Bidirectional Temporal Association of Device-Detected Atrial Fibrillation and Sleep Apnea
Authors, Journal, Affiliations, Type, DOI
- Authors: Philipp Spitaler, Valentin Bilgeri (corresponding author), Gudrun Feuchtner, Patrick Rockenschaub, Fabian Barbieri, Birkan Tugrul, Agne Adukauskaite, Markus Stuhlinger, Bernhard Erich Pfeifer, Peter Willeit, Axel Bauer, and Wolfgang Dichtl
- Affiliations: Department of Internal Medicine III and Department of Radiology, Medical University of Innsbruck; Institute of Clinical Epidemiology, Public Health, Health Economics, Medical Statistics and Informatics, Innsbruck; Department of Cardiology, Angiology and Intensive Care Medicine, Charité–Universitätsmedizin Berlin; Helmholtz-Zentrum Hereon, Teltow; Landesinstitut für Integrierte Versorgung Tirol; UMIT-Tirol, Hall in Tirol; Department of Public Health and Primary Care, University of Cambridge
- Journal: Heart Rhythm, volume 23, number 11, November 2026, pages 2342–2350
- Type: Prospective observational cohort analysis (ACaSA study) using time-stratified, intraindividual case-crossover methods
- DOI: 10.1016/j.hrthm.2026.07.026
- Registration: NCT05127720 (ACaSA)
- Funding: Unrestricted educational grant from MicroPort and a grant from the Österreichischer Herzfonds; the funder had no role in design, analysis, interpretation or writing
Overview
This study asks whether atrial fibrillation and sleep apnea drive each other day by day, using continuous pacemaker telemetry in 218 patients and 112,242 analysed patient-days. It separates two things prior work bundled: the onset of a new AF episode and the continuation of one already running. All three pathways were positive — a higher preceding-day AF burden preceded incident severe sleep apnea (OR 2.52 for ≥12 hours versus <6 minutes), severe sleep apnea preceded new AF episode onset (OR 1.35), and severe sleep apnea preceded continuation of an ongoing episode (OR 2.39). The onset association disappeared when a full-day lag was imposed (OR 0.98), suggesting a short-acting mechanism rather than a durable one. The findings are the first high-resolution evidence for the reverse direction — AF worsening sleep-disordered breathing — which earlier device studies had looked for and not found.
Keywords
Atrial fibrillation, sleep apnea, sleep-disordered breathing, respiratory disturbance index, device-detected atrial fibrillation, remote monitoring, pacemaker, case-crossover, bidirectional association
Key Takeaways
Rationale
- AF and sleep apnea frequently coexist and each is independently associated with adverse cardiovascular outcomes; sleep apnea is common in AF and may reduce the success of rhythm control
- The temporal direction has been incompletely defined, and prior work focused almost entirely on the pathway from sleep apnea to AF — including VARIOSA-AF, which linked greater sleep-apnea severity to subsequent AF
- The proposed mechanisms for that direction are intermittent hypoxaemia, intrathoracic pressure swings, autonomic activation and acute atrial stretch
- The reverse direction has a mechanistic rationale — AF-related haemodynamic change, reduced cardiac output and altered volume status could destabilise respiration — but high-resolution longitudinal data were lacking
- Cardiac implantable electronic devices measure both daily AF burden and a respiratory disturbance index continuously, which permits day-level temporal analysis that polysomnography-based cohorts cannot support
Study Design
- Prospective observational analysis of the ACaSA study (NCT05127720) at the Medical University of Innsbruck; data from September 2020 to August 2025
- Patients implanted with MicroPort BOREA DR or TEO DR dual-chamber pacemakers, transmitting data weekly; permanent AF, single-chamber devices and fewer than 90 days of valid follow-up were excluded, as were the first 4 weeks after implantation to remove any surgical effect
- Device definitions:
- Severe device-detected sleep apnea (DDSA): respiratory disturbance index ≥20 events/h, measured midnight to 5:00 AM by thoracic impedance. Apnea was an absent respiratory cycle >10 seconds; hypopnea a sustained >10-second fall in respiratory amplitude ≥50% of the mean minute ventilation of preceding validated cycles
- Device-detected AF (DDAF) episode day: daily ≥6 minutes of mode-switch burden, the threshold adopted from the 2024 ESC guideline definition of clinically relevant atrial high-rate episodes. Mode switch triggers when atrial rate exceeds 120 bpm and is ≥25% faster than the previous eight beats, confirmed if ≥28 of 32 or ≥36 of 64 beats are fast
- Three case-crossover pathways, each a strict three-step sequence of baseline state → exposure → next-day outcome, analysed with conditional logistic regression:
- Preceding-day DDAF burden → incident severe DDSA, restricted to days after a non-severe night
- Preceding-night severe DDSA → incident DDAF episode onset, restricted to days after a quiescent day (<6 minutes)
- Preceding-night severe DDSA → continuation of an ongoing DDAF episode across midnight
- The design's key property: time-stratified matching of each case day to control days within the same calendar month of the same year, which controls long-term trends and seasonality, combined with an intraindividual comparison that balances all static patient characteristics
- Secondary generalised linear mixed-effects models with a patient random intercept provided absolute marginal probabilities and continuous dose–response curves (natural cubic splines with knots at 1 and 12 hours for DDAF burden, and at 20 and 40 events/h for RDI)
- Sensitivity analyses: a 1-day lag to address measurement-window overlap, longer onset thresholds (≥1 and ≥12 hours), stratification by baseline heart failure, and same-weekday matching to address behavioural confounding
Cohort and Data Volume
- 280 enrolled, 218 analysed — 46 excluded for permanent AF or a VVI device, 16 for fewer than 90 days of follow-up
- Median age 76 years (IQR 69–81); 36.7% female; median BMI 26.2 kg/m²
- Pacing indication: sick sinus syndrome 52.1%, third-degree AV block 28.9%, Mobitz II second-degree block 15.0%, Wenckebach 3.3%
- Comorbidity: hypertension 66.8%, non-permanent AF 38.5%, coronary artery disease 37.8%, diabetes 18.0%, previous myocardial infarction 11.5%
- Median LVEF 58.2% (IQR 53.5–63.5); median NT-proBNP 325 ng/L; beta-blocker 28.9%; NOAC 31.7%
- 160,599 follow-up nights recorded after a 28-day post-implant blanking period; 47,798 (29.8%) invalidated by the device's own quality control (sensor noise, excessive pacing-related instability), leaving 112,801 valid nights and 112,242 fully matched patient-days
- Median follow-up 484 days per patient (IQR 262–738)
- Sensor invalidation was not driven by AF burden: mean daily DDAF burden was comparable on valid and invalid nights (125.4 vs 117.0 minutes) and the invalid-night rate did not differ between high- and low-burden patients (30.5% vs 30.6%). It was related to pacing dynamics — valid nights had higher right atrial pacing (42.0% vs 33.5%) and lower right ventricular pacing (39.1% vs 50.2%)
Exposure Frequency
- DDAF (≥6 minutes) occurred in 117 patients (53.7%), and on 11.7% of analysed patient-days
- Severe DDSA (RDI ≥20) occurred on at least one night in 190 patients (90.5%), and on 28.1% of analysed patient-nights; the median RDI on those nights was 28.8 events/h
- The two conditions coincided in 107 patients (49.1%) at least once during follow-up
Pathway 1 — Preceding-Day DDAF and Incident Severe DDSA
| Preceding-day DDAF burden | OR (95% CI) | P | Absolute probability of severe DDSA |
|---|---|---|---|
| <6 minutes (reference) | — | — | 14.2% |
| 6 minutes to 1 hour | 1.38 (1.10–1.74) | .005 | 21.8% |
| 1–12 hours | 1.81 (1.52–2.15) | <.001 | 22.5% |
| ≥12 hours | 2.52 (1.97–3.22) | <.001 | 29.7% |
- Graded dose–response, with continuous spline modelling showing a non-linear rise in the odds of severe DDSA as preceding-day DDAF burden increases
- This is the direction the literature had done least to establish
Pathway 2 — Severe DDSA and Incident DDAF Episode Onset
- OR 1.35 (95% CI 1.10–1.64; P=.003) for onset after a severe DDSA night versus a non-severe one
- Absolute daily probability: 0.20% → 0.28% across the whole cohort
- In the 82 patients with a known history of paroxysmal or persistent AF, 1.04% → 1.45% — the authors' argument that the population-average absolute risk is diluted by bradycardia-only pacemaker recipients who have no arrhythmic propensity
- Graded by RDI severity relative to <20: 1.28 (1.04–1.58) for 20–29, 1.52 (1.13–2.04) for 30–39, and 1.76 (1.24–2.52) for ≥40
- Longer onset definitions retained the association: episodes ≥1 hour OR 1.33 (1.04–1.70; P=.021); ≥12 hours OR 1.95 (1.01–3.79; P=.048)
Pathway 3 — Severe DDSA and Continuation of an Ongoing DDAF Episode
- OR 2.39 (95% CI 1.47–3.88; P<.001) — the largest sleep-apnea effect of the three pathways
- Absolute probability of continuing into the next day: 7.3% → 14.1%, roughly doubled
- Graded by RDI: 2.21 (1.32–3.70) for 20–29, 2.36 (1.19–4.68) for 30–39, 4.43 (2.03–9.66) for ≥40
- In the 26 patients with baseline heart failure the point estimate was 9.90 (2.53–38.79), which the authors themselves flag as unreliable given the subgroup size and low transition frequency — the very wide confidence interval is the message, not the point estimate
Sensitivity Analyses
- A 1-day lag abolished the onset association: with a minimum 19-hour separation between the sleep measurement and the AF outcome, OR 0.98 (0.80–1.20; P=.834). The authors read this as evidence of a transient, short-acting mechanism rather than a persistent one
- Same-weekday matching preserved both associations — OR 1.42 for onset and 2.62 for continuation — arguing against weekend behavioural shifts (alcohol, sleep timing) as the explanation
- Baseline heart-failure stratification left the associations intact in the larger non-heart-failure subgroup
Interpretation
- The authors describe a bidirectional, same-day-timescale interaction: sleep apnea raises the odds of both starting and sustaining AF, and AF burden raises the odds of the following night's respiratory disturbance
- The proposed mechanism for the sleep-apnea-to-AF direction is existing (hypoxaemia, pressure swings, autonomic activation, atrial stretch); the proposed mechanism for the AF-to-sleep-apnea direction is AF-related haemodynamic change, loss of atrial contribution to ventricular filling, and altered volume status affecting respiratory stability
- The clinical inference is explicitly tentative: a patient undergoing AF ablation may remain vulnerable to recurrence if severe sleep apnea is untreated, and a patient receiving CPAP may continue to have nocturnal respiratory instability if AF repeatedly disrupts cardiopulmonary haemodynamics — neither proposition is tested here
- The authors are careful about causality: observational analysis cannot establish it, shared mechanisms (acute volume overload, abrupt autonomic imbalance) could drive both, and temporal ordering alone does not exclude reverse causality
Limitations of the document
- The device measures a respiratory disturbance index, not polysomnography, and cannot distinguish obstructive from central apnea. This is the most consequential limitation: the AF-to-sleep-apnea pathway is precisely the direction in which a central mechanism is most plausible, because an AF episode can transiently reduce cardiac output and raise left atrial filling pressure — known triggers for central apnea. The study therefore cannot say whether preceding AF produces obstructive or central events
- No concurrent polysomnography to validate the RDI in this cohort, and no CPAP adherence data — a dynamic therapy that affects both respiratory stability and arrhythmia burden, and therefore a potentially significant time-varying confounder
- Residual confounding by other acute daily factors — physical activity, intercurrent illness, alcohol, fluid status, medication non-adherence — cannot be excluded despite the design
- Highly selected population: elderly (median 76 years), dual-chamber pacemakers, single centre, one manufacturer (MicroPort). The device-derived RDI and the 5:00 AM measurement window are specific to that platform
- Nearly 30% of nights were discarded by the device's quality control. The authors show this was unrelated to AF burden, but the dependence on pacing dynamics means the analysed nights are those with stable pacing behaviour, which may not be representative
- Many secondary and sensitivity analyses without multiplicity control, which the authors acknowledge makes them exploratory and hypothesis-generating
- The absolute risks are small — a 0.08-percentage-point daily increase in AF onset in the whole cohort. The reanalysis in patients with prior AF (1.04% → 1.45%) is the clinically meaningful framing, but it is a post hoc subgroup
- The 1-day lag result cuts both ways: it is presented as evidence of a transient mechanism, but it also means the association is confined to measurement windows that overlap in the early morning — the RDI is measured midnight to 5:00 AM and the AF "day" begins at midnight — so temporal precedence between the two pathways is less clean than the three-step framing implies
Key Concepts Mentioned
- concepts/OSA-Arrhythmogenic-Substrate — the bidirectional substrate this study gives day-level temporal support to
- concepts/Sleep-Disordered-Breathing — the wider taxonomy, diagnosis and treatment evidence
- concepts/Subclinical-AF — device-detected AF is the exposure and outcome throughout
- concepts/AF-Burden — the 6-minute threshold and device-detected burden measurement
- concepts/Catheter-Ablation-AF — the clinical inference about untreated sleep apnea and arrhythmia recurrence
- concepts/Remote-Patient-Monitoring — the telemonitoring infrastructure that made the study possible
Key Entities Mentioned
- entities/Obstructive-Sleep-Apnea — the obstructive phenotype the device cannot isolate
- entities/Atrial-Fibrillation — the arrhythmia under study
- entities/Heart-Failure — baseline heart failure in 26 patients, with an unstable but large point estimate for episode continuation
- MicroPort — pacemaker manufacturer and unrestricted educational grant funder. No entity page exists and none is warranted
Wiki Pages Updated
wiki/sources/ddaf-ddsa-hrs-2026.md— created (this page)wiki/concepts/OSA-Arrhythmogenic-Substrate.md— new bidirectional day-level section; the VARIOSA-AF "directionality confirmed" claim revised; two contradictions addedwiki/concepts/Sleep-Disordered-Breathing.md— epidemiology section revised; contradiction addedwiki/entities/Obstructive-Sleep-Apnea.md— key facts and contradiction addedwiki/sourceindex.md— index rowwiki/wikiindex.md— OSA rows revisedlog.md— ingest entry