleep Sci 2024 (PMID 38846592); Rosenberg, Kryger et al. Sleep Med 2019 (PMID 30803831); Zhang et al. Chin Med J 2020 (PMID 33009018); Berik Safçi. Sleep Breath 2024 (PMID 38308750); Chen et al. Respiration 2026 (PMID 41269910); Mallampalli et al. Sleep Breath 2026 (PMID 41902991); Barateau, Arnulf et al. Respir Med Res 2024 (PMID 38861872); Tanayapong, Thakkinstian et al. CNS Drugs 2025 (PMID 40208562).