henhaobio
Sign in

bioRxivNeurosciencePreprint

Vigilance profiles at sleep onset in patients with difficulty initiating sleep, with and without depressive or anxiety symptoms

Laura ABBATTISTA, Benjamin WacquieriD, Mélanie StraussiD

PDFDOI

In-depth summary

Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 3 (16 Oct 2026) · CC BY 4.0

In 2,952 sleep-clinic patients, difficulty initiating sleep (DIS) came with longer sleep latency and later deep-sleep build-up; in unmedicated DIS patients, depressive symptoms accompanied a less wake-like sleep onset.

Conventional polysomnography (PSG) collapses the wake-to-sleep transition into one 30-s epoch, prior sleep-onset studies were small and used low-resolution measures, and DSM-5 and ICSD-3 dropped insomnia subtypes for lack of reliable criteria. This work applies continuous 3-s EEG vigilance markers, including a per-patient calibrated wakefulness probability, to 2,952 clinic recordings, starting from the complaint rather than a diagnosis.

Key results

  1. ShownSelf-reported difficulty initiating sleep (DIS) has calibration-independent objective correlates: longer sleep latency from lights-off and slower build-up of deep sleep over the night.

    DIS (n = 1,328) vs no DIS (n = 1,624): median sleep latency 48 vs 41 min. Hypnodensity time to reach 5% of the cumulative N2×N3 probability product 30 vs 28.5 min (adjusted OR = 1.003, 95% CI 1.001–1.005, pFDR < 0.01) and 10% 44 vs 42.8 min (OR = 1.002, pFDR < 0.05); the cumulative W×N1 product did not differ. · 3.2. Whole-night sleep; 2.2. Questionnaire and clinical information

  2. SuggestedThe delay sits mainly before the first scored sleep epoch: the sleep-onset period (SOP), spanning the last 3 min of scored wake before the first sleep epoch to 3 min into early consolidated N2/N3 sleep, is barely longer in DIS, though its EEG content may be slightly more wake-like.

    Median SOP 11.5 min in both groups; adjusted OR = 1.004 per min (95% CI 1.002–1.007, pFDR < 0.001), driven by wake and N1 time, with N2 time unchanged. Mean intrinsic Vigilance Score (iVS) in the SOP 47.5 vs 45.5 (OR = 1.01, pFDR < 0.01); wake-like epochs 40.0 vs 38.2%. Micro-sleep rate per minute and hypnodensity stage probabilities in the SOP did not differ. · 3.3. Sleep Onset Period; 2.4.1. Wakefulness, Sleep Onset Period and Early…; 4. Discussion

  3. ShownIn unmedicated DIS patients, depressive symptoms (Beck Depression Inventory-II, BDI-II ≥ 14) went with a less wake-like SOP despite higher insomnia severity and no difference in wake amount; anxiety symptoms (Spielberger trait anxiety, STAI-T ≥ 45) showed a weaker, non-significant pattern.

    Depressive vs not (n = 252 vs 304): SOP mean iVS 45.9 vs 48.4 (OR = 0.98, 95% CI 0.97–0.99, pFDR < 0.01); sleep-like epochs 35.3 vs 30.8%. Fixed-model hypnodensity agreed: wake probability 59.9% vs 64.1%, N1 17.7% vs 15.2%, N2 12.7% vs 10.6%. Insomnia Severity Index 18 vs 15; sleep efficiency 70.1 vs 68.0%, not different. Anxiety: iVS 46.2 vs 49.3, pFDR = 0.58. · 3.3. Sleep Onset Period; 3.1. Demographic characteristics and clinical…; 3.2. Whole-night sleep; 2.2. Questionnaire and clinical information

  4. SuggestedAcross the night and within every American Academy of Sleep Medicine (AASM) stage, the individually calibrated iVS was slightly higher in DIS, suggesting EEG at a given stage lies a little closer to wakefulness.

    Whole-recording mean iVS 32.7 vs 29.0 (OR = 1.02, 95% CI 1.02–1.03); W 82.2 vs 80.9, N1 19.5 vs 17.9, N2 6.92 vs 6.16 (OR = 1.06), N3 1.73 vs 1.42, REM 10.1 vs 9.43; all pFDR < 0.001. Wake-like epochs (iVS 66-100) 27.9 vs 24.7%, sleep-like (iVS 0–20) 55.9 vs 60.1%. Depressive or anxiety symptoms did not change stage-wise iVS. · 3.2. Whole-night sleep

  5. SuggestedIn the fixed 3 min of scored wakefulness before sleep onset, DIS patients combined slightly higher vigilance with more frequent brief vigilance drops; the iVS detected these drops but the theta/alpha ratio did not.

    Mean iVS 80.5 vs 79.7 (OR = 1.009, 95% CI 1.002–1.02, pFDR < 0.05); wake-like epochs 76.7 vs 75.0%; iVS-detected micro-sleep episodes 1.33 vs 1.00 per min, also longer and larger; no group difference using theta/alpha-ratio detection, and no depressive or anxiety effect in this window. · 3.4. AASM Pre-sleep Wakefulness; 4. Discussion

Methods, data and reproducibility

Single-centre retrospective PSG analysis grouping patients by Insomnia Severity Index item 1 (≥ 2 = DIS) and, within unmedicated DIS, by BDI-II and STAI-T cut-offs. Vigilance on 3-s C4-A1 epochs came from the iVS (decile-binned delta, theta, alpha, beta power mapped to each patient's own probability of scored wake), a normalized theta/alpha ratio and micro-sleep detection, plus 30-s YASA hypnodensities.

  • Overnight clinical PSG recordings from the Sleep Functional Unit, Hôpital Universitaire de Bruxelles (2017 to 2023), used for the DIS vs no-DIS comparison · human (sleep-clinic patients) · n = 2,952 recordings (DIS 1,328; no DIS 1,624)
  • Unmedicated DIS subsample with BDI-II and STAI questionnaires, used for the depressive and anxiety symptom analyses · human (sleep-clinic patients) · n = 556 depression (252 vs 304); 555 anxiety (361 vs 194)
Methods & tools
MNE-Python v1.8.0, YASA, Stata 14.2, jamovi 2.6, Insomnia Severity Index (ISI), Beck Depression Inventory (BDI-II), Spielberger State-Trait Anxiety Inventory (STAI), AASM sleep scoring criteria, intrinsic Vigilance Score (iVS)
Reagents & environment
MNE-Python v1.8.0 for spectral analysis; Stata version 14.2 and jamovi version 2.6 for statistics; no requirements file or container given.
Reproducibility gaps
Data available upon reasonable request; no code is shared for the iVS, theta/alpha-ratio or micro-sleep detection pipelines.
Limitations (6)
  • DIS was defined from a single self-reported ISI item rather than an insomnia disorder diagnosis, and most DIS patients also reported maintenance difficulties, so findings concern a symptom dimension. · stated by the authors
  • Effect sizes were small and their clinical significance is not established. · stated by the authors
  • Depressive and anxiety symptoms rest on questionnaire cut-offs rather than structured interviews; the BDI-II itself contains sleep and fatigue items. · stated by the authors
  • Only one in-laboratory night was analyzed, with no measure of perceived sleep latency on that night. · stated by the authors
  • The iVS is calibrated on each patient's own AASM scoring and its 20 and 66 cut-offs were set by ROC in the same data; it has no external validation, and the cohort is single-centre and retrospective. · noted in this summary
  • The comparison group is sleep-clinic patients (median ISI 10, more obstructive sleep apnea), not good sleepers, and group imbalances in age, sex and BMI are handled only by covariate adjustment. · noted in this summary

Read this if you study insomnia phenotyping or sleep-onset EEG and want large-sample evidence on how a falling-asleep complaint, and depressive symptoms within it, relate to continuous vigilance markers beyond 30-s staging.

Citations
1

Abstract

Study Objectives: Difficulty initiating sleep (DIS) is a core insomnia complaint, frequently associated with depressive and anxiety symptoms, yet its objective correlates remain elusive. Insomnia subtypes were merged into a single diagnosis in DSM-5 and ICSD-3 because their criteria lacked reliability, leaving open whether objective markers could help characterize insomnia heterogeneity. Conventional polysomnography (PSG) reduces the wake-sleep transition to a single 30-s epoch. We combined conventional and continuous EEG-based vigilance markers to characterize sleep onset in patients with and without a DIS complaint, and in patients with DIS with and without depressive or anxiety symptoms. Methods: We retrospectively analyzed PSG recordings from 2,952 sleep-clinic patients, grouped by the self-reported complaint of difficulty falling asleep (Insomnia Severity Index item 1 [≥] 2: DIS, n = 1,328; no DIS, n = 1,624). In unmedicated patients with DIS (n = 556), depressive (BDI-II [≥] 14) and anxiety (STAI-T [≥] 45) symptoms were assessed. Vigilance was quantified on 3-s epochs with an intrinsic Vigilance Score (iVS), a probability-of-wakefulness index calibrated on each participant's own scored night, alongside the theta/alpha ratio and micro-sleep detection, and on 30-s epochs with automated hypnodensities. Results: Patients with DIS showed a longer sleep latency and slightly higher vigilance markers during both wakefulness and sleep, whereas the duration of the SOP itself was only marginally different (median 11.5 min in both groups). Deep sleep also accumulated later across the night (hypnodensity). In wakefulness preceding sleep onset, higher vigilance coexisted with more frequent brief vigilance drops. Among patients with DIS, depressive symptoms were associated with a less wake-like sleep-onset period despite more severe insomnia complaints, a finding not explained by differences in wake amount and convergent with hypnodensity. Anxiety symptoms showed a similar but weaker, non-significant pattern. Conclusions: In patients complaining of difficulty initiating sleep, the delay lies mainly before the first sleep epoch: once it is reached, progression to consolidated sleep is only marginally delayed, although its vigilance content may differ. Continuous EEG markers detect such differences, which are not captured by sleep staging. The association of depressive symptoms with a less vigilant sleep onset, despite worse complaints, suggests that similar insomnia complaints may rest on partly different physiological substrates, and that continuous vigilance markers could help revisit insomnia heterogeneity.

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISStephan '21Ujma '26Hacohen '26Snipes '24Sanders '19Iijima '23
67% semantic matchNeuroscience
Aurélie Stephan, Sandro Lecci, Jacinthe Cataldi+1 more·bioRxiv·Apr 6, 2021· 1 citations