A patient with a PSG showing AHI 28 is told “you have sleep apnea” and is prescribed CPAP. The prescription may be correct. It may also be wrong — fatal-level wrong in rare cases — if the events were predominantly central, or if they convert to central on CPAP initiation. The difference between obstructive, central, and complex sleep apnea is not a pedantic classification; it determines which device to buy, what target residual AHI looks like, and in a meaningful minority of patients, whether the standard CPAP therapy is actually safe.
This article covers the physiological distinction between the three, how a PSG tells them apart, what “treatment-emergent central sleep apnea” actually is, and the specific indications for stepping up from CPAP to BiPAP S-T or ASV (adaptive servo-ventilation).
The three apnea types, physiologically
All three share the same surface feature: airflow at the nose and mouth stops (or drops substantially) for at least 10 seconds. What differs is the mechanism producing that flow cessation.
Obstructive sleep apnea (OSA). The upper airway collapses — typically at the level of the soft palate, tongue base, or epiglottis. Respiratory effort continues (in fact, increases, as the patient works against the closed airway), but no air moves. Chest-wall and abdominal inductance belts show rising effort; flow is zero. Intrathoracic pressures swing wildly. On arousal, airway muscle tone returns, the airway opens, and breathing resumes — often with a loud snort or gasp. This is the standard OSA phenotype, driven by anatomic predisposition (mandibular retrognathia, tongue size, neck adiposity) and muscle-tone dysregulation.
Central sleep apnea (CSA). The brainstem respiratory centres fail to generate the drive to breathe. Effort disappears along with flow — chest and abdomen are still. This happens transiently at sleep onset in healthy individuals (so-called “sleep-onset centrals”) but pathologically in three distinct contexts:
- Cheyne-Stokes respiration (CSR) — a crescendo-decrescendo breathing pattern, classically seen in congestive heart failure and in some stroke patients. The physiology: delayed feedback between lung CO₂ and chemoreceptor response produces oscillating ventilation, with periodic central apneas at the trough of the oscillation.
- Idiopathic CSA — no identified cause; rare.
- Secondary CSA — from opioid use, high-altitude residence, brainstem lesions, severe hypothyroidism, renal failure, or congenital central hypoventilation syndrome.
Complex / treatment-emergent central sleep apnea (TECSA). Centrals that were not present on the diagnostic PSG but emerge when CPAP is applied. The mechanism is related to the loop-gain instability that CSR represents — when CPAP is introduced, the stabilised airway allows hyperventilation, PaCO₂ drops below the apneic threshold, and respiratory drive intermittently ceases. In most patients this is a transient, self-resolving phenomenon that fades over 4–8 weeks. In a minority it persists and constitutes treatment failure with CPAP.
The distinction between these three is scored during PSG from the effort channels. It cannot be scored from a Type III HSAT as reliably (effort belt signals are there, but scoring without EEG is harder), and home CPAP devices infer it using proxy signals like the forced-oscillation technique.
How PSG distinguishes the three
On a sleep-lab PSG, the scoring rules are:
- Obstructive apnea — ≥ 90% airflow drop for ≥ 10 s, with continued or increased thoracoabdominal effort. The effort belts show rhythmic chest/abdominal motion against the zero-flow baseline. Paradoxical motion (chest in, abdomen out) is characteristic.
- Central apnea — ≥ 90% airflow drop for ≥ 10 s, with absent thoracoabdominal effort. Both belts flat.
- Mixed apnea — starts with absent effort (central pattern), transitions mid-event to effort against a closed airway (obstructive pattern). Scored as its own category but clinically managed as OSA in most cases.
- Cheyne-Stokes respiration — a crescendo-decrescendo flow pattern with at least 3 consecutive cycles, each ≥ 40 seconds in duration, with at least 5 CSR cycles per hour. Central apneas appear at the troughs. (AASM Practice Guidelines)
Hypopneas are scored as obstructive or central by similar physiological criteria. An obstructive hypopnea shows flow reduction with preserved or increased effort and often with snoring or inspiratory-flow flattening. A central hypopnea shows reduced flow with reduced effort in rough proportion.
A few practical points:
- Belt artefact. Effort signals can be unreliable if the RIP belts are poorly fitted or if the patient is very obese — the belts may not fully capture the effort at the lower rib cage. This produces spurious “centrals” on scoring. An experienced technologist recognises this and discounts the artefact.
- REM variability. Effort pattern can look briefly reduced during REM-phasic muscle atonia, producing what look like centrals but are artefacts of REM tone. A reliable classification requires consistent pattern across multiple events.
- Mixed apneas are common. Pure OSA and pure CSA are cleaner teaching examples than clinical reality. Many patients have predominantly OSA with a minority of centrals (< 10%) — this is still classified as OSA for treatment purposes.
Treatment-emergent CSA prevalence and natural history
TECSA is clinically important because many patients starting CPAP will develop it transiently and this is often misinterpreted — either by the patient (who sees it as a sign CPAP is making things worse) or by a less-experienced clinician (who prematurely escalates therapy).
Key figures from the published literature:
- Prevalence at CPAP initiation: approximately 5–15% of OSA patients starting CPAP develop a Central Apnea Index (CAI) ≥ 5/hour in the first nights.
- Prevalence at 2–3 months: approximately 2–4% of CPAP-treated patients still have clinically significant persistent centrals. The natural history for the remainder is resolution.
- Risk factors for persistent TECSA: male, older age, coronary artery disease, heart failure, opioid use, and higher baseline CPAP pressure requirement.
A patient’s first-week device report showing AHI 8–12 with most events scored as central is, statistically, likely to be transient TECSA. The right clinical response is:
- Confirm the data is not leak-artefactual (leak < 24 L/min 95th-percentile on ResMed).
- Confirm the pressure is not at the top of the range every night (i.e., the APAP is not chronically maxed out, which can precipitate centrals).
- Observe over 4–6 weeks with continued therapy.
- Re-review the data at 6–8 weeks. If centrals have resolved to CAI < 5 and overall AHI < 5, the therapy is working.
- If centrals persist beyond 8 weeks at clinically significant rates, escalation becomes appropriate.
Premature escalation to ASV at the first week is expensive (ASV units cost 2–4× CPAP), technically more complex to titrate, and unnecessary in the majority of these patients.
When to escalate: CPAP → BiPAP S-T → ASV
CPAP suppresses OSA by splinting the airway with constant positive pressure. It does nothing for central events — a central apnea is an absence of drive, and no amount of CPAP makes the brainstem generate breaths. For pure OSA, CPAP is sufficient.
BiPAP S-T (spontaneous/timed) adds a backup rate. If the patient does not trigger a breath within a set interval, the device delivers a breath at a preset rate. This prevents prolonged central apneas and addresses pure CSA (especially non-Cheyne-Stokes CSA, or hypoventilation-driven CSA) and mixed picture where a backup rate is enough. S-T is the right escalation for:
- Central sleep apnea from opioid use (where the stable pattern can be ventilated through with a fixed rate).
- Hypoventilation picture with associated centrals (more the TVAPS/AVAPS territory when V_T targeting is needed).
- Neuromuscular weakness with periodic central pauses.
ASV (adaptive servo-ventilation) is a more sophisticated mode that monitors the patient’s breath-by-breath ventilation and delivers variable pressure support to stabilise a moving average of ventilation. It targets the instability that drives Cheyne-Stokes and loop-gain-unstable TECSA. ASV is the right escalation for:
- Persistent TECSA after 8+ weeks of adequate CPAP.
- Cheyne-Stokes respiration with CSR-index ≥ 10/hour (but see the heart-failure caveat below).
- Idiopathic CSA where CPAP and BiPAP-S-T have failed.
Critical caveat: the 2015 SERVE-HF trial showed increased cardiovascular mortality with ASV therapy in patients with symptomatic heart failure (NYHA II–IV) with LVEF ≤ 45% and predominantly central apneas. In this patient population, ASV is contraindicated. For patients with heart failure and predominantly central apneas, the evidence supports optimising guideline-directed heart-failure therapy (ARNI, beta-blocker, MRA, SGLT2 inhibitor) — which itself often resolves the CSR — rather than ASV. In patients without heart failure or with OSA-predominant AHI and preserved LVEF, ASV remains an appropriate treatment.
Clinical and operational guidance
A firm set of rules that work in practice:
1. The diagnostic PSG should classify events by type. A report that just says “AHI 28” without specifying obstructive/central/mixed breakdown is under-documented. Demand the breakdown. Centrals > 20% of events on a diagnostic PSG warrants further investigation before CPAP initiation — echocardiogram if not already done, medication review (opioids), and consideration of whether BiPAP-S-T is a better first-line than CPAP.
2. First-month CPAP review should report Central Apnea Index. ResMed AirSense reports “ClearAirway” events — these are the centrals. Philips reports “CA” events separately. If the CAI is > 5/hour at 4 weeks, do not assume this will resolve without active review; adjust pressure range (narrower range, lower top pressure if maxed out), verify leak is controlled, and set a clear review at 8 weeks.
3. Do not escalate to ASV at 1–4 weeks. Unless centrals are severe (CAI > 15) and symptomatic (witnessed apneas, desaturations the patient or partner notices), continued CPAP with observation is the right call.
4. Opioid-related CSA is its own category. A patient on chronic opioids for pain, cancer, or addiction treatment may develop a central picture that does not resolve on CPAP because the opioid itself is suppressing drive. Tapering the opioid (where clinically feasible) is the first intervention. If the opioid is clinically necessary, BiPAP-S-T with a backup rate is often adequate. ASV in this population has not shown the mortality signal seen in heart failure.
5. Pure OSA patients on CPAP who develop new centrals years into stable therapy need evaluation for new cardiovascular disease or medication change (new opioid, new CNS depressant). This is not a normal drift.
Indian-context specifics
1. PSG access limits the clinical granularity. In tier-2 and tier-3 Indian cities, patients often receive a Type III HSAT rather than Type I PSG. Type III can detect centrals if effort belts are included, but the scoring is less reliable than Type I. A patient with an elevated CAI on Type III should be referred for confirmatory Type I if clinically feasible.
2. Heart-failure screening in Indian OSA patients is inconsistent. Given the SERVE-HF mortality signal for ASV in LVEF-reduced HF, any patient being considered for ASV should have a recent echocardiogram. In practice, this step is sometimes skipped. A firm rule: no ASV prescription without echocardiogram documentation.
3. Opioid-related CSA is under-recognised. Indian chronic-pain and palliative-care prescribing of opioids has grown. A patient on morphine, methadone, or sustained-release oxycodone who presents with symptoms of sleep-disordered breathing should be evaluated specifically for CSA, and the CSA should be attributed to the opioid until proven otherwise.
4. ASV units are expensive and less widely distributed in India. ResMed AirCurve ASV and Philips DreamStation ASV retail at ₹2,50,000–₹3,50,000 in Indian distributor channels, which is an order-of-magnitude jump from a ₹60,000 APAP. The prescribing decision is therefore weighty, and premature escalation is not without consequence. Conversely, the existence of this price barrier should not be a reason to deny ASV to patients who genuinely need it — underservice in this niche is real.
5. BiPAP S-T is more widely available than ASV and is the right next step after CPAP for many Indian patients with persistent centrals, particularly those with opioid exposure or hypoventilation. ResMed Lumis VPAP ST, Philips DreamStation BiPAP S-T, and BMC G3 B30VT all deliver S-T at roughly ₹1,20,000–₹2,20,000.
Closing
The three apnea types have different physiology, different treatment, and different prognosis. A patient with OSA does well on CPAP. A patient with pure CSA needs a backup rate. A patient with complex sleep apnea usually resolves with continued CPAP; the minority who don’t need ASV — with the SERVE-HF caveat. A diagnostic PSG should classify events; a CPAP data review should monitor for TECSA; an 8-week review confirms whether escalation is needed.
The error pattern to avoid: treating all AHI numbers as interchangeable and missing the event-type distinction. A patient put on CPAP for “AHI 25” that was actually 80% central will not improve on CPAP, and the failure will be misattributed to adherence or mask fit rather than the wrong initial prescription.
Consult your sleep physician for any decision to switch modes or escalate therapy.
References: AASM Manual for the Scoring of Sleep and Associated Events v3; Javaheri S et al, J Clin Sleep Med 2014; Morgenthaler TI et al, Sleep 2006; Cowie MR et al, NEJM 2015 (SERVE-HF) [CITATION].