Finding the minimum effective continuous positive airway pressure to eliminate obstructive events
Before any pressure is applied, a full attended diagnostic polysomnogram (PSG) documents the untreated severity of obstructive sleep apnea (OSA). Recurrent partial or complete collapse of the retropalatal and retroglossal airway produces apneas, hypopneas, and oxygen desaturations that define the Apnea-Hypopnea Index (AHI) — the core severity metric that determines whether treatment is indicated and how aggressively it must be pursued.
Obstructive sleep apnea results from a mechanical failure of upper airway patency during sleep, driven by the loss of the wakefulness stimulus to pharyngeal dilator muscle tone. Three anatomic segments are most vulnerable:
• Retropalatal airway (behind the soft palate/uvula): the most common site of primary collapse, particularly in supine position where gravity pulls the palate posteriorly • Retroglossal airway (behind the tongue base): genioglossus muscle relaxation during sleep allows the tongue to fall backward, especially pronounced in REM sleep when skeletal muscle atonia is most profound • Lateral pharyngeal walls: parapharyngeal fat pads and enlarged tonsils narrow the lumen independent of anteroposterior collapse
The Starling resistor model describes the pharynx as a collapsible tube between two rigid segments (nasopharynx and trachea). When the critical closing pressure (Pcrit) of the airway exceeds the intraluminal pressure generated by inspiratory effort, the airway collapses. Pcrit in severe OSA is often only slightly negative or even positive (airway collapses even without inspiratory effort), whereas in normal individuals Pcrit is strongly negative (airway remains patent under substantial negative pressure).
AHI severity thresholds (AASM): Normal <5/hr, Mild OSA 5–15/hr, Moderate OSA 15–30/hr, Severe OSA >30/hr. A single diagnostic night with baseline AHI of 30–40/hr is typical of the patients who proceed directly to attended in-lab CPAP titration.
A standard diagnostic PSG simultaneously records: EEG (sleep staging N1/N2/N3/REM), EOG (eye movements for REM detection), chin and leg EMG (muscle tone and periodic limb movements), ECG, nasal pressure transducer and oronasal thermal airflow sensors, chest and abdominal respiratory inductance plethysmography belts (effort), pulse oximetry (SpO₂), body position sensor, and snore microphone.
AASM scoring rules (2012, updated 2020) define: • Obstructive apnea: ≥90% drop in airflow for ≥10 seconds with continued or increased inspiratory effort against the closed airway • Central apnea: ≥90% drop in airflow for ≥10 seconds with absent inspiratory effort — the brainstem fails to signal a breath rather than the airway mechanically obstructing • Hypopnea: ≥30% drop in airflow for ≥10 seconds associated with either a ≥3% oxygen desaturation or an EEG arousal • RERA (respiratory effort-related arousal): increasing respiratory effort or airflow flattening for ≥10 seconds ending in arousal, without meeting apnea/hypopnea criteria
The Apnea-Hypopnea Index (AHI) = (obstructive apneas + central apneas + hypopneas) / total sleep time in hours. The Respiratory Disturbance Index (RDI) additionally includes RERAs and is generally used to guide titration decisions in the sleep lab.
Continuous positive airway pressure works by delivering a constant flow of pressurized air through a sealed nasal, oronasal, or full-face mask, acting as a pneumatic splint that mechanically holds the pharyngeal walls apart — preventing the negative intraluminal pressure of inspiration from collapsing the airway. Titration always begins at a low, comfortable starting pressure to allow acclimation before pressure is progressively increased.
CPAP does not "push air into the lungs" like a ventilator — it maintains a constant positive baseline pressure throughout both inspiration and expiration. This positive intraluminal pressure counteracts the negative pressure generated by the diaphragm during inspiration (which is what collapses a vulnerable airway) and mechanically distends the pharyngeal soft tissue, increasing cross-sectional airway area at the retropalatal and retroglossal segments.
The physical principle is analogous to inflating a floppy tube from the inside: at 4 cmH₂O the airway of a patient with a Pcrit near 0 remains only marginally more patent than unsplinted, and obstructive events typically persist. Effective splinting generally requires intraluminal pressure to exceed the patient-specific Pcrit by 2–4 cmH₂O of margin, which is precisely why pressure must be individually titrated rather than fixed.
Interface choice materially affects both comfort and delivered pressure fidelity:
• Nasal mask / nasal pillows: preferred first-line interface; lower leak rates, less claustrophobia, but ineffective in mouth-breathers unless combined with a chin strap • Full-face (oronasal) mask: indicated for chronic mouth breathing, nasal obstruction, or high pressure requirements (>15 cmH₂O) where nasal-only interfaces cannot maintain effective seal • Nasal pillow: minimal contact area, favored for claustrophobic patients, but more sensitive to pressure-related air leak at higher settings
The titration technologist starts most adult patients at 4 cmH₂O, the lowest pressure permitted under the AASM titration protocol, and allows a brief acclimation period while confirming mask seal, monitoring comfort, and beginning continuous nasal pressure/flow signal review before initiating stepwise increases in Stage 3.
With the patient asleep and the mask sealed, the sleep technologist follows a standardized, protocol-driven algorithm: pressure is raised in defined 1 cmH₂O increments whenever a minimum number of obstructive apneas, hypopneas, RERAs, or a sustained period of loud snoring is observed, with a mandatory minimum interval between increases to allow each new pressure level to be adequately assessed before the next change.
The technologist continuously reviews the nasal pressure/flow waveform and respiratory effort belts in real time, watching specifically for flow-limitation flattening (a plateaued rather than rounded inspiratory flow curve, indicating early airway narrowing before frank apnea/hypopnea occurs), snoring on the microphone channel, and desaturation events on the oximeter.
Each time the AASM-defined event threshold is met within the observation window at the current pressure, the pressure is raised by exactly 1 cmH₂O and the observation window resets. This continues until: (a) the residual AHI/RDI falls to target and no further events, flow limitation, or snoring are observed for a sustained period across all sleep stages and body positions, or (b) the maximum protocol pressure (20 cmH₂O, or higher only under a physician-approved extended protocol) is reached without full resolution, at which point bilevel or supplemental therapies may be considered instead.
Titration is deliberately conservative and stepwise rather than jumping straight to a high pressure: overshooting pressure introduces new problems (arousals from pressure discomfort, aerophagia, central apnea induction, and worsened mask leak) that can themselves fragment sleep and obscure the true optimal setting.
Obstructive events are consistently most frequent and most severe in the supine position (gravity displaces the tongue and soft palate posteriorly) and during REM sleep (skeletal muscle atonia removes nearly all pharyngeal dilator muscle tone, including the genioglossus). A pressure that appears fully effective in lateral N2 sleep may still be inadequate once the patient rolls supine and enters REM.
AASM protocol therefore explicitly requires the titration to capture — whenever the recording night allows — a period of supine REM sleep at the candidate optimal pressure with no residual respiratory events, oxygen desaturations, or arousals, before that pressure can be confidently designated as the prescription pressure. If REM or supine sleep cannot be captured on a single-night titration, a split-night or repeat full-night titration study is often required.
| Product | Indication | Trial Design | Key Result |
|---|---|---|---|
| Obstructive apnea | ≥2 events observed | Complete airflow cessation with persistent inspiratory effort at current pressure | +1 cmH₂O, reassess after ≥5 min |
| Hypopnea | ≥3 events observed | Partial airflow reduction ≥30% with desaturation or arousal | +1 cmH₂O, reassess after ≥5 min |
| RERA | ≥5 events observed | Flow limitation/effort increase ending in arousal, below hypopnea threshold | +1 cmH₂O, reassess after ≥5 min |
| Loud/unambiguous snoring | ≥3 min cumulative | Audible snoring on microphone channel without frank flow reduction | +1 cmH₂O, reassess after ≥5 min |
Raising pressure is not a purely one-directional fix. Above a patient-specific threshold, two new problems commonly emerge: treatment-emergent central sleep apnea (complex sleep apnea), in which the brainstem's ventilatory drive becomes unstable at higher splinting pressures, and progressively worsening unintentional mask leak, which paradoxically reduces the effective pressure actually reaching the airway and can invalidate the titration data being collected.
A subset of patients who show clear resolution of obstructive events at an effective pressure will simultaneously develop new central apneas — periods of absent airflow with no inspiratory effort — that were not present, or were minimal, during the untreated diagnostic study. This phenomenon, termed treatment-emergent central sleep apnea (TECSA) or "complex sleep apnea," is thought to result from CPAP-induced changes in CO₂ clearance: as the airway is mechanically opened, tidal volume increases and CO₂ is washed out below the apneic threshold, triggering brainstem-driven central apnea via loop-gain instability.
Management: TECSA identified during titration should not automatically be treated by further increasing CPAP pressure, since higher pressure can worsen rather than resolve central events. Most TECSA resolves spontaneously with continued conventional CPAP use over 6–8 weeks as ventilatory control stabilizes; persistent cases refractory to CPAP may require adaptive servo-ventilation (ASV) or bilevel therapy with a backup rate.
A rising AHI with a falling obstructive-event count as pressure increases is the key titration-night signature of emerging central events — the technologist must distinguish this from simple undertitration, where all event types would still be falling together.
Two categories of leak occur during CPAP use:
• Intentional (vent) leak: engineered exhalation ports built into the mask/circuit that continuously vent CO₂; this leak is expected, accounted for in device pressure algorithms, and not a fault • Unintentional leak: air escaping around a poorly sealed mask cushion, an open mouth (with a nasal-only interface), or a damaged mask/tubing connection — this leak is clinically significant
As set pressure rises, unintentional leak tends to worsen because higher intraluminal pressure pushes harder against any imperfect seal. This creates a destructive cycle: rising leak drops the actual pressure delivered to the pharynx below the set pressure, causing apparent "non-response" to titration, prompting further pressure increases, which further increases leak. Device software and the titration technologist both monitor total leak continuously; unintentional leak above roughly 24 L/min degrades pressure delivery and event-detection accuracy, and leak above ~40 L/min frequently renders the recorded respiratory data unscoreable, requiring an in-study mask refit, chin strap addition, or interface change before titration can validly continue.
The titration concludes once a single pressure is identified that eliminates obstructive apneas, hypopneas, RERAs, and snoring, keeps oxygen saturation stable, and holds across every sleep stage and body position tested — most critically, supine REM sleep, the condition of greatest airway vulnerability. This pressure, not the highest pressure trialed, becomes the patient's CPAP prescription.
AASM defines a graded hierarchy of titration adequacy:
• Optimal: AHI reduced to <5/hr for at least a 15-minute period that includes supine REM sleep, without significant leak, at a pressure below the maximum protocol pressure • Good: AHI reduced to ≤10/hr, or by ≥50% if the baseline AHI was <15/hr, including supine REM sleep or ruled out as unobtainable • Adequate: AHI not reduced to ≤10/hr but reduced by ≥75% from baseline, or optimal/good criteria met but without a supine REM period captured • Unacceptable: none of the above criteria are met
Only an Optimal or Good titration is typically used to set a fixed CPAP prescription with confidence; Adequate studies are often followed by a repeat titration or transition to an auto-titrating (APAP) device that can self-adjust pressure breath-by-breath within a prescribed range.
The guiding titration principle is minimum effective pressure, not maximum achievable pressure. Once the lowest pressure that fully controls respiratory events across all positions and sleep stages is found, further increases offer no additional therapeutic benefit and instead increase the risk of aerophagia (air swallowing, bloating), mask leak, arousals from pressure discomfort, dry mouth/nasal congestion, and reduced long-term adherence.
Following an in-lab manual titration, many patients are alternatively managed with an auto-titrating PAP (APAP) device set to a pressure range (e.g., 4–16 cmH₂O) that algorithmically tracks flow limitation and snoring breath-by-breath and adjusts pressure in real time — approximating the in-lab titration process continuously at home, though attended manual titration remains the reference standard for complex cases, high pressure requirements, or when central apnea/leak issues need direct technologist troubleshooting.
A patient whose obstructive events fully resolve at 9 cmH₂O should be prescribed 9 cmH₂O — not 14 cmH₂O simply because it was also "successful" during the titration night. Overtitration is one of the most common preventable causes of poor long-term CPAP adherence.