Central Sleep Apnea

Central sleep apnea (CSA) comprises recurrent pauses or reductions in breathing during sleep caused by reduced ventilatory drive. Respiratory effort is absent throughout a central apnea. In obstructive sleep apnea (OSA), airflow stops despite continued effort. A mixed apnea contains an initial central portion followed by effort against an obstructed airway.

A single central event during a sleep transition does not establish a CSA syndrome. Diagnosis requires the appropriate sleep-study pattern and clinical context. Central hypopneas involve reduced ventilation and effort rather than complete cessation.

CSA is less common than OSA in the general adult population. It becomes more relevant in heart failure, chronic opioid exposure and selected neurological disorders. Older age, male sex and atrial fibrillation are associated features. Prevalence estimates vary substantially with population, cardiac treatment and scoring definitions; a population-wide estimate should not be applied to a heart-failure clinic.

Patients need not be obese or markedly sleepy. Conversely, obesity does not exclude CSA, and OSA can coexist with it. In Hong Kong practice, ask about prescribed analgesics, methadone and recent travel to high altitude as well as cardiovascular disease; there is no need to assume a locally measured prevalence where data are unavailable.

Medullary respiratory networks, including the pre-Bötzinger complex, generate respiratory rhythm. Pontine networks modulate breathing, and chemoreceptor feedback adjusts ventilation to metabolic demand. Central chemoreceptors respond to CO₂-related changes in brain extracellular pH; peripheral chemoreceptors respond to hypoxaemia, CO₂ and pH. Phrenic motor output activates the diaphragm.

During sleep, the wakefulness contribution to breathing diminishes. In NREM sleep, a fall in PaCO₂ below the apneic threshold can transiently suppress respiratory output. The difference between the sleeping eupneic PaCO₂ and this threshold is the CO₂ reserve. A narrow reserve makes an overshoot in ventilation more likely to provoke apnea.

Loop gain describes the size of the corrective ventilatory response relative to a disturbance. A high response can overshoot rather than settle. Controller gain reflects chemosensitivity; plant gain reflects the blood-gas change produced by a change in ventilation; circulatory delay affects feedback timing. In heart failure, heightened chemosensitivity, congestion-related hyperventilation and delayed circulation favour instability. It is an oversimplification to say that low cardiac output invariably increases every gain component. 1

Neuromuscular weakness, obesity hypoventilation syndrome and congenital central hypoventilation syndrome principally raise concern for sustained inadequate ventilation, not simply recurrent discrete central apneas. They belong in the differential and can coexist with central events. An effort belt may also underestimate very weak effort. Assess CO₂ and respiratory muscle function when appropriate instead of classifying every patient with weakness as having CSA.

Congenital central hypoventilation is associated with PHOX2B variants and requires specialist assessment; apnea of prematurity has a separate paediatric framework. Adult event criteria and treatment recommendations should not be applied to infants.

• Night: witnessed breathing pauses, waxing and waning respiration, fragmented sleep, insomnia, awakening breathless, or non-restorative sleep. Snoring can occur if obstruction coexists.
• Day: fatigue, sleepiness, poor concentration and reduced function. Some heart-failure patients have little subjective sleepiness despite substantial sleep-disordered breathing.
• Cause-specific clues: orthopnoea, oedema or palpitations; opioid/sedative exposure; focal brainstem signs; or progressive weakness with morning headache suggesting hypoventilation.
• Examination: assess fluid status, pulse/rhythm, blood pressure, cardiac and lung findings, body habitus, upper airway and neurological function. Neither BMI nor symptoms can reliably separate CSA from OSA.

Cheyne–Stokes breathing often predominates in NREM sleep. Sleep stage and position may influence central events, but neither provides a diagnosis without the airflow and effort recordings.

• Heart failure with paroxysmal nocturnal dyspnoea: orthopnoea, oedema and congestion support the diagnosis. CSA and cardiac PND can coexist. Clinical cardiac assessment and echocardiography are important; a raised BNP is not specific to left-heart failure and can occur with renal dysfunction, AF or right-heart strain.
• Arrhythmia or myocardial ischaemia: palpitations, syncope or chest pain require appropriate ECG/cardiac assessment. Do not attribute new chest pain or acute pulmonary oedema to a sleep disorder.
• Asthma, COPD or interstitial lung disease: cough, wheeze, exertional symptoms or lung examination findings direct spirometry and other pulmonary investigations. Nocturnal hypoxaemia alone cannot identify CSA.
• Reflux or nocturnal panic: choking, cough, acid regurgitation or panic symptoms can mimic awakenings from sleep-disordered breathing, and can coexist with it.

1. Clarify sleep opportunity, symptom timing and bed-partner observations.
2. Review heart failure, neurological disease, medications and altitude exposure.
3. Examine the cardiorespiratory system, airway and neuromuscular function.
4. Use attended PSG when CSA or hypoventilation is suspected; add CO₂ monitoring for suspected underventilation.
5. Interpret clinical symptoms alongside the study. A normal sleepiness score or absent snoring does not exclude disease.

Review sedative prescribing carefully, particularly when combined with opioids or when hypoventilation is possible. Avoid assuming that all insomnia in heart failure is CSA, or that every hypnotic has identical effects.

Suspected CSA requires a study capable of measuring sleep, airflow and respiratory effort. Attended PSG is the preferred investigation, particularly with heart failure, chronic opioids, neurological disease or possible hypoventilation. Limited home tests are designed principally for uncomplicated suspected OSA and should not be used to exclude CSA or sustained hypoventilation. 1

• Apnea: airflow-signal amplitude falls by ≥90% for ≥10 seconds.
• Central apnea: inspiratory effort is absent throughout the apnea.
• Obstructive apnea: effort persists throughout; paradoxical motion is supportive but not required.
• Mixed apnea: effort is initially absent, then resumes before airflow recovers.
• Hypopnea: ≥30% airflow reduction for ≥10 seconds with ≥3% desaturation or EEG arousal under the recommended AASM rule. Some reporting systems use a 4% desaturation rule; document the convention.

When classifying a hypopnea, snoring, inspiratory flow limitation and thoracoabdominal paradox favour obstruction. A weak or dislodged effort signal can mimic a central event. 2

CAI counts central apneas only. CAHI counts central apneas plus central hypopneas. Neither should be confused with the overall AHI, which includes all apneas and hypopneas per hour of sleep.

For primary CSA and several secondary syndromes, the usual respiratory threshold is at least 5 central apneas/hypopneas per hour, constituting more than half of all apneas/hypopneas. Symptoms, causative context and exclusion criteria vary by subtype. Do not turn this into a universal rule that every subtype requires CAI ≥5 or that a number alone establishes the diagnosis. Primary CSA additionally requires compatible symptoms, no nocturnal hypoventilation and no better explanation. 3

AHI bands often reported are 5 to <15 mild, 15 to <30 moderate and ≥30 severe. In CSA, also report central burden, desaturation, CO₂, symptoms and underlying disease; the AHI band alone does not determine treatment.

For context, adult sleep hypoventilation can be scored when CO₂ exceeds 55 mmHg for ≥10 minutes, or rises ≥10 mmHg above awake supine baseline to a value >50 mmHg for ≥10 minutes, using an appropriate arterial measure or surrogate. 2

Review usage, mask leak, pressures, residual event pattern and symptoms. Device-reported central events are a screening clue, not a definitive PSG diagnosis. Reassess treatment settings and comorbidities; consider an attended titration if persistent events or symptoms would change treatment.

The Epworth scale measures subjective sleepiness, while STOP-BANG estimates OSA risk. Neither diagnoses CSA nor reliably excludes it. Use the clinical context to decide on testing.

Management depends on the mechanism, comorbidities, gas exchange and treatment burden. Optimise contributing disease while evaluating sleep symptoms; there is no universal requirement to wait 3–6 months before treating clinically important CSA. Improving the AHI does not by itself establish improved survival.

• Heart failure: optimise guideline-directed treatment and congestion management. For HFrEF this includes the appropriate renin–angiotensin system/ARNI treatment, evidence-based beta blocker, MRA and SGLT2 inhibitor as tolerated, plus diuretics for fluid overload. CRT, ICD or advanced HF therapy should follow cardiac indications, not be prescribed solely for CSA. Inotropes are not routine treatment for all acute decompensated HF.
• Opioids: coordinate a gradual reduction or alternative analgesic strategy with the prescriber when feasible. Avoid abrupt unsupervised cessation. Buprenorphine can also cause sleep-disordered breathing; rotation is not a guaranteed remedy.
• Neurological disease or muscle weakness: treat the underlying disorder and assess whether the main problem is sustained hypoventilation requiring ventilatory support.
• Altitude: consider descent, acclimatisation and severity of associated altitude illness. Severe symptoms or suspected high-altitude pulmonary/cerebral oedema need urgent altitude-illness management.
• Review alcohol, sedatives and interacting respiratory depressants. Address sleep opportunity and advise against driving or operating machinery when sleepy or otherwise unsafe.

The 2025 AASM recommendations are conditional, reflecting limited certainty. CPAP, ASV and acetazolamide are options across several adult CSA causes. Bilevel PAP with a backup rate is suggested for primary, medication-related, treatment-emergent and medical-condition CSA; heart-failure CSA is not included in that recommendation. Bilevel PAP without backup is discouraged. Low-flow oxygen is suggested for heart-failure and altitude-associated CSA. Transvenous phrenic stimulation is an option for primary and heart-failure CSA, generally after less invasive treatments. 1

Acetazolamide dosing is individualised by indication and renal function. Monitor renal function, bicarbonate and electrolytes; adverse effects include paraesthesia, taste change, acidosis, hypokalaemia and stones. Check product contraindications and relevant allergy history. Theophylline is not a routine contemporary CSA treatment because of limited evidence and risks including arrhythmia and seizures.

The FDA indication for the remedē system is moderate-to-severe CSA in adults; a universal CAI ≥20/h requirement is incorrect. Eligibility depends on the sleep study, device labelling and specialist assessment. Potential harms include infection, lead or implant complications and stimulation discomfort. A survival benefit has not been established. 3

The 2025 AASM guideline conditionally supports ASV, including in HF, with shared decisions focused on symptom/quality-of-life improvement. In HFrEF it restricts treatment to experienced centres with close monitoring. 1

An LVEF above 45% is not by itself proof of suitability, and it is not synonymous with HFpEF. Sustained hypercapnic respiratory failure generally calls for a ventilation strategy designed to support ventilation, rather than treating ASV as a substitute for all forms of NIV.

If symptoms are improving and central burden is acceptable, continue appropriately adjusted CPAP with close follow-up, often over the next few weeks to months. Correct excessive leak and review pressure settings and reversible contributors. Do not instruct every patient to continue unchanged for a fixed interval. Persistent substantial events, poor tolerance, hypoxaemia or symptoms warrant earlier reassessment and possible titration of another modality, with the ASV precautions above.

At follow-up, assess sleep quality, daytime function, residual respiratory events, oxygenation/CO₂ where relevant, mask leak and treatment tolerance. PAP should cover all sleep, including naps; a reimbursement threshold such as four hours on 70% of nights is not the therapeutic goal. Repeat PSG or titration when findings would change care, and reassess underlying disease if CSA persists or worsens.

Morning headache, sustained nocturnal desaturation or daytime hypercapnia should prompt evaluation for coexisting hypoventilation rather than attributing every symptom to discrete central events.

Cheyne–Stokes breathing can reflect impaired cardiac function and circulatory delay. Observational studies associate CSA with adverse outcomes in HF, but its role as a modifiable cause versus a marker of disease remains incompletely resolved. 1

AF and ventricular arrhythmias may coexist with CSA. OSA also has an important association with AF; it is incorrect to assign AF exclusively to CSA or to diagnose a respiratory subtype from the arrhythmia. Palpitations, syncope or concerning rhythms warrant standard cardiac assessment.

Unlike obstructive apnea, a central apnea lacks sustained inspiratory effort against an occluded airway. Large negative intrathoracic pressure swings during obstruction should not be taught as the defining mechanical insult of a central apnea.

• CANPAP: CPAP improved several physiological measures in HF-associated CSA but did not improve transplant-free survival overall. A favourable post-hoc subgroup association does not prove a survival benefit from suppressing AHI. 2
• SERVE-HF and ADVENT-HF: different ASV trials had different safety findings, and neither establishes routine ASV as a mortality-reducing therapy. The Mx tab describes the current guideline and device-specific restrictions.
• Clinical follow-up should therefore assess symptoms, tolerability, gas exchange and the underlying disease, not promise longer survival from a normalised AHI.

Hypertension, stroke, metabolic disease, glaucoma and erythrocytosis should not all be listed as proven direct complications of isolated CSA merely because they appear in general sleep-apnoea notes. Likewise, significant pulmonary hypertension or cor pulmonale requires assessment for left-heart disease, lung disease, hypoventilation and other causes. Hypoxaemia may contribute, but CSA alone should not be assumed to explain these findings.

• Opioids: respiratory depression and overdose risk require coordinated medication review. Coexisting hypoventilation, sedative combinations or escalating doses increase concern. Acute reduced consciousness with slow or absent breathing is an emergency and requires immediate resuscitation/overdose management, not an elective sleep-study pathway. 3
• Neuromuscular disease: progressive ventilatory failure, ineffective cough or aspiration arise from muscle/bulbar dysfunction. These are complications of the underlying disorder, not automatically consequences of CSA.
• Heart failure: worsening congestion, arrhythmia or recurrent admissions may require urgent HF treatment regardless of the sleep-study result.
• Treatment: mask discomfort and leaks, oxygen-related CO₂ retention in susceptible patients, medication adverse effects and implant complications need monitoring. ASV-specific HF safety is discussed in Management.

Outcome depends on aetiology, comorbidities and response to treatment. Altitude-related or treatment-emergent central events may resolve; chronic cardiac or neurological disease may require ongoing surveillance. Reassess persistent symptoms or changing respiratory events and reconsider whether the original subtype still explains the findings.